al'Protectio
Integrated Science Assessment for
Sulfur Oxides - Health Criteria
September 2008
ISA: EPA/600/R-08/047F
Annexes: EPA/600/R-08/047FA
Contains Errata Sheet created on 3/4/2009
-------
Disclaimer
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
National Center for Environmental Assessment - RTP
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC
-------
Table of Contents
Table of Contents i
List of Tables v
List of Figures vii
Acronyms and Abbreviations x
Authors and Contributors xxii
SOx Project Team xxvi
Clean Air Scientific Advisory Committee for Sulfur Oxides Primary NAAQS Review Panel xxviii
Preface xxx
Chapter 1. Introduction 1-1
1.1. Document Development 1-2
1.2. Document Organization 1-2
1.3. EPA Framework for Causal Determination 1-2
1.3.1. Scientific Evidence Used in Establishing Causality 1-3
1.3.2. Association and Causation 1-4
1.3.3. Evidence for Going beyond Association to Causation 1-4
1.3.4. Multifactorial Causation 1-6
1.3.5. Uncertainty 1-7
1.3.5.1. Types of Uncertainty 1-7
1.3.5.2. Approaches to Characterizing Uncertainty 1-8
1.3.6. Application of Framework for Causal Determination 1 -9
1.3.7. First Step—Determination of Causality 1-11
1.3.8. Second Step—Evaluation of Population Response 1-12
1.3.9. Concepts in Evaluating Adversity of Health Effects 1-12
1.4. Conclusions 1-13
Chapter 2. Source to Dose 2-1
2.1. Sources of Sulfur Oxides 2-1
2.2. Atmospheric Chemistry 2-3
2.3. Measurement Methods and Associated Issues 2-5
2.3.1. Sources of Positive Interference 2-5
2.3.2. Sources of Negative Interference 2-6
2.3.3. Other Techniques for Measuring SO2 2-6
2.4. Monitoring Site Characteristics 2-7
2.4.1. Design Criteria for the NAAQS SO2 Monitoring Networks 2-7
2.4.1.1. Horizontal and Vertical Placement 2-7
2.4.1.2. Spacing from Minor Sources 2-7
2.4.1.3. Spacing from Obstructions 2-8
2.4.1.4. Spacing from Trees 2-8
2.4.2. Locations of SO2 Monitors in Selected Metropolitan Areas 2-9
2.4.3. Ambient SO2 Concentrations in Relation to SO2 Sources 2-23
2.5. Environmental Concentrations of SOx 2-32
2.5.1. Spatial and Temporal Variability of Ambient SO2 Concentrations 2-32
2.5.2. Five-Minute Sample Data in the Monitoring Network 2-41
2.5.3. Policy Relevant Background Contributions to SO2 Concentrations 2-46
2.6. Issues Associated with Evaluating SO2 Exposure 2-50
2.6.1. General Considerations for Personal Exposure 2-50
2.6.2. Methods Used for Monitoring Personal Exposure 2-53
2.6.3. Relationship between Personal Exposure and Ambient Concentration 2-53
2.6.3.1. Indoor Versus Outdoor SO2 Concentrations 2-54
2.6.3.2. Relationship of Personal Exposure to Ambient Concentration 2-56
2.6.4. Exposure Errors in Epidemiologic Studies 2-60
i
-------
2.6.4.1. Community Time-Series Studies 2-61
2.6.4.2. Short-Term Panel Studies 2-63
2.6.4.3. Long-Term Cohort Studies 2-63
2.6.4.4. Summary of Evaluation of Exposure Error in Epidemiologic Studies 2-63
2.7. Dosimetry of Inhaled Sulfur Oxides 2-64
2.7.1. Respiratory Gas Deposition 2-64
2.7.2. Particles and Sulfur Oxide Mixtures 2-66
2.7.3. Distribution and Elimination of SOx 2-67
Chapter 3. Integrated Health Effects 3-1
3.1. Respiratory Morbidity Associated with Short-Term Exposure 3-2
3.1.1. Summary of Findings from the Previous Review 3-2
3.1.2. Potential Mode of Action for Respiratory Health Effects 3-3
3.1.3. Respiratory Effects Associated with Peak (5-10 min) Exposure 3-4
3.1.3.1. Respiratory Symptoms 3-5
3.1.3.2. Lung Function 3-5
3.1.3.3. Airway Inflammation 3-8
3.1.3.4. Mixtures and Interactive Effects 3-8
3.1.3.5. Summary of Evidence on the Effect of Peak Exposure on Respiratory Health 3-9
3.1.4. Respiratory Effects Associated with Short-Term (> 1 h) Exposure 3-11
3.1.4.1. Respiratory Symptoms 3-11
3.1.4.2. Lung Function 3-17
3.1.4.3. Airway Inflammation 3-19
3.1.4.4. Airway Hyperresponsiveness and Allergic Sensitization 3-19
3.1.4.5. Respiratory Illness-Related Absences 3-21
3.1.4.6. Emergency Department Visits and Hospitalizations for Respiratory Diseases 3-21
3.1.4.7. SO2-PM Interactions and Other Mixture Effects 3-30
3.1.4.8. Summary of Evidence on the Effect of Short-Term (> 1 h) Exposure on Respiratory Health 3-30
3.1.5. Evidence of the Effects of SO2 on Respiratory Morbidity from Intervention Studies 3-32
3.1.6. Summary of Evidence of the Effect of Short-Term SO2 Exposure on Respiratory Health 3-33
3.2. Systemic Morbidity Associated with Short-Term SO2 Exposure 3-34
3.2.1. Summary of Findings from the Previous Review 3-34
3.2.2. Cardiovascular Effects Associated with Short-Term Exposure 3-34
3.2.2.1. Heart Rate and Heart Rate Variability 3-35
3.2.2.2. Repolarization Changes 3-36
3.2.2.3. Cardiac Arrhythmias 3-37
3.2.2.4. Blood Pressure 3-37
3.2.2.5. Blood Markers of Cardiovascular Risk 3-38
3.2.2.6. Acute Myocardial Infarction 3-39
3.2.2.7. Emergency Department Visits and Hospitalizations for Cardiovascular Diseases 3-39
3.2.2.8. Summary of Evidence on the Effects of Short-Term SO2 Exposure on Cardiovascular Health 3-42
3.2.3. Other Effects Associated with Short-Term SO2 Exposure 3-42
3.3. Mortality Associated with Short-Term SO2 Exposure 3-43
3.3.1. Summary of Findings from the Previous Review 3-43
3.3.2. Mortality and Short-Term SO2 Exposure in Multicity Studies and Meta-Analyses 3-43
3.3.2.1. Multicity Studies 3-44
3.3.2.2. Meta-Analyses of Air Pollution-Related Mortality Studies 3-47
3.3.3. Evidence of the Effect of SO2 on Mortality from an Intervention Study 3-48
3.3.4. Summary of Evidence on the Effects of Short-Term SO2 Exposure on Mortality 3-49
3.4. Morbidity Associated with Long-Term SO2 Exposure 3-52
3.4.1. Summary of Findings from the Previous Review 3-52
3.4.2. Respiratory Effects Associated with Long-Term Exposure to SO2 3-53
3.4.2.1. Asthma, Bronchitis, and Respiratory Symptoms 3-53
3.4.2.2. Lung Function 3-55
3.4.2.3. Morphological Effects 3-56
3.4.2.4. Lung Host Defense 3-56
3.4.2.5. Summary of Evidence on the Effects of Long-Term Exposure on Respiratory Health 3-57
3.4.3. Carcinogenic Effects Associated with Long-Term Exposure 3-57
3.4.4. Cardiovascular Effects Associated with Long-Term Exposure 3-59
3.4.5. Prenatal and Neonatal Outcomes Associated with Long-Term Exposure 3-60
3.4.6. Other Organ System Effects Associated with Long-Term Exposure 3-63
3.5. Mortality Associated with Long-Term SO2 Exposure 3-63
ii
-------
3.5.1. Summary of Findings from the Previous Review 3-63
3.5.2. Associations of Mortality and Long-Term Exposure in Key Studies 3-64
3.5.2.1. U.S. Cohort Studies 3-64
3.5.2.2. European Cohort Studies 3-67
3.5.2.3. Cross-Sectional Analysis Using Small Geographic Scale 3-67
3.5.3. Summary of Evidence on the Effect of Long-Term Exposure on Mortality 3-68
Chapter 4. Public Health Impact 4-1
4.1. Assessment of Concentration-Response Function and Potential Thresholds 4-1
4.1.1. Evidence from Human Clinical Studies 4-1
4.1.2. Evidence from Epidemiologic Studies 4-4
4.1.3. Summary of Evidence on Concentration-Response Functions and Thresholds 4-7
4.2. Susceptible and Vulnerable Populations 4-7
4.2.1. Pre-existing Disease 4-8
4.2.1.1. Pre-existing Respiratory Diseases 4-8
4.2.1.2. Pre-existing Cardiovascular Diseases 4-9
4.2.2. Genetic Factors for Oxidant and Inflammatory Damage from Air Pollutants 4-10
4.2.3. Age-Related Susceptibility 4-12
4.2.4. Other Potentially Susceptible Populations 4-14
4.2.5. Factors that Potentially Increase Vulnerability to SO2 4-14
4.2.6. Summary of Potentially Susceptible and Vulnerable Populations 4-15
Chapter 5. Summary and Conclusions 5-1
5.1. Emissions and Ambient Concentrations of SO2 5-1
5.2. Health Effects of S02 5-2
5.3. Integration of the Evidence 5-8
5.4. Susceptible and Vulnerable Populations 5-10
5.5. Conclusions 5-10
ANNEXES
Annex A. Literature Selection A-1
A.1. Literature Search and Retrieval A-1
A.2. General Criteria for Study Selection A-1
A.2.1. Criteria for Selecting Epidemiologic Studies A-1
A.2.2. Criteria for Selecting Animal and Human Toxicological Studies A-2
A.3. Other Approaches to the Causal Determination A-4
A.3.1. Surgeon General's Report: The Health Consequences of Smoking A-4
A.3.2. EPA: Guidelines for Carcinogen Risk Assessment A-6
A.3.3. Improving the NAS/IOM Presumptive Disability Decision-Making Process for Veterans Report A-9
A.3.4. National Acid Precipitation Assessment Program Guidelines A-14
A.3.5. IARC Guidelines for Scientific Review and Evaluation Categories A-16
A.3.6. NTP: Report on Carcinogens A-22
Annex B. Additional Information on the Atmospheric Chemistry of SOx B-1
B.1. Introduction B-1
B.1.1. Multiphase Chemical Processes Involving SOx and Halogens B-2
B.1.2. Mechanisms for the Aqueous Phase Formation of Sulfate B-4
B.1.3. Multiphase Chemical Processes Involving SOx and NH3 B-5
B.2. Transport of SOx in the Atmosphere B-5
B.3. Emissions of SO2 B-6
B.4. Methods Used to Calculate SOx and Chemical Interactions in the Atmosphere B-8
B.5. Chemical-transport Models B-9
B.5.1. Regional Scale Chemical-Transport Models B-9
B.5.2. Intra-urban Scale Dispersion Modeling B-13
B.5.3. Global-scale CTMs B-13
B.5.4. Modeling the Effects of Convection B-14
B.5.5. CTM Evaluation B-15
B.6. Sampling and Analysis of SOx B-15
B.6.1. Sampling and Analysis for SO2 B-15
B.6.1.1. Other Techniques for Measuring SO2 B-16
B.6.2. Sampling and Analysis for SO42-, NO3, and NH4"* B-16
iii
-------
Annex C. Modeling Human Exposure C-1
C.1. Introduction C-1
C.2. Population Exposure Models: Their Evolution and Current Status C-5
C.3. Ambient Concentrations of SO2 and Related Air Pollutants C-7
C.4. Characterization of Microenvironmental Concentrations C-8
C.4.1. Characterization of Activity Events C-9
C.4.2. Characterization of Inhalation Intake and Uptake C-9
Annex D. Controlled Human Exposure D-1
Annex E. Toxicological Studies E-3
Annex F. Epidemiologic Studies F-1
References
iv
-------
List of Tables
Table 1-1. Aspects to aid in judging causality. 1-9
Table 1-2. Weight of evidence for causal determination. 1-11
Table 2-1. Proximity to SO2 monitors for the total population by city. Percentages are given with respect to the total population in
each city. 2-15
Table 2-2. Proximity to SO2 monitors for children aged 0-4 yr by city. Percentages are given with respect to the total population in
the age group in each city. 2-15
Table 2-3. Proximity to SO2 monitors for children aged 5-17 yr by city. Percentages are given with respect to the total population in
the age group in each city. 2-16
Table 2-4. Proximity to SO2 monitors for adults aged 65 yr and over by city. Percentages are given with respect to the total
population in the age group in each city. 2-16
Table 2-5. Monitor counts for California and San Diego County, 2005. 2-23
Table 2-6. Monitor counts for Ohio and Cuyahoga County, 2005. 2-23
Table 2-7. Mean ambient concentrations of SO2 and SO42" in different regions of the U.S. averaged over 2003-2005. 2-32
Table 2-8. Concentration distributions of SO2 inside and outside CMSAs from 2003-2005. 2-33
Table 2-9. Range of mean annual SO2 concentrations and Pearson correlation coefficients in urban areas having at least four
regulatory monitors, 2003-2005. 2-35
Table 2-10. Locations, counts, sampling periods and statistics for monitors reporting hourly maximum 5-min SO2 values, 1997-
2007. 2-41
Table 2-11. Locations, counts, sampling periods and statistics for monitors reporting all twelve 5-min SO2 values, 1997-2007. 2-42
Table 2-12. Pearson correlation coefficient between maximum 5-min and 1-h avg SO2 concentrations at the 16 sites reporting all
twelve 5-min SO2 values. 2-43
Table 2-13. Relationships of indoor to outdoor SO2 concentrations. 2-55
Table 2-14. Association between personal exposure concentration and ambient concentration (longitudinal correlation coefficients). 2-57
Table 2-15. Association between personal exposure concentration and ambient concentration (pooled correlation coefficients). 2-58
Table 3-1. Percentage of asthmatic adults in controlled human exposures experiencing SO2 induced decrements in lung function. 3-10
Table 4-1. Factors Potentially Contributing to Susceptibility or Vulnerability to Air Pollution 4-8
Table 5-1. Key health effects of short-term exposure to SO2 observed in human clinical studies. 5-3
Table 5-2. Key respiratory health effects of exposure to SO2 in animal toxicological studies. 5-4
Table 5-3. Key findings on the health effects of SO2 exposure 5-11
Table 5-4. Effects of short-term exposure to SO2 on respiratory symptoms among children. 5-14
Table 5-5. Effects of short-term SO2 exposure on emergency department visits and hospital admissions for respiratory outcomes. 5-16
Table B-1. Atmospheric lifetimes of SO2 and reduced sulfur species with respect to reaction with OH, NO3, and CI radicals. B-1
Table B-2. Relative contributions of various reactions to the total S(IV) oxidation rate within a sunlit cloud, 10 min after cloud
formation. B-3
Table B-3 Emissions of NOx, NH3, and SO2 in the U.S. by source and category, 2002. B-6
v
-------
Table C-1. The Essential Attributes of the pNEM, HAPEM, APEX, SHEDS, and MENTOR-1A C-6
Table D-1. Effects of medications on S02-induced changes in lung function among human subjects. D-1
Table D-2. Summary of new studies of controlled human exposure to SO2. D-2
Table E-1. Respiratory System - Effects of SO2 on lung function. E-3
Table E-2. Respiratory System - Inflammatory responses following SO2 exposure. E-4
Table E-3. Respiratory System - Effects of SO2 exposure on airway responsiveness and allergic sensitization. E-5
Table E-4. Respiratory System - Effects of SO2 layered on metallic or carbonaceous particles. E-6
Table E-5. Respiratory System - Effects of mixtures containing SO2 and O3. E-9
Table E-6. Respiratory System - Effects of SO2 and sulfate mixtures. E-10
Table E-7. Respiratory System - Effects of actual or simulated air pollution mixtures. E-11
Table E-8. Effects of meteorological conditions on SO2 effects. E-12
TableE-9. Cardiovascular effects of SO2 and metabolites. E-13
Table E-10. Hematological effects of SO2. E-14
Table E-11. Carcinogenic effects of SO2. E-15
Table E-12. Nervous system effects of SO2 and metabolites. E-16
Table E-13. Reproductive and developmental effects of SO2. E-19
Table E-14. Endocrine system effects of SO2. E-20
Table E-15. Liver and gastrointestinal effects of SO2. E-20
Table E-16. Renal effects of SO2. E-22
Table E-17. Respiratory System - Effect of S02on morphology. E-22
Table E-18. Respiratory System - Effects of SO2 exposure on host lung defenses. E-23
Table E-19. Genotoxic effects of SO2 and metabolites. E-24
Table E-20. Respiratory System - Effects of SO2 and metabolites on biochemistry. E-26
Table E-21. Lymphatic system effects of SO2 and SO2 mixtures. E-28
Table F-1. Short-term exposure to SO2 and respiratory morbidity in field/panel studies. F-1
Table F-2. Short-term exposure to SO2 and emergency department visits and hospital admissions for respiratory diseases. F-20
Table F-3. Short-term exposure to SO2 and cardiovascular morbidity in field/panel studies. F-59
Table F-4. Short-term exposure to SO2 and emergency department visits and hospital admissions for cardiovascular diseases. F-65
Table F-5. Short-term exposure to SO2 and mortality. F-78
Table F-6. Long-term exposure to SO2 and respiratory morbidity. F-90
Table F-7. Long-term exposure to SO2 and lung cancer incidence and mortality. F-102
Table F-8. Long-term exposure to SO2 and prenatal and neonatal outcomes. F-104
Table F-9. Long-term exposure to SO2 and mortality. F-111
vi
-------
List of Figures
Figure 1 -1. Potential relationships of SOx with adverse health effects. 1 -6
Figure 2-1. 2001 County-level SO2 emissions densities (tons per square mile) from off-road mobile and other transportation
sources. 2-2
Figure 2-2. Location of SO2 monitors with respect to population density in the Atlanta, GA MSA. 2-9
Figure 2-3. Location of SO2 monitors with respect to population density in the Cincinnati, OH MSA. 2-10
Figure 2-4. Location of SO2 monitors with respect to population density in the Cleveland, OH MSA. 2-11
Figure 2-5. Location of SO2 monitors with respect to population density in the Los Angeles/Riverside, CA MSA. 2-12
Figure 2-6. Location of SO2 monitors with respect to population density in the New York City, NY/Philadelphia, PA MSA. 2-13
Figure 2-7. Location of SO2 monitors with respect to population density in the St. Louis, MO MSA. 2-14
Figure 2-8. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), California, 2005. 2-17
Figure 2-9. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Ohio, 2005. 2-18
Figure 2-10. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Arizona, 2005. 2-19
Figure 2-11. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Pennsylvania, 2005. 2-20
Figure 2-12. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), New York, 2005. 2-21
Figure 2-13. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Massachusetts, 2005. 2-22
Figure 2-14. Location of SO2 monitors within a 15 km buffer zone with respect to combustion sources and highways in the Atlanta,
GA MSA. 2-24
Figure 2-15. Location of SO2 monitors within a 15 km buffer zone with respect to combustion sources and highways in the
Cincinnati, OH MSA. 2-25
Figure 2-16. Location of SO2 monitors within a 15 km buffer zone with respect to combustion sources and highways in the
Cleveland, OH MSA. 2-26
Figure 2-17. Location of SO2 monitors within a 15 km buffer zone with respect to combustion sources and highways in the Los
Angeles/Riverside, CA MSA. 2-27
Figure 2-18. Location of SO2 monitors within a 15 km buffer zone with respect to combustion sources and highways in the New York
City, NY/Philadelphia, PA MSA. 2-28
Figure 2-19 Location of SO2 monitors within a 15 km buffer zone with respect to combustion sources and highways in the St. Louis,
MO MSA. 2-29
Figure 2-20. State-level SO2 emissions, 1990-2005. 2-30
Figure 2-21. Annual mean ambient SO2 concentration, 1989 through 1991 (A), and 2003 through 2005 (B). 2-30
Figure 2-23. Annual mean ambient SO2 emissions for Acid Rain Program cooperating facilities, 2006. 2-31
Figure 2-24. Diel variation in SO2 concentration across all monitoring sites reporting into AOS for 2005. 2-34
Figure 2-25. Steubenville, OH, 2003-2005. 2-36
Figure 2-26. Philadelphia, 2003-2005. 2-37
Figure 2-27. Los Angeles, 2003-2005. 2-38
vii
-------
Figure 2-28. Riverside, CA, 2003-2005. 2-39
Figure 2-29. Phoenix, 2003-2005. 2-40
Figure 2-30. SO2 monitors reporting maximum or continuous 5-min avg values for any period, 1997-2007. 2-42
Figure 2-31. Time series and frequency distributions of voluntarily reported maximum 5-min SO2 concentrations from 6 monitors
located in Iowa, Missouri, Pennsylvania and West Virginia. 2-44
Figure 2-32. Time series of hourly maximum 5-min SO2 data showing a 24 h (upper panels) and 1 week (lower panels) time window
centered on the peak value for the two sites with the lowest (IA) and highest (PA) maximum values in the preceding
figure. 2-45
Figure 2-33. Annual mean model-predicted concentrations of SO2 (ppb). 2-47
Figure 2-34.15-min avg ambient SO2 concentrations measured at 1 Hawaii Volcanoes National Park monitoring site (Jaggar
Museum), March 12,13, and 15,2007. 2-48
Figure 2-35.15-min avg ambient SO2 concentrations measured at 2 Hawaii Volcanos National Park monitoring sites on September
29, 2007. 2-49
Figure 2-36. Percentage of time spent in various environments in the U.S. 2-50
Figure 2-37. Average annual indoor and outdoor SO2 concentrations for each of the six cities included in the Harvard six-cities study
analysis. 2-54
Figure 3-1. Distribution of individual airway sensitivity to SO2. 3-7
Figure 3-2. Odds ratios (95% CI) for incidence of morning asthma symptoms of 846 asthmatic children from the National
Cooperative Inner-City Asthma Study. 3-11
Figure 3-3. Odds ratios (95% CI) for daily asthma symptoms of 990 asthmatic children from the Childhood Asthma Management
Program Study. 3-12
Figure 3-4. Odds ratios (95% CI) for incidence of cough among children, grouped by season. 3-14
Figure 3-5. Odds ratios (95% CI) for the incidence of lower respiratory tract or asthma symptoms among children, grouped by
season. 3-15
Figure 3-6. Relative risks (95% CI) of S02-associated emergency department visits and hospitalizations for all respiratory causes
among all ages and separated by age group. 3-23
Figure 3-7. Relative risks (95% CI) of S02-associated emergency department visits and hospitalizations for asthma among all ages
and age-specific groups. 3-26
Figure 3-8. Relative risks (95% C I) of S02-associated emergency department visits (*) and hospitalizations for all respiratory
causes and asthma, with and without copollutant adjustment. 3-29
Figure 3-9. Relative risks (95% CI) of S02-associated emergency department visits (*) and hospitalizations for all cardiovascular
causes, arranged by age group. 3-40
Figure 3-10. All cause mortality excess risk estimates for SO2 from the National Morbidity, Mortality, and Air Pollution Study. 3-44
Figure 3-11. Relative risks (95%CI) of S02-associated all-cause (nonaccidental) mortality, with and without copollutant adjustment,
from multicity and meta-analysis studies. 3-50
Figure 3-12. Relative risks (95%CI) of S02-associated mortality for all (nonaccidental), respiratory, and cardiovascular causes from
multicity studies. 3-51
Figure 3-13. Relative risks (95% CI) for low birth weight, grouped by trimester of SO2 exposure. 3-61
Figure 3-14. Relative risks (95%CI) of S02-associated all-cause (nonaccidental) mortality, with and without adjustment for sulfate,
from longitudinal cohort studies. 3-69
viii
-------
Figure 4-1. Percent of mild and moderate asthmatics yE = 40-50 L/min) experiencing an S02-induced increase in (a) sRaw of
> 100% or a decrease in (b) FEVi of > 15%, adjusted for effects of moderate to heavy exercise in clean air. 4-2
Figure 4-2. S02-induced increase in sRaw among mild and moderate asthmatics following 10 min exposures with moderate to
heavy exercise (Ve = 40-50 L/min). 4-3
Figure 4-3. S02-induced decrease in FEVi among mild and moderate asthmatics following 10 min exposures with moderate to
heavy exercise (Ve = 40-50 L/min). 4-4
Figure 4-4. Adjusted odds ratios of asthma hospitalizations by groupings of 24-h avg SO2 concentrations in Bronx County, New
York. 4-5
Figure 4-5. Relative odds ratio of incidence of lower respiratory tract symptoms smoothed against 24-h avg SO2 concentrations on
the previous day, controlling for temperature, city, and day of week. 4-6
Figure 4-6. Relative risks (95% CI) of age-specific associations between short-term exposure to SO2 and respiratory ED visits* and
hospitalizations. 4-11
Figure 4-7. Summary density curves of the relative risks of age-specific associations between short-term exposure to SO2 and ED
visits and hospitalizations for all respiratory causes. 4-13
Figure 4-8. Summary density curves of the relative risks of age-specific associations between short-term exposure to SO2 and ED
visits and hospitalizations for asthma. 4-13
Figure 5-1. Odds ratios (95% CI) for the association between short-term exposures to ambient SO2 and respiratory symptoms in
children. 5-6
Figure 5-2. Relative risks (95% CI)for the association between short-term exposures to ambient SO2 and emergency department
(ED) visits/hospitalizations for all respiratory diseases and asthma in children. 5-7
Figure A-1. Selection process for studies included in the ISA. A-3
Figure A-2. Focusing on unmeasured confounders/covariates, or other sources of spurious association from bias. A-12
Figure A-3. Example posterior distribution for the determination of Sufficient. A-12
Figure A-4. Example posterior distribution for the determination of Equipoise and Above. A-13
Figure A-5. Example posterior distribution for the determination of Against. A-14
Figure B-1. Transformations of sulfur compounds in the atmosphere. B-2
Figure B-2. Comparison of aqueous-phase oxidation paths. B-4
Figure B-3. Sulfate wet deposition (mg S/m2/yr) of the mean model versus measurements for the National Atmospheric Deposition
Program (NAD P) network. B-14
Figure C-1. Schematic description of a general framework identifying the processes (steps or components) involved in assessing
inhalation exposures and doses for individuals and populations. C-4
ix
-------
Acronyms and Abbreviations
a alpha
(3 beta; the calculated health effect parameter
YN2O5 reaction potential coefficient (gamma) for N2O5
A delta, difference; change
t tau; atmospheric lifetime
|jeq microequivalent
8-OHdG 8-hydroxy-2N-deoxyguanosine
ACCENT Atmospheric Composition Change: the European NeTwork of excellence (European
Union Project)
ACS American Cancer Society
ADS annular denuder system
AERMOD AMS/EPA Regulatory Model (steady-state plume model)
AHH aryl hydrocarbon hydroxylase
AHR airways hyper responsiveness
AIRMoN Atmospheric Integrated Research Monitoring Network
AirPEx Air Pollution Exposure (model)
AirQUIS Air Quality Information System (model)
AIRS Atmospheric Infrared Sounder (instrument)
ALSC Adirondack Lake Survey Corporation
ALT alanine-amino-transferase
AM alveolar macrophages
AMMN N-nitroso-acetoxymethylmethylamine
AMS American Meteorological Society
AP alkaline phosphatase
APEX Air Pollution Exposure (model)
APHEA Air Pollution on Health: a European Approach (study)
APIMS atmospheric pressure ionization mass spectrometer
AQCD Air Quality Criteria Document
AQS Air Quality System (database)
ARIC Atherosclerosis Risk in Communities (study)
ARP Acid Rain Program
ASG Atmospheric Studies Group of TRC
ASI Acid Stress Index
asl above sea level
AST aspartate-amino-transferase
atm atmosphere
ATMOS Atmospheric Trace Molecule Spectroscopy
ATTILA type of Lagrangian model
B[a]P benzo[a]pyrene
Ba barium
BAL bronchoalveolar lavage
BC black carbon
BCS base-cation surplus
BHPN N-bis(2-hydroxypropyl) nitrosamine
BHR bronchial hyperresponsiveness
x
-------
BME Bayesian Maximum Entropy (framework)
Br bromine
Br bromine ion
Br2 molecular bromine
BrCI bromine chloride
BrO bromine oxide
BS black smoke
bw body weight
C carbon or carbon black particles
C2H6 ethane
CsHs isoprene
Ca ambient air concentration
Ca calcium
CA chromosome aberrations
Ca(N03)2 calcium nitrate
Ca(OH)2 calcium hydroxide
Ca2+ calcium ion
CaCb calcium chloride
CaC03 calcium carbonate
CALPUFF Advanced non-steady-state meteorological and air quality modeling system developed
byASG scientists and distributed by TRC. Used by the EPA for assessing long range
transport of pollutants.
CAMP
Childhood Asthma Management Program
CAMx
urban multi-scale grid based model
CARB
California Air Resources Board
CASAC
Clean Air Scientific Advisory Committee (CA
CASTNet
Clean Air Status and Trends Network
CAT
catalase
CB4
Carbon Bond 4 chemical mechanism model
CDC
Centers for Disease Control and Prevention
CFD
computational fluid dynamics (modeling)
CFR
Code of Federal Regulations
CH2I2
diiodomethane
CH2O
formaldehyde
(CH3)2SO
dimethyl sulfoxide, DMSO
CH3C(0)
acetyl radical
CH3C(0)00
acetyl peroxy radical
CH3CHO
acetaldehyde
CHsHg
methylmercury, MeHg
CH3OOH
methyl hydroperoxide
CH3-S-CH3
dimethylsulfide, DMS
CH3-S-H
methyl mercaptan
CH3SO3H
methanesulfonic acid
CH3-S-S-CH3
dimethyl disulfide, DMDS
CH4
methane
CHAD
Consolidated Human Activities Database
CHF
congestive heart failure
chl a
chlorophyll a
Choi
cholesterol
xi
-------
CHS Children's Health Study
CI confidence interval
Ci interstitial air concentration
CIMS chemical ionization mass spectroscopy
CI chlorine
CL critical load
CI" chlorine ion
Cb molecular chlorine
CLaMS type of Lagrangian model
CMAQ Community Multiscale Air Quality (modeling system)
CMD count median diameter
CMSA consolidated metropolitan statistical area
CO carbon monoxide
CO2 carbon dioxide
CO3" carbonate ion
CoH coefficient of haze
CONUS contiguous United States
COPD chronic obstructive pulmonary disease
CS2 carbon disulfide
CTM chemical transport model
Cu copper
CVD cardiovascular disease
CYP cytochrome P450
Dae aerodynamic diameter
DEcCBP DEP extract coated carbon black particles
DEN diethylnitrosamine
DEP diesel exhaust particles
DEP+C diesel exhaust particle extract adsorbed to C
dG 2N-deoxyguanosine
DMBA 7,12-dimethylbenzanthracene
DMDS dimethyl disulfide, CH3-S-S-CH3
DMS dimethyl sulfide, CH3-S-CH3
DMSO dimethylsulfoxide
DNS direct numerical simulation (approach)
DOAS differential optical absorption spectroscopy
DON dissolved organic nitrogen
EC elemental carbon
ECG electrocardiography; electrocardiogram
ED emergency department
EDXRF energy dispersive X-ray fluorescence
EE energy expenditure (average EE rate)
EIB exercise-induced bronchial reactivity
ELF epithelial lining fluid
EMAP Environmental Monitoring and Assessment Program
EMECAM Spanish Multicentre Study on Air Pollution and Mortality
EMEP Co-operative Programme for Monitoring and Evaluation of the Long-range Transmission
of Air Pollutants in Europe
EOS Earth Observation System
EPA U.S. Environmental Protection Agency
xii
-------
ESA
European Space Agency
ET
extrathoracic
Fe
iron
FEM(s)
Federal Equivalent Method(s)
FeP04
iron phosphate
FeS
iron sulfide
FEVo.75
forced expiratory volume in 0.75 second
FEVi
forced expiratory volume in 1 second
F-factor
fraction of the change in mineral acid anions that is neutralized by base cation release
FHLC
fetal hamster lung cells
FLEXPART
type of Lagrangian model
FPD
flame photometric detector
FR
Federal Register
FRM
Federal Reference Method
FUR
Fourier Transform Infrared Spectroscopy
FVC
forced vital capacity
G6PD
glucose-6-phosphate dehydrogenase
GAM
Generalized Additive Model(s)
GAW
Global Atmospheric Watch (program)
GCE
Goddard Cumulus Ensemble (model)
GCS
y-glutamylcysteine synthetase
GEOS
Goddard Earth Observing System
GE0S-1DAS
Goddard Earth Observing System Data Assimilation System
GEOS-Chem
Goddard Earth Observing System (with global chemical transport model)
GFED
Global Fire Emissions Database
GHG
greenhouse gas
GIS
Geographic Information System
GLM
Generalized Linear Model(s)
GOES
Geostationary Operational Environmental Satellites
GOME
Global Ozone Monitoring Experiment
GPx
glutathione peroxidase
GRed
glutathione reductase
GSD
geometric standard deviation
GSH
glutathione; reduced glutathione
GSSG
glutathione disulfide
GSSO3H
glutathione S-sulfonate
GST
glutathione-S-transferase
GT
y-glutamyl transpeptidase
h
hour
H
hydrogen; hydrogen atom
H+
hydrogen ion
H2O
water
H2O2
hydrogen peroxide
H2S
hydrogen sulfide
H2SO3
sulfurous acid
H2SO4
sulfuric acid
HAP(s)
hazardous air pollutant(s)
HAPEM
Hazardous Air Pollutant Exposure Model
xiii
-------
HAPEM6
HAPEM, version 6
HC
hydrocarbon
HCHO
formaldehyde
HCI
hydrochloric acid
HEADS
Harvard-EPAAnnular Denuder System
HEI
Health Effects Institute
HF
high frequency
Hg
mercury
HN02, HONO
nitrous acid
HNOs, HOONO
nitric acid
HNO4
pernitric acid
H02
hydroperoxyl; hydroperoxyl radical
HO2NO2
peroxynitric acid
HOBr
hypobromous acid
HOCI
hypochlorous acid
HOX
hypohalous acid
HP
hydrolyzed protein
HR
heart rate
HRV
heart rate variability
HSO3
hydrogen sulfite, bisulfite
hso4~
bisulfate ion
hso4~
sulfuric acid ion
hi/
solar ultraviolet photon with energy at wavelength v
HVA-ICa
high-voltage activated calcium currents
/'
microenvironment
I iodine
IA Integrated Assessment
IARC International Agency for Research on Cancer (WHO)
IBEM individual based exposure model(s)
IC ion chromatography
ICARTT International Consortium for Atmospheric Research on Transport and Transformation
ICD9 International Classification of Diseases, Ninth Revision
ICDs implanted cardioverter defibrillators
Ig immunoglobulin (e.g., IgA, IgE, IgG)
IgG immunoglobulin
IHD ischemic heart disease
NASA International Institute for Applied Systems Analysis (an international research
organization)
IL interleukin (e.g., IL-4, IL-6, IL-8)
IMPROVE Interagency Monitoring of Protected Visual Environments (network)
10 iodine oxide
IOM Institute of Medicine
IPC International Cooperative Programme
IPCC Intergovernmental Panel on Climate Change
IPCC-AR4 IPCC 4th Assessment Report
IQR interquartile range
IR infrared
ISA Integrated Science Assessment
ISAAC International Study of Asthma and Allergies in Children
xiv
-------
ISC3 steady-state Gaussian plume dispersion model used to assess pollutant concentrations
from industrial sources
IUGR intrauterine growth retardation
i.v. intravenous (injection route)
JPL Jet Propulsion Laboratory
K potassium
K- potassium ion
Ka, Kb dissociation constant(s)
Kh Henry's Law constant in M/atm
KNO3 potassium nitrate
Kw ion product of water
LC0.01 lethal concentration at which 0.01 % of exposed animals die
LD33 lethal dose at which 33% of exposed animals die
LDH lactate dehydrogenase, lactic acid dehydrogenase
LES Large Eddy Simulation (approach)
LF low frequency
LIDAR Light Detection and Ranging (remote sensing system)
LIF laser-induced fluorescence
LIMS Limb Infrared Monitor of the Stratosphere
LOD limit of detection
LOEL lowest-observed-effect level
LRD lower respiratory disease
LRS lower respiratory symptoms
LRTAP Long Range Transport of Air Pollution
LTM Long-Term Monitoring (project)
M molar
MAD median aerodynamic diameter
MAQSIP Multiscale Air Quality Simulation Platform (model)
MAX-DOAS multiple axis differential optical absorption spectroscopy
MBL marine boundary layer
MCh methacholine
ME microenvironmental (factors)
MEF5o% maximal midexpiratory flow at 50% of forced vital capacity
MeHg methylmercury, ChbHg
MEM model ensemble mean
MENTOR Modeling Environment for Total Risk
MENTOR-1A MENTOR for One-Atmosphere (model)
MET metabolic equivalent of tasks
Mfg manufacturing
Mg magnesium
Mg2+ magnesium ion
MgO magnesium oxide
Ml myocardial infarction
MIMS membrane inlet mass spectrometry
min minute
MM5 Penn State/NCAR (National Center for Atmospheric Research) Mesoscale Model,
version 5
MMAD mass median aerodynamic density
MMD mass median diameter
xv
-------
MMEF maximal midexpiratory flow
Mn manganese
MN micronudei
MNPCE micronudeated PCE
Mo molybdenum
MOBILE6 Highway Vehide Emission Factor Model
MODIS Moderate Resolution Imaging Spectroradiometer
MONICA Monitoring Trend and Determinants in Cardiovascular Disease (registry)
MOPITT Measurement of Pollution in the Troposphere
MOZART Model for Ozone and Related Chemical Tracers
MOZART-2 Model for Ozone and Related Chemical Tracers, version 2
MPAN peroxymethacrylic nitrate
MSA metropolitan statistical area
Mt million tons
N nitrogen
N, n number of observations
N2 molecular nitrogen; nonreactive nitrogen
N2O nitrous oxide
N2O5 dinitrogen pentoxide
NA not available; insufficient data
Na sodium
Na+ sodium ion
Na2Mo04 sodium molybdate
Na2S04 sodium sulfate
NAAQS National Ambient Air Quality Standards
NaCI sodium chloride
NaC03 sodium carbonate
NADP National Atmospheric Deposition Program
NAMS National Air Monitoring Stations
NAPAP National Acid Precipitation Assessment Program
NARSTO North American Regional Strategy for Atmospheric Ozone
NAS National Academy of Sciences
NASA National Aeronautics and Space Administration
NATTS National Air Toxics Trends (network)
NCAR National Center for Atmospheric Research
NCICAS National Cooperative Inner-City Asthma Study
NCore National Core Monitoring Network
NDMA N-nitroso-dimethylamine
NEI National Emissions Inventory
NEM National Exposure Model
NEM/pNEM NEMandpNEM
NERL National Exposure Research Laboratory
NH2 amino (chemical group)
NH3 ammonia
NH4+ ammonium ion
(NH4)2S04 ammonium sulfate
NH4CI ammonium chloride
NH4NO3 ammonium nitrate
xvi
-------
NHANES National Health and Nutrition Examination Survey
NHAPS National Human Activity Pattern Survey
NHx nitrogen category label for NH3 (ammonia) plus NH4+(ammonium)
NHy total reduced nitrogen (from ammonia and ammonium)
Ni nickel
NILU Norwegian Institute for Air Research
nitro-PAH nitro-polycyclic aromatic hydrocarbon
NMBzA N-nitrosomethylbenzylamine
NMMAPS National Morbidity, Mortality, and Air Pollution Study
NO nitric oxide
NO2 nitrogen dioxide
NO2 nitrite ion
NO3 nitrate, nitrate radical
NO3 nitrate, nitrate ion
NOAA U.S. National Oceanic and Atmospheric Administration
NOAA-ARL U.S. National Oceanic and Atmospheric Administration Air Resources Laboratory
NOAEL no-observed-adverse-effect level
NOEL no-observed-effect level
NOx oxides of nitrogen; sum of NO and NO2
NOy sum of NOx and NOz; odd nitrogen species; total oxidized nitrogen
NOz sum of all inorganic and organic reaction products of NOx (HONO, HNO3, HNO4,
organic nitrates, particulate nitrate, nitro-PAHs, etc.)
NPS National Park Service
NR not reported
Nr reactive nitrogen
NRC National Research Council
NS nonsignificant
NSF National Science Foundation
nss non-sea salt
NSTC National Science and Technology Council
NTN National Trends Network
NTP National Toxicology Program
1602 isotope of oxygen
02 molecular oxygen
03 ozone
OAQPS Office of Air Quality Planning and Standards (U.S. EPA)
OC organic carbon
OCS carbonyl sulfide
OH hydroxyl radical
OR odds ratio
P phosphorus
P, p probability value
Pi 1st percentile
Ps 5th percentile
P95 95th percentile
P99 99th percentile
PAARC Air Pollution and Chronic Respiratory Diseases (study)
PAH(s) polycyclic aromatic hydrocarbon(s)
PAMS Photochemical Assessment Monitoring Stations
xvii
-------
PAN(s) peroxyacyl nitrate(s) (e.g. most common PAN: peroxyacetyl nitrate)
Pb lead
PBEM population based exposure model(s)
PBL planetary boundary layer
PC(S02) provocative concentration of SO2 that produces a 100% increase in specific airway
resistance
PCB(s) polychlorinated biphenyl compound(s)
PCE polychromatic erythrocytes
PD100 provocative dose that produces a 100% increase in sRAW
PD20 provocative dose that produces a 20%decrease in FEV1
PD20FEV1 20% decrease in forced expiratory volume in 1 second
pdf bi-Gaussian (probability density function)
PEACE Pollution Effects on Asthmatic Children in Europe (study)
PEC pulmonary endocrine cells
PEF peak expiratory flow
PEMs personal exposure monitors
PF pulsed fluorescence
pH relative acidity
PIXE proton induced X-ray emission
PKA cyclic AMP-dependent protein kinase A
pKa dissociation constant
PKI synthetic peptide inhibitor of PKA
PL phospholipids
PM particulate matter
PM10 particulate matter with a 50% upper cut point at 10 [jm aerodynamic diameter and a
collection efficiency curve as defined in the Code of Federal Regulations
PM2.5 particulate matter with a 50% upper cut point at 2.5 |jm aerodynamic diameter and a
collection efficiency curve as defined in the Code of Federal Regulations; surrogate for
fine PM
PM10-2.5 particulate matter with a 50% upper cut point at 10 |jm aerodynamic diameter, a 50%
lower cut point at 2.5 |jm aerodynamic diameter, and collection efficiency curves
identical to those for PM10 and PM2.5; surrogate for thoracic coarse PM (does not
include fine PM)
PM13 particulate matter with a 50% upper cut point at 13 |jm aerodynamic diameter
PM-CAMx Comprehensive Air Quality Model with extensions and with particulate matter chemistry
PMT photomultiplier tube
PNC particle number concentration
pNEM probabilistic National Exposure Model
PnET Photosynthesis and EvapoTranspiration (model)
PnET-BGC Photosynthesis and EvapoTranspiration-BioGeoChemical (model)
PnET-CN Photosynthesis and EvapoTranspiration model of C, water, and N balances
PnET-N-DNDC Photosynthesis and EvapoTranspiration-Denitrification-Decomposition (model)
PNO3" particulate nitrate
PO4", PO43" phosphate
POPs persistent organic pollutants
ppb parts per billion
ppbv parts per billion by volume
ppm parts per million
PPN peroxypropionyl nitrate
ppt parts per trillion
xviii
-------
pptv parts per trillion by volume
PRB policy relevant background
pS042" particulate sulfate
PTFE Polytetrafluoroethylene Filters (Teflon filter for air sampling)
Q flow rate; discharge
Q10 temperature coefficient
QAPP Quality Assurance Project Plan
QT interval measure of the time interval between the start of the Q wave and the end of the T wave
in the heart's electrical cycle
R- generic organic group attached to a molecule
R, r correlation coefficient
R2,r2 coefficient of determination
Ra aerodynamic resistance
RACM Regional Atmospheric Chemistry Mechanism
RADM Regional Acid Deposition Model
RAMS Regional Atmospheric Modeling System
RANS Reynolds Averaged Numerical Simulation (approach)
RAPS Regional Air Pollution Study
RAR rapidly activating receptor
Raw airway resistance
Rb boundary layer resistance
RBC red blood cell or erythrocyte
Rc internal resistance
R-C(0)00 organic peroxy radical
R-COO(s) strongly acidic organic anion(s)
RD BM S relational database management system
RDT Recovery Delay Time
REMAP Regional Environmental Monitoring and Assessment Program
RH relative humidity
RMR resting metabolic rate
RMSE root mean squared error
r-MSSD root mean square of successive differences in R-R intervals.
R-O2 organic peroxyl; organic peroxy
R-O2NO2 peroxynitrate
R-ONO2 organic nitrate
RR risk ratio; relative risk
RRx lognormal-transformed response ratio
RuBisCO ribulose-1,5-bisphosphate carboxylase/oxygenase
32S sulfur-32, stable isotope of sulfur
34S sulfur-34, stable isotope of sulfur
35S sulfur-35, radioactive isotope of sulfur
86Sr strontium-86, stable isotope of strontium
87Sr strontium-87, stable isotope of strontium
S sulfur
S2- sulfur radical
S2* electronically excited sulfur molecule
S2O disulfur monoxide
SAB Science Advisory Board
SAPALDIA Study of Air Pollution and Lung Diseases in Adults
xix
-------
SAPRAC Statewide Air Pollution Research Center
SAPRC Stratospheric Processes and their Role in Climate
SAVIAH Small-Area Variation in Air Pollution and Health (study)
SBL Stable Boundary Layer
SBUV Solar Backscatter Ultraviolet Spectrometer
s.c. subcutaneous (route of injection)
SC safe concentration
SCAQS Southern California Air Quality Study
SCE sister chromatid exchanges
SC IAMACHY Scanning Imaging Absorption Spectrometer for Atmospheric Cartography
SD standard deviation
SDNN standard deviation of normal R-R intervals
Se selenium;
SE, se; SEM, sem standard error; standard error of mean
SEARCH Southeastern Aerosol Research and Characterization Study (monitoring program)
SEPs somatosensory-evoked potentials
SES socioeconomic status
SGV sub grid variability
SHEDS Simulation of Human Exposure and Dose System
Si silicon
SIDS sudden infant death syndrome
SIP State Implementation Plan
SLAMS State and Local Air Monitoring Stations
SMOKE Spare-Matrix Operator Kernel Emissions
SNP single nucleotide polymorphism
SO sulfur monoxide
502 sulfur dioxide
503 sulfur trioxide
SO32- sulfite ion
SO42" sulfate ion
SOB shortness of breath
SOD superoxide dismutase
SOx sulfur oxides
SPARROW SPAtially Referenced Regressions on Watershed Attributes (model)
SPF specific pathogen free
SPM suspended particulate matter
SQCA squamous cell carcinoma
Sr strontium
sRaw specific airway resistance
SRB sulfate-reducing bacteria
SRP soluble reactive phosphorus
SSO seabuckthorn seed oil
SSWC Steady State Water Chemistry (model)
STE stratospheric-tropospheric exchange
STN Speciation Trends Network
STRF Spatio-Temporal Random Field (theory)
SUM06 seasonal sum of all hourly average concentrations > 0.06 ppm
SV40 simian virus 40
xx
-------
SVOC semivolatile organic compound
T, t time; duration of exposure
TAF Tracking and Analysis Framework (model)
T air air temperature
TAR Third Assessment Report
TBARS thiobarbituric acid reactive substances
TC total carbon
TDLAS Tunable Diode Laser Absorption Spectrometer
TEA triethanolamine
Tg teragram
TIME Temporally Integrated Monitoring of Ecosystems (program)
TNF tumor necrosis factor (e.g., TNF-a)
TOC potassium channel transient outward currents
TOR thermal-optical reflectance (method)
TRACE-P Transport and Chemical Evolution over the Pacific
TSP total suspended particles
TSS total suspended solids
TTX tetrodotoxin
TTX-R tetrodotoxin-resistant
TTX-S tetrodotoxin-sensitive
Twater water temperature
U.S. United States of America
UMD-CTM University of Maryland Chemical Transport Model
UNECE United Nations Economic Commission for Europe
URI upper respiratory infections
URS upper respiratory symptoms
USDA U.S. Department of Agriculture
USFS U.S. Forest Service
USGS U.S. Geological Survey
UV ultraviolet
UV-A ultraviolet radiation of wavelengths from 320 to 400 nm
UV-B ultraviolet radiation of wavelengths from 280 to 320 nm
Ve minute ventilation
Vd deposition rate
VEPs visual-evoked potentials
VOC volatile organic compound
W tungsten
WHO World Health Organization
WMO World Meteorological Organization
WRF Weather Research and Forecasting (model)
wt % percent by weight
XNO3 nitrate halogen-X salt
XO halogen-X oxide
XRF X-ray fluorescence
yr year
Zn zinc
ZnO zinc oxide
xxi
-------
Authors and Contributors
Authors
Dr. Jee Young Kim (SOx Team Leader)—National Center for Environmental Assessment,
U.S. Environmental Protection Agency, Research Triangle Park, NC
Dr. Jeffrey Arnold—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. James S. Brown—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Barbara Buckley—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Ila Cote—National Center for Environmental Assessment, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. Douglas Johns—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Ellen Kirrane—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Dennis Kotchmar—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Thomas Long—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Thomas Luben—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Qingyu Meng—Oak Ridge Institute for Science and Education, Postdoctoral Research Fellow to
National Center for Environmental Assessment, U.S. Environmental Protection Agency, Research
Triangle Park, NC
Dr. Anu Mudipalli—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Joseph Pinto—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Mary Ross—National Center for Environmental Assessment, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. David Svendsgaard—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Lori White—National Center for Environmental Assessment, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. William Wilson—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
xxii
-------
Dr. Brett Grover—National Exposure Research Laboratory, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. Douglas Bryant—Intrinsik Science, Mississauga, Ontario, Canada
Dr. Arlene Fiore—Geophysical Fluid Dynamics Laboratory/National Oceanographic & Atmospheric
Administration, Princeton, NJ
Dr. Panos Georgopoulos—Computational Chemodynamics Laboratory, Environmental and Occupational
Health Sciences Institute, Piscataway, NJ
Dr. Vic Hasselblad—Duke University Medical Center, Durham, NC
Dr. Larry Horowitz—Geophysical Fluid Dynamics Laboratory/National Oceanographic and Atmospheric
Administration, Princeton University Forrestal Campus, Princeton, NJ
Ms. Annette Ianucci—Sciences International, Alexandria, VA
Dr. Kazuhiko Ito—Department of Environmental Medicine, New York University School of Medicine,
Tuxedo, NY
Dr. Jane Koenig— Department of Environmental and Occupational Health Sciences, University of
Washington, Seattle, WA
Dr. Therese Mar— Department of Environmental and Occupational Health Sciences, University of
Washington, Seattle, WA
Dr. James Riddle—Sciences International, Alexandria, VA
Contributors
Ms. Christina Cain—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Ms. Rebecca Daniels—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Steven J. Dutton—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Elizabeth Oesterling —Oak Ridge Institute for Science and Education, Postdoctoral Research Fellow
to National Center for Environmental Assessment, U.S. Environmental Protection Agency, Research
Triangle Park, NC
Dr. Jennifer Richmond-Bryant—National Center for Environmental Assessment, U.S. Environmental
Protection Agency, Research Triangle Park, NC
Mr. Jason Sacks—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Dale Allen—Department of Atmospheric and Oceanic Sciences, University of Maryland, College
Park, MD
Ms. Louise Camalier—Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC
xxiii
-------
Dr. Russell Dickerson—Department of Atmospheric and Oceanic Science, University of Maryland,
College Park, MD
Dr. Tina Fan—Environmental and Occupational Health Sciences Institute, Piscataway, NJ
Dr. William Keene—Department of Environmental Sciences, University of Virginia, Charlottesville, VA
Dr. Randall Martin—Department of Physics and Atmospheric Science, Dalhousie University, Halifax,
Nova Scotia, Canada
Dr. Maria Morandi—Department of Environmental Sciences, School of Public Health, University of
Texas - Houston Health Science Center, Houston, TX
Dr. William Munger—Division of Engineering and Applied Sciences, Harvard University, Cambridge,
MA
Mr. Charles Piety—Department of Meteorology, University of Maryland, College Park, MD
Dr. Sandy Sillman—Department of Atmospheric, Ocean, and Space Sciences, University of Michigan,
Ann Arbor, MI
Dr. Helen Suh—Department of Environmental Health, Harvard School of Public Health, Boston, MA
Dr. Charles Wechsler—Environmental and Occupational Health Sciences Institute, Piscataway, NJ
Dr. Clifford Weisel—Environmental and Occupational Health Sciences Institute, Piscataway, NJ
Dr. Jim Zhang—Environmental and Occupational Health Sciences Institute, Piscataway, NJ
Reviewers
Dr. Tina Bahadori—American Chemistry Council, Arlington, VA
Dr. Tim Benner—Office of Science Policy, U.S. Environmental Protection Agency, Washington, DC
Dr. Daniel Costa—National Program Director for Air, U.S. Environmental Protection Agency, Research
Triangle Park, NC
Dr. Robert Devlin—National Health and Environmental Effects Research Laboratory, U.S. Environmental
Protection Agency, Chapel Hill, NC
Dr. Judy Graham—American Chemistry Council, Arlington, VA
Dr. Stephen Graham—Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Ms. Beth Hassett-Sipple—Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Gary Hatch—National Health and Environmental Effects Research Laboratory, U.S. Environmental
Protection Agency, Research Triangle Park, NC
Dr. Scott Jenkins—Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. David Kryak—National Exposure Research Laboratory, U.S. Environmental Protection Agency,
Research Triangle Park, NC
xxiv
-------
Mr. John Langstaff—Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Morton Lippmann—Department of Environmental Medicine, New York University School of
Medicine, Tuxedo, NY
Dr. Karen Martin—Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. William McDonnell—William F. McDonnell Consulting, Chapel Hill, NC
Dr. Dave McKee—Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Ms. Connie Meacham—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Lucas Neas—National Health and Environmental Effects Research Laboratory, U.S. Environmental
Protection Agency, Chapel Hill, NC
Dr. Russell Owen—National Health and Environmental Effects Research Laboratory, U.S. Environmental
Protection Agency, Research Triangle Park, NC
Dr. Haluk Ozkaynak—National Exposure Research Laboratory, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. Jennifer Peel—Department of Environmental and Radiological Health Sciences, Colorado State
University, Fort Collins, CO
Mr. Harvey Richmond—Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Mr. Steven Silverman—Office of General Counsel, U.S. Environmental Protection Agency, Washington,
DC
Dr. Michael Stewart—Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Ms. Susan Stone—Office of Air Quality Planning and Standards, Office of Air and Radiation, U.S.
Environmental Protection Agency, Research Triangle Park, NC
Ms. Chris Trent—Office of Air Quality Planning and Standards, Office of Air and Radiation, U.S.
Environmental Protection Agency, Research Triangle Park, NC
Dr. John Vandenberg—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Alan Vette—National Exposure Research Laboratory, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Ms. Debra Walsh—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Mr. Ron Williams—National Exposure Research Laboratory, U.S. Environmental Protection Agency,
Research Triangle Park, NC
xxv
-------
SOx Project Team
Executive Direction
Dr. Ila Cote (Acting Director)—National Center for Environmental Assessment-RTP Division, U.S.
Environmental Protection Agency, Research Triangle Park, NC
Ms. Debra Walsh (Deputy Director)—National Center for Environmental Assessment-RTP Division, U.S.
Environmental Protection Agency, Research Triangle Park, NC
Dr. Mary Ross (Branch Chief)—National Center for Environmental Assessment, U.S. Environmental
Protection Agency, Research Triangle Park, NC
Scientific Staff
Dr. Jee Young Kim (SOx Team Leader)—National Center for Environmental Assessment, U.S.
Environmental Protection Agency, Research Triangle Park, NC
Dr. Jeffrey Arnold—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. James S. Brown—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Barbara Buckley—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Steven J. Dutton—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Douglas Johns—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Ellen Kirrane—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Dennis Kotchmar—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Thomas Long—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Thomas Luben—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Qingyu Meng—Oak Ridge Institute for Science and Education, Postdoctoral Research Fellow to
National Center for Environmental Assessment, U.S. Environmental Protection Agency, Research
Triangle Park, NC
Dr. Anu Mudipalli—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
xxvi
-------
Dr. Elizabeth Oesterling —Oak Ridge Institute for Science and Education, Postdoctoral Research Fellow
to National Center for Environmental Assessment, U.S. Environmental Protection Agency, Research
Triangle Park, NC
Dr. Joseph Pinto—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Jennifer Richmond-Bryant—National Center for Environmental Assessment, U.S. Environmental
Protection Agency, Research Triangle Park, NC
Mr. Jason Sacks—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. David Svendsgaard—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Dr. Lori White—National Center for Environmental Assessment, U.S. Environmental Protection Agency,
Research Triangle Park, NC
Dr. William Wilson—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Technical Support Staff
Ms. Ellen Lorang—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Ms. Connie Meacham—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Ms. Deborah Wales—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
Mr. Richard Wilson—National Center for Environmental Assessment, U.S. Environmental Protection
Agency, Research Triangle Park, NC
xxvii
-------
Clean Air Scientific Advisory Committee for
Sulfur Oxides Primary NAAQS Review Panel
Chairperson
Dr. Rogene Henderson*, Senior Scientist Emeritus, Lovelace Respiratory Research Institute,
Albuquerque, NM
Members
Mr. Ed Avol, Professor, Preventive Medicine, Keck School of Medicine, University of Southern
California, Los Angeles, CA
Dr. John R. Balmes, Professor, Department of Medicine, Division of Occupational and Environmental
Medicine, University of California, San Francisco, CA
Dr. Ellis Cowling*, University Distinguished Professor At-Large Emeritus, Colleges ofNatural
Resources and Agriculture and Life Sciences, North Carolina State University, Raleigh, NC
Dr. James D. Crapo*, Professor of Medicine, Department of Medicine, National Jewish Medical and
Research Center, Denver, CO
Dr. Douglas Crawford-Brown*, Professor and Director, Department of Environmental Sciences and
Engineering, Carolina Environmental Program, University of North Carolina, Chapel Hill, NC
Dr. Terry Gordon, Professor, Environmental Medicine, NYU School of Medicine, Tuxedo, NY
Dr. Dale Hattis, Research Professor, Center for Technology, Environment, and Development, George
Perkins Marsh Institute, Clark University, Worcester, MA
Dr. Donna Kenski, Director of Data Analysis, Lake Michigan Air Directors Consortium, Rosemont, IL
Dr. Patrick Kinney, Associate Professor, Department of Environmental Health Sciences, Mailman
School of Public Health, Columbia University, New York, NY
Mr. Steven Kleeberger, Professor and Lab Chief, Laboratory of Respiratory Biology, National Institute
of Environmental Health Sciences, National Institutes of Health, Research Triangle Park, NC
Dr. Timothy Larson, Professor, Department of Civil and Environmental Engineering, University of
Washington, Seattle, WA
Dr. Kent Pinkerton, Professor, Regents of the University of California, Center for Health and the
Environment, University of California, Davis, CA
Mr. Richard L. Poirot*, Environmental Analyst, Air Pollution Control Division, Department of
Environmental Conservation, Vermont Agency ofNatural Resources, Waterbury, VT
Dr. Edward Postlethwait, Professor and Chair, Department of Environmental Health Sciences, School of
Public Health, University of Alabama, Birmingham, AL
Dr. Armistead (Ted) Russell*, Professor, Department of Civil and Environmental Engineering, Georgia
Institute of Technology, Atlanta, GA
xxviii
-------
Dr. Jonathan M Samet, Professor and Chair, Department of Epidemiology, Bloomberg School of Public
Health, Johns Hopkins University, Baltimore, MD
Dr. Richard Schlesinger, Associate Dean, Department of Biology, Dyson College, Pace University, New
York, NY
Dr. Christian Seigneur, Vice President, Atmospheric & Environmental Research, Inc., San Ramon, CA
Dr. Elizabeth A. (Lianne) Sheppard, Research Professor, Biostatistics and Environmental &
Occupational Health Sciences, Public Health and Community Medicine, University of Washington,
Seattle, WA
Dr. Frank Speizer*, Edward Kass Professor of Medicine, Channing Laboratory, Harvard Medical
School, Boston, MA
Dr. George Thurston, Professor, Environmental Medicine, NYU School of Medicine, New York
University, Tuxedo, NY
Dr. James Ultman, Professor, Chemical Engineering, Bioengineering Program, Pennsylvania State
University, University Park, PA
Dr. Ronald Wyzga, Technical Executive, Air Quality Health and Risk, Electric Power Research Institute,
P.O. Box 10412, Palo Alto, CA
Science Advisory Board Staff
Dr. Holly Stallworth, Designated Federal Officer, 1200 Pennsylvania Avenue, NW 1400F, Washington,
DC, 20460, Phone: 202-343-9981, Fax: 202-233-9867, (stallworth.holly@epa.gov)
* Members of the statutory Clean Air Scientific Advisory Committee (CAS AC) appointed by the EPA Administrator
xxix
-------
Preface
Legislative Requirements
Section 109, Title 42 (U.S. Code, 2003) directs the U.S. Environmental Protection Agency (EPA)
Administrator to propose and promulgate "primary" and "secondary" National Ambient Air Quality
Standards (NAAQS) for pollutants listed under section 108. Section 109(b)(1) defines a primary standard
as one "the attainment and maintenance of which in the judgment of the Administrator, based on such
criteria and allowing an adequate margin of safety, are requisite to protect the public health."1 A
secondary standard, as defined in section 109(b)(2), must "specify a level of air quality the attainment and
maintenance of which, in the judgment of the EPA Administrator, based on such criteria, is required to
protect the public welfare from any known or anticipated adverse effects associated with the presence of
[the] pollutant in the ambient air."2 The requirement that primary standards include an adequate margin of
safety was intended to address uncertainties associated with inconclusive scientific and technical
information available at the time of standard setting. It was also intended to provide a reasonable degree
of protection against hazards that research has not yet identified. See Lead Industries Association v. EPA,
647 F.2d 1130, 1154 (D.C. Cir, 1980), cert, denied. 449 U.S. 1042 (1980); American Petroleum Institute v.
Costle, 665 F.2d 1176, 1186 (D.C. Cir, 1981) cert, denied. 455 U.S. 1034 (1982). The aforementioned
uncertainties are components of the risk associated with pollution at levels below those at which human
health effects can be said to occur with reasonable scientific certainty. Thus, in selecting primary
standards that include an adequate margin of safety, the Administrator is seeking not only to prevent
pollution levels that have been demonstrated to be harmful but also to prevent lower pollutant levels that
may pose an unacceptable risk of harm, even if the risk is not precisely identified as to nature or degree.
In selecting a margin of safety, the EPA considers such factors as the nature and severity of the
health effects involved, the size of sensitive or vulnerable population(s) at risk, and the kind and degree of
the uncertainties that must be addressed. The selection of any particular approach to providing an
adequate margin of safety is a policy choice left specifically to the Administrator's judgment. See Lead
Industries Association v. EPA, supra. 647 F.2d at 1161-62.
In setting standards that are "requisite" to protect public health and welfare, as provided in section
109(b), EPA's task is to establish standards that are neither more nor less stringent than necessary for
these purposes. In so doing, EPA may not consider the costs of implementing the standards. See Whitman
v. American Trucking Associations, 531 U.S. 457, 465472, 475-76 (U.S. Supreme Court 2001).
Section 109(d)(1) requires that "not later than December 31, 1980, and at 5-year intervals
thereafter, the Administrator shall complete a thorough review of the criteria published under section 108
and the national ambient air quality standards... and shall make such revisions in such criteria and
standards and promulgate such new standards as may be appropriate..." Section 109(d)(2) requires that an
independent scientific review committee "shall complete a review of the criteria.. .and the national
primary and secondary ambient air quality standards... and shall recommend to the Administrator any
new... standards and revisions of existing criteria and standards as may be appropriate..." Since the early
1980s, this independent review function has been performed by the Clean Air Scientific Advisory
Committee (CASAC) of EPA's Science Advisory Board (SAB).
1 The legislative history of section 109 indicates that a primary standard is to be set at "the maximum permissible ambient air level. . which will
protect the health of any [sensitive] group of the population," and that for this purpose "reference should be made to a representative sample of
persons comprising the sensitive group rather than to a single person in such a group" [S. Rep. No. 91-1196, 91st Cong., 2d Sess. 10 (1970)].
(Tsai et al., 2006)
2 Welfare effects as defined in section 302(h) [42 U.S.C. 7602(h)] include, but are not limited to, "effects on soils, water, crops, vegetation, man-
made materials, animals, wildlife, weather, visibility and climate, damage to and deterioration of property, and hazards to transportation, as well
as effects on economic values and on personal comfort and well-being."
XXX
-------
History of Reviews of the Primary NAAQS for Sulfur Oxides
On April 30, 1971, the EPA promulgated primary NAAQS for sulfur oxides (SOx). These primary
standards, which were based on the findings outlined in the original 1969 Air Quality Criteria for Sulfur
Oxides, were set at 0.14 parts per million (ppm) averaged (avg) over a 24-hour (h) period, not to be
exceeded more than once per year, and 0.030 ppm annual arithmetic mean with sulfur dioxide (S02) as
the indicator. In 1982, EPA published the Air Quality Criteria for Particulate Matter and Sulfur Oxides
(U.S. EPA, 1982) along with an addendum of newly published controlled human exposure studies, which
updated the scientific criteria upon which the initial standards were based. In 1986, a second addendum
was published presenting newly available evidence from epidemiologic and controlled human exposure
studies (U.S. EPA, 1986a). In 1988, EPA published a proposed decision not to revise the existing
standards (53 FR 14926). However, EPA specifically requested public comment on the alternative of
revising the current standards and adding a new 1-h primary standard of 0.4 ppm.
As a result of public comments on the 1988 proposal and other post-proposal developments, EPA
published a second proposal in the Federal Register (FR) on November 15, 1994 (59 FR 58958). The
1994 re-proposal was based in part on a supplement to the second addendum of the criteria document,
which evaluated new findings on short-term S02 exposures in asthmatics (U.S. EPA, 1994c). As in the
1988 proposal, EPA proposed to retain the existing 24-h and annual standards. The EPA also solicited
comment on three regulatory alternatives to further reduce the health risk posed by exposure to high 5-
minute (min) peaks of S02 if additional protection were judged to be necessary. The three alternatives
were: 1) revising the existing primary S02NAAQS by adding a new 5-min standard of 0.60 ppm S02; 2)
establishing a new regulatory program under section 303 of the Act to supplement protection provided by
the existing NAAQS, with a trigger level of 0.60 ppm S02, one expected exceedance; and 3) augmenting
implementation of existing standards by focusing on those sources or source types likely to produce high
5-min peak concentrations of S02. On May 22, 1996, EPA's final decision, that revisions of the NAAQS
for SOx were not appropriate at that time, was announced in the Federal Register (61 FR 25566).
xxxi
-------
Chapter 1. Introduction
The Integrated Science Assessment (ISA) is a concise review, synthesis, and evaluation of the most
policy-relevant science, and communicates critical science judgments relevant to the NAAQS review. As
such, the ISA forms the scientific foundation for the review of the primary (health-based) NAAQS for
SOx-1 The primary NAAQS for SOx, with S02 serving as the indicator, is set at 0.14 ppm, averaged over
a 24-h period, not to be exceeded more than once per year, and 0.030 ppm annual arithmetic mean. The
ISA accurately reflects "the latest scientific knowledge useful in indicating the kind and extent of
identifiable effects on public health which may be expected from the presence of [a] pollutant in ambient
air" (Clean Air Act, Section 108, 2003). Key information and judgments formerly contained in the Air
Quality Criteria Document (AQCD) for SOx are incorporated in this assessment. Additional details of the
pertinent scientific literature published since the last review, as well as selected older studies of particular
interest, are included in a series of annexes. This ISA thus serves to update and revise the information
available at the time of the previous review of the NAAQS for SOx in 1996.
S02 is the most important of the gas-phase sulfur oxides (SOx) for both atmospheric chemistry and
health effects. SOx is usually defined to include sulfur trioxide (S03) and gas-phase sulfuric acid (H2S04)
as well, but neither is present in the atmosphere in concentrations significant for human exposures.
Descriptions of the atmospheric chemistry of SOx include both gaseous and particulate species; a
meaningful analysis would not be possible otherwise. Most studies on the health effects of gaseous SOx
focus on S02; effects of other gaseous species are considered as information is available. The health
effects of particulate SOx are included in the review of the NAAQS for particulate matter (PM). In
evaluating the health evidence, this ISA considers possible influences of other atmospheric pollutants,
including interactions of S02 with other co-occurring pollutants such as PM, nitrogen oxides (NOx),
carbon monoxide (CO), and ozone (03).
The Integrated Plan for Review of the Primary NAAQS for SOx (U.S. EPA, 2007) identifies key
policy-relevant questions that provide a framework for this review of the scientific evidence. These
questions frame the entire review of the NAAQS, and thus are informed by both science and policy
considerations. The ISA organizes and presents the scientific evidence such that, when considered along
with findings from risk analyses and policy considerations, will help the EPA address these questions
during the NAAQS review.
¦ How has new information altered/substantiated the scientific support for the occurrence of
health effects following short- and/or long-term exposure to levels of SOx found in the
ambient air?
¦ How does new information influence conclusions from the previous review regarding the
effects of SOx on susceptible populations?
¦ At what levels of SOx exposure do health effects of concern occur?
¦ How has new information altered conclusions from previous reviews regarding the
plausibility of adverse health effects caused by SOx exposure?
¦ To what extent have important uncertainties identified in the last review been reduced?
Have new uncertainties emerged?
¦ What are the air quality relationships between short-term and long-term exposures to SOx?
'A review of the secondary S02 NAAQS, in conjunction with a review of the secondary NAAQS for NOx, is underway independently, as is a
review of the primary NAAQS for NOx and a review of the primary and secondary NAAQS for PM.
1-1
-------
1.1. Document Development
EPA initiated the current formal review of the NAAQS for SOx on May 15, 2006 with a call for
information from the public (FR, 2006). In addition to the call for information, publications are identified
through an ongoing literature search process that includes extensive computer database mining on specific
topics. Additional publications were identified by EPA scientists in a variety of disciplines by combing
through relevant, peer-reviewed scientific literature obtained through these ongoing literature searches,
reviewing previous EPA reports, and a review of reference lists from important publications. All relevant
epidemiologic, human clinical, and animal toxicological studies, including those related to exposure-
response relationships, mechanism(s) of action, or susceptible subpopulations published since the last
review were considered. Added to the body of research were EPA's analyses of air quality and emissions
data, studies on atmospheric chemistry, transport, and fate of these emissions, as well as issues related to
exposure to SOx. Further information was acquired from consultation with content and area experts and
the public. Annex A has more discussion of search strategies and criteria for study selection.
1.2. Document Organization
The ISA is composed of five chapters. This introductory chapter presents background information,
and provides an overview of EPA's framework for making causal judgments. Chapter 2 highlights key
concepts or issues relevant to understanding the atmospheric chemistry, sources, exposure, and dosimetry
of SOx, following a "source-to-dose" paradigm. Chapter 3 evaluates and integrates epidemiologic, human
clinical, and animal toxicological information relevant to the review of the primary NAAQS for SOx.
Chapter 4 has information related to the public health impact of ambient SOx exposure, with emphasis on
potentially susceptible and vulnerable population groups. Finally, Chapter 5 presents key findings and
conclusions from the atmospheric sciences, ambient air data analyses, exposure assessment, dosimetry,
and health effects for consideration in the review of the NAAQS for SOx.
A series of annexes supplement this ISA. The annexes provide additional details of the pertinent
literature published since the last review, as well as selected older studies of particular interest. These
annexes contain information on:
¦ atmospheric chemistry of SOx as well as the sampling and analytic methods for
measurement of SOx;
¦ environmental concentrations and human exposure to SOx;
¦ toxicological studies of health effects in laboratory animals;
¦ human clinical studies of health effects related to peak (5-10 min) and short-term (1-h or
longer) exposure to SOx; and
¦ epidemiologic studies of health effects from short- and long-term exposure to SOx.
Detailed information about methods and results of health studies is summarized in tabular format, and
generally includes information about: concentrations of SOx and averaging times; study methods
employed; results and comments; and quantitative results for relationships between effects and exposure
to SOx.
1.3. EPA Framework for Causal Determination
It is important to have a consistent and transparent basis to evaluate the causal nature of air
pollution-induced health effects. The framework described below establishes uniform language
1-2
-------
concerning causality and brings more specificity to the findings. It draws standardized language from
across the federal government and wider scientific community, especially from the recent National
Academy of Sciences (NAS) Institute of Medicine (IOM) document, Improving the Presumptive
Disability Decision-Making Process for Veterans (IOM, 2007), the most recent comprehensive work on
evaluating the causality of health effects. This section:
¦ describes the kinds of scientific evidence used in establishing a general causal relationship
between exposure and health effects;
¦ defines cause, in contrast to statistical association;
¦ discusses the sources of evidence necessary to reach a conclusion about the existence of a
causal relationship;
¦ highlights the issue of multifactorial causation;
¦ identifies issues and approaches related to uncertainty; and
¦ provides a framework for classifying and characterizing the weight of evidence in support
of a general causal relationship.
Approaches to assessing the separate and combined lines of evidence (e.g., epidemiologic, human
clinical, and animal toxicological studies) have been formulated by a number of regulatory and science
agencies, including the IOM of the National Academies of Science (IOM, 2008), International Agency for
Research on Cancer (IARC, 2006), EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005),
Centers for Disease Control and Prevention (CDC, 2004), and National Acid Precipitation Assessment
Program (NAPAP, 1991). Highlights or excerpts from the various decision framework documents are
included in Annex A.
These formalized approaches offer guidance for assessing causality. The frameworks are similar in
nature, although adapted to different purposes, and have proven effective in providing a uniform structure
and language for causal determinations. Moreover, these frameworks must support decision-making under
conditions of uncertainty.
1.3.1. Scientific Evidence Used in Establishing Causality
Causality determinations are based on the evaluation and synthesis of evidence from across
scientific disciplines; the type of evidence that is most important for such determinations will vary by
pollutant or assessment. The most compelling evidence of a causal relationship between pollutant
exposures and human health effects comes from human clinical studies. This type of study experimentally
evaluates the health effects of administered exposures in humans under highly-controlled laboratory
conditions.
In epidemiologic or observational studies of humans, the investigator does not control exposures or
intervene with the study population. Broadly, observational studies can describe associations between
exposures and effects. These studies fall into several categories: cross-sectional, prospective cohort, and
time-series studies. "Natural experiments" offer the opportunity to investigate changes in health with a
change in exposure; these include comparisons of health effects before and after a change in population
exposures, such as closure of a pollution source.
Experimental animal data complement the clinical and observational data; these studies can help
characterize effects of concern, exposure-response relationships, sensitive subpopulations and modes of
action. In the absence of clinical or epidemiologic data, animal data alone may be sufficient to support a
likely causal determination, assuming that humans respond similarly to the experimental species.
1-3
-------
1.3.2. Association and Causation
"Cause" is a significant, effectual relationship between an agent and an associated disorder or
disease. "Association" is the statistical dependence among events, characteristics, or other variables. An
association is prima facie evidence for causation; alone, however, it is insufficient proof of a causal
relationship between exposure and disease. Unlike an association, a causal claim supports the creation of
counterfactual claims; that is, a claim about what the world would have been like under different or
changed circumstances (IOM, 2007). Much of the newly available health information evaluated in this
ISA comes from epidemiologic studies that report a statistical association between ambient exposure and
health outcome.
Many of the health outcomes reported in these studies have complex etiologies. The diseases, such
as asthma, coronary heart disease or cancer are typically initiated by a web of multiple agents. Outcomes
depend on a variety of factors, such as age, genetic susceptibility, nutritional status, immune competence,
and social factors (Gee and Payne-Sturges, 2004; IOM, 2007). Further, exposure to a combination of
agents could cause synergistic or antagonistic effects. Thus, the observed risk represents the net effect of
many actions and counteractions.
1.3.3. Evidence for Going beyond Association to Causation
Moving from association to causation involves elimination of alternative explanations for the
association. Human clinical studies are experiments conducted in a controlled laboratory setting using
fixed concentrations of air pollutants under carefully regulated environmental conditions and subject
activity levels. Results of human clinical studies may provide evidence of potential mechanisms for
observed effects and a direct quantitative assessment of the exposure-health response relationship among
individuals. In a randomized crossover study design, subjects in a population are exposed to a pollutant
and a sham at different time points, and the responses to the two types of exposures are compared. This
study design is effective at controlling for potential confounders, since the subject is serving as his/her
own control. The results are assessed by rigorous comparison of changes in appropriate outcomes
between the pollutant and sham exposures. By assigning exposure randomly, the study design attempts to
remove the effect of any factor that might influence exposure. Done properly, and setting aside the role of
chance, only a causal relationship between exposure and health outcome should produce observed
associations in randomized clinical trials.
A lack of observation of effects from human clinical studies does not necessarily mean that a causal
relationship does not exist. Human clinical studies are often limited because the study population is
generally small, which restricts the ability to discern statistically significant findings in the health
outcomes of interest between exposure to varying concentrations of air pollutant and clean air and to
precisely characterize the exposure-response relationship. In addition, the most susceptible individuals
may be explicitly excluded (for ethical reasons), and other susceptible individuals or groups, such as those
with preexisting health conditions, may not be included. Study subjects must either be healthy, or have a
level of illness which does not preclude them from participating in the study. Asthmatics who are unable
to withhold the use of bronchodilators for at least 6 hours prior to exposure and subjects with a recent
history of upper respiratory tract infections are typically excluded from clinical studies of air pollution
exposure. While human clinical studies provide important information on the biological plausibility of
associations observed between S02 exposure and health outcomes in epidemiologic studies, observed
effects in these studies may underestimate the response in certain subpopulations for these reasons.
Epidemiologic studies provide important information on the associations between health effects
and exposure of human populations to ambient air pollution. These studies also help to identify
susceptible or vulnerable subgroups and associated risk factors. However, associations observed between
specific air pollutants and health outcomes in epidemiologic studies may be confounded by copollutants
1-4
-------
or meteorological conditions. Extensive discussion of these issues is provided in the 2004 AQCD for PM
(U.S. EPA, 2004) and the 2006 AQCD for Ozone and Related Photochemical Oxidants (U.S. EPA,
2006b), and therefore presented only briefly below.
Inferring causation requires consideration of potential confounders. In confounding, the apparent
effect of the exposure of interest is distorted because the effect of an extraneous factor is mistaken for or
mixed with the actual exposure effect, which may be null (Rothman and Greenland, 1998). When
associations are found in epidemiologic studies, one approach to remove spurious associations from
possible confounders is to control for characteristics that may differ between exposed and unexposed
persons; this is frequently termed "adjustment." Multivariable regression models constitute one tool for
estimating the association between exposure and outcome after adjusting for characteristics of participants
that might confound the results. The use of multipollutant regression models has been the prevailing
approach for controlling potential confounding by copollutants in air pollution health effects studies.
Finding the likely causal pollutant from multipollutant regression models is made difficult by the
possibility that one or more air pollutants may be acting as a surrogate for an unmeasured or poorly-
measured pollutant or for a particular mixture of pollutants. Further, the correlation between the air
pollutant of interest and various copollutants may show temporal and spatial discongruities that can
influence exposures and health effects. Thus, results of models that attempt to distinguish gaseous and
particle effects must be interpreted with caution. Despite these limitations, the use of multipollutant
models is still the prevailing approach employed in most air pollution epidemiologic studies, and may
provide some insight into the potential for confounding or interaction among pollutants.
Another way to adjust for potential confounding is through stratified analysis, i.e., examining the
association within homogeneous groups with respect to the confounding variable. Stratified analysis can
also be used to examine potential effect modification. The use of stratified analyses has an additional
benefit: it allows examination of effect modification through comparison of the effect estimates across
different groups. If investigators successfully measured characteristics that distort the results, adjustment
of these factors help separate a spurious from a true causal association. Appropriate statistical adjustment
for confounders requires identifying and measuring all reasonably expected confounders. Deciding which
variables to control for in a statistical analysis of the association between exposure and disease depends
on knowledge about possible mechanisms and the distributions of these factors in the population under
study. Identifying these mechanisms makes it possible to control for potential sources that may result in a
spurious association.
Measurement error is another problem encountered when adjusting for spurious associations.
Controlling for confounders, whether by adjustment or stratification, is only successful when the
confounder is well-measured. Considered together, the effects of a well-measured covariate may be
overestimated in contrast to a covariate measured with greater error. There are several components that
contribute to exposure measurement error in these studies, including the difference between true and
measured ambient concentrations, the difference between avg personal exposure to ambient pollutants
and ambient concentrations at central monitoring sites, and the use of average population exposure rather
than individual exposure estimates.
Confidence that unmeasured confounders are not producing the findings is increased when multiple
studies are conducted in various settings using different subjects or exposures; each of which might
eliminate another source of confounding from consideration. Thus, multi-city studies which use a
consistent method to analyze data from across locations with different levels of covariates can provide
insight on potential confounding in associations. The number and degree of diversity of covariates, as
well as their relevance to the potential confounders, remain matters of scientific judgment. Intervention
studies, because of their experimental nature, can be particularly useful in characterizing causation.
In addition to clinical and epidemiologic studies, the tools of experimental biology have been
valuable for developing insights into human physiology and pathology. Laboratory tools have been
extended to explore the effects of putative toxicants on human health, especially through the study of
model systems in other species. Background knowledge of the biological mechanisms by which an
exposure might or might not cause disease can prove crucial in establishing, or negating, a causal claim.
1-5
-------
At the same time, species can differ from each other in fundamental aspects of physiology and anatomy
(e.g., metabolism, airway branching, hormonal regulation) that may limit extrapolation. Testable
hypotheses about the causal nature of proposed mechanisms or modes of action are central to utilizing
experimental data in causal determinations.
Direct Causa! Effect
set
t
>- I Risk for outcome
Source
Surrogate
SOv
Other
Pollutants
t
Risk for outcome
Mediated Effect
SOx —•> PM >¦ I Risk for outcome
t,
Confounding
sex
1
Confounder
t
Risk for outcome
Figure 1 -1. Potential relationships of SOx with adverse health effects.
1.3.4. Multifactorial Causation
Scientific judgment is needed regarding likely sources and magnitude of confounding, together
with judgment about how well the existing constellation of study designs, results, and analyses address
this potential threat to inferential validity. One key consideration in this review is evaluation of the
potential contribution of SOx to health effects, when it is a component of a complex air pollutant mixture.
1-6
-------
There are multiple ways by which S0X might cause or be associated with adverse health effects, as
illustrated in Figure 1-1. First, the reported SOx effect estimates in epidemiologic studies may reflect
independent SOx effects on respiratory health. Second, ambient SOx may be serving as an indicator of
complex ambient air pollution mixtures that share the same source as SOx (i.e., combustion of sulfur-
containing fuels or metal smelting). Finally, copollutants may mediate the effects of SOx or SOx may
influence the toxicity of copollutants. Epidemiologists use the term "interaction" or "effect modification"
to denote the departure of the observed joint risk from what might be expected based on the separate
effects of the factors. These possibilities are not necessarily exclusive. In addition, confounding can result
in the production of an association between adverse health effects and SOx that is actually attributable to
another factor that is associated with SOx in a particular study. Multivariate models are the most widely
used strategy to address confounding in epidemiologic studies, but such models are not always easily
interpreted when assessing effects of covarying pollutants such as PM, 03, and nitrogen dioxide (N02).
1.3.5. Uncertainty1
In estimating the causal influence of an exposure on health outcomes, it is recognized that scientific
findings incorporate uncertainty. Uncertainty can be defined as a state of having limited knowledge where
it is impossible to exactly describe an existing state or future outcome - the lack of knowledge about the
correct value for a specific measure or estimate.2 Uncertainty characterization and uncertainty assessment
are two activities that lead to different degrees of sophistication in describing uncertainty. Uncertainty
characterization generally involves a qualitative discussion of the thought processes that lead to the
selection and rejection of specific data, estimates, scenarios, etc. The uncertainty assessment is more
quantitative. The process begins with simpler measures (i.e., ranges) and simpler analytical techniques
and progresses, to the extent needed to support the decision for which the assessment is conducted, to
more complex measures and techniques. Data will not be available for all aspects of an assessment and
those data that are available may be of questionable or unknown quality. In these situations, evaluation of
uncertainty can include professional judgment or inferences based on analogy with similar situation. The
net result is that the assessments will be based on a number of assumptions with varying degrees of
uncertainty. Additionally, uncertain information from different sources of different quality must be
combined. It is important that uncertainty be qualitatively and, to the extent possible, quantitatively
described.
1.3.5.1. Types of Uncertainty
Uncertainty in assessment can be classified into three broad categories: uncertainty regarding
missing or incomplete information needed to fully define the exposure and dose (scenario uncertainty);
uncertainty regarding some parameters (parameter uncertainty); and uncertainty regarding gaps in
scientific theory required to make predictions on the basis of causal inferences (model uncertainty).
Identification of the sources of uncertainty is the first step toward eventually determining the type of
action necessary to reduce that uncertainty. The following sections will discuss sources of uncertainty and
approaches for analyzing uncertainty.
1 This discussion on uncertainty was adapted from EPA's Guidelines for Exposure Assessment, May 29, 1992, Federal Register 57(104):22888-
22938, EPA/600/Z-92/001.
2 Variability is distinguished from uncertainty in that the former is an actual difference in some property of exposure or response; an intrinsic
property that can be better understood with additional information but not reduced or eliminated. In contrast, uncertainty can be reduced or
eliminated with additional information.
1-7
-------
Scenario Uncertainty
The sources of scenario uncertainty include descriptive errors, aggregation errors, errors in
professional judgment, and incomplete analysis. Descriptive errors include errors in information, such as
misclassification of disease or pollutant exposure. Aggregation errors arise as a result of lumping
approximations. Included among these are assumptions of homogeneous populations, and spatial and
temporal approximations such as assumptions of steady-state conditions. Errors in professional judgment
also are a source of uncertainty. A potentially serious source of uncertainty arises from incomplete
analysis. For example, the lack of experimental data from clinical studies on severe asthmatics to
pollutant exposures limits the qualitative and quantitative analyses of health effects.
Parameter Uncertainty
Sources of parameter uncertainty include measurement errors, sampling errors, variability, and use
of generic or surrogate data. Measurement errors can be random or systematic. Random error results from
imprecision in the measurement process. Systematic error is a bias or tendency away from the true value.
Sampling errors concern sample representativeness. The purpose of sampling is to make an inference
about the nature of the whole from a measurement of a subset of the total population. The inability to
characterize the inherent variability in various parameters is a major source of uncertainty. The use of
generic or surrogate data is common when site-specific data are not available, for example generalized
descriptions of environmental settings. The approach to characterizing uncertainty in parameter values
will vary. It can involve an order-of-magnitude bounding of the parameter range when uncertainty is high,
or a description of the range for each of the parameters including the lower- and upper-bound and the best
estimate values and justification for these based on available data or professional judgment. In some
circumstances, characterization can take the form of a probabilistic description of the parameter range.
The appropriate characterization will depend on several factors, including whether a sensitivity analysis
indicates that the results are significantly affected by variations within the range. When the results are
significantly affected by a particular parameter an attempt should be made to reduce the uncertainty by
developing a description of the likely occurrence of particular values within the range. If enough data are
available, standard statistical methods can be used to obtain a meaningful representation.
Model Uncertainty
Model uncertainty can be defined as gaps in scientific theory required to make predictions on the
basis of causal inferences. Modeling errors are due to models being simplified representations of reality;
hence, rationales for model selection should be discussed. Even after the most appropriate model has been
selected for the purpose at hand, one is still faced with the question of how well the model represents the
real situation. This question is compounded by the overlap between modeling uncertainties and other
uncertainties, e.g., natural variability in inputs, representativeness of the modeling scenario, and
aggregation errors. A dilemma is that many existing models (particularly the very complex ones) and the
hypotheses contained within them cannot be fully tested, although certain components of the model may
be tested. Even when a model has been validated under a particular set of conditions, uncertainty will
exist in its application to situations beyond the test system.
1.3.5.2. Approaches to Characterizing Uncertainty
Various approaches to characterizing uncertainty can be described, such as classical statistical
methods, sensitivity analysis, or probabilistic uncertainty analysis, in order of increasing complexity and
data requirements.
1-8
-------
Classical statistical methods are used to analyze uncertainty in measured values (Robinson, 1989).
Given a data set of measured exposure values for a series of individuals, the population distribution may
be estimated directly, provided that the sample design was developed properly to capture a representative
sample. The measured exposure values also may be used to directly compute confidence interval
estimates for percentiles of the distribution. When the distribution is estimated from measured exposures
for a probability sample of population members, confidence interval estimates for percentiles of the
exposure distribution are the primary uncertainty characterization.
Sensitivity analysis is the process of changing one variable while leaving the others constant and
determining the effect on the output. The procedure involves fixing each uncertain quantity, one at a time,
at its credible lower-bound and then its upper-bound (holding all others at their medians), and then
computing the outcomes for each combination of values. These results are useful to identify the variables
that have the greatest effect on exposure and to help focus further information gathering. The results do
not provide any information about the probability of a quantity's value being at any level within the
range; therefore, this approach is most useful at the screening level when deciding about the need and
direction of further analyses.
Probabilistic uncertainty analysis is generally considered the next level of complexity (Eggleston,
1993). The most common example is the Monte Carlo technique where probability density functions are
assigned to each parameter, then values from these distributions are randomly selected and inserted into
the equation. After this process is completed many times, a distribution of predicted values results that
reflects the overall uncertainty in the inputs to the calculation. By averaging over many different
competing models, Bayesian Model Averaging incorporates model uncertainty into conclusions about
parameters and prediction.
1.3.6. Application of Framework for Causal Determination
EPA uses a two-step approach to evaluate the scientific evidence on health effects of criteria
pollutants. The first step determines the weight of evidence in support of causation and characterizes the
strength of any resulting causal classification. The second step includes further evaluation of the
quantitative evidence regarding the concentration-response relationships and the levels, duration and
pattern of exposures at which effects are observed.
To aid judgment, various "aspects"1 of causality have been discussed by many philosophers and
scientists. The most widely cited aspects of causality in epidemiology, and public health in general, were
articulated by Sir Austin Bradford Hill in 1965 and have been widely used (CDC, 2004; U.S. EPA, 2005;
IARC, 2006; IOM, 2007). These nine aspects (Hill, 1965) have been modified (below) for use in causal
determinations specific to health and environmental effects and pollutant exposures.2
Table 1-1. Aspects to aid in judging causality.
1. Consistency of the observed association. An inference of causality is strengthened when a
pattern of elevated risks is observed across several independent studies. The reproducibility
of findings constitutes one of the strongest arguments for causality. If there are discordant
lrThe "aspects" described by Hill (1965) have become, in the subsequent literature, more commonly described as "criteria." The original term
"aspects" is used here to avoid confusion with "criteria" as it is used, with different meaning, in the Clean Air Act.
2 The Hill apects were developed for use with epidemiology data. They have been modified here for use with a broader array of data, i.e.,
epidemiologic, controlled human exposure, and animal toxicological studies, as well as in vitro data, and to be more consistent with EPA's
Guidelines for Carcinogen Risk Assessment.
1-9
-------
results among investigations, possible reasons such as differences in exposure,
confounding factors, and the power of the study are considered.
2. Strength of the observed association. The finding of large, precise risks increases
confidence that the association is not likely due to chance, bias, or other factors. A modest
risk, however, does not preclude a causal association and may reflect a lower level of
exposure, an agent of lower potency, or a common disease with a high background level.
3. Specificity of the observed association. As originally intended, this refers to increased
inference of causality if one cause is associated with a single effect or disease (Hill, 1965).
Based on current understanding this is now considered one of the weaker guidelines for
causality; for example, many agents cause respiratory disease and respiratory disease has
multiple causes. The ability to demonstrate specificity under certain conditions remains,
however, a powerful attribute of experimental studies. Thus, although the presence of
specificity may support causality, its absence does not exclude it.
4. Temporal relationship of the observed association. A causal interpretation is strengthened
when exposure is known to precede development of the disease.
5. Biological gradient (exposure-response relationship). A clear exposure-response
relationship (e.g., increasing effects associated with greater exposure) strongly suggests
cause and effect, especially when such relationships are also observed for duration of
exposure (e.g., increasing effects observed following longer exposure times). There are,
however, many possible reasons that a study may fail to detect an exposure-response
relationship. Thus, although the presence of a biologic gradient may support causality, the
absence of an exposure-response relationship does not exclude a causal relationship.
6. Biological plausibility. An inference of causality tends to be strengthened by consistency
with data from experimental studies or other sources demonstrating plausible biological
mechanisms. A lack of biologic understanding, however, is not a reason to reject causality.
7. Coherence. An inference of causality may be strengthened by other lines of evidence (e.g.,
clinical and animal studies) that support a cause-and-effect interpretation of the association.
The absence of other lines of evidence, however, is not a reason to reject causality.
8. Experimental evidence (from human populations). The strongest evidence for causality
can be provided when a change in exposure brings about a change in adverse health effect
or disease frequency in either clinical or observational studies.
9. Analogy. Structure-activity relationships and information on the agent's structural analogs
can provide insight into whether an association is causal. Similarly, information on mode
of action for a chemical, as one of many structural analogs, can inform decisions regarding
likely causality.
While these aspects provide a framework for assessing the evidence, they do not lend themselves to
being considered in terms of simple formulas or fixed rules of evidence leading to conclusions about
causality (Hill, 1965). For example, one cannot simply count the number of studies reporting statistically
significant results or statistically nonsignificant results for health effects and reach credible conclusions
about the relative weight of the evidence and the likelihood of causality. Rather, these important
considerations are taken into account with the goal of producing an objective appraisal of the evidence,
informed by peer and public comment and advice, which includes weighing alternative views on
controversial issues. Additionally, it is important to note that the principles in Table 1-1 cannot be used as
a strict checklist, but rather to determine the weight of the evidence for inferring causality. In particular,
not meeting one or more of the principles does not automatically exclude a study from consideration (e.g.,
see discussion in U.S. Surgeon General's Report, CDC, 2004).
1-10
-------
1.3.7. First Step—Determination of Causality
In the ISA, EPA assesses results of recent relevant publications, in light of evidence available
during the previous NAAQS review, to draw conclusions on the causal relationships between relevant
pollutant exposures and health outcomes. This ISA uses a five-level hierarchy that classifies the weight of
evidence for causation, not just association1; that is, whether the weight of scientific evidence makes
causation at least as likely as not, in the judgment of the reviewing group. In developing this hierarchy,
EPA has drawn on the work of previous evaluations, most prominently the IOM's Improving the
Presumptive Disability Decision-Making Process for Veterans (IOM, 2007), EPA's Guidelines for
Carcinogen Risk Assessment (U.S. EPA, 2005), and the U.S. Surgeon General's smoking reports (CDC,
2004). Excerpts from these reports are presented in Annex A. In the ISA, EPA uses a series of five
descriptors to characterize the weight of evidence for causality. This weight of evidence evaluation is
based on various lines of evidence from epidemiologic, controlled human exposure and animal studies, or
other mechanistic, toxicological, or biological sources. These separate judgments are integrated into a
qualitative statement about the overall weight of the evidence and causality. The five descriptors for
causal determination are described in Table 1-2.
Table 1-2. Weight of evidence for causal determination.
Relationship Description
Causal relationship Evidence is sufficient to conclude that there is a causal relationship between relevant pollutant exposures and the health
outcome. That is, a positive association has been observed between the pollutant and the outcome in studies in which
chance, bias, and confounding could be ruled out with reasonable confidence. Evidence includes, for example, controlled
human exposure studies; or observational studies that cannot be explain by plausible alternatives or are supported by
other lines of evidence (e.g. animal studies or mechanism of action information). Evidence includes replicated and
consistent high-quality studies by multiple investigators.
Likely to be a causal relationship Evidence is sufficient to conclude that a causal relationship is likely to exist between relevant pollutant exposures and the
health outcome but important uncertainties remain. That is, a positive association has been observed between the
pollutant and the outcome in studies in which chance and bias can be ruled out with reasonable confidence but potential
issues remain. For example: a) observational studies show positive associations but copollutant exposures are difficult to
address and/or other lines of evidence (controlled human exposure, animal, or mechanism of action information) are
limited or inconsistent; or b) animal evidence from multiple studies, sex, or species is positive but limited or no human
data are available. Evidence generally includes replicated and high-quality studies by multiple investigators.
Suggestive of a causal relationship Evidence is suggestive of a causal relationship between relevant pollutant exposures and the health outcome, but is
limited because chance, bias and confounding cannot be ruled out. For example, at least one high-quality study shows a
positive association but the results of other studies are inconsistent.
Inadequate to infer a causal Evidence is inadequate to determine that a causal relationship exists between relevant pollutant exposures and the
relationship health outcome. The available studies are of insufficient quantity, quality, consistency or statistical power to permit a
conclusion regarding the presence or absence of an association between relevant pollutant exposure and the outcome.
Suggestive of no causal relationship Evidence is suggestive of no causal relationship between relevant pollutant exposures and the health outcome Several
adequate studies, covering the full range of levels of exposure that human beings are known to encounter and
considering sensitive subpopulations, are mutually consistent in not showing a positive association between exposure
and the outcome at any level of exposure. The possibility of a very small elevation in risk at the levels of exposure
studied can never be excluded.
1 It should be noted that the CDC and IOM frameworks use a four-category hierarchy for the strength of the evidence. A five-level hierarchy is
used here to be consistent with the EPA Guidelines for Carcinogen Risk Assessment and to provide a more nuanced set of categories.
1-11
-------
1.3.8. Second Step—Evaluation of Population Response
Beyond judgments regarding causality are questions relevant to quantifying risks to populations.
The fundamental issue is to understand the relationships between pollutant exposures and health effects in
the human population. In addressing this issue several important questions must be considered:
¦ What is the concentration-response or dose-response relationship in the human population?
¦ What is the interrelationship between incidence and severity of effect?
¦ What exposure conditions (dose or exposure, duration and pattern) are important?
¦ What subpopulations appear to be differentially affected i.e., more susceptible to effects?
To address these questions the second step of the framework evaluates the entirety of policy-
relevant quantitative evidence regarding the concentration-response relationships including levels of
pollutant and exposure durations at which effects are observed, and subpopulations that differ from the
general population. This integration of evidence results in identification of a study or set of studies that
best approximates the concentration-response relationship for the U.S. population, given the current state
of knowledge and the uncertainties that surround these estimates.
To accomplish this integration, evidence from multiple and diverse types of studies is considered.
Response is evaluated over a range of observations which is determined by the type of study and methods
of exposure or dose and response measurements. Results from different protocols are compared and
contrasted. Extensive human concentration-response data exist for all criteria pollutants, unlike most other
environmental pollutants. Animal data also inform evaluation of concentration-response, particularly
relative to dosimetry, mechanisms of action, and characteristics of susceptible subpopulations. For some
health outcomes, the probability and severity of health effects and associated uncertainties can be
characterized. Chapter 5 presents the integrated findings informative for evaluation of population risks.
An important consideration in characterizing the public health impacts associated with exposure to
a pollutant is whether the concentration-response relationship is linear across the full concentration range
encountered, or if nonlinear relationships exist along any part of this range. Of particular interest is the
shape of the concentration-response curve at and below the level of the current standards. The shape of
the concentration-response curve varies, depending on the type of health outcome, underlying biological
mechanisms and dose. At the human population level, however, various sources of variability and
uncertainty tend to smooth and "linearize" the concentration-response function (such as the low data
density in the lower concentration range, possible influence of measurement error, and individual
differences in susceptibility to air pollution health effects). Additionally, many chemicals and agents may
act by perturbing naturally occurring background processes that lead to disease, which also linearizes
population concentration-response relationships (Clewell and Crump, 2005; Crump et al., 1976; Hoel,
1980). These attributes of population dose-response may explain why the available human data at ambient
concentrations for some environmental pollutants (e.g., 03, lead [Pb], PM, secondhand tobacco smoke,
radiation) do not exhibit evident thresholds for cancer or noncancer health effects, even though likely
mechanisms of action include nonlinear processes for some key events. These attributes of human
population dose-response relationships have been extensively discussed in the broader epidemiologic
literature (e.g., Rothman and Greenland, 1998).
1.3.9. Concepts in Evaluating Adversity of Health Effects
In evaluating the health evidence, a number of factors can be considered in determining the extent
to which health effects are "adverse" for health outcomes such as changes in lung function. What
constitutes an adverse health effect may vary between populations. Some changes in healthy individuals
may not be considered adverse while those of a similar type and magnitude are potentially adverse in
more susceptible individuals.
1-12
-------
The American Thoracic Society (ATS) published an official statement titled "What Constitutes an
Adverse Health Effect of Air Pollution?" (ATS, 2000). This statement updated the guidance for defining
adverse respiratory health effects that had been published 15 years earlier (ATS, 1985), taking into
account new investigative approaches used to identify the effects of air pollution and reflecting concern
for impacts of air pollution on specific susceptible groups. In the 2000 update, there was an increased
focus on quality of life measures as indicators of adversity and a more specific consideration of
population risk. Exposure to air pollution that increases the risk of an adverse effect to the entire
population is viewed as adverse, even though it may not increase the risk of any identifiable individual to
an unacceptable level. For example, a population of asthmatics could have a distribution of lung function
such that no identifiable single individual has a level associated with significant impairment. Exposure to
air pollution could shift the distribution such that no one individual experiences clinically relevant effects;
however, this shift toward decreased lung function would be considered adverse because individuals
within the population would have diminished reserve function and, therefore, would be at increased risk if
affected by another agent.
1.4. Conclusions
The ISA is a concise review, synthesis, and evaluation of the most policy-relevant science, and
communicates critical science judgments relevant to the NAAQS review. It reviews the most policy
relevant evidence from epidemiologic, human clinical, and animal toxicological studies, including
mechanistic evidence from basic biological science. Annexes to the ISA provide additional details of the
literature published since the last review. A framework for making critical judgments concerning causality
is presented in this chapter. It relies on a widely accepted set of principles and standardized language to
express evaluation of the evidence. This approach can bring rigor and clarity to the current and future
assessments. This ISA should assist EPA and others, now and in the future, to represent accurately what is
presently known—and what remains unknown—concerning the effects of SOx on human health.
1-13
-------
Chapter 2. Source to Dose
This chapter contains basic information about concepts and findings in atmospheric sciences,
human exposure assessment, and human dosimetry. It is meant to serve as a prologue for the detailed
discussions of health effects data in Chapters 3 and 4. Section 2.1 gives an overview of the sources of
S02. Atmospheric chemistry processes involved in the oxidation of S02 and those involved in the
production of S02 from reduced sulfur gases in the atmosphere are discussed in Section 2.2. A description
of S02 measurement methods and monitor siting are presented in Sections 2.3 and 2.4. Data for ambient
S02 concentrations are characterized in Section 2.5. Policy relevant background (PRB) concentrations of
S02, i.e., those concentrations defined to result from uncontrollable emissions, are also presented in
Section 2.5. Factors related to personal exposure to S02 are discussed in Section 2.6. Finally, Section 2.7
covers the dosimetry of S02 in the respiratory tract. This organization generally follows that given in the
National Research Council (NRC) paradigm for integrating air pollutant research (NRC, 1998).
2.1. Sources of Sulfur Oxides
Anthropogenic emissions of S02 in the U.S. are mainly due to combustion of fossil fuels by
electrical utilities (-66 %) and industry (-29%). Transportation-related sources contribute only -5%
based on 2002 statistics (U.S. EPA, 2006c). Thus, most S02 emissions originate from point sources.
Annex B has a detailed breakdown of emissions by source category. Almost all of the S in fuel is released
as volatile components (S02 or S03) during combustion. Hence, based on S content in fuel stocks, sulfur
emissions can be calculated to a higher degree of accuracy than other pollutants such as NOx or primary
PM. However, these estimates are national averages and may not accurately reflect the contribution of
specific local sources for determining individual exposure to S02 at a particular location and time.
An additional source of S02 emissions of concern in particular locations not immediately obvious
from national-scale averages and totals are transit and in-port activities in areas with substantial shipping
traffic (Wang et al., 2007). Because of the importance of these S02 emissions, the ports of Long Beach
and Los Angeles, CA, for example, are part of a Sulfur Emissions Control Area in which S contents of
fuels are not to exceed 1.5%. Figure 2-1 shows S02 emissions densities combined for all non- or off-road
transportation-related emitters in which coastal areas with ports, and shipping routes, such as the
Mississippi River, are easily discerned. In Los Angeles County, CA, for example, off-road transportation
including shipping and port traffic contributed 1.4 of the total 4.1 tons of S02 per square mile in 2001; in
King County (including the city of Seattle), WA, the off-road transportation fraction was 42% of the total
S02 emissions density, or 1.2 of the total 2.8 tons per square mile. Emissions data more specific to the
ports are not available in the routine emissions inventories. Increased uncertainty accompanies estimates
of the actual S02 loads from these sources. Modeling studies by Vutukuru and Dabdub (2008) for
southern California ports, for example, have shown that ships contribute >1 ppb to the 24-h avg S02
concentration at the coast, in Long Beach, CA, with < 10% of that (<0.1 ppb) farther inland.
The largest natural sources of S02 are volcanoes and wildfires. Although S02 constitutes a
relatively minor fraction (0.005% by volume) of total volcanic emissions (Holland, 1978), concentrations
in volcanic plumes can be in the range of several to tens of ppm. Volcanic sources of S02 in the U.S. are
limited to the Pacific Northwest, Alaska, and Hawaii. Sulfur is a component of amino acids in vegetation
and is released during combustion. Gaseous S emissions from this source are mainly in the form of S02.
In addition to its role as an emitted primary pollutant, S02 is also produced by the photochemical
oxidation of reduced S compounds such as dimethyl sulfide (CH3-S-CH3, or DMS), hydrogen sulfide
(H2S), carbon disulfide (CS2), carbonyl sulfide (OCS), methyl mercaptan (CH3-S-H), and dimethyl
disulfide (CH3-S-S-CH3). The sources for these compounds are mainly biogenic (see Annex Table B-3).
2-1
-------
Emissions of reduced S species are associated typically with marine organisms living either in pelagic or
coastal zones, and with anaerobic bacteria in marshes and estuaries. Emissions of DMS from marine
plankton represent the largest single atmospheric source of reduced sulfur species (Berresheim et al.,
1995). OCS is lost mainly by photolysis (e-folding lifetime1, [t] ~6 months). Other reduced S species are
lost mainly by reaction with hydroxyl radical (OH) and N03 radicals, and are relatively short-lived;
lifetimes range from a few hours to a few days (see Annex Table B-l). Reaction with N03 radicals at
night most likely represents the major loss process for DMS and CH3-S-H. Although the mechanisms for
the oxidation of DMS are not completely understood, excess sulfate in marine aerosol appears related
mainly to the production of S02 from the oxidation of DMS. Emissions of S from natural sources are
small compared to industrial emissions within the U.S. However, important exceptions occur locally as
the result of volcanic activity, wildfires and in certain coastal zones as described above.
2001 County Emissions Density (tons per square mile) of SO2
0 _ >0-0.016 0.016-0.041 0.041-0.087
0.087-0.18 HO.18-0.5 H0.5+
Source: EPA, 2006
Figure 2-1. 2001 County-level SO2 emissions densities (tons per square mile) from off-road
mobile and other transportation sources.
Because OCS is relatively long-lived, it can survive oxidation in the troposphere and be transported
upward into the stratosphere. Crutzen (1976) proposed that its oxidation to sulfate in the stratosphere
serves as the major source of the stratospheric aerosol layer. However, Myhre et al. (2004) proposed that
S02 transported upward from the troposphere by deep convection is the most likely source, since the flux
of OCS is too small. In addition, in situ measurements of the isotopic composition of S in stratospheric
1 The e-folding lifetime is the time needed to reduce the concentration to i/e of its initial value.
2-2
-------
sulfate do not match those of OCS (Leung et al., 2002). Thus, anthropogenic S02 emissions could be
important precursors to the formation of the stratospheric aerosol layer.
2.2. Atmospheric Chemistry
The only gaseous forms of SOx of interest in tropospheric chemistry are S02 and S03. S03 can be
emitted from the stacks of power plants and factories; however, it reacts extremely rapidly with water
(H20) in the stacks or immediately after release into the atmosphere to form H2S04, which mainly
condenses onto existing particles when particle loadings are high; it can nucleate to form new particles
under lower concentration conditions. Thus, only S02 is present in the tropospheric boundary layer at
concentrations of concern for human exposures. The gas phase oxidation of S02 is initiated by the
reaction
sa2 + OH + M -» HSO 3 + M
Reaction 2-1
where M is an atmospheric constituent such as N2 and 02 that helps stabilize the reaction product.
Reaction 2-1 is followed by
hso3 + o2->so3 + HO2
Reaction 2-2
so3 + h2o -> h2so4
Reaction 2-3
Because the saturation vapor pressure of H2S04 is extremely low, it will be removed rapidly from the gas
phase by transfer to aerosol particles and cloud drops. Depending on atmospheric conditions and
concentrations of ambient particles and gaseous species that can participate in new particle formation, it
can also nucleate to form new particles. Rate coefficients for the reactions of S02 with either the
hydroperoxyl radical (H02) or N03 are too low to be significant (Jet Propulsion Laboratory, 2003). Note:
for the subsequent discussion that sulfates such as H2S04 are referenced using the S(VI) notation and
intermediate sulfites such as S032" are denoted by S(IV).
S(VI) species, including the bisulfate ion (HS04) and sulfate (S042 ) are the dominant S species in
clouds. Intermediate S(IV) products, such as hydrogen sulfite (HS03) and sulfite (S032 ) are also present
in clouds following dissolution of S02 in water but preceding transformation to S(VI) products. The chief
species capable of oxidizing S(IV) to S(VI) in cloud water are 03, peroxides (either hydrogen peroxide
(H202) or organic peroxides), hydroxyl (OH) radicals, and ions of transition metals such as iron (Fe),
manganese (Mn) and copper (Cu) that can catalyze the oxidation of S(IV) to S(VI) by 02. The basic
mechanism of the aqueous phase oxidation of S02 has long been studied and can be found in numerous
texts on atmospheric chemistry, e.g., Seinfeld and Pandis (1998), Finlayson-Pitts and Pitts (1999), Jacob
(1999), and Jacobson (2002). Following Jacobson (2002), the steps involved in the aqueous phase
oxidation of S02 can be summarized as:
2-3
-------
Dissolution of S02
S02(g)oS02(aq)
Reaction 2-4
The formation and dissociation of H2S03
SO2 (aq) + H20(aq) <=> H2 S03 <=> H+ + HS03 <=>2H h + SO,
Reaction 2-5
In the pH range commonly found in rainwater (pH 2 to 6), the most important reaction converting
S(IV) to S(VI) is
HS03 +H2 02 + Hi <^>S042 + H20 + 2HI
Reaction 2-6
because S032 is much less abundant than HS03 .
For pH up to about 5.3, H202 is the dominant oxidant. At pH > 5.3, 03 becomes dominant,
followed by Fe(III), according to Seinfeld and Pandis (1998). However, differences in concentrations of
oxidants result in differences in the pH at which this transition occurs. It should also be noted that the
oxidation of S02 by 03 and 02 tends to be self-limiting: as sulfate is formed, the pH decreases and the
rates of these reactions decrease. Higher pH levels are expected to be found mainly in marine aerosols.
However, in marine aerosols, the chloride-catalyzed oxidation of S(IV) may be more important (Hoppel
and Caffrey, 2005; Zhang and Millero, 1991). Because the ammonium ion (NH4) is so effective in
neutralizing acidity, it affects the rate of oxidation of S(IV) to S(VI) and the rate of dissolution of S02 in
particles and cloud drops.
A comparison of the relative rates of oxidation by gas and aqueous phase reactions by Warneck
(1999) indicates that on average only about 20% of S02 is oxidized by gas phase reactions; the remainder
is oxidized by aqueous phase reactions. In areas away from strong pollution sources, the S02 x is ~7 days
with respect to gas phase oxidation, based on measurements of the rate constant for Reaction 2-1 (Jet
Propulsion Laboratory, 2003) and a nominal concentration for the OH radical of 106/cm3. However, the
mechanism of S02 oxidation at a particular location depends on local environmental conditions. For
example, oxidants such as OH radicals are depleted near stacks. Under this condition, almost no S02 is
oxidized in the gas phase. Further downwind, as the plume is diluted with background air, the gas phase
oxidation of S02 increases in importance. Conditions in the plume can become more oxidizing than in
background air as distance from the stack increases. This can cause the S02 oxidation rate to exceed that
in background air. S02 in the planetary boundary layer is also removed from the atmosphere by dry
deposition to moist surfaces, resulting in an atmospheric x with respect to dry deposition of approximately
1 day to 1 week. Wet deposition of S naturally depends on the variable nature of rainfall, but in general
results in x ~7 days for S02. Oxidation and deposition together lead to an overall lifetime of S02 in the
atmosphere of < 1 to 4 days, depending on altitude and location and meteorological conditions. Smaller
values are expected near the surface.
2-4
-------
2.3. Measurement Methods and Associated Issues
Currently, ambient S02 is measured using instruments based on pulsed ultraviolet (UV)
fluorescence. The UV fluorescence monitoring method for atmospheric S02 was developed as an
improvement to the flame photometric detection (FPD) method, which in turn had replaced the
pararosaniline wet chemical method. This latter method is still the EPA's Federal Reference Method
(FRM) for atmospheric S02, but it is rarely used because it is complex and has slow response even in its
automated forms. Both the UV fluorescence and FPD methods are designated as Federal Equivalent
Methods (FEMs) by EPA, but UV fluorescence has largely supplanted the FPD approach because the UV
method is inherently linear and the FPD method needs consumable hydrogen gas.
In the UV fluorescence method, S02 molecules absorb UV light at one wavelength and emit UV
light at longer wavelengths through excitation of the S02 molecule to a higher energy (singlet) electronic
state. Once excited, the molecule decays nonradiatively to a lower-energy electronic state from which it
then decays to the original electronic state by emitting a photon of light at a longer wavelength (i.e., a
lower-energy photon) than the original, incident photon. The intensity of the emitted light is thus
proportional to the number of S02 molecules in the sample gas.
In commercial analyzers, light from a high-intensity UV lamp passes through a bandwidth filter to
allow only photons with wavelengths around the S02 absorption peak (near 214 nanometers [nm]) to
enter the optical chamber. The light passing through the source bandwidth filter is collimated using a UV
lens and passes through the optical chamber, where it is detected on the opposite side of the chamber by
the reference detector. A photomultiplier tube (PMT) is offset from and placed perpendicular to the light
path to detect the S02 fluorescence. Since the S02 fluorescence at 330 nm is different from its excitation
wavelength, an optical bandwidth filter is placed in front of the PMT to filter out any stray light from the
UV lamp. A lens is located between the filter and the PMT to focus the fluorescence onto the active area
of the detector and optimize the fluorescence signal. The limit of detection (LOD) for a non-trace level
S02 analyzer is required to be 10 ppb (40 CFR 53.23.c). However, most commercial analyzers have
detection limits of about 3 ppb; many monitors might have lower effective detection limits. The EPA,
through its National Core (NCore) initiative (U.S. EPA, 2005b) is in the process of supporting state, local,
tribal, and federal networks in the implementation of newer trace-level S02 instrumentation. These new
trace-level instruments have detection limits of 0.1 ppb or lower. More information related to S02
sampling and analysis is in Annex Section B.6.
2.3.1. Sources of Positive Interference
The most common sources of interference to the UV fluorescence method for S02 are other gases
that fluoresce in a similar fashion when exposed to UV radiation. The most significant of these are
poly cyclic aromatic hydrocarbons (PAHs), of which naphthalene is a prominent example. Xylene is
another common hydrocarbon that can cause fluorescent interference. Consequently, any such aromatic
hydrocarbons in the optical chamber can act as positive interference. To remove this source of
interference, high-sensitivity S02 analyzers, such as those to be used in the NCore network (U.S. EPA,
2005b), have hydrocarbon scrubbers to remove these compounds from the sample stream before the
sample air enters the optical chamber.
Luke (1997) reported positive artifacts of a modified pulsed fluorescence detector generated by the
coexistence of nitric oxide (NO), CS2, and a number of highly fluorescent aromatic hydrocarbons such as
benzene, toluene, o-xylene, m-xylene, /^-xylene, m-cthyltolucne. ethylbenzene, and 1,2,4-
trimethylbenzene. The positive artifacts could be reduced by using a hydrocarbon "kicker" membrane. At
a flow rate of 300 standard cc/min and a pressure drop of 645 torr across the membrane, the interference
from ppm levels of many aromatic hydrocarbons was eliminated. NO fluoresces in a spectral region close
to that of S02. However, in high-sensitivity S02 analyzers, the bandpass filter in front of the PMT is
2-5
-------
designed to prevent NO fluorescence from being detected at the PMT. Care must be exercised when using
multicomponent calibration gases containing both NO and S02, so that the NO rejection ratio of the S02
analyzer is sufficient to prevent NO interference.
The most common source of positive bias (as contrasted with positive spectral interference) in
high-sensitivity S02 monitoring is stray light in the optical chamber. Since S02 can be electronically
excited by a broad range of UV wavelengths, any stray light with an appropriate wavelength that enters
the optical chamber can excite S02 in the sample and increase the fluorescence signal. Furthermore, stray
light at the wavelength of the S02 fluorescence that enters the optical chamber may impinge on the PMT
and increase the fluorescence signal. Several design features minimize stray light, including the use of
light filters, dark surfaces, and opaque tubing.
Nicks and Benner (2001) reported a sensitive S02 chemiluminescence detector based on a
differential measurement: response from ambient S02 is determined by the difference between air
containing S02 and air scrubbed of S02 when both air samples contain other detectable sulfur species.
Assuming monotonic efficiency of the sulfur scrubber, all positive artifacts should also be reduced with
this technique.
2.3.2. Sources of Negative Interference
Nonradiative deactivation (quenching) of excited S02 molecules can occur from collisions with
common molecules in air, including nitrogen, oxygen, and water. During collisional quenching, the
excited S02 molecule transfers energy, kinetically allowing the S02 molecule to return to the original
lower energy state without emitting a photon. Collisional quenching results in a decrease in the S02
fluorescence and, hence, an underestimation of S02 concentration in the air sample. Of particular concern
is the variable water vapor content of air. Luke (1997) reported that the response of the detector could be
reduced by an amount of ~7 to 15% at water vapor mixing ratios of 1 to 1.5 mole percent (relative
humidity [RH] = 35 to 50% at 20 to 25 °C and 1 atmosphere [atm] for a modified pulsed fluorescence
detector [Thermo Environmental Instruments, Model 43s]). Condensation of water vapor in sampling
lines must be avoided, as water on the inlet surfaces can absorb S02 from the sample air. The simplest
approach to avoid condensation is to heat sampling lines to a temperature above the expected dew point
and to within a few degrees of the controlled optical bench temperature. At very high S02 concentrations,
reactions between electronically excited S02 and ground state S02 might occur, forming S03 and SO
(Calvert et al., 1978). However, the possibility that this artifact might be affecting measurements at very
high S02 levels has not been examined.
2.3.3. Other Techniques for Measuring SO2
More sensitive techniques for measuring S02 are available, but most of these systems are too
complex and expensive for routine monitoring applications. However, techniques such as those described
by Luke (1997) can be used to improve the sensitivity of ambient S02 monitors by eliminating sources of
common interference. See descriptions in Annex B.
2-6
-------
2.4. Monitoring Site Characteristics
2.4.1. Design Criteria for the NAAQS SO2 Monitoring Networks1
Trace level S02 monitoring is currently required at the approximately 75 proposed NCore sites, as
noted in CFR 40 Part 58 Appendices C and D. Continued operation of existing State and Local Air
Monitoring Sites (SLAMS) for S02 using Federal Reference Methods (FRM) or Federal Equivalent
Methods (FEM) is required until discontinuation is approved by the EPA Regional Administrator. Where
SLAMS S02 monitoring is required, at least one of the sites must be a maximum concentration site for
that specific area. In 2007, there were -500 S02 monitors reporting values to the EPA Air Quality System
database (AQS). The AQS contains measurements of air pollutant concentrations in the 50 states, plus the
District of Columbia, Puerto Rico, and the Virgin Islands, for the 6 criteria air pollutants and hazardous
air pollutants.
The appropriate spatial scales for S02 SLAMS monitoring are the microscale, middle, and possibly
neighborhood scales.
¦ Micro (-5 -100 meters(m) and middle scale (-100 - 500 m)—Some data uses associated
with microscale and middle scale measurements for S02 include assessing the effects of
control strategies to reduce concentrations (especially for the 3-h and 24-h averaging
times), and monitoring air pollution episodes.
¦ Neighborhood scale (-500 m-4 km)—This scale applies where there is a need to collect
air quality data as part of an ongoing S02 stationary source impact investigation. Typical
locations might include suburban areas adjacent to S02 stationary sources, for example, or
for determining background concentrations as part of studies of population responses to
S02 exposure.
2.4.1.1. Horizontal and Vertical Placement
The probe, or at least 80 percent of the monitoring path, must be located between 2 and 15 m above
ground level for all S02 monitoring sites. The probe, or at least 90 percent of the monitoring path, must be
positioned at least 1 m vertically or horizontally from any supporting structure, walls, parapets,
penthouses, etc., and away from dusty or dirty areas. If the probe, or a significant portion of the
monitoring path, is located near the side of a building, it should be located on the windward side relative
to the prevailing wind direction during the season of highest concentration potential for the pollutant
being measured.
2.4.1.2. Spacing from Minor Sources
Local minor sources of a primary pollutant such as S02 can affect concentrations of that particular
pollutant at a monitoring site. If the objective for that site is to investigate these local primary pollutant
emissions, then the site should be located where the spatial and temporal variability in these emissions
can be captured. This type of monitoring site would likely be the microscale type. If a monitoring site is
to be used to determine air quality over a much larger area, such as a neighborhood or city, a monitoring
1 This section is adapted from Code of Federal Regulations 40 CFR Parts 53 and 58 and Appendix E to Part 58, as revised: Vol. 71, No. 200 / 17
October 2006
2-7
-------
agency should avoid placing a monitor probe, path, or inlet near local, minor sources. The plume from the
local minor sources should not be allowed to inappropriately influence the air quality data collected.
To minimize these potential interferences, the probe, or at least 90 percent of the monitoring path,
must be placed away from the furnace, incineration flues, or other minor sources of S02. The separation
distance should take into account the heights of the flues, type of waste or fuel burned, and the S content
of the fuel.
2.4.1.3. Spacing from Obstructions
Buildings and other obstacles may possibly scavenge S02 and can act to restrict airflow for any
pollutant. To avoid this interference, the probe, inlet, or at least 90 percent of the monitoring path must
have unrestricted airflow and be located away from obstacles. The distance from the obstacle to the probe,
inlet, or monitoring path must be at least twice the height of the obstruction's protrusion. An exception
can be made for measurements taken in street canyons or at source-oriented sites where buildings and
other structures are unavoidable. Generally, a probe or monitoring path located near or along a vertical
wall is undesirable, because air moving along the wall may be subject to possible removal mechanisms. A
probe, inlet, or monitoring path must have unrestricted airflow in an arc of at least 180 degrees. This arc
must include the predominant wind direction for the season of greatest pollutant concentration potential.
Special consideration must be devoted to the use of open path analyzers, due to their inherent
potential sensitivity to certain types of interferences, or optical obstructions. A monitoring path must be
clear of all trees, brush, buildings, plumes, dust, or other optical obstructions, including potential
obstructions that may move due to wind, human activity, growth of vegetation, etc. Temporary optical
obstructions, such as rain, particles, fog, or snow, should be considered when locating an open path
analyzer. Any temporary obstructions that are of sufficient density to obscure the light beam will affect
the ability of the open path analyzer to measure pollutant concentrations continuously. Transient, but
significant obscuration of especially longer measurement paths could occur because certain
meteorological conditions (e.g., heavy fog, rain, snow) and/or aerosol levels are of sufficient density to
prevent the analyzer's light transmission. If certain compensating measures are not otherwise
implemented at the onset of monitoring (e.g., shorter path lengths, higher light source intensity), data
recovery during periods of greatest primary pollutant potential could be compromised. For instance, if
heavy fog or high particulate levels are coincident with periods of projected NAAQS-threatening
pollutant potential, the resulting data may not be representative for reflecting maximum pollutant
concentrations, despite the fact that the site may otherwise exhibit an acceptable, even exceedingly high
overall valid data capture rate.
2.4.1.4. Spacing from Trees
Trees can provide surfaces for S02 adsorption or reactions and surfaces for particle deposition.
Trees can also act as obstructions in cases where they are located between the air pollutant sources or
source areas and the monitoring site, and where the trees are of sufficient height and leaf canopy density
to interfere with normal airflow around the probe, inlet, or monitoring path. To reduce possible
interference, the probe, inlet, or at least 90 percent of the monitoring path must be at least 10 meters or
further from the drip line of trees.
For microscale sites, no trees or shrubs should be located between the probe and the source under
investigation, such as a roadway or a stationary source.
2-8
-------
2.4.2. Locations of SO2 Monitors in Selected Metropolitan Areas
Figures 2-2 through 2-7 display S02 monitor density with respect to population density for the six
metropolitan regions analyzed. The locations of S02 monitors in selected areas where air pollution-health
outcome studies have been conducted are characterized in this section. The studies are described in later
chapters of the ISA. The study areas included six regions comprising eight metropolitan statistical areas
(MSAs), as defined by the U.S. Census Bureau (http://www.census.gov/): Atlanta, Cincinnati, Cleveland,
Los Angeles/Riverside, New York City/Philadelphia, and St. Louis. S02 monitor location data (i.e.,
latitude/longitude) for 2004 were obtained from EPA's AirData website (http://www.epa.gov/air/data/).
Monitors were mapped for a particular region if they were contained by the MSA or if they were within
15 km of its boundary. The total population and populations of three sensitive subgroups (under age 5,
age 5 to 17, and those over age 65) were calculated for those areas using the population data contained
within the census block maps.
Atlanta Metropolitan Statistical Area
Figure 2-2. Location of SO2 monitors with respect to population density in the Atlanta, GA MSA.
2005 Population Density
I ISO2 Monitors (5 km buffer)
Population per Sq Mile
M 0 - 230
H 231 - 460
461 - 2300
2301 - 4600
H 4601 - 11500
M 11501 - 45900
3 Kilometers
2-9
-------
Cincinnati Metropolitan Statistical Area
0 5 10
20
30
40
~ Kilometers
50
o
2005 Population Density
I IS02 Monitors (5 km buffer)
Population per Sq Mile
HO-146
¦¦ 147 - 292
293- 1460
1461 - 2920
H 2921 - 7300
M 7301 -29175
~ Kilometers
0 15 30 60 90 120 150
Figure 2-3. Location of SO2 monitors with respect to population density in the Cincinnati, OH
MSA.
2-10
-------
Cleveland Metropolitan Statistical Area
4
5 10
I Kilometers
20 30 40
2005 Population Density
I IS02 Monitors (5 km buffer)
Population per Sq Mile
M 0 - 202
203 - 404
405 - 2020
2021 - 4040
4041 - 10100
10101 - 40380
0 15 30
60
90
i Kilometers
120
Figure 2-4. Location of SO2 monitors with respect to population density in the Cleveland, OH
MSA.
2-11
-------
Los Angeles/Riverside Metropolitan Statistical Areas
Kilometers
10 20
2005 Population Density
I I SOj Monitors (5 km buffer)
Population per Sq. Mile
M 0 - 702
m703 - 1404
1405 -7022
7023 - 14043
M14044-35108
¦ 35109 - 140433
0 50 100 200 300
1 Kilometers
400
Figure 2-5. Location of SO2 monitors with respect to population density in the Los
Angeles/Riverside, CA MSA.
2-12
-------
New York City/Philadelphia Metropolitan
Statistical Areas
^ ¦¦—Kiln meters
0 10 20 40 60 80
2005 Population Density
I ISO2 Monitors (5 km buffer)
Population pe Sq Mile
¦¦ 0-2154
¦ 2155 -4309
4310 - 21545
21546 - 43090
¦ 43091 - 107725
¦ 107726 - 430900
¦ ¦ ¦ Kilometers
0 25 50 100 150 200
Figure 2-6. Location of SO2 monitors with respect to population density in the New York City,
NY/Philadelphia, PA MSA.
2-13
-------
St. Louis Metropolitan Statistical Area
Kilo meters
10 20 30 40
i Kilometers
0 25 50 100 150 200
Figure 2-7. Location of SO2 monitors with respect to population density in the St. Louis, MO MSA.
Tables 2-1 through Table 2-4 break down the population density around S02 monitors for the total
population and for sub-populations of children aged 0-4 yr and 5-17 yr, and adults aged 65 yr and over,
for each region. Variation in percentage within a certain radius of the monitor was fairly low for each city
across the three age groups. However, between-city disparities in population density were larger. The
New York City/Philadelphia region had the highest population density in all three age groups and overall
2005 Population Density
I IS02 Monitors (5 km buffer)
Population per Sq Mile
0 - 141
M 142 - 282
283 - 1409
1410-2818
M 2819 -7044
M 7045 -28175
2-14
-------
with the highest or near highest proportion, -72-73%, of each population within 15 km of the monitors.
The mid-western cities of Cincinnati, Cleveland, and St. Louis also had a large proportion, -66-75%, of
their population within 15 km of a monitor. This suggests that these sub-populations, as well as the total
populace, are well represented by the monitoring networks in these regions. Los Angeles and Atlanta had
lower proportions of their populations within 15 km of an S02 monitor, with Atlanta having only -22%
for the total population within 15 km, with similar proportions of children and a slightly higher (-27%) of
the elder population represented. These latter figures likely reflect the lower density of local S02 sources
in these regions.
Table 2-1. Proximity to SO2 monitors for the total population by city. Percentages are given with
respect to the total population in each city.
Proximity Region
to S02
Monitor
(km)
Atlanta
Cincinnati
Cleveland
Los Angeles
New York/
Philadelphia
St. Louis
n
%
n
%
n
%
n
%
n
%
n
%
0-1
13,389
0.3
14,745
0.7
29,733
1.4
72,465
0.4
432,621
1.7
42,211
1.5
0-5
214,238
4.3
225,274
10.8
382,849
17.8
1,663,990
9.9
6,985,553
27.8
799,941
28.5
0-10
610,742
12.3
733,054
35.2
1,006,787
46.8
6,030,847
35.8
13,727,129
54.7
1,561,159
55.6
0-15
1,080,472
21.7
1,367,658
65.6
1,553,286
72.3
9,694,760
57.6
18,304,364
72.9
1,897,492
67.6
Total MSA
4,980,447
100.0
2,085,092
100.0
2,149,472
100.0
16,839,035
100.0
25,094,739
100.0
2,807,659
100.0
Note that population in proximity to the monitor is cumulative (i.e. those within 5 km includes those within 1 km) and the total MSA population includes those living
in the region beyond the 15 km radius of the monitor.
Table 2-2. Proximity to SO2 monitors for children aged 0-4 yr by city. Percentages are given with
respect to the total population in the age group in each city.
Proximity to
SO2 Monitor
(km)
Region
Atlanta Cincinnati Cleveland Los Angeles New York/ St. Louis
Philadelphia
n
%
n
%
n
%
n
%
n
%
n
%
0-1
425
0.1
1,057
0.8
2,409
1.7
4,548
0.4
28,421
1.7
2,485
1.4
0-5
14,216
4.5
15,606
11.0
30,903
21.9
126,764
10.4
453,267
27.7
50,937
28.5
0-10
37,440
11.8
48,942
34.6
71,852
50.8
451,417
37.1
890,406
54.4
100,856
56.4
0-15
68,231
21.5
94,851
67.0
105,809
74.8
721,458
59.2
1,198,850
73.3
121,685
68.0
Total MSA
317,949
100.0
141,537
100.0
141,425
100.0
1,218,227
100.0
1,635,831
100.0
178.868
100.0
Note that population in proximity to the monitor is cumulative (i.e. those within 5 km includes those within 1 km) and the total MSA population includes those living
in the region beyond the 15 km radius of the monitor.
2-15
-------
Table 2-3. Proximity to SO2 monitors for children aged 5-17 yr by city. Percentages are given with
respect to the total population in the age group in each city.
Region
Proximity to
SO2 Monitor (km) Atlanta Cincinnati Cleveland Los Angeles New York/ St. Louis
Philadelphia
n
%
n
%
n
%
n
%
n
%
n
%
0-1
833
0.1
2,629
0.7
5,639
1.4
11,611
0.4
70,589
1.6
8,512
1.6
0=5
33,552
4.1
41,083
10.5
77,974
19.3
321,665
9.9
1,185,810
26.9
151,317
28.7
0-10
92,715
11.4
133,500
34.2
194,495
48.2
1,146,231
35.4
2,375,339
53.9
295,508
56.0
0-15
168,430
20.5
258,819
66.2
291,335
72.2
1,853,488
57.2
3,212,239
72.9
356,514
67.6
Total USA
813,107
100.0
390,704
100.0
403,465
100.0
3,238,473
100.0
4,406,226
100.0
527,773
100.0
Note that population in proximity to the monitor is cumulative (i.e. those within 5 km includes those within 1 km) and the total MSA population includes those living
in the region beyond the 15 km radius of the monitor.
Table 2-4. Proximity to SO2 monitors for adults aged 65 yr and over by city. Percentages are
given with respect to the total population in the age group in each city.
Region
Proximity to
SO2 Monitor (km) Atlanta Cincinnati Cleveland Los Angeles New York/ St. Louis
Philadelphia
n
%
n
%
n
%
n
%
n
%
n
%
0-1
279
0.1
1,826
0.8
3,425
1.1
7,197
0.5
52,315
1.7
6,929
2.0
0-5
17,237
5.3
29,363
12.5
46,067
14.8
138,838
7.0
805,226
25.9
123,954
35.4
0-10
53,503
16.4
90,437
38.5
144,465
46.4
511,431
33.0
1,668,273
53.6
219,972
62.9
0-15
88,867
27.2
164,229
69.9
233210
74.9
852,535
55.0
2,243,853
72.1
257,891
73.7
Total MSA
326,858
100.0
235,116
100.0
311,579
100.0
1,549,138
100.0
3,113,768
100.0
349,881
100.0
Note that population in proximity to the monitor is cumulative (i.e. those within 5 km includes those within 1 km) and the total MSA population includes those living
in the region beyond the 15 km radius of the monitor.
Figures 2-8 through 2-13 illustrate the 2005 geospatial locations of monitors for S02, N02, CO,
particulate matter < 10 |im (PM10), particulate matter < 2.5 |im (PM2 5), and 03. These locations, sited in
several cities in six states, were selected as relevant for S02 health effects studies presented in Chapter 3;
see the discussion of intracity S02 correlations that follows. For each state, Map A shows locations of
each monitor for all six pollutants; and Map B shows only the S02 monitor locations. Totals for each
monitor type are included. These figures demonstrate the important point that not all S02 monitors in any
Consolidated Metropolitan Statistical Area (CMSA) are co-located with monitors for other pollutants.
2-16
-------
Sacrarrn
San Francisco:
San Josi
San Francis)
Sacram
Francisco.
San Josi
Highway
Monitor Location
Interstate
+
CO (86)
Federal
A
N02 (105)
State
V
03 (176)
O
PM10 (177)
~
PM25 (177)
~
S02 (35)
San Diego
Source: US EPA Office of Air and Radiation AQS Database
Figure 2-8. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), California, 2005.
Shaded counties have at least one monitor.
2-17
-------
Cleveland
Toledo
Clevelan
incinnati
Highway
Monitor Location
Interstate
+
CO (15)
Federal
A
N02 (4)
State
V
03 (49)
O
PM-io (49)
~
PM2 5 (49)
~
S02 (31)
Source: US EPA Office of Air and Radiation AQS Database
Figure 2-9. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Ohio, 2005.
Shaded counties have at least one monitor.
2-18
-------
Flagstaff
Tucson
Tucson
Highway
Monitor Location
Interstate
+
CO (20)
Federal
A
N02 (13)
State
V
03 (45)
O
PM10 (67)
~
PM25(16)
~
S02 (7)
Source: US EPA Office of Air and Radiation AQS Database
Figure 2-10. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Arizona, 2005.
Shaded counties have at least one monitor.
2-19
-------
Allentown
/ & /
risburg
* Philadelp
i
'Allen^ayvn
Highway
Monitor Location
— Interstate
+
CO (25)
Federal
A
N02 (29)
State
V
03 (47)
O
CD
0
0-
~
PM25 (49)
~
S02 (42)
Source: US EPA Office of Air and Radiation AQS Database
Figure 2-11. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Pennsylvania,
2005. Shaded counties have at least one monitor.
2-20
-------
Buffalo
Albany
Rocheste
Highway
Monitor Location
Interstate
+
CO (11)
Federal
A
N02 (9)
State
V
03 (34)
O
PM10(10)
~
PM2 5 (28)
~
S02 (25)
New York
Source: US EPA Office of Air and Radiation AQS Database
Figure 2-12. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), New York, 2005.
Shaded counties have at least one monitor.
Buffalo
2-21
-------
Boston
Boston
Highway
Monitor Location
- Interstate
+
CO (5)
Federal
A
N02 (13)
State
V
03 (16)
O
PM10 (11)
~
CN
CN,
5*
CL
~
S02 (10)
Source: US EPA Office of Air and Radiation AQS Database
Figure 2-13. Criteria pollutant monitor locations (A) and SO2 monitor locations (B), Massachusetts,
2005. Shaded counties have at least one monitor.
Table 2-5 lists the totals for all criteria air pollutant monitors (except Pb) in California, as well as
the subset of these monitors in San Diego County. At each of the four sites where S02 was measured in
San Diego county, N02, CO, PMi0, PM2 s, and 03 were also measured, with the exception of PM2 5 at one
site (AQS ID 060732007) in Otay Mesa, CA. Table 2-6 lists the totals for all criteria air pollutant
monitors (except Pb) in Ohio, as well as the subset of monitors in Cuyahoga County. In Cuyahoga
County, PM;) and PM2 s were measured at all four sites where S02 was also measured in 2005, but 03 and
CO were not measured at any of those four sites; N02 was only measured at one site (AQS ID 39050060)
near Cleveland's city center and -0.5 km from the intersection of Interstate Highways 77 and 90.
2-22
-------
Table 2-5.
Monitor counts for California and San Diego County, 2005.
S02 NO2 03 CO
PM10
PM2.5
California (all)
San Diego County
35 105 176 86
4 9 10 6
177
7
97
7
Table 2-6.
Monitor counts for Ohio and Cuyahoga County, 2005.
SO2 NO2 03 CO
PM10
PM2.5
Ohio (all)
Cuyahoga County
31 4 49 15
4 2 3 4
49
6
49
7
2.4.3. Ambient SO2 Concentrations in Relation to SO2 Sources
Figures 2-14 through 2-19 show the locations of S02 monitors in relation to S02 point sources from
electricity generation for the Atlanta, Cincinnati, Cleveland, Los Angeles/Riverside, New York
City/Philadelphia, and St. Louis MSAs. Information on point sources (i.e., fossil fueled electricity
generating units and industrial point sources) for 2002 was obtained from the National Emissions
Inventory (NEI) (EPA, 2006). Point sources were mapped for a particular region if they were contained
by the MSA or if they were within 30 km of its boundary, and S02 monitors were mapped with buffer
zones of 1, 5, 10, and 15 km constructed around the monitor locations. The figures show that the S02
monitors are placed in relative proximity to the majority of point sources of S02 with the exception of
Atlanta, where most point sources are found outside the city center. In all cases, at least 1 monitor is in the
vicinity of the most populated urban center and at least one is sited for background S02 concentrations.
S02 data collected from the State and Local Monitoring System (SLAMS) and National Air
Monitoring System (NAMS) networks, like those illustrated in Figure 2-2 through Figure 2-7 show that
the decline in S02 emissions from electric generating utilities has substantially improved air quality. No
monitored exceedance of the S02 annual ambient air quality standard in the lower 48 States of the U.S.
has been recorded between 2000 and 2005, according to the EPA Acid Rain Program (ARP) 2005
Progress Report (EPA, 2006b). EPA's trends data (http://www.epa.gov/airtrends/') reveal that the national
composite avg S02 annual mean ambient concentration decreased by 48% from 1990 to 2005; the largest
single-year reduction was 1994-95, the ARP's first operating year (EPA, 2006b). Figure 2-20 depicts data
for S02 emissions in the contiguous United States (CONUS) during those years, with state-level totals.
These emissions data trends are consistent with the trends in the observed ambient concentrations
from the Clean Air Status and Trends Network (CASTNet). Following implementation of the Phase I
controls on ARP sources between 1995 and 2000, significant reductions in S02 and ambient S042
concentrations were observed at CASTNet sites throughout the eastern U.S.. The mean annual
concentrations of S02 and S042 from CASTNet's long-term monitoring sites can be compared using two
3-year periods, 1989-1991 and 2003-2005, shown in Figure 2-21 for S02 and Figure 2-22 for S042 .
2-23
-------
Atlanta Metropolitan Statistical Area
~ Kilometers
0 5 10 20 30 40 50
Distance to S02 Monitors
I I 1 km
5 km
I I 10 km
I I 15 km
I External Combustion Boilers
Internal Combustion Engines
— Major Highways
I I Atlanta
] Kilometers
0 20 40 80 120 160 200
Figure 2-14. Location of SO2 monitors within a 15 km buffer zone with respect to combustion
sources and highways in the Atlanta, GA MSA.
2-24
-------
Cincinnati Metropolitan Statistical Area
M m i Kilometers
0 5 10 20 30 40 50
Distance to S02 Monitors
I I 1 km
5 km
I I 10 km
l l 15 km
1 External Combustion Boilers
Internal Combustion Engines
Major Highways
I I Cincinnati
: Kilometers
0 15 30 60 90 120 150
Figure 2-15. Location of SO2 monitors within a 15 km buffer zone with respect to combustion
sources and highways in the Cincinnati, OH MSA.
2-25
-------
Cleveland Metropolitan Statistical Area
0 2.5 5
I Kilo meters
10 15 20
Distance to S02 Monitors
I I 1 km
5 km
I I 10 km
I I 15 km
I External Combustion Boilers
Internal Combustion Engines
Major Highways
I I Cleveland
0 15 30
60
90
iKilometers
120
Figure 2-16. Location of SO2 monitors within a 15 km buffer zone with respect to combustion
sources and highways in the Cleveland, OM MSA.
2-26
-------
Los Angeles/Riverside Metropolitan Statistical Areas
Distance to S02 Monitors
I I 1 km
5 km
I I 10 km
I I 15 km
I External Combustion Boilers
Internal Combustion Engines
Major Highways
I I Los Angeles/Riverside
m m ^¦ Kilometers
0 50 100 200 300 400
Figure 2-17. Location of SO2 monitors within a 15 km buffer zone with respect to combustion
sources and highways in the Los Angeles/Riverside, CA MSA.
2-27
-------
New York City/Philadelphia Metropolitan
Statistical Areas
i Kilometers
0 10 20 40 60 80
i Kilometers
0 25 50 100 150 200
Figure 2-18. Location of SO2 monitors within a 15 km buffer zone with respect to combustion
sources and highways in the New York City, NY/Philadelphia, PA MSA.
Distance to S02 Monitors
1 I 15 km
I External Combustion Boilers
Internal Combustion Engines
Major Highways
I I New York City/Philadelphia
2-28
-------
St. Louis Metropolitan Statistical Area
Kilometers
Distance to S02 Monitors
Kilometers
0 25 50 100 150 200
Figure 2-19 Location of SO2 monitors within a 15 km buffer zone with respect to combustion
sources and highways in the St. Louis, MO MSA.
2-29
-------
S02 Emissions in 1990
~ S02 Emissions in 1995
I I S02 Emissions in 2000
CH S02 Emissions in 2005
Scale: Largest bar equals
2.2 million tons of S02
emissions in Ohio, 1990
Source: Environmental Protection Agency Clean Air Markets Division httpJ/wmi.epa.qov/airmarkets/index.htmI
Figure 2-20. State-level SO2 emissions, 1990-2005.
SO2
(jtg/m3)
SO2
0*g/m3)
Source: U.S. EPA CASTNet
Figure 2-21. Annual mean ambient SO2 concentration, 1989 through 1991 (A), and 2003 through
2005(B).
2-30
-------
Under 5,000 tons
Source: U.S. EPA CASTNet
Figure 2-22. Annual mean ambient SO42" concentration, 1989 through 1991 (A), and 2003 through
2005(B).
• 50.000 to 100.000 tons
• 100,000 to 207.000 tons
Source: Environmental Protection Agency, Clean Air Markets Division http://www.epa.qov/airmarkets/index.htmI
Figure 2-23. Annual mean ambient SO2 emissions for Acid Rain Program cooperating facilities,
2006. Dots represent monitoring sites. Lack of shading for Southern Florida indicates
lack of monitoring coverage.
2-31
-------
From 1989 through 1991— the years prior to implementation of the ARP Phase I— the highest
ambient mean concentrations of S02 and S042 were observed in western Pennsylvania and along the
Ohio River Valley: > 20 (ig/m3 (~8 ppb) S02 and > 15 (ig/m3 S042 . These reductions are shown in Figure
2-21 and Figure 2-22, respectively. In the years since the ARP controls were enacted, both the magnitude
of S042 concentrations and their areal extent have been significantly reduced, with the largest decreases
again along the Ohio River Valley.
Figure 2-23 depicts the magnitude and spatial distribution of S02 emissions in 2006 from sources
in the ARP for the CONUS. This depiction clearly shows the continuing predominance of S02 sources in
the U.S. east of the Mississippi River with even stronger magnitude in the central Ohio River Valley, as
evident in the smoothed concentration plots in Figure 2-21. As shown in Table 2-7, regional distributions
of S02 and S042 concentrations averaged for 2003-2005 reflect this geospatial emissions source
difference as well.
2.5. Environmental Concentrations of SOx
2.5.1. Spatial and Temporal Variability of Ambient SO2 Concentrations
S02 concentrations have been falling throughout all regions of the CONUS, as demonstrated by the
CASTNet data reviewed above. In and around most individual CMS As, the trends are also toward lower
S02 levels. Table 2-7 shows that many annual and 1-h mean concentrations for the years 2003 through
2005 were consistently at or below the operating LOD of ~3 ppb for the standard sensitivity UV
fluorescence S02 monitors deployed in the regulatory networks. Table 2-8 shows that the 1-h avg, 24-h
avg, and aggregate mean value for the years 2003-2005 for inside and outside (all CMS As) were
operating just above the limit of detection (LOD) of ~ 3 ppb.
Table 2-7. Mean ambient concentrations of SO2 and SO42- in different regions of the U.S.
averaged over 2003-2005.
Mid-Atlantic
Midwest
Northeast
Southeast
Region
SO2 (ppb)
3.3
2.3
1.2
1.3
Concentration
S042" (jjg/m3)
4.5
3.8
2.5
4.1
2-32
-------
Table 2-8. Concentration distributions of SO2 inside and outside CMSAs from 2003-2005. Values
shown are in ppb.
Averaging Time
Monitor Locations
Number of
Samples
Mean
Percentiles
1 5 10 25 30 50 70 75 90 95 99
Max
1-H MAX CONCENTRATION
Inside CMSAs
332405
13 1 1
1 3 4
7
13
16
30
45
92
714
Outside CMSAs
53417
13 1 1
1 1 2
5
10
13
31
51
116
636
1-HAVG CONCENTRATION
Inside CMSAs
7408145
4 1 1
1 1 1
2
4
5
10
15
34
714
Outside CMSAs
1197179
4 1 1
1 1 1
2
3
3
7
13
36
636
24-HAVG CONCENTRATION
Inside CMSAs
327918
4 1 1
1 1 2
3
5
6
10
13
23
148
Outside CMSAs
52871
4 1 1
1 1 1
2
3
4
8
12
25
123
ANNUAL AVG CONCENTRATION
Inside CMSAs
898
4 1 1
1 1 2
4
5
6
8
10
12
15
Outside CMSAs
143
4 1 1
1 1 2
3
4
5
8
9
13
14
AGGREGATE 3-YR AVG CONCENTRATION, 2003-2005
Inside CMSAs
283
4 1 1
1 2 3
3
5
5
8
10
12
14
Outside CMSAs
42
4 1 1
1 2 2
3
4
5
8
9
13
13
Figure 2-24 shows the composite diel variation in hourly S02 concentrations for the mean and
selected percentile values from all monitors reporting S02 data into AQS. The AQS contains
measurements of air pollutant concentrations in the 50 states plus the District of Columbia, Puerto Rico,
and the Virgin Islands for the six criteria air pollutants and hazardous air pollutants.
Figure 2-24 shows hourly mean concentrations and the 50th, 75th, 95th and 99th percentile values
from all monitors reporting data into AQS. All of these metrics show clear daytime maxima and nighttime
minima. The magnitude of the day-night difference increases with increasing concentrations. Note:
concentrations given in AQS are rounded to the nearest ppb, and thereby flatten the variability at lower
concentrations, as shown in the lower panel of Figure 2-24. The day-to-night variability likely reflects the
entrainment of S02 emitted by elevated point sources into the mixed layer that is growing by convection
during the morning and early afternoon. The effect of higher concentration values on the means is shown
by the 1 to 3 ppb difference between mean and 50th percentile values during the day.
The strong west-to-east increasing gradient in S02 emissions described above is well-replicated in
the observed concentrations in the AQS data set. For example, Table 2-9 shows the mean annual
concentrations from 2003-2005 for the 12 CMSAs with four or more S02 regulatory monitors. Values
ranged from a reported low of ~1 ppb in Riverside, CA and San Francisco, CAto a high of-12 ppb in
Pittsburgh, PA and 14 pbb in Steubenville, OH, in the highest S02 source region.
2-33
-------
50
40 -
3#
0"° •<>¦.,
O.
— 95th Percentile
O 99th P ercentile
O.
•O
o.
¦<>
<>
o
O-O-OO-OO'
o.
o.
'O-O-O-hO'"0
a.
a.
20
10 -
O
= ;
5
4 -
3 -
V V -V • V - V- -V - v - V ¦ V V - V- -V V
¦I - VV-V-VV-..V
V-.-W-V-V
~i 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r
0 I 2 3 4 5 0 7 8 9 10 11 12 13 14 15 16 17 18 19 2 0 21 2 2 2 3
Sample Hour
Source: EPA AQS
Figure 2-24. Diel variation in SO2 concentration across all monitoring sites reporting into AQS for
2005. The upper panel shows 95th, and 99th percentile values and the lower panel
shows mean, 50th, and 75th percentile values.
2-34
-------
Table 2-9. Range of mean annual SO2 concentrations and Pearson correlation coefficients in
urban areas having at least four regulatory monitors, 2003-2005.
CMSA (# Monitors)
Mean SO2 Concentration (ppb)
Pearson Correlation Coefficient
Philadelphia, PA (10)
3.6-5.9
0.37-0.84
Washington, DC (5)
3.2-6.5
0.30-0.68
Jacksonville, FL (5)
1.7-3.4
-0.03-0.51
Tampa, FL (8)
2.0-4.6
-0.02-0.18
Pittsburgh, PA (10)
6.8-12
0.07-0.77
Steubenville, OH (13)
8.6-14
0.11 -0.88
Chicago, IL (9)
2.4-6.7
0.04-0.45
Salt Lake City, UT (5)
2.2-4.1
0.01-0.25
Phoenix, AZ (4)
1.6-2.8
-0.01 -0.48
San Francisco, CA (7)
1.4-2.8
-0.03-0.60
Riverside, CA (4)
1.3-3.2
-0.06-0.15
Los Angeles, CA (5)
1.4-4.9
-0.16-0.31
The Pearson correlation coefficients (r) for concentration data from multiple monitors taken as
pairs in these CMSAs were generally very low for all cities, especially at the lower end of the observed
concentration range. Some negative correlation coefficients were observed on the West Coast and Florida
(see Table 2-9). This reflects strong heterogeneity in S02 ambient concentrations within a given CMSA
and therefore indicates possibly different exposures of spatially distinct subgroups in these CMSAs. At
higher concentrations, the r values were also higher. In some CMSAs, this heterogeneity may result from
meteorological effects, whereby a generally well-mixed subsiding air mass containing one or more S02
plumes with relatively high concentration would be more uniformly spread than faster-moving plumes
with lower concentrations. However, instrument error may also play a role because the highest r values,
i.e., those >0.7, correspond to the highest S02 concentrations, i.e., >6 and > 10 ppb. Since the lowest S02
concentrations are at or below the operating LOD and demonstrate the lowest correlation across monitors
that share at least some air mass characteristics most of the year, the unbiased instrument error in this
range may be confounding interpretation of any possible correlation. This could be because the same
actual ambient concentration would be reported differently by different monitors (with different error
profiles) in the CMSA for this lowest concentration range.
To improve characterization of the extent and spatiotemporal variance of S02 concentrations within
each of the CMSAs having four or more S02 monitors, the means, minima, and maxima were computed
from daily mean data across all available monitors for each month for the years 2003 through 2005.
Because many of these CMSAs with S02 monitors also reported S042 , it is possible to compute the
degree of correlation between S02, the emitted species, and S042 , the most prominent oxidized product
from S02. Although S042 values are averaged over all available data at each site, they are generally
available at their monitoring sites on a schedule of only 1 in 3 days or 1 in 6 days. Furthermore, S02 and
S042 monitors are not all co-located throughout the CMSAs. For each of the five example CMSAs in
Figure 2-25 through Figure 2-29, monthly aggregated values are depicted for: (a) the monthly mean,
minimum, and maximum S02 concentrations; (b) the monthly mean, minimum and maximum S042
concentrations; and (c) a scatterplot of S02 versus S042 concentrations.
2-35
-------
In Steubenville, OH (Figure 2-25), the area of highest S02 concentrations of all 12 CMSAs with
more than four monitors, all monthly mean S02 concentrations were substantially <30 ppb, although max
daily means in some months were often > 60 ppb, or even > 90 ppb. S042 data at Steubenville were
insufficient to make meaningful comparisons, although the 12 months of available S042 data suggest no
correlation with S02.
(a)120 t
100 ¦¦
(b) 40 T
36 ;*
32 ;¦
^ 28 ¦¦
E 24 ¦¦
-------
(a) 50
45
40 ¦¦
35 ""
"g. 30 --
% 25
C/3
20
15
10
5
III II
N> >> V 4>J . >J >J >} >J JO . >J >J M
^ ^ ^ ^ cJp ^ ^ ^ ^ ^ ^ ^ ^
(b) 70 --
60 ¦¦
50 -•
E
q> 40 --
O
30 •¦
20 +
10
0
illll
!ll
.. ¦ . jliiilj::: I..
fi
S? 5?" jS1 .5?" S? S? S?1 Si1* .52^ 5^ S^1 5§i sS3 sS3 ^
¦#¦* ^ ^ ^ cf& ^ ^ ^ ^ ^ ^ ^ ^ ^ i> ^
(c) 25
20 ;¦
O
E 15 ; ¦
10 ¦¦
5 :-
0
O)
-
O
cn
o
o
O O
o
o
oo<> *^oo o° * o
o . o
qO
o ooo
S02(ppb)
10
12
Figure 2-26. Philadelphia, 2003-2005. (a) Monthly mean, minimum, and maximum SO2
concentrations, (b) Monthly mean, minimum, and maximum SO42- concentrations, (c)
Monthly mean SO42- concentrations as a function of SO2 concentrations.
S02 and S042" trends differ substantially in Philadelphia, PA (Figure 2-26) from those in
Steubenville, OH. S02 in Philadelphia, PA (Figure 2-26). S02is present at roughly one-half the monthly
mean concentrations in Steubenville, OH, and demonstrates a strong seasonality with S02 concentrations
peaking in winter. By contrast, S042 concentrations in Philadelphia (Figure 2-26) peak in the three
summer months, with pronounced wintertime minima. This seasonal anticorrelation still contains
considerable monthly scatter.
Los Angeles, CA (Figure 2-27), presents a special case since its size, power requirements, and role
as a port city place a larger number of S02 emitters in or near the city than would otherwise be expected
on the West Coast. Concentrations of S02 demonstrate weak seasonality in these 3 years, with
2-37
-------
summertime means of ~3 to 4 ppb, and maxima generally higher than wintertime ones, although the
highest means and maxima occur during the winter of 2004-2005. S042 at Los Angeles shows stronger
seasonality, most likely because the longer summer days of sunny weather allow for additional oxidation
of S02 to S042 than could happen in winter. Weak seasonal effects in S02 likely explain the complete
lack of correlation between S02 and S042 here.
(a) 16 ¦¦
14 ¦¦
12
s- 10 ¦¦
a
s 8 +
N
O „
to 6 +
mm
y? jS" y?> J? 5? y? yf* & & of" d* y}* ^ „
^ ^ ^ ^ ^ ^ ^ ^ ^ ^ i> c$ ^
& J>
(b) 45 3-
40 i-
35
*C 30
20 ¦ ¦
O
in 15
10 f
5
0
I
f I
ii
{
h
S* s-y
sr jr
£ ^ \-S"
v>
$ ^ #
^ of ^
(c) 20 -j-
18
16
14
£ 12 -
I 10 ¦¦
•% 8 -¦
O
w 6 ¦ -
4 --
2
0 --
o «
o
O o O $
o <*>
o
O O
<>
0%
3
SO, (ppb)
Figure 2-27. Los Angeles, 2003-2005. (a) Monthly mean, minimum, and maximum SO2
concentrations, (b) Monthly mean, minimum, and maximum SO42- concentrations, (c)
Monthly mean SO42- concentrations as a function of SO2 concentrations.
2-38
-------
(a) 16 ¦-
14 --
12 --
3" 10 "
o.
S 8 +
O
to
6 +
4
-IIIIU
ll ll 1 III I
ill
.5?" .5?" .J? ^ .5^" ,5^" .S^" vS?1
^ s£> $ <£
^ ^ i* cJ& ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ c$
(b) 18 --
16 ¦¦
14 -•
10 ¦¦
O
to
8 -•'
:-I-I
ll
I
>o )o >o no >o \j \j \y ;o jo )o )o )o jo >o >o vo >o
^ # cf ^ sf ^ ^ i> cf ^ /¦ ^ ^ # cf ^
(c) 10 ;-
9 ¦¦
8 ¦•
- 7 ;¦
E 6"
oi
3 5 ¦•
fN
4 -•
3 ••
2 ¦¦
1 -•
0 •-
o
w
8 °°
°<> O
o O $ ~
O
-------
Phoenix, AZ was the CMSA with the lowest monthly mean S02 and S042 concentrations
examined here (Figure 2-29). In Phoenix, nearly all monthly mean S02 values were at or below the
regulatory monitors" operating LOD of ~3 ppb. S042 concentrations were equivalently low, roughly one-
half the concentrations seen in Riverside, CA. The monthly mean data show strong summertime peaks for
even these very low-level S042 observations, which at ~1 to 3 (.ig/nr1. were generally one-half of those in
Philadelphia. This suggests some seasonality in S02, though anticorrelated with S042 : however, the trend
is very weak, as the correlation scatterplot shows.
(a) 10 T
8 ¦¦
(b) 6 T
0>
0.0 0.5 1.0 1.5 2.0 2.5 3.0
SO, (ppb)
3.5 4.0 4.5 5.0
Figure 2-29. Phoenix, 2003-2005. (a) Monthly mean, minimum, and maximum SO2 concentrations.
(b) Monthly mean, minimum, and maximum SO42- concentrations, (c) Monthly mean
SO42- concentrations as a function of SO2 concentrations.
2-40
-------
2.5.2. Five-Minute Sample Data in the Monitoring Network
Although the number of monitors across the CONUS varies somewhat from year to year, in 2006
there were -500 S02 monitors in the NAAQS monitoring network (http://www.epa.gov/air/data/). The
state and local agencies responsible for these monitors are required to report 1-h avg concentrations to the
EPA AQS. However, a small number of sites, only 98 total from 1997 to 2007, and not the same sites in
all years—voluntarily reported 5-min block avg data to AQS. Of these, only 16 reported all twelve 5-min
avg in each hour at least part of the time between 1997 and 2007. The remainder reported only the
maximum 5-min avg in each hour. See Table 2-10 and Table 2-11 for a breakdown of these monitoring
locations and sampling periods, and Figure 2-30 for the distribution of these sites across the CONUS.
Table 2-10. Locations, counts, sampling periods and statistics for monitors reporting hourly
maximum 5-min SO2 values, 1997-2007.
State Number of Counties
Number of Monitors
Number of Years Years Operating
Mean
GM1
GSD2
Ol
O
953
993
Max
CM
q:
<
3
11
1997-2007
4
3
2
3
10
37
659
CO 1
1
10
1997-2006
8
4
3
4
29
57
216
DE 1
1
2
1997- 1998
17
6
4
5
97
184
381
DC 1
1
6
2000 - 2007
9
6
2
6
23
42
482
FL 1
1
4
2002 - 2005
8
2
3
1
40
106
473
IA 6
9
5
2001 - 2005
4
2
3
1
12
45
307
LA 1
1
4
1997-2000
12
5
3
5
41
131
857
MO 1
14
11
1997-2007
9
3
3
2
32
146
4367
MT 1
7
10
1997-2006
8
3
4
2
35
77
843
NC 2
2
8
1997-2004
9
3
4
4
34
108
805
ND 11
19
11
1997-2007
3
1
2
1
11
54
499
PA 8
23
11
1997-2007
14
8
3
8
48
105
1099
sc 1
10
3
2000 - 2002
3
2
2
2
10
28
277
UT 1
1
2
1997- 1998
3
2
2
1
9
21
209
WV 2
5
7
2001 - 2007
12
7
3
7
38
80
856
1 Geometric mean
2 Geometric standard deviation
3 Percentile
2-41
-------
Table 2-11. Locations, counts, sampling periods and statistics for monitors reporting all twelve
5-min SO2 values, 1997-2007.
State
Number of Counties
Number of Monitors
Number of Years
Years Operating
Mean
GM1
GSD2
Ol
O
953
993
Max
DC
1
1
1
2007
5
4
2
4
12
19
400
FL
1
1
4
2002 - 2005
4
2
3
1
18
62
473
MO
1
2
5
2003 - 2007
3
2
2
1
11
43
259
MT
1
4
1
2002
3
2
2
1
12
34
843
NC
1
1
4
1999-2002
5
2
3
1
22
77
805
PA
2
5
6
2002 - 2007
11
5
4
5
39
83
921
WV
2
2
5
2001 - 2005
9
5
3
5
29
58
508
1 Geometric mean
2 Geometric standard deviation
3 Percentile
nra
Counties
5-Minute S02
Ambient Monitors
¦ Max 5-Minute Values
CI All 5-Minute Values
Figure 2-30. SO2 monitors reporting maximum or continuous 5-min avg values for any period,
1997-2007.
EPA guidance on quality assurance practices for S02 monitoring by state/local monitoring agencies
is aimed at ensuring adequate quality of 1-h S02 concentration data. Measurement of 5-min avg
concentrations may involve quality assurance challenges that are not addressed by current EPA guidance.
This possibility has not been specifically investigated to date, so the information presented here should be
considered to be of uncharacterized uncertainty at this time. Furthermore, the voluntary nature of this
2-42
-------
reporting results in irregular coverage in both space and time. For example, some sites reported data for
some years while others did not. Because the 5-min data were reported voluntarily by the cooperating
states from a subset of monitors in the national data network, no information is available to judge their
degree of representativeness relative to the national network.
When maximum 5-min concentrations were reported, the absolute highest concentration over the
ten-year period exceeded 4 ppm, but the 99th percentile of the maximum 5-min concentrations were all
below 200 ppb. Medians from reported data ranged from 1 ppb to 8 ppb, and the avg for each maximum
5-min level ranged from 3 ppb to 17 ppb. Delaware, Pennsylvania, Louisiana, and West Virginia had
mean values for maximum 5-min data exceeding 10 ppb. Among aggregated within-state data for the 16
monitors from which all 5-min avg data were reported, the median values ranged from 1 ppb to 5 ppb,
and the means ranged from 3 ppb to 11 ppb. The highest reported concentration was 921 ppb, but the 99th
percentile values for aggregated within-state data were all below 90 ppb. It should be emphasized that
monitoring was not continual during the ten-year reporting period, and monitoring was not performed
simultaneously among the sites. For these reasons, caution must be taken when comparing the
distributions among the various sites.
Despite these limitations, distributions of the available 5-min data and comparisons with their
respective 1-h avg can provide some insight into the temporal behavior of short-duration S02
concentrations at the monitoring stations where these data are available. Table 2-12 summarizes
correlations between the maximum 5-min avg and the corresponding 1-h avg computed from the 5-min
data for the monitors reporting all twelve 5-min avg. The correlations are high with only one monitor
observing a correlation coefficient under 0.9.
Table 2-12. Pearson correlation coefficient between maximum 5-min and 1-h avg SO2
concentrations at the 16 sites reporting all twelve 5-min SO2 values.
State
Site ID
Correlation Coefficient
DC
110010041
0.87
FL
120890005
0.92
MO
290770026
0.92
MO
290770037
0.93
MT
301110066
0.95
MT
301110079
0.93
MT
301110082
0.93
MT
301110083
0.92
NC
371290006
0.91
PA
420030021
0.94
PA
420030064
0.95
PA
420030116
0.96
PA
420033003
0.95
PA
420070005
0.93
WV
540990002
0.95
WV
541071002
0.93
2-43
-------
IA- 190330018
ft §
2002
2005
2004
2005
00
0.5
1.0
N 32769
rww 2
max 166
99% 18
95% 4
50%: 1
—
15
—I
20
ft I
MO - 290770026
~r
nm
~r
2000
T
2002
T~
20(X
~r
2006
r
200&
CN _
© -
I
i 0
—r—
1.5
N 124442
meetfi 7
max: 852
99% 80
95% 41
50% 1
—1—
20
-1
3.0
ft 1
MO - 29077003-7
we
2000
2002
2004
2006
20oe
I—
00
-I—
05
1 0
T—
1 5
N 124623
mean 7
max. 48D
99%: 109
95% 31
50%: 2
-1—
20
—1
25
MO - 290990034
• • t! „* ¦ * .. j •
r,£» .-*!¦' .
A/kC t !V<>> 1. il
—T—
05
LLlliiuki
M: 33809
mean 20
max 1000
99% 258
95% 93
50% 4
—I—
20
—T—
25
n
30
PA - 420070005
i i r r 1 r
1993 2000 2002 2004 2006 2006
N t18062
mean 19
max 1099
99% 16D
95% 70
1 1 1 it
50% 7
00 0 5 1 0 1.5 20 2.5 3 0
§
I I
§ § -
WV - 541071£»2
~T~
2005
N 68482
mean: 14
max. 508
99% 106
95% 48
50%: 7
I 1 T 1 1 1
00 05 1.0 1.5 2.0 25
Figure 2-31. Time series and frequency distributions of voluntarily reported maximum 5-min SO2
concentrations from 6 monitors located in Iowa, Missouri, Pennsylvania and West
Virginia. Frequency distributions are shown in terms of the log of the concentration
on the x-axis.
Figure 2-31 shows the time series and distribution of maximum 5-min S02 data per hour for six
monitors located in Iowa (site ID = 190330018), Missouri (290770026, 290770037 and 290990004),
Pennsylvania (420070005) and West Virginia (541071002). These sites were selected to represent the
available 5-min data. The highest observed 5-min values for these six monitors range from 166 ppb at the
Iowa monitor (Mason City, IA) to 1099 ppb at the Pennsylvania monitor (Beaver Co, PA). In general,
2-44
-------
Figure 2-31 demonstrates that the higher range of mean values compared with the medians reflects a few
high concentration events that skew the means upward with the mean concentration between 2 and 7
times greater than the median of these data.
A - 190330018
3
IN
jQ
•Q.
5
O
W
o
o
PA - 420070005
2003-10-04
1997-10-13
IA- 190330018
&
Cl
O
cft
§
Oi
O
S
o
§
o
-
to
o
o
o
PA - 420070005
2003-09-30
2003-10-03
2003-10-06
T 1 1 1 T
1997-10-09 1997-10-12 1997-10-15
Figure 2-32. Time series of hourly maximum 5-min SO2 data showing a 24 h (upper panels) and 1
week (lower panels) time window centered on the peak value for the two sites with the
lowest (IA) and highest (PA) maximum values in the preceding figure.
Excursions of high 5-min concentrations should not be expected to be confined to one 5-min
interval in any given hour as time scales for the meteorological conditions responsible for the excursions
are much longer. During transport of emissions from their source to a receptor at the surface, turbulent
2-45
-------
mixing and dilution of the plume occur, further extending the area at the surface that is affected. Figure
2-32 shows the time series of 5 min data centered on time of occurrence of the value (lower panel) for the
two sites with the highest maximum value (PA) and lowest maximum value (IA) in the preceding figure.
As can be seen for these two sites at least, high levels of S02 can be sustained at the surface for a few
hours and the time for these excursions to affect the surface can differ from site to site and during
different periods at the same site. Note also that these events do not occur in isolation as can be seen at the
PA site, at which several excursions above the 200 ppb level occurred within a week. Further analysis
would be required to determine characteristic time scales for the durations of these excursions across the
U.S., using emissions and meteorological data.
2.5.3. Policy Relevant Background Contributions to SO2
Concentrations
Background concentrations used for purposes of informing decisions about the NAAQS are
referred to as PRB concentrations; those concentrations that would occur in the U.S. in the absence of
anthropogenic emissions in continental North America (defined here as the U.S., Canada, and Mexico).
PRB concentrations include contributions from natural sources everywhere in the world, and from
anthropogenic sources outside these three countries. Background levels so defined facilitate separation of
cases where pollution levels can be controlled by U.S. regulations (or through international agreements
with neighboring countries), from cases where pollution is generally uncontrollable by the U.S.. EPA
assesses risks to human health and environmental effects from S02 levels in excess of PRB
concentrations.
Contributions to PRB concentrations include natural emissions of S02 and photochemical reactions
involving reduced sulfur compounds of natural origin, as well as their long-range transport from outside
of North America from any source. As an example, transport of S02 from Eurasia across the Pacific
Ocean or the Arctic Ocean would carry PRB S02 into the U.S. Annex B contains a schematic diagram
showing the major photochemical processes involved in the sulfur cycle, including natural sources of
reduced sulfur species from anaerobic microbial activity in wetlands and volcanic activity. Volcanoes and
wildfires are the major natural source of S02. Biogenic emissions from agricultural activities are not
considered in the formation of PRB concentrations. Discussions of the sources and estimates of emissions
are given in Annex Section B.6.
The MOZART-2 global model of tropospheric chemistry (Horowitz et al., 2003) is used to estimate
the PRB contribution to S02 concentrations. The model setup for the present-day simulation, i.e.,
including all sources in the U.S., Canada, and Mexico, was published in a series of papers from a recent
model inter-comparison (Dentener et al., 2006a; van Noije et al., 2006). MOZART-2 is driven by the
National Oceanic and Atmospheric Administration's National Center for Environmental Prediction
(NOAA/NCEP) meteorological fields and the International Institute for Applied Systems Analysis
(IIASA) 2000 emissions at a resolution of 1.9° x 1.9° with 28 o (sigma) levels in the vertical and includes
gas- and aerosol-phase chemistry. Results shown in Figure 2-33 are for the meteorological year 2001. An
additional PRB simulation was conducted in which continental North American anthropogenic emissions
were set to zero.
2-46
-------
Total
50°N
45°N
4C°N
35°N
30°N
25°N
120°W
100°W
ao°w
< 0.01 1.21 2.41 3.60 4.80 6.00 ppb
Background
120°W ioo°w eo°w
< 0.001 0.0 0 6 0.011 0.015 0.02 0 0.025 ppb
Percent Background Contribution
50°N
45°N
40°N
35°N
30°N
25°N
120°W 100°W SO°W
10 15 20 25
Source: NOAA Geophysical Fluid Dynamics Laboratory.
Figure 2-33. Annual mean model-predicted concentrations of SO2 (ppb).
2-47
-------
The role of PRB in contributing to S02 concentrations in surface air is examined first. Figure 2-33
shows the annual mean predicted S02 concentrations in surface air in the simulation including all sources,
or the "base case" (top panel); the PRB simulation (middle panel); and the percentage contribution of the
background to the total base case S02 (bottom panel). Maximum concentrations in the base case
simulation, > 5 ppb, occur along the Ohio River Valley (upper panel). Background S02 concentrations are
orders of magnitude smaller, below 10 parts per trillion (ppt) over much of the U.S. (middle panel).
Maximum PRB concentrations of S02 are 30 ppt. In the Northwest where there are geothermal sources of
S02, the contribution of PRB to total S02 is 70 to 80%, however absolute S02 concentrations are still of
the order of ~2 ppb or less. With the exception of the West Coast where volcanic S02 emissions cause
high PRB concentrations, PRB contributes < 1% to present-day S02 concentrations in surface air (bottom
panel).
Ambient SO., - Jaggar Museum - 15 minute averages
2400 -
March
March
March
2000 -
1 600 -
CL
Q.
O 1 200 -
800 -
400 -
00:00
04:00
08:00
1 2:00
1 6:00
20:00
24:00
Hour
Source: National Park Servic
Figure 2-34. 15-min avg ambient SO2 concentrations measured at 1 Hawaii Volcanoes National
Park monitoring site (Jaggar Museum), March 12,13, and 15,2007.
When estimating background concentrations, it is instructive to consider measurements of S02 at
relatively remote monitoring sites, i.e., sites located in sparsely populated areas not subject to obvious
local sources of pollution. Berresheim et al. (1993) used a type of atmospheric pressure ionization mass
spectrometer (APIMS) at Cheeka Peak, WA (48.30EN 124.62EW, 480 m asl) in April 1991 during a field
study for DMS oxidation products. S02 concentrations ranged between 20 and 40 ppt. Thornton et al.
(2002) have also used an APIMS with an isotopically labeled internal standard to determine background
2-48
-------
S02 levels. S02 concentrations of 25 to 40 ppt were observed in northwestern Nebraska in October, 1999
at 150 m above ground using the National Center for Atmospheric Research (NCAR)'s C-130 research
aircraft. These data are comparable to remote central South Pacific convective boundary layer S02 data
(Thornton, 1999).
As noted earlier, volcanic sources of S02 in the U.S. are found in the Pacific Northwest, Alaska,
and Hawaii. The greatest potential domestic effects from volcanic S02 occurs on the island of Hawaii.
Nearly continuous venting of S02 from Mauna Loa and Kilauea produces S02 in high concentrations (see
Figure 2-34 and Figure 2-35) at two National Park sites near the Kilauea caldera and the nearby east rift
zone. The latter emits several times as much S02 as the Kilauea caldera. The two measurement sites
within the National Park are < 3 km from the summit emission source and ~10 km from the east rift
source and are affected by the two sources during southerly and easterly winds. A number of communities
and population centers are within the same distance from the east rift gas source that affects these two
monitoring sites. When the normal trade wind flows are disrupted, emissions from the sources can be
brought directly to these various communities. Since these communities are located at a similar distance
from the large east rift emission source as the National Park monitoring stations, it is probable that these
communities experience S02 concentrations as high as those measured within Hawaii Volcanoes National
Park.
Since 1980, the Mount St. Helens volcano (46.20°N, 122.18°W, summit 2549 m asl) in the
Washington Cascade range has been a variable source of S02. Its major effects came in the explosive
eruptions of 1980, which primarily affected the northwestern U.S. The Augustine volcano near the mouth
of the Cook Inlet in southwestern Alaska (59.363°N, 153.43°W, summit 1252 m asl) has emitted variable
quantities of S02 since its last major eruptions in 1986. Volcanoes in the Kamchatka peninsula in far
eastern Siberia do not particularly affect the surface concentrations in northwestern North America.
Overall, the background contribution to S02 over the U.S. is relatively small, with a max PRB of 0.030
ppb S02, except for areas with volcanic activity.
5000
4000 -
^ 3000 -
Cl
? 2000 -|
iN
m 1000
Sept 29 2007 -15-rninute averages
Kilauea Visitor Center
— ¦ — Ja ggar Museum
U
— J ^
5000
4000
3000
2000
1000
1 1 1 1 1 1 1 1 1 1 1
0 200 400 600 BOO 1000 1 200 1400 1600 1800 2000 2200 2400
Time
Source: National Park Service
Figure 2-35. 15-min avg ambient SO2 concentrations measured at 2 Hawaii Volcanos National Park
monitoring sites on September 29,2007.
2-49
-------
2.6. Issues Associated with Evaluating SO2 Exposure
2.6.1. General Considerations for Personal Exposure
Human exposure to an airborne pollutant consists of contact between the human and the pollutant
at a specific concentration for a specified period of time. People spend various amounts of time in
different microenvironments characterized by different pollutant concentrations. The integrated exposure
of a person to a given pollutant is the sum of the exposures over all time intervals for all
microenvironments. Figure 2-36 represents a composite average of activity patterns across all age groups
in the U.S., based on data collected in the National Human Activity Pattern Survey (NHAPS) (Klepeis et
al., 2001). The demographic distribution of the respondents was designed to be similar to that of overall
U.S. Census data. Different cohorts, e.g., the elderly, young and middle-aged working adults, and children
exhibit different activity patterns.
NHAPS - Nation, Percentage Time Spent
Total n = 9,196
TOTAL TIME SPENT
INDOORS (86.9%)
IN A RESIDENCE (68.7%)
- OUTDOORS (7.6%)
IN A VEHICLE (5.5%)
OTHER INDOOR LOCATION (
BAR-RESTAURANT (1.8%>
OFFICE-FACTORY (5.4%)
Source: Klepeis et al. (2001)
Figure 2-36. Percentage of time spent in various environments in the U.S.1
1 For example, the cohort of working adults between the ages of 18 and 65 yr represents —50% of the population. Of this total, about 60% work
outside the home, spending —24% (40 h/168 h) of their time in factory/office environments. Thus, this cohort is likely to spend considerably
more time in offices and factories than shown in the figure (5.4%), which reflects the entire population, and is also likely to spend less time in a
residence than small children or the elderly.
2-50
-------
In a given microenvironment, the ambient component of a person's microenvironmental exposure
to a pollutant is determined by the following physical factors:
¦ ambient concentration Ca
¦ air exchange rate a,
¦ pollutant specific penetration coefficient P,
¦ pollutant specific decay rate k,
¦ fraction of time an individual spends in the microenvironment yt
These factors are in turn affected by the following exposure factors:
¦ environmental conditions, such as weather and season
¦ dwelling conditions such as: proximity to sources; the amount of natural ventilation (e.g.,
open windows and doors, and the "draftiness" of the dwelling); and the ventilation system
¦ personal activities (e.g., the time spent cooking, commuting, or exercising [See
Section 2.7.1])
¦ indoor sources and sinks of a pollutant
Microenvironmental exposures can also be influenced by the individual-specific factors such as
age, gender, health or socioeconomic status (SES).
A person's exposure to a pollutant, such as S02, can be represented by:
n
Jljt ~ w I;
i=i
Equation 2-1
where ET is an individual's total personal exposure for a specific time period, n is the total number of
microenvironments encountered, C, is the average concentration, and /, is the time spent in the 7th
microenvironment. A person's exposure can be characterized as: an instantaneous exposure, a peak
exposure such as might occur during a short-term event such as cooking, an average exposure, or an
integrated exposure over all environments encountered. These distinctions are important because health
effects caused by long-term low-level exposures may differ from those caused by short-term peak
exposures.
An individual's total exposure (ET) can also be represented by:
Et — Ea + Ena — {y0 + X V/ [/J/dj/(u^ + &/)] }Ena— \y0 + X.V, Ejnf} Ca + Ena
i i ' '
Equation 2-2
subject to the constraint
y0 + X >', = 1
i
Equation 2-3
where Ea is the ambient component of personal exposure, Ena is the nonambient component of personal
exposure, yD is the fraction of time spent outdoors, and v, is the fraction of time spent in microenvironment
F-
i. ah and kt are the infiltration factor, penetration coefficient, air exchange rate, and decay rate,
respectively for microenvironment In the case where an exposure occurs mainly in one
microenvironment, Equation 2-2 may be approximated by Equation 2-4 where y is the fraction of time
spent outdoors, and a is the ratio of personal exposure from a pollutant of ambient origin to the pollutant's
ambient concentration (or the ambient exposure factor). Other symbols have the same definitions as in
Equations 2-2 and 2-3.
2-51
-------
et Ea + Ena {y + (/ y) \Pa/(ci + A')]} Ca + Ena ccCa + Ena
Equation 2-4
If concentrations in a single microenvironment are considered, then Equation 2-4 can be reduced to
= Ca+ Cna = [Fa /(a + k)]Ca + S/[V(a + *)]
Equation 2-5
where Cme is the concentration in a microenvironment, Ca and Cna are the contributions to Cme from
ambient and nonambient sources, S is the microenvironmental source strength, and V is the volume of the
microenvironment. (Bracketed symbols are same as Equation 2-2.) In this equation, it is assumed that
microenvironments do not exchange air with each other but only with the ambient environment.
Microenvironments in which people are exposed to air pollutants such as S02 typically include
residential indoor environments, other indoor locations, near-traffic outdoor environments, other outdoor
locations, and in vehicles, as shown in Figure 2-36. Indoor combustion sources such as gas stoves and
space heaters need to be considered when evaluating exposures to S02. However, in the U.S., the only
important indoor sources of S02 are kerosene heaters, which are not widely used. Exposure
characterization is improved when microenvironment-level exposures are considered to estimate the
ambient component of personal exposure and to describe the relationship between ambient air pollution
exposures and health outcomes.
Time-activity diaries, completed by study participants, are used to compile activity patterns for
input to exposure models and assessments. The EPA's National Exposure Research Laboratory (NERL)
has consolidated the majority of the most significant human activity databases into one comprehensive
database called the Consolidated Human Activity Database (CHAD). Eleven different human activity
pattern studies were evaluated to obtain over 22,000 person-days of 24-h human activities in CHAD
(McCurdy et al., 2000). These data can be useful in assembling population cohorts to be used in exposure
modeling and analysis.
In general, the relationship between personal exposures and ambient concentrations is modified by
pollutant behavior in microenvironments. During infiltration, ambient pollutants can be lost through
chemical and physical processes. Therefore, the ambient component of a pollutant's microenvironmental
concentration is less than its ambient concentration, and can be represented as the product of the
ambient concentration and the infiltration factor (F,„/or a [if people spend 100% of their time indoors]).
In addition, exposure to nonambient, microenvironmental sources modifies the relationship between
personal exposures and ambient concentrations.
In practice, it is extremely difficult to characterize community exposures by measurements of each
individual's personal exposures. Instead, the distribution of personal exposures in a community, or the
population exposure, is simulated by extrapolating measurements of personal exposure using various
techniques or by stochastic, deterministic or hybrid exposure modeling approaches such as APEX,
SHEDS, and MENTOR (see Annex Section C.2 for a description of modeling methods). Variations in
community-level personal exposures are determined by cross-community variations in ambient pollutant
concentrations and the physical and exposure factors mentioned above. These factors also determine the
strength of the association between population exposure to S02 of ambient origin and ambient S02
concentrations.
Of major concern is the ability of S02 measured by ambient monitors to serve as a reliable
indicator of personal exposure to S02 of ambient origin. The key question is what errors are associated
with using S02 measured by ambient monitors as a surrogate for personal exposure to ambient S02 and/or
its oxidation products in epidemiologic studies. There are three aspects to this issue: (1) ambient and
personal sampling issues; (2) the spatial variability of ambient S02 concentrations as related to exposures;
and (3) the associations between ambient concentrations and personal exposures as influenced by
2-52
-------
exposure factors, e.g., indoor sources and time spent indoors and outdoors. Items (1) and (3) are discussed
individually in the following sections; item (2) was discussed previously in Section 2.4.2.
2.6.2. Methods Used for Monitoring Personal Exposure
Three basic methods of analysis have been used as personal exposure monitors (PEMs) to measure
personal exposure to S02. The Harvard-EPA annular denuder system (HEADS) was initially developed to
measure particles and acid gases simultaneously (Brauer et al., 1999; Koutrakis et al., 1988b). The aerosol
is initially sampled at 10 L/min through an impactor that is attached to an annular denuder to remove
particles. Subsequently, the aerosol is sampled through an annular denuder coated with sodium carbonate
(Na2C03). This denuder is used to trap S02, nitric acid (HN03), and nitrous acid (HN02). Following
sampling, the denuder is extracted with ultrapure water and analyzed by ion chromatography. Collection
efficiencies of S02 in the denuder are typically around 0.993, which compares well with predicted values.
For a study conducted in Baltimore, MD, Chang et al. (2000) developed and employed a personal
roll-around system (RAS), which is an active sampling system designed to measure short-term personal
exposure concentrations of several atmospherically relevant species, including S02. For the measurement
of S02, the RAS employed an N02/S02 sorbent denuder worn on a vest by the study participant. The
hollow glass denuder, encased in an aluminum jacket, is coated with triethanolamine (TEA) for the
collection of S02 and N02, and aerosol is sampled through the denuder at 100 cc/min. Following
sampling, the denuder can be extracted and analyzed for S02 concentrations by ion chromatography. The
detection limit for 1-h sampling of S02 was reported to be 62 ppb, which resulted in many of the 1-h
samples being below the LOD.
The most commonly employed S02 PEM method for personal exposure studies is the passive
badge sampler. A personal multipollutant sampler has been developed to measure particulate and gaseous
pollutants simultaneously (Demokritou et al., 2001). A single elutriator, operating at 5.2 L/min, is
employed to sample particulate pollutants. A passive S02 badge is attached diametrically to the elutriator,
which has been coated with Teflon to minimize reactive gas losses. The passive badge sample is coated
with TEA for the collection of S02 and N02. Because wind speed can affect the collection rate of the
passive badge sampler, this system employs a constant face velocity across the passive badge sampler. For
24-h sampling times, the estimated limit of detection (LOD) for S02 is 5 ppb.
Currently, limits exist for using PEM systems to measure personal exposure to S02. Because S02
concentrations have been declining annually in the U.S., little focus has been placed on improving the
methods of analysis. LODs for S02 PEMs (-5-10 ppb for 24-h sampling) are often greater than the
concentrations of S02 that are typically observed in urban ambient environments. However, much lower
detection limits can be achieved by extending the sampling time (Kasper-Giebl et al., 1999). Personal
exposure monitoring studies often suffer from having many of the daily S02 samples (e.g., 30 to 70%)
collected below the sampler's LOD (see Table 2-14 and Table 2-15). Because of these issues, current
methods cannot characterize hourly or shorter exposures and their relationship to ambient concentrations
unless these values are in the range of several tens to hundreds of ppb.
2.6.3. Relationship between Personal Exposure and Ambient
Concentration
Because S02 concentrations have declined markedly over the past few decades, relatively few
recent personal exposure studies have focused on S02. Another consideration is that current indoor and
outdoor levels in many areas are often beneath detection limits for passive personal S02 monitors.
2-53
-------
2.6.3.1. Indoor Versus Outdoor SO2 Concentrations
Several studies in the U.S., Canada, Europe, and Asia have examined the relationships of indoor,
outdoor, and personal concentrations of S02 to ambient S02 concentrations. Perhaps the most
comprehensive set of indoor-outdoor data was obtained by Spengler et al. (1979) during the Harvard Six
Cities Study. These data are shown in Figure 2-37. 24-h ambient and indoor S02 concentrations were
measured every sixth day for 1 yr in a minimum of 10 homes or public facilities for each of the cities
studied.
50
ST 40
E
D)
3 30
CM
o
W 20
10
¦
Outdoor
¦
Indoor
20
15 s
Q.
Q.
CM
10 O
V)
PORT TOPE
KING
WAT
STL
STEU
Source: Adapted from Spengler et al. (1979)
Figure 2-37. Average annual indoor and outdoor SO2 concentrations for each of the six cities
included in the Harvard six-cities study analysis. PORT = Portage, Wl / TOPE =
Topeka, KS / KING = Kingston, TN / WAT = Watertown, MA/ STL = St. Louis, MO / STEU
= Steubenville, OH.
As can be seen from Table 2-13, a wide range is found in the ratio of indoor to outdoor
concentrations among the different studies. These differences among studies could be due in part to
differences in building characteristics (e.g., forced ventilation, building age, and building function such as
residences, schools, or other public buildings), in activities affecting air exchange rates, and in analytical
capabilities. In several studies, high values for R2 were found, suggesting that indoor levels were largely
driven by outdoor levels. A few studies found higher levels of S02 indoors than outdoors in some
samples. This situation could have arisen if there were indoor sources or as a result of analytical
measurement issues. One would expect to find lower concentrations indoors than outdoors because S02 is
consumed by reactions on indoor surfaces, especially those that are moist. Chao (2001) acknowledged
this point but could not account for the findings of this study. It was noted that two samples had unusually
high indoor to outdoor ratios and that the mean ratios would have been much lower otherwise. Winter-
summer differences in the indoor:outdoor ratio are consistent with seasonal differences in air exchange
rates, as noted by Brauer et al. (1991).
2-54
-------
Table 2-13. Relationships of indoor to outdoor SO2 concentrations.
Study
Location
Indoor to Outdoor Ratio
(# samples)
Notes
Spengler et al. (1979)
Portage, Wl
Topeka, KS
Kingston, TN
Watertown, MA
St. Louis. MO
Steubenville, OH
0.67 (349)
0.50 (389)
0.08 (425)
0.33 (486)
0.31 (543)
0.39 (499)
One year during Harvard Six Cities Study. West-Gaeke
method.
Stock et al. (1985)
Houston, TX
0.54 (2425)
May to October, continuous FRM for indoor and
outdoor.
Merangerand Brule (1987)
Antigonish, NS, Canada
0.84 (8)
Early spring, 1 wk avg in 1 house with oil furnace,
FPD-TA
Brauer et al. (1989)
Boston, MA
0.23 (24)
Summer, HEADS
Li and Harrison (1990)
Essex, UK
0.22
Summer
Brauer et al. (1991)
Boston, MA
0.39 (geom. mean) (29), R2 = 0.89
0.05 (geom. mean) (23), R2 = 0.73
Summer, HEADS
Winter, HEADS
Chan et al. (1994)
Taipei, Taiwan
0.24(15)
0.23 (37)
Summer, PS
Winter, PS
Lee etal. (1999b)
Hong Kong
0.92, R2= 0.56
Winter, PF
Patterson and Eatough (2000)
Lindon, UT
0.027 ± 0.0023, R2= 0.73
Winter, ADS, all samples
Kindzierski and Sembaluk
Boyle, Alberta, Canada
0.12(12)
Late Fall, PS
(2001)
Sherwood Park, Alberta,
Canada
0.14(13)
Chao (2001)
Hong Kong
1.01 ±0.78(10)
Summer. Windows mainly kept closed, PS
Kindzierski and Ranganathan
(2006)
Fort McKay, Alberta, Canada
0.35 (30)
Fall. All indoor levels < LOD and set =1/2 LOD, PS
FPD-TA = Flame Photometric Detection-Thermal Analysis
HEADS = Harvard-EPA Annular Denuder System
PS = passive sampler
PF = pulsed fluorescence
FRM = Federal Reference Method
ADS = Annular Denuder System
Indoor, or nonambient, sources of S02 could complicate the interpretation of associations between
personal exposure to ambient S02 in exposure studies. The source of indoor S02 is combustion of sulfur-
containing fuels, and higher levels are expected when emissions are poorly vented. Brauer et al. (2002b)
noted that only one study (Biersteker et al., 1965) conducted inferential analyses of potential determinants
of exposure to indoor S02 levels. In the Biersteker et al. study, conducted in the Netherlands, indoor
levels increased with oil, coal, and gas heating, as well as smoking in homes and increased outdoor levels.
Triche et al. (2005) measured S02 levels in homes in which secondary heating sources (fireplaces,
kerosene heaters, gas space heaters, and wood stoves) were used. They found elevated indoor levels of
S02 when kerosene heaters were in use. Median levels of S02 when kerosene heaters were used (6.4 ppb)
were much higher than when they were not in use (0.22 ppb). The maximum S02 level associated with
kerosene heater use was 90.5 ppb. They did not find elevated S02 levels when the other secondary heating
sources were in use.
2-55
-------
2.6.3.2. Relationship of Personal Exposure to Ambient Concentration
A few studies have evaluated the association of personal S02 exposure to ambient concentrations
(Brauer et al., 1989; Chang et al., 2000; Sarnat et al., 2000; 2001; 2005; 2006a). However, no studies have
characterized the relationship between community avg exposures and ambient concentrations of S02.
Some of these personal exposure studies were conducted under the Health Effects Institute's
Characterization of Particulate and Gas Exposures of Sensitive Subpopulations Living in Baltimore and
Boston research plan (Koutrakis et al., 2005). However, the focus of many of these studies has been
exposure to particles with acid gases included to evaluate confounder or surrogate issues.
Table 2-14 summarizes the longitudinal correlation coefficients between personal S02 exposures
and ambient concentrations of S02, and Table 2-15 shows the pooled correlation coefficients. Most of the
studies examined lack the ability to quantify 24-h avg personal S02 exposures as a result of the low
ambient S02 concentrations and the limitations of passive sampling, except two studies conducted by
Brauer et al. (1989) and Sarnat et al. (2006a) in which the sampling systems can adequately quantify 24-h
personal exposures.
Brauer et al. (1989) determined the slope of the regression line between personal and ambient
concentrations to be 0.13 ± 0.02, R2 = 0.43, based on 44 measurements made in Boston, MA during the
summer of 1988. This study had the highest proportion of personal samples above the detection limit
among the studies considered and the best regression fit between personal exposure and ambient
concentrations. Most if not all of the data points obtained using the HEADS appeared to be above the
working detection limits as defined by the authors in their publications (Brauer et al., 1989; Koutrakis et
al., 1988b). Note that calculating detection limits in this way could result in lower apparent detection
limits than if field blanks were used. The authors reported significance at the p < 0.001 level for the slope
but not the intercept. Since the stationary monitoring site was located at an elevation of 250 m above
street level, the use of data from this ambient monitoring site will overestimate personal exposure, as the
concentration of S02 increases with height because it is emitted mainly by elevated point sources. Indeed,
the ambient concentrations are about a factor of two higher than the concentrations measured outside
residences. Sarnat et al. (2006a) reported that ambient S02 was observed to be significantly associated
with personal S02 exposure concentrations during the fall (slope = 0.08 for overall population) in a study
in Steubenville, OH. The authors also observed the effect of ventilation on the association between
personal exposure concentrations and ambient concentrations (slope = 0.07 for subjects in buildings with
low ventilation rates, and 0.13 for subjects in buildings with high ventilation rates).
The associations between personal exposure and ambient concentration cannot be examined in the
other studies because almost all the personal exposure concentrations were beneath detection limits. For
example, Chang et al. (2000) tested a new personal active sampling device (a RAS with a TEA-based
denuder) on volunteer participants to measure hourly personal exposure to S02. However, the method
detection limit was too high for S02 (62 ppb for 1-h sampling) to generate a robust S02 exposure dataset
to perform further analysis, and so the authors did not use the S02 data.
2-56
-------
Table 2-14. Association between personal exposure concentration and ambient concentration
(longitudinal correlation coefficients).
Study
Study Design
Season
Mean Cone.
(PPb)
Statistics
Comments
Sarnat Longitudinal, Baltimore, 20 senior, healthy, nonsmoking people (avg
etal. age 75), summer of 1998 and winter of 1999,1 day averaged sample,
(2000) for 12 consecutive days for each subject; four to six subjects were
measured concurrently during each 12-day monitoring period.
Winter
Ambient:
6.6-10.2
Personal:
-0.8-1.2
Slope: NR
Intercept: NR
Correlation
Coefficient (r):
-0.75 to 0.65 with
a median of 0.02
(14 subjects)
The LOD for 24-h sampling
was 6.5 ppb. All personal
samples were below LOD.
Sarnat Longitudinal, Baltimore, 56 seniors, schoolchildren, and people with
et al. COPD, summer of 1998 and winter of 1999,14 of 56 subjects
(2001) participated in both sampling seasons; all subjects were monitored for
12 consecutive days (24-h avg samples) in each of the one or two
seasons, with the exception of children who were measured for 8
consecutive days during the summer.
Winter Ambient: Slope: 1) Concentrations are
4-17 -0.05 (N=487 with estimated from Figure 1 in the
Personal: 45 subjects) paper.
-2-3 Intercept: 0.54* 2) Correlation coefficients are
(N=487 with 45 estimated from Figure 2 in the
subjects) paper.
r = -0.75 to 0.6 3) LOD was referred to Sarnat
with a median of - et al. (2000), which was 6.5
0.1 (45 subjects) ppb. Therefore, all personal
samples were below LOD.
Sarnat
et al.
(2005)
Longitudinal, Boston, 43 seniors and schoolchildren, summer of 1999
and winter of 2000, Similar study design as Sarnat et al. (2001).
Summer
Winter
Ambient:
Slope: 0.00
2.8-4.5
(N-335)
Personal:
Intercept: NR
0.3-0.5
r = -0.60 to 0.70
with a median of
0.00,
(Sample size: NR)
Ambient:
Slope: -0.02
4.9-10.7
(N = 299)
Personal: -
Intercept: NR
0.3-1.9
r = -0.55 to 0.60
with a median of
0.10 (Sample size
NR)
1) Correlation coefficients are
estimated from Figure 1 in the
paper.
2) LOD was 2.3 ppb, and
96.5% of personal samples
were below LOD.
1) Correlation coefficients are
estimated from Figure 1 in the
paper.
2) LOD was 3.2 ppb, and
95.4% of personal samples
were below LOD.
NR: not reported
* p < 0.05
In the context of determining the effects of ambient pollutants on human health, the association
between the ambient component of personal exposures and ambient concentrations is more relevant than
the association between personal total exposures (ambient component + nonambient component) and
ambient concentrations. As described in Equations 2-2 and 2-4, personal total exposure can be
decomposed into two parts: an ambient and a nonambient component. Usually, the ambient component of
personal exposure is not directly measureable. However, it can be estimated by exposure models, or the
personal total exposure can be regarded as the personal exposure of ambient origin if there are no indoor
or nonambient sources. It is expected that the association between ambient concentrations and the
ambient component of personal exposures would be stronger than the association between ambient
concentrations and personal total exposures as long as the ambient and nonambient component of
personal total exposure are independent. None of the studies examined indoor sources. However, indoor
sources are not expected to be present for S02. The correlation coefficients between personal ambient S02
2-57
-------
exposures and ambient S02 concentrations in different types of exposure studies are relevant to different
types of epidemiologic studies.
Table 2-15. Association between personal exposure concentration and ambient concentration
(pooled correlation coefficients).
Study
Study Design
Season
Mean Cone.
(PPb)
Statistics
Comments
Brauer et al. Pooled, Boston, study population was NR, the
(1989) number of participants was estimated to be 48,
July and August of 1988 for 24 days, 1 day
averaged sample, two subjects were monitored
each day.
Summer Ambient: Slope: 0.13* 1) Concentrations estimated from
2.5 - 9.5 (N=44) Figure 2 in the paper.
Personal: Intercept: Not 2) Central site monitor was 250 m
0.4-1.8 significant above the ground level.
R2: 0.43 3) LOD for personal samples was
-0.19 ppb based on the method of
determining the LOD for an active
sampling system.
Sarnat et al.
(2006a)
Steubenville, 15 senior subjects, summer and fall Summer
of 2000, two consecutive 24-h samples were
collected for each subject for each wk, 23 wks
total. Correlation coefficients were calculated in
the pooled data set.
Ambient:
2.7 ±3.9
Personal:
1.5 ±3.3
Slope: 0.03
(N=106)
Intercept: NR
R2: 0.00
LOD was 5.5 ppb; 53.5% of
personal samples were below LOD.
Fall
Ambient:
5.4 ±9.6
Personal:
0.7 ±1.9
Slope: 0.08*
(N=152)
Intercept: NR
R2: 0.15
LOD was 3.8 ppb, and 31.6% of
personal samples were below LOD.
* significant at a = 0.05 level
There are three types of correlations generated from different study designs and ways to analyze
the data from exposure studies: longitudinal, "pooled," and daily-avg correlations (U.S. EPA, 2004).
Longitudinal correlations1 are calculated when data from a study includes measurements over multiple
days for each subject (longitudinal study design). Longitudinal correlations describe the temporal
relationship between daily personal S02 exposure or microenvironment concentration and daily ambient
S02 concentration for the same subject. The longitudinal correlation coefficient can differ between
subjects (i.e., each person may have a different correlation coefficient). The distribution of correlations for
each subject across a population could be obtained with this type of data (e.g., Sarnat et al., 2000; 2001;
2005). A longitudinal correlation coefficient between the ambient component of personal exposures and
ambient concentrations is relevant to the panel epidemiologic study design. In Table 2-14, most
longitudinal studies reported the association between personal total exposures and ambient concentrations
for each subject. For some subjects the associations were strong and for some subjects the associations
were weak. The weak personal and ambient associations do not necessarily mean that ambient
{n - 1)5
where "r" is the longitudinal correlation coefficient between personal exposure and ambient concentration, "a"
represents the ambient concentration, "x" represents exposure, represents the ith subject, "/' represents the yth measurement (with the
averaging time ranging from two days to two weeks for SO2 measurement), "s" represents the standard deviation, and in the longitudinal
studies is the number of measurements for each subject. The ambient concentration aj could be measured by one ambient monitor or the average
of several ambient monitors.
2-58
-------
concentrations are not a good surrogate for personal exposures, because the weak associations could have
resulted from the day-to-day variation in the nonambient component of total personal exposure. These
types of correlations can have a substantial effect on the value of the resultant correlation coefficient.
Mage et al. (1999) showed that very low correlations between personal exposure and ambient
concentrations could be obtained when people with very different nonambient exposures are pooled, even
though their individual longitudinal correlations are high.
Pooled correlations1 are calculated when a study involves one or only a few measurements per
subject and when different subjects are studied on subsequent days. Pooled correlations combine
individual-subject/individual-day data for the calculation of correlations. Pooled correlations describe the
relationship between daily personal S02 exposure and daily ambient S02 concentration across all subjects
in the study (e.g., Brauer et al., 1989; Sarnat et al., 2006a).
Daily-avg correlations2 are calculated by averaging exposure across subjects for each day. Daily-
avg correlations then describe the relationship between the daily avg exposure and daily ambient pollutant
concentration. This type of correlation (i.e., the association between the ambient component of
community avg exposures and ambient concentrations) is more directly relevant to community time-series
studies, in which ambient concentrations are used as a surrogate for community avg exposure to
pollutants of ambient origin. However, exposure of the population to S02 of ambient origin has not been
reported in any of the studies examined.
Not only does the exposure study design determine the meaning of the correlation coefficients in
the context of exposure assessment in epidemiologic studies, but the type of correlation calculation also
affects the strength of the association between personal exposures and ambient concentrations. The
strength of the association between personal exposures with ambient and/or outdoor concentrations for a
population is determined by variations in several physical factors: indoor or other local sources, air
exchange rate, penetration, decay rate of the pollutant in different microenvironments, and the time
people spend in different microenvironments with different pollutant concentrations. For different types
of correlation coefficients, the components of the variance of these physical factors are different, and
therefore the strength of different types of correlation coefficients is different. Longitudinal correlation
coefficients reflect the interpersonal variations of these physical factors. Pooled correlation coefficient
reflect both inter- and intra- personal variations of these physical factors. For the association between
community avg exposures and ambient concentrations, interpersonal variations of these physical factors
are reduced by averaging personal exposures across a community. Therefore, the strength of the
associations between personal exposures and ambient concentrations may not be directly comparable,
although these associations are determined by the same set of physical factors (but affected in different
ways).
Since correlations are standardized quantities that depend on multiple features of the data not only
is the linear "relatedness" (covariance) of the two quantities important in a correlation, but so is the
variability of each. That variability can be affected by exposure factors in various ways. In the following
assessments, the effects of these physical factors on the strength of correlation coefficients are primarily
Zk v)f" ")
r°*=—(
1 1 fl — IK 5
v > x a where "r" is the pooled correlation coefficient and is the number of paired measurements of exposure and
ambient concentration, and all other symbols are defined the same way as those in the longitudinal correlation coefficient.
Y$Fj-xXaj-a)
r_ =J.
(/7-l)s-Sa
2 Xj , where "r" is the daily-average correlation coefficient and "ft" is the number of measurement period, during each of
which the exposure for all subjects are measured, and all other symbols are defined the same way as those in the longitudinal correlation
coefficient.
2-59
-------
examined within a study, and the purpose of the inter-study comparison is to examine the consistency of
the effects across different types of studies.
The strength of the associations between personal exposures and ambient concentrations could also
be affected by the quality of the data collected during the exposure studies. There are at least six aspects
associated with the quality of the data: method precision, method accuracy (compared with FRM), percent
of data above method detection limits (based on field blanks), completeness of the data collection, sample
size, and soundness of the quality assurance/quality control procedures. Unfortunately, not all studies
reported the six aspects of the data quality issue. The fraction of data below the detection limit might be a
concern for some studies (Sarnat et al., 2000; 2001; 2005). Correlation coefficients would be biased low if
data used in their calculation are below detection limits. Sampling interferences associated with both
ambient (see Section 2.3) and personal sampling (see Section 2.6.2) could also affect data quality.
Therefore, caution must be exercised when interpreting the results in Table 2-14 and Table 2-15. Sarnat
et al. (2001; 2005; 2006a) examined the associations between ambient S02 concentrations and ambient or
personal co-pollutant concentrations. Sarnat et al. (2001) reported that during the winter of 1999, ambient
S02 was significantly associated (p <0.05) with personal exposure to fine particulate matter (PM2 5)
(slope = - 0.24), personal exposure to S042 (slope = - 0.03), and personal exposure to PM2 5 of ambient
origin (slope = - 0.16). However, it should be noted that all the slopes are negative perhaps as the result
of measurement error. Sarnat et al. (2005) reported that significant associations between ambient S02 and
either personal exposures or ambient concentrations of other pollutants were found for personal S042
(winter, slope = 0.06); (summer, slope = 0.39); personal PM2 5 (summer, slope = 1.68), ambient S042
(winter, slope = 0.19); and ambient PM2 5 (winter, slope = 0.80). In Sarnat et al. (2006a), ambient S02 was
observed to be significantly associated with ambient PM2 5, ambient S042 , and ambient elemental carbon
(EC) during the fall (R2 = 0 .22, 0.33, and 0.34 respectively). It was significantly associated with personal
PM2 5 during the summer, and personal S042 and personal EC during the fall (R2 = 0.07, 0.06, and 0.05
respectively).
Of significant concern is the ability of currently available techniques for monitoring either personal
exposures or ambient concentrations to measure S02 concentrations that are typically found in most urban
environments. In some studies, most data might be beneath detection limits. This is especially true for
personal exposure and indoor data. Indeed, in one study (Chang et al., 2000), the investigators had to
discard data for S02 because the values were mostly beneath detection limits. In the study of Kindzierski
and Ranganathan (2006), all indoor concentration data were beneath detection limits. In Sarnat et al.
(2000), -70% of personal measurements were beneath detection limits, and -33% of personal
measurements returned apparent negative concentration values. In such situations, associations between
ambient concentrations and personal exposure are inadequately characterized. When personal exposure
concentrations are above detection limits, a reasonably strong association is observed between personal
exposures and ambient concentrations.
2.6.4. Exposure Errors in Epidemiologic Studies
This assessment considered the errors that result from using the ambient concentration of an air
pollutant as an exposure indicator rather than using the actual personal exposure to that air pollutant in the
epidemiologic statistical analysis. Such errors change both the health effects estimate, expressed as the
relative risk factor, (3, and the standard error of (3. There are many assumptions made in going from the
available measurement of a pollution indicator to an estimate of the personal exposure. The importance of
these assumptions and their effect on (3 depend on the type of epidemiologic study.
The considerations of exposure error for S02 are simplified compared to those for N02 and PM.
The only experimental measure available is the ambient concentration of S02. In addition, indoor and
other non-ambient sources of S02 are not thought to be important in population studies, lessening
concerns about the possible influence of exposures other than to ambient S02. However, because S02 is
2-60
-------
rapidly removed by interaction with surfaces (more slowly than 03 but more rapidly than N02 or PM), the
ratio of indoor concentrations to outdoor concentrations will be lower, and perhaps more variable than in
the case of N02, PM and CO (which is relatively un-reactive with surfaces).
2.6.4.1. Community Time-Series Studies
This section applies primarily to studies on the association of daily avg S02 concentrations with
daily measures of mortality or morbidity in a community. The following three exposure issues are of
primary concern with respect to S02 time-series epidemiologic analysis: (1) the relationship of the
measured concentration of S02 to the true concentration; (2) the relationship of day-to-day variations in
the concentrations of S02, as measured at a central monitoring site, with the corresponding variations in
the avg concentration of S02 over the geographic area from which the health measurements are drawn;
and (3) the relationship of the community avg concentration of S02 to the avg personal exposure to
ambient S02. These three issues are described below.
Relationship of Measured SO2 to the True Concentration
Since there is always a random component to instrumental measurement error, the correlation of
the measured S02 with the true S02, on either a 24-h or 1-h basis, will be less than 1. Sheppard et al.
(2005) indicate that instrument error in the individual or daily avg concentrations have "the effect of
attenuating the estimate of a." Zeger et al. (2000) suggest that instrument error has both Berkson and non-
Berkson error components. However, the authors state that the "instrument error in the ambient levels is
close to the Berkson type." In order for this error to cause substantial bias in (3, the error term (the
difference between the true concentrations and the measured concentrations) must be strongly correlated
with the measured concentrations. Zeger et al. (2000) suggest that, "further investigations of this
correlation in cities with many monitors are warranted." Averaging across multiple unbiased ambient
monitors in a region should reduce the instrument measurement error (Sheppard et al., 2005; Wilson and
Brauer, 2006; Zeger et al., 2000). There are concerns about the precision and accuracy of the ambient
concentration measurements because S02 concentrations are much lower now than when the S02
standards were first promulgated. Typical ambient concentrations of S02 in the contiguous U.S. are nearly
all at or beneath the detection limit of the monitors currently used in the regulatory network. Thus, greater
relative error is most often observed at the lower ambient concentrations compared to the less frequent
higher concentration exposures, as might occur because of plume downwash near local point sources or
entrainment of plumes downwind from large power plants or smelters. It is unclear how uncertainties in
the true concentrations of S02, i.e., instrument measurement error, will change (3.
Relationship of Day-to-day Variations in the Ambient Concentration of SO2 to Variations in
the Community Average
There has been little analysis of the spatial variation of S02 across communities. S02 emissions
arise mainly from coal fired power plants (see Annex Table B-4). Newer power plants and smelters in the
U.S. are no longer located within urban centers. However, some older power plants and industrial
facilities are located in many urban areas, especially in the Midwest and Northeast. Downwash from the
plumes emitted from these facilities can contribute to elevated levels of S02 at the surface in these cities.
However, it is anticipated that S02 will behave largely as a regional pollutant in most areas.
Site-to-site correlations of S02 concentrations, as shown for several cities in Table 2-9 vary from
very low to very high values. This suggests that the concentration of S02, measured at any given
monitoring site, may not be highly correlated with the avg community concentration in some areas. There
are a number of possible reasons for these findings: local sources that cause the S02 to be unevenly
distributed spatially; a monitoring site being chosen to represent a nearby source; terrain features that
2-61
-------
divide the community into several sub-communities that differ in the spatial and temporal pattern of
pollution; and errors in the measurement of the low concentrations of S02 present at most sites. To the
extent that the correlation of the ambient concentration with the community avg concentration is < 1, (3
will be reduced. Similarly, (3 will be reduced if there are subareas of the community where the correlation
between the local avg concentrations and the concentrations measured at the ambient monitoring site is
< 1. If concentrations in an area of a community impacted by plumes from local S02 sources might be
higher than, and not well-correlated with, the concentrations at the ambient monitor, and if such high
concentrations affect a sizable portion of the population exposed to emissions from a local source, that
community might not be suitable for time-series epidemiologic analyses. On the other hand, if the plume
impacts the ambient monitor, the high concentration of S02 not accompanied by a corresponding high
effect in the entire community will bias (3 toward the null.
An additional complication will arise if location near a source is correlated with sensitivity. For
example, if poverty causes sensitivity due to poor nutrition, and land prices decline the closer one gets to
a major source, then exposure and sensitivity will be correlated. If the day to day variations in the
concentrations due to the source are correlated with the day to day variations in the ambient
concentrations used in the epidemiologic analysis, (3 will be increased above the (3 that would be found if
the exposure and sensitivity were not correlated. However, it is more likely that the variations in the
concentrations related to the source are not correlated with the ambient concentrations. In this case the
health effects from the source S02 would not contribute to (3 and |3 would be decreased compared to a
uniform distribution of sensitivity.
Relationship of Community Average Concentration of SO2 to Average Personal Exposure to
Ambient SO2
People spend much of their time indoors, and in the absence of indoor sources, indoor
concentrations are lower than outdoor concentrations. This is very likely the case with S02, since the only
known significant indoor source of S02 in the U.S. is the use of kerosene heaters, which is not thought to
be widespread enough to influence population studies. Differences in infiltration factors among homes
can also result in differences among individuals' personal exposures. It is necessary to consider how this
difference between the ambient concentration, which is used in epidemiologic analyses, and the personal
ambient exposure (which includes exposure to the full outdoor concentration while outdoors and exposure
of only a fraction of the outdoor concentrations while indoors) will affect the calculated (3. Personal
exposure to ambient S02 is given by Ka = a • Ca where Ea is exposure to ambient S02, a is the ambient
exposure factor with values between 0 and 1, and Ca is the ambient S02 concentration as measured at a
community monitoring site. Zeger et al. (2000) noted that for community time-series epidemiology, which
analyzes the association between health effects and potential causal factors at the community scale rather
than the individual scale, it is the correlation of the daily avg ambient concentrations with the daily
community average personal exposures that is important, not the correlation between the daily avg
ambient concentrations and the individual personal exposures. Thus, as mentioned in Section 2.6.3, the
low correlation between daily avg ambient concentrations and individual personal exposures, as
frequently found in pooled panel exposure studies, is not relevant to community time-series
epidemiologic analysis. Unfortunately, no studies provide adequate information about the community avg
personal exposure to S02.
There has also been concern with the variation of a. Zeger et al. (2000) suggested (for PM) that
variations in the individual daily values of a would be a Berkson error and would not change the point
estimate of (3, although such variations would increase the standard error of (3. Sheppard et al. (2005) used
simulations to confirm this for nonreactive pollutants. However, such variations increase the standard
error. Day-to-day variations in the population avg fraction of ambient exposure will not change the point
estimate of (3 unless the population avg fraction of ambient exposure is correlated with seasonal trends in
ambient concentration, according to Sheppard et al. (2005).
2-62
-------
In the case of a correlation of location near a source with sensitivity, the health effects of the higher
concentrations will be attributed to the avg concentration and (3 will be increased. A similar but lesser
effect on (3 will occur anytime there are high concentrations resulting from localized sources that are not
included in the long-term avg concentration and that vary from city to city.
Both Zeger et al. (2000) and Sheppard et al. (2005) show that if P is the health effect parameter
that would be obtained with a time-series analysis using the ambient exposure and (3C is the health effect
parameter that would be obtained with a time-series analysis using the ambient concentration, then (3C =
a • P'4. Thus, time-series studies yield different parameters depending on whether they use concentration
or exposure. However, the two parameters are related by a.
2.6.4.2. Short-Term Panel Studies
Panel epidemiology refers to studies that follow a relatively small number of subjects for a
relatively short time. Panel studies typically examine the association between symptoms or health
outcomes of individuals and either ambient concentrations or personal exposures. Personal exposures to
S02 usually are not measured. Rather, ambient concentrations are more often used in panel studies.
Similar types of exposure error as discussed for community time series apply to panel studies.
The ambient exposure factor (a) may differ for each person and each day leading to error in the
exposure estimate. If a panel is composed of subjects who live in similar housing and have similar activity
pattern, and the study is limited to a single season, the variation in a over time and individual subjects
may be small. However, if the panels are composed of more diverse subjects or extend for more than one
season, values of a may be quite variable. Such variability will affect the estimate of exposure for each
subject.
2.6.4.3. Long-Term Cohort Studies
For long-term exposure epidemiologic studies, concentrations are integrated over time periods of a
year or more and usually for spatial areas the size of a city, county, or metropolitan statistical area (MSA),
although integration over smaller areas may be feasible. These studies focus on spatial variations in
concentrations. Health effects are then regressed in a statistical model against the avg concentrations in
the series of cities (or other areas). In time-series studies, a constant difference between the measured and
the true concentration (instrument offset) will not affect (3, nor will variations in the daily average a or the
daily avg nonambient exposure, unless the variations are correlated with the daily variations in
concentrations. However, in long-term exposure epidemiologic studies, if instrument measurement errors,
long-term avg values of a, or long-term averages of nonambient exposure differ for different cities (or
other areas used in the analysis), the city-to-city long-term ambient S02 concentrations will not be
perfectly correlated with the long-term avg exposure to either ambient or total S02. This lack of
correlation would be expected to bias the point estimate of (3.
2.6.4.4. Summary of Evaluation of Exposure Error in Epidemiologic Studies
Exposure error caused by using ambient concentrations of S02 as a surrogate for exposure to
ambient S02 affect (3 in different ways, depending upon the type of epidemiologic study. In community
time-series and short-term panel epidemiologic studies, the nonambient source component of personal
exposure and the variation in the ambient exposure factor caused by building ventilation practices and
personal behaviors generally will not change the estimate of (3. But, the spatial variation of S02 or the
representativeness of the ambient monitor might bias the estimate of (3 toward null. Therefore, (3 observed
in S02 community time-series or panel epidemiologic studies would be stronger and less uncertain if
exposure errors had been adjusted and/or controlled. In long-term cohort epidemiologic studies,
2-63
-------
instrument measurement errors, factors that influence exposure to ambient S02, or long-term averages of
nonambient exposure may differ for different cities, which may bias the estimate of |3, but the extent and
direction of this bias is unclear.
2.7. Dosimetry of Inhaled Sulfur Oxides
This section is intended to present an overview of general concepts related to the dosimetry of S02
in the respiratory tract. Dosimetry of S02 refers to the measurement or estimation of the amount of S02 or
its reaction products reaching and persisting at specific respiratory tract and systemic sites after exposure.
One of the principal effects of inhaled S02 is that it stimulates bronchial epithelial irritant receptors and
initiates a reflexive contraction of smooth muscles in the bronchial airways. The compound most directly
responsible for health effects may be the inhaled S02, and/or its chemical reaction products. Complete
identification of the causative agents and their integration into S02 dosimetry is a complex issue that has
not been thoroughly evaluated. Few studies have investigated S02 dosimetry in the interval since the 1982
AQCD and the 1986 Second Addendum.
2.7.1. Respiratory Gas Deposition
The major factors affecting the transport and fate of gases and aerosols in the respiratory tract are:
the morphology of the respiratory tract; the physiochemical properties of the mucus and surfactant layers;
respiratory functional parameters such as tidal volume, flow rate, and route of breathing; physicochemical
properties of the gas; and the physical processes that govern gas transport. Physicochemical properties of
S02 relevant to respiratory tract uptake include its solubility and diffusivity in epithelial lining fluid
(ELF), as well as its reaction-rate with ELF constituents. Henry's law relates the gas phase and liquid
phase interfacial concentrations at equilibrium, and is a function of temperature and pressure. Henry's law
shows that the amount of S02 in the aqueous phase is directly proportional to the partial pressure or
concentration of S02 in the gas phase. Although the solubility of most gases in mucus and surfactant is
not known, the Henry's law constant is known for many gases in water. Inversely proportional to
solubility, the Henry's law constant for S02 is 0.048 (mole/liter) air / (mole/liter) water at 37° C and 1
atm. For comparison, the value for 03 is 6.4 under the same conditions (Kimbell and Miller, 1999). In
general, the more soluble a gas is in biological fluids, the more rapid and proximal its absorption will be
in the respiratory tract. When the partial pressure of S02 on mucosal surfaces exceeds that of the gas
phase, such as during expiration, some desorption of S02 from the ELF may be expected.
Because S02 is highly soluble in water, it is expected to be almost completely absorbed in the nasal
passages of both humans and laboratory animals under resting conditions. The dosimetry of S02 can be
contrasted with the lower solubility gas, 03, for which the predicted tissue doses (03 flux to liquid-tissue
interface) are very low in the trachea and increase to a maximum in the terminal bronchioles or first
airway generation in the pulmonary region (see Chapter 4, U.S. EPA, 2006b). Similar to 03, the nasal
passages remove S02 more efficiently than the oral pathway (Brain, 1970; Melville, 1970; Nodelman and
Ultman, 1999). With exercise, the pattern of S02 absorption shifts from the upper airways to the
tracheobronchial airways in conjunction with a shift from nasal to oronasal breathing and increased
ventilatory rates. As a result of its effect on delivery and uptake, mode of breathing is also recognized as
an important determinant of the severity of S02-induced bronchoconstriction. The greatest responses
occur during oral breathing followed by oronasal breathing, and the smallest responses were observed
during nasal breathing.
Andersen et al. (1974) measured the nasal absorption of S02 (25 ppm) in 7 volunteers during
inspiration at an avg inspired flow of 23 L/min (i.e., eucapnic hyperpnea [presumably] to simulate light
2-64
-------
exertion). These investigators reported that the oropharyngeal S02 concentration was below their limit of
detection (0.25 ppm), implying that at least 99% of S02 was absorbed in the nose of subjects during
inspiration. Speizer and Frank (1966) also measured the absorption of S02 (16.1 ppm) in 7 healthy
subjects at an avg ventilation of 8.5 L/min (i.e., at rest). They reported that 14% of the inhaled S02 was
absorbed within the first 2 cm into the nose. The concentration of S02 reaching the pharynx was below
the limit of detection, suggesting that at least 99% was absorbed during inspiration. Melville (1970)
measured the absorption of S02 (1.5 to 3.4 ppm) during nasal and oral breathing in 12 healthy volunteers.
Total respiratory tract absorption of S02 was significantly greater (p < 0.01) during nasal than oral
breathing (85 versus 70%, respectively) and was independent of the inspired concentration. Melville
(1970) did not report respired flow rates, so the effect of flow on the S02 absorption could not be
discerned. However, it may be noted that the total respiratory absorption during nasal breathing reported
by Melville (1970) was clearly less than the nasal absorption reported by both Andersen et al. (1974) and
Speizer and Frank (1966). The disparity in nasal absorption between these studies is, in part, due to
desorption of S02 during expiration as discussed in Section 2.7.3.
Frank et al. (1969) and Brain (1970) investigated the oral and nasal absorption of S02 in the
surgically isolated upper respiratory tract of anesthetized dogs. Radiolabeled S02 (35S02) at the
concentrations of 1, 10, and 50 ppm was passed separately through the nose and mouth at the steady
unidirectional flows of 3.5 and 35 L/min for 5 min. The nasal absorption of S02 (1 ppm) was 99.9% at 3.5
L/min and 96.8% at 35 L/min. The oral absorption of S02 (1 ppm) was 99.56% at 3.5 L/min, but only
34% at 35 L/min. The nasal absorption of S02 at 3.5 L/min increased with concentration at 1, 10, and 50
ppm and was reported to be 99.9, 99.99, and 99.999%, respectively. This increase in absorption with
concentration was hypothesized to be due to increased mucus secretion and increased nasal resistance at
the higher S02 concentrations. The increased mucus was thought to provide a larger reservoir for S02
uptake. The increased nasal resistance may increase turbulence in the airflow and, thereby, decrease the
boundary layer between the gas and liquid phases. Dissimilar to the nose, S02 absorption in the mouth
decreased from 99.56 to 96.3% when the concentration was increased from 1 to 10 ppm at 3.5 L/min.
Frank et al. (1969) noted that the aperture of the mouth may vary considerably, and that this variation may
affect S02 uptake in the mouth. Although S02 absorption was dependent on inhaled concentration, the
rate and route of flow had a greater effect on the magnitude of S02 absorption in the upper airways.
Strandberg (1964) studied the uptake of S02 in the respiratory tract of rabbits. A tracheal cannula
with two outlets was utilized to allow sampling of inspired and expired air, and S02 absorption was
observed to depend on inhaled concentration. The absorption during maximal inspiration was 95% at high
concentrations (100 to 700 ppm), reflecting an increased S02 removal in the extrathoracic (ET) region,
whereas it was only 40% at low concentrations (0.05 to 0.1 ppm). On expiration, the total S02 absorbed
(i.e., inspiratory removal in the ET region plus removal in the lower airways) was 98% at high
concentrations and only 80% at the lower concentrations.
Amdur (1966) examined changes in airway resistance in guinea pigs due to S02 exposure. Guinea
pigs were exposed for 1-h to 0.1 to 800 ppm S02 during natural unencumbered breathing or to 0.4 to 100
ppm while breathing through a tracheal cannula. At concentrations of 0.4- to 0.5 ppm S02, route of
administration did not affect the airway resistance response, whereas at concentrations of > 2 ppm, the
responses were greater in animals exposed by tracheal cannula. Based on the concentration-dependent
absorption of S02 in the ET region observed by Strandberg (1964), Amdur (1966) concluded that the
airway resistance responses at low-exposure concentrations were independent of method of
administration, because the lung received nearly the same concentration with or without the cannula as
evidenced by minimal ET absorption.
More recently, Ben-Jebria et al. (1990) investigated the absorption of S02 in excised porcine
tracheae. Absorption was monitored over a 30-min period following the introduction of S02 (0.1 to 0.6
ppm, inlet concentration) at a constant flow (2.7 to 11 L/min). The data were analyzed using diffusion-
reactor theory. An overall mass transfer coefficient (KS02) was determined and separated into its
contributions due to gas (convection and diffusion) and tissue phase (diffusivity, solubility, and reaction
rates) resistances. S02 in the liquid phase was assumed to form HSO, rapidly, in proportion with the gas
2-65
-------
phase S02 concentration. HSO , then diffused down the concentration gradient into the tissues where it
reacted irreversibly with biochemical substrates. Initially, KS02 was limited only by gas phase resistance,
but it decreased exponentially over the first 5 to 10 min of S02 exposure to a smaller steady-state value
that was due to tissue resistance to S02 absorption. The initial and steady-state KS02 values were found
to be independent of inlet S02 concentration, i.e., for a given flow, the fractional absorption of S02 did
not depend on S02 concentration. An increased KS02 (initial and steady-state) was observed with an
increasing flow that was thought to be due to a decrease in the boundary layer near the walls of the
trachea for radial S02 transport. This is in agreement with Aharonson et al. (1974), who also reported that
the transfer rate coefficient for S02 increases with increasing flow. However, the initial molar flux of S02
across the gas-tissue interface appears to increase purely as a function of the increase in mass transport
occurring with increasing flow (see Figure 5 in Ben-Jebria et al., 1990). Given that the steady-state KS02
remained stable during the 10 to 30 min of exposure and that no S02 leakage through the tissue was
identified, the authors concluded that there was an irreversible sink for S02 within the tissue.
Mathematical modeling specific to the regional respiratory uptake of S02 is unavailable for humans
and laboratory animals. More generally, the influence of age on gas dosimetry in humans during light
activity (on average) was examined by Ginsberg et al. (2005) using the U.S. EPA reference concentration
methodology (U.S. EPA, 1994a). For a highly soluble gas, such as S02, they predicted that the majority of
gas uptake would occur in the ET region and that the fractional uptake in these airways would be
modestly greater in a 3-month-old infant than an adult. The rate of gas uptake per surface, however, in the
ET region and large bronchial airways was not markedly different between infants and adults. The smaller
bronchial airways of adults were predicted to receive a greater dose (i.e., uptake per unit time and surface
area) relative to infants, although the majority of the inhaled S02 would be removed proximal to these
airways.
In summary, inhaled S02 is readily absorbed in the upper airways of both humans and laboratory
animals. During nasal breathing, the majority of available data suggests 95% or greater S02 absorption
occurs in the nasal passages, even under ventilation levels comparable to exercise. Somewhat less S02 is
absorbed in the oral passage than in the nasal passages. The difference in S02 absorption between the
mouth and the nose is highly dependent on respired flow rates. With an increase in flow from 3.5 to 35
L/min, nasal absorption is relatively unaffected, whereas oral absorption is reduced from 100 to 34%.
Thus, the rate and route of breathing have a great effect on the magnitude of S02 absorption in the upper
airways and so the penetration of S02 to the lower airways. Overall, the available data clearly show that
the pattern of S02 absorption that shifts from the upper airways to the tracheobronchial airways in
conjunction with a shift from nasal to oronasal breathing and associated increased ventilatory rates in
exercising humans. Mode of breathing is also recognized as an important determinant of the severity of
S02 induced bronchoconstriction, with the greatest responses occurring during oral breathing followed by
oronasal breathing and the smallest responses observed during nasal breathing.
2.7.2. Particles and Sulfur Oxide Mixtures
As already discussed, inhaled S02 is readily absorbed in the upper airways, particularly during
nasal breathing. It has been suggested that SOx may become absorbed to particles and subsequently
transported to more distal lung regions. Depending on atmospheric conditions, S02 can be transformed to
secondary sulfate particles and acid aerosols (H2S04) and can adsorb onto particulate matter (see Section
2.2). Jakab et al. (1996) observed that the conversion of S02 to S042 on the surface of carbon black
aerosols was dependent on high relative humidity (> 85%) and S02 concentration. These investigators
suggested that fine carbon black particles can be an effective vector for delivery of S042 to the peripheral
lung. This is not believed to be a mechanism for bringing sulfite into the deep lung. Other studies
investigating the effects of S02 coated aerosols are briefly discussed in Section 3.1.4.7.
2-66
-------
Sulfate aerosols are hygroscopic and grow in the respiratory tract. The implications of hygroscopic
growth on deposition have been reviewed extensively by Morrow (1986) and Hiller (1991). In general,
compared to nonhygroscopic particles of the same initial size, the deposition of hygroscopic aerosols in
different regions of the lung may be higher or lower, depending on the initial size. For particles with
initial sizes larger than 0.5 |im (aerodynamic diameter), the influence of hygroscopicity would be to
increase total deposition with a shift in regional deposition from the distal to larger proximal airways; for
smaller particles deposition would tend to decrease. A thorough review of respiratory deposition and
clearance of particulate matter is available elsewhere (U.S. EPA, 1996, 2004). The intent herein was to
briefly mention some issues specific to SOx.
2.7.3. Distribution and Elimination of SOx
When S02 contacts the fluids lining the airway, it dissolves into the aqueous fluid and forms
hydrogen (H ) ions and bisulfite (HS03 ) and sulfite (S032 ) anions (ATS, "Health effects of outdoor air
pollution. Committee of the Environmental and Occupational Health Assembly of the American Thoracic
Society," 1996). The majority of anions are expected to be present as HSO, at a concentration
proportional to the gas phase concentration of S02 (Ben-Jebria et al., 1990). Because of the chemical
reactivity of these anions, various reactions are possible, leading to the oxidation of S032 to S042 (see
Section 12.2.1, U.S. EPA, 1982). Clearance of S032 from the respiratory tract may involve several
intermediate chemical reactions and transformations. Gunnison and Benton (1971) identified ^-sulfonate
in blood as a reaction product of inhaled S02. Following inhalation of S02, the clearance half-time of 4.1
days for ^-sulfonate in rabbits has been reported (Gunnison and Palmes, 1973).
Some S02 is also removed by desorption of from the respiratory tract. Desorption is expected when
the partial pressure of S02 in airway lining fluids exceeds that of the air. Speizer and Frank (1966) found
that on expiration, 12% of the S02 absorbed during inspiration was desorbed into the expired air. During
the first 15 min after the 25- to 30-min S02 exposure, another 3% was desorbed. In total, 15% of the
amount of S02 originally inspired and absorbed was desorbed from the nasal mucosa. Frank et al. (1969)
reported that up to 18% of the S02 was desorbed within -10 min after exposure.
2-67
-------
Chapter 3. Integrated Health Effects
This integrated discussion is structured to provide a coherent framework for the assessment of
health risks associated with human exposure to ambient S02 in the U.S.. The main goals of this chapter
are: (1) to integrate newly available epidemiologic, human clinical, and animal toxicological evidence
with consideration of key findings and conclusions from the 1982 AQCD for Sulfur Oxides and First
Addendum (U.S. EPA, 1982), 1986 Second Addendum (U.S. EPA, 1986b), and 1994 Supplement to the
Second Addendum, (U.S. EPA, 1994c); and (2) to draw conclusions about the causal role of S02 in
relation to a variety of health effects. These causal determinations utilize the framework outlined in
Chapter 1.
This chapter is organized to present morbidity and mortality associated with short-term exposures
to S02, followed by morbidity and mortality associated with long-term exposures. Human clinical studies
examining the effect of peak exposures (less than 1-h, generally 5-10 min) of S02 on respiratory
symptoms and lung function are discussed first. Later sections describe the findings of epidemiologic
studies that examine the association between short-term (generally 24-h avg) and long-term (generally
months to years) ambient S02 exposure and health outcomes, such as respiratory symptoms in children
and asthmatics, emergency department (ED) visits and hospital admissions for respiratory and
cardiovascular diseases, and premature mortality. The human clinical and epidemiologic evidence are
presented with relevant animal toxicological data, when available.
The evaluation of epidemiologic evidence involves consideration of sources of uncertainty, as
discussed in Chapter 1, including exposure error, potential confounders or effect modifiers, statistical
modeling issues, publication bias, and multiple testing. Efforts have been made to assess the impact of
these uncertainties in the evaluation of the epidemiologic literature. For example, in studies examining
multiple single-day lag models, the pattern of association across the various lags was evaluated.
Additional focus was placed on results from distributed and moving avg lags as they are able to examine
multiday effects. Both single- and multiple-pollutant models were considered and examined for
robustness of results. Additional analyses of multiple model specifications for adjustment of temporal or
meteorological trends were regarded as sensitivity analyses. Further, the evaluation of the epidemiologic
evidence also considered study population and sample size, with particular emphasis placed on multicity
studies that by their very nature can reduce uncertainty related to publication bias. Other factors
considered were study location (North America versus other regions), meaningfiilness and validity of the
health endpoint measurements, and appropriateness of the statistical analyses methods used. These
considerations led to emphasis of certain studies in the chapter text, tables, and figures.
Animal toxicological studies may provide further evidence for the potential mechanism of an
observed effect; however, most of these studies have been conducted at concentrations vastly exceeding
current ambient conditions. In discussing the mechanisms of SOx toxicity, studies conducted under
atmospherically relevant conditions are emphasized whenever possible; studies at higher levels are also
considered, due to species-to-species differences and potential differences in sensitivity between study
subjects and especially susceptible human populations.
This chapter focuses on important recent scientific studies, with emphasis on those conducted at or
near current ambient concentrations. Given their respective strengths and limitations, evidence from
human clinical, epidemiologic, and animal toxicological studies was considered in order to evaluate the
causality of SOx-health effects associations. The annexes supplement the information included here by
presenting more details of the literature.
3-1
-------
3.1. Respiratory Morbidity Associated with Short-Term
Exposure
3.1.1. Summary of Findings from the Previous Review
The majority of the S02 human clinical studies discussed in the 1982 AQCD for SOx evaluated
respiratory effects of S02 exposure in healthy adults, with some limited data from clinical studies of
adults with asthma. S02-related respiratory effects such as increased airway resistance and decreased
forced expiratory volume in 1 s (FEVi) were observed in healthy individuals at concentrations > 1.0-5.0
ppm, and in asthmatics at concentrations <1.0 ppm. The 1986 Second Addendum (U.S. EPA, 1986b) and
1994 Supplement to the Second Addendum (U.S. EPA, 1994c) reviewed several additional controlled
studies involving both healthy and asthmatic individuals. In general, these studies found no pulmonary
effects of S02 exposure in healthy subjects exposed to concentrations <1.0 ppm (Bedi et al., 1984;
Folinsbee et al., 1985; Kulle et al., 1984; Stacy et al., 1983). However, in exposures of asthmatic adults,
respiratory effects were observed following short-term exposures (5-10 min) to levels <1.0 ppm (Balmes
et al., 1987; Horstman et al., 1986; Linn et al., 1987).
Only a few epidemiologic studies reviewed in the 1982 AQCD were useful in determining the
concentration-response relationship of respiratory health effects from short-term exposure to S02. The
most notable study was by Lawther et al. (1970), which examined the association between air pollution
and worsening health status in bronchitic patients residing in London, England. It was concluded in the
1982 AQCD that worsening of health status among chronic bronchitic patients was associated with daily
black smoke (BS) levels of 250-500 (ig/m3 in the presence of S02 levels in the range of 191-229 ppb. In
the 1986 Second Addendum, additional studies investigated morbidity associated with short-term
exposure to S02. The most relevant study was by Dockery et al. (1982), which examined pulmonary
function in school children in Steubenville, OH, as part of the Harvard Six Cities Study. This study found
that small but statistically significant reversible decrements in forced vital capacity (FVC) and forced
expiratory volume in 0.75 s (FEV0 75) were associated with increases in 24-h avg concentrations of total
suspended particles (TSP) at levels ranging up to 220-420 |_ig/m3 and S02 at levels ranging up to 107-
176 ppb. However, it was impossible to separate the relative contributions of TSP and S02, and no
threshold level for the observed effects could be discerned from the wide range of exposure levels.
Epidemiologic evidence for an association between S02 and respiratory morbidity, as indicated by
increased use of ED facilities or increased hospital admissions for respiratory diseases, was also reported
in the 1982 AQCD. Overall, these results suggested increased upper respiratory tract morbidity during
episodic marked elevations of PM or S02 (400-500 ppb), especially among older adults. The 1982 AQCD
further concluded that the studies reviewed provided essentially no evidence for an association between
asthma attacks and acute exposures at typical ambient PM or S02 levels in the U.S. (the mean annual avg
S02 concentrations from 1972 to 1977 was approximately 6 ppb, with 90th percentile values ranging from
15 to 20 ppb).
The 1982 AQCD reported numerous effects on the respiratory system in animals exposed to S02.
Effects were generally observed at levels exceeding those found in the ambient environment, and
included morphological changes, altered pulmonary function, lipid peroxidation, and changes in host lung
defenses. The immediate effect of acute S02 exposure in animals was increased pulmonary resistance to
airflow, a measure of bronchoconstriction. Bronchoconstriction was reported to be the most sensitive
indicator of lung function effects.
Collectively, the human clinical, epidemiologic, and animal toxicological, studies provided
biological plausibility and coherent evidence of an adverse effect of ambient S02 on respiratory health.
Since the 1982 AQCD, 1986 Second Addendum, and 1994 Supplement to the Second Addendum,
additional studies have been conducted to determine the relationship between short-term exposures to
3-2
-------
ambient S02 and adverse respiratory health effects, including respiratory symptoms, lung function, airway
inflammation, airway hyperresponsiveness (AHR), lung host defenses, and ED visits and hospitalizations
for respiratory causes. The epidemiologic, human clinical, and animal toxicological evidence on the
effects of S02 on these various endpoints are discussed below. The findings of the previous review are
integrated below with the current literature.
3.1.2. Potential Mode of Action for Respiratory Health Effects
The 1982 AQCD (U.S. EPA, 1982) gave background information on the biochemistry of S02,
chemical reactions of bisulfite (HS03 ), metabolism of S02, and the activating or inhibiting effects of
bisulfite on various enzymes. S02 readily dissolves in water, rapidly becoming hydrated to form sulfurous
acid, which at physiological pH substantially dissociates to form bisulfite and sulfite (S032 ) ions. In vitro
studies have shown that S02 and/or bisulfite readily react with nucleic acids, proteins, lipids, and other
classes of biomolecules. Bisulfite participates in three important types of reactions with biomolecules:
sulfonation (sulfitolysis), autooxidation with generation of free radicals, and addition to cytosine.
Products of sulfonation reactions have been shown to be long-lived in vivo and may be highly reactive.
Products of autooxidation may be responsible for the initiation of lipid peroxidation, which, among other
effects, could damage plasma membranes. In addition, bisulfite can react with nucleic acids to convert
cytosine to uracil, thus resulting in mutational events. A principal mechanism of detoxification of S02
(and sulfite/bisulfite) occurs through the enzymatic activity of sulfite oxidase, resulting in the production
of sulfate. Sulfite oxidase is a molybdenum-containing enzyme, and the 1982 AQCD noted that depleting
its activity in an animal model through a low-molybdenum diet supplemented with the competitive
inhibitor tungsten resulted in a significant lowering of the LD50 for intraperitoneally injected bisulfite. It
was also noted that while in vitro exposure to S02 or sulfite/bisulfite had been shown to either activate or
inhibit a variety of enzymes, no such effects had yet been demonstrated for in vivo exposure.
As discussed in the 1982 AQCD, the immediate effect of acute S02 exposure in animals is
bronchoconstriction. Reactions of S02 with respiratory tract fluids can result in the production of
bisulfite, sulfite, and a lowering of the pH, which may be involved in the bronchoconstrictive response. It
is now widely appreciated that bronchoconstriction following S02 exposure is mediated by
chemosensitive receptors in the tracheobronchial tree. Rapidly activating receptors (RARs) and sensory
C-fiber receptors found at all levels of the respiratory tract are sensitive to irritant gases such as S02
(Coleridge and Coleridge, 1994; Widdicombe, 2006). Activation of these vagal afferents stimulates
central nervous system reflexes resulting in bronchoconstriction, mucus secretion, mucosal vasodilation,
cough, apnea followed by rapid shallow breathing, and effects on the cardiovascular system such as
bradycardia and hypotension or hypertension (Coleridge and Coleridge, 1994; Widdicombe and Lee,
2001; Widdicombe 2003).
Early experiments demonstrated that S02-induced reflexes were mediated by cholinergic
parasympathetic pathways involving the vagus nerve and inhibited by atropine (Grunstein et al., 1977;
Nadel et al., 1965a; 1965b). Bronchoconstriction was found to involve smooth muscle contraction since
(3-adrenergic agonists such as isoproterenol reversed the effects (Nadel et al., 1965a; 1965b). Histamine
was also thought to be involved in S02-induced bronchoconstriction (U.S. EPA, 1982).
More recent experiments in animal models conducted since 1982 have demonstrated that both
cholinergic and noncholinergic mechanisms may be involved in S02-induced effects. In two studies
utilizing bilateral vagotomy, vagal afferents were found to mediate the immediate ventilatory responses to
S02 (Wang et al., 1996), but not the prolonged bronchoconstrictor response (Barthelemy et al., 1988).
Other studies showed that atropine failed to block S02-induced bronchoconstriction, and that a local axon
reflex resulting in C-fiber secretion of neuropeptides (i.e., neurogenic inflammation) was responsible for
the effect (Atzori et al., 1992; Hajj et al., 1996). Neurogenic inflammation has been shown to play a key
role in animal models of airway inflammatory disease (Groneberg et al., 2004).
3-3
-------
In humans, the mechanisms responsible for S02-induced bronchoconstriction are not fully
understood. In non-asthmatics, near complete attenuation of bronchoconstriction has been demonstrated
using the anticholinergic agents atropine and ipratropium bromide (Snashall and Baldwin, 1982; Tan et
al., 1982; Yildirim et al., 2005). However, in asthmatics, these same anticholinergic agents (Field et al.,
1996; Myers et al., 1986), as well as short- and long-acting ^-adrenergic agonists (Gong et al., 1996;
Linn et al., 1988), theophylline (Koenig et al., 1992), cromolyn sodium (Myers et al., 1986), nedocromil
sodium (Bigby and Boushey, 1993) and leukotriene receptor antagonists (Gong et al., 2001; Lazarus et al.,
1997) only partially blocked S02-induced bronchoconstriction (see Annex Table D-l, (U.S. EPA, 1994c).
That none of these therapies have been shown to completely attenuate the effects of S02 implies the
involvement of both parasympathetic pathways and inflammatory mediators in asthmatics. Strong
evidence of this was borne out in a study by Myers et al. (1986), in which asthmatic adults were exposed
to S02 following pretreatment with cromolyn sodium (a mast cell stabilizer), atropine (a muscarinic
receptor antagonist), and the two medications together. While both treatments individually provided some
protection against the bronchoconstrictive effects of S02, there was a much stronger and statistically
significant effect following concurrent administration of the two medications.
It has been proposed that inflammation contributes to the enhanced sensitivity to S02 seen in
asthmatics by altering autonomic responses (Tunnicliffe et al., 2001), enhancing mediator release (Tan et
al., 1982) and/or sensitizing C-fibers and RARs (Lee and Widdicombe, 2001). Whether local axon
reflexes also play a role in S02-induced bronchoconstriction in asthmatics is not known (Groneberg et al.,
2004; Lee and Widdicombe, 2001; Widdicombe, 2003). However, differences in respiratory tract
innervation between rodents and humans suggest that C-fiber mediated neurogenic inflammation may be
unimportant in humans (Groneberg et al., 2004; Widdicombe and Lee, 2001; 2003).
3.1.3. Respiratory Effects Associated with Peak (5-10 min) Exposure
S02-induced respiratory effects among exercising asthmatics are well-documented, and have been
consistently observed following peak exposures (defined here as 5-10 min exposures to relatively high
concentrations, e.g., 0.2-1.0 ppm) (Balmes et al., 1987; Bethel et al., 1985; Horstman et al., 1986; 1988;
Linnetal., 1984; 1987; 1988; 1990; Schachter et al., 1984; Sheppard et al., 1981). S02-induced
decrements in lung function have been observed in asthmatics at concentrations as low as 0.1 ppm when
S02 is administered via mouthpiece (Koenig et al., 1990; Sheppard et al., 1981). However, these
exposures cannot be directly compared to exposures occurring among freely breathing subjects as a larger
fraction of administered S02 reaches the tracheobronchial airways during oral breathing (see Section
2.7.1, Kirkpatrick et al., 1982; Linn et al., 1983a). Since the publication of the 1994 Supplement, several
additional human clinical studies have been published that provide supportive evidence of S02-induced
decrements in lung function and increases in respiratory symptoms among exercising asthmatics (see
Annex Table D-2). Descriptions of older studies were presented in the 1994 Supplement, and are not
described in great detail in this document. However, based in part on recent guidance from the American
Thoracic Society (ATS) regarding what constitutes an adverse health effect of air pollution (ATS, 2000),
key older studies described in the 1994 Supplement were reviewed and analyzed along with studies
published since 1994. In their official statement, the ATS concluded that an air pollution-induced shift in a
population distribution of a given health-related endpoint (e.g., lung function in asthmatic children)
should be considered adverse, even if this shift does not result in the immediate occurrence of illness in
any one individual in the population. The ATS also recommended that transient loss in lung function with
accompanying respiratory symptoms attributable to air pollution should be considered adverse. However,
it is important to note that symptom perception is highly variable among asthmatics even during severe
episodes of asthmatic bronchoconstriction. An asymptomatic decrease in lung function may pose a
significant health risk to asthmatic individuals as it is less likely that these individuals will seek treatment
(Eckert et al., 2004; Fritz et al., 2007). Therefore, whereas the conclusions in the 1994 Supplement were
3-4
-------
based on S02 exposure concentrations which resulted in large decrements in lung function along with
moderate to severe respiratory symptoms, the current review of data from human clinical studies focused
on moderate to large S02-induced decrements in lung function along with respiratory symptoms ranging
from mild (perceptible wheeze or chest tightness) to severe (breathing distress requiring the use of a
bronchodilator).
3.1.3.1. Respiratory Symptoms
The 1994 Supplement to the Second Addendum described in detail several studies that evaluated
respiratory symptoms following controlled human exposures to S02. Briefly, following 5-min exposures
to 0, 0.2, 0.4, and 0.6 ppm S02 during moderate to heavy levels of exercise (yE = 48 L/min), Linn et al.
(1983b) reported that the severity of respiratory symptoms (i.e., cough, chest tightness, throat irritation)
among asthmatics increased with increasing S02 concentration. Relative to clean air exposures, exposures
to S02 resulted in statistically significant increases in respiratory symptoms at concentrations of 0.4 and
0.6 ppm. In a subsequent study, Linn et al. (1987) observed a significant effect of S02 on respiratory
symptoms in asthmatics who were engaged in slightly lower levels of exercise (yE = 40 L/min) for a
duration of 10 min. Clear increases in respiratory symptoms were observed at concentrations of 0.6 ppm,
with 43% of asthmatic subjects experiencing S02-induced symptoms. Some evidence of S02-induced
increases in respiratory symptoms was also demonstrated at concentrations as low as 0.4 ppm, with 15%
of subjects experiencing symptoms (Smith, 1994). It was also observed that these symptoms abated < 1 h
after exposure. Balmes et al. (1987) reported that 7 out of 8 asthmatic adults developed respiratory
symptoms, including wheezing and chest tightness, following 3-min exposures to 0.5 ppm S02 via
mouthpiece during eucapnic hyperpnea (yE = 60 L/min).
Additional human clinical studies published since the 1994 Supplement to the Second Addendum
have provided support for previous conclusions regarding the effect of peak exposures to S02 on
respiratory symptoms. In a human clinical study of S02-sensitive asthmatics, Gong et al. (1995) reported
that respiratory symptoms (i.e., shortness of breath, wheeze, and chest tightness) increased with
increasing S02 concentration (0, 0.5, and 1.0 ppm S02) following exposures of 10 min with varying
levels of exercise. It was also observed that exposure to 0.5 ppm S02 during light exercise evoked a more
severe symptomatic response than heavy exercise in clean air. Trenga et al. (1999) observed a significant
correlation between decreases in FEVi and increases in respiratory symptoms following 10 min exposures
via mouthpiece to 0.5 ppm S02.
3.1.3.2. Lung Function
In controlled exposures of healthy human subjects to S02, respiratory effects including increased
respiration rates, decrements in peak flow, bronchoconstriction, and increased airway resistance have
been observed at concentrations > 1 ppm (Abe, 1967; Amdur et al., 1953; Andersen et al., 1974; Frank et
al., 1962; Lawther, 1955; 1975; Sim and Pattle, 1957; Snell and Luchsinger, 1969). S02-induced
decrements in lung function can be potentiated by increasing ventilation rate, either through eucapnic
hyperpnea or by performing exercise during exposure. This effect is likely due to an increased uptake of
S02 resulting from both the increase in minute ventilation as well as a shift from nasal breathing to
oronasal breathing.
It has been clearly established that subjects with asthma are more sensitive to the respiratory effects
of S02 exposure than healthy individuals without asthma. Asthmatic individuals exposed to S02
concentrations as low as 0.2-0.3 ppm for 5-10 min during exercise have been shown to experience
moderate or greater bronchoconstriction, measured as an increase in specific airway resistance (sRaw) of
> 100% or decrease in FEVi of > 15% after correction for exercise-induced responses in clean air (Bethel
etal., 1985; Linn et al., 1983b; 1984; 1987; 1988; 1990). It has been consistently demonstrated that these
3-5
-------
decrements in lung function are more pronounced following exposures to higher concentrations of S02
(0.4-0.6 ppm), with a greater fraction of asthmatics affected (Linn et al., 1983b; 1987; 1988; 1990;
Magnussen et al., 1990; Roger et al., 1985). Gong et al. (1995) demonstrated a concentration-response
relationship between S02 and lung function by exposing 14 unmedicated, S02-sensitive asthmatics to 0,
0.5, and 1 ppm S02 under 3 different levels of exercise. It was shown that increasing S02 concentration
had a greater effect on sRaw and FEVi than increasing exercise level. In some cases, bronchoconstrictive
responses to S02 can occur in as little as 2 min after the start of exposure (Balmes et al., 1987; Horstman
et al., 1988). S02-induced decrements in lung function appear to be transient, and the magnitude of effect
has not been observed to increase with repeat exposures. There is evidence of a diminished response to
S02 when repeat exposures (10 min) occur within 5 h of the initial exposure. However, when exposure to
S02 occurs during a 30-min period with continuous exercise, the response to S02 develops rapidly and is
maintained throughout the 30-min exposure (Kehrl et al., 1987; Linn et al., 1984; 1987). Although the
majority of human clinical studies have been conducted at 20-25°C and 70-85% relative humidity, there is
some evidence that the respiratory effects of S02 are exacerbated when exposure occurs in cold or dry
ambient conditions (Bethel et al., 1984; Linn et al., 1985b).
Since some of the studies involving asthmatic subjects have used change in sRaw as the endpoint
of interest while others measured changes in FEV, or both, a comparison of FEV, and sRaw based on
data from Linn et al. (1987, 1990) was provided in the 1994 Supplement to the Second Addendum. Based
on simple linear interpolation of the data from these two studies, a 100% increase in sRaw corresponded
to a 12 to 15% decrease in FEV, and a 200% increase in sRaw corresponded to a 25 to 30% decrease in
FEVJ.
One of the aims of the Linn et al. (1987) study was to determine how the intensity of response
varied with asthma severity or status. In this study, 24 normal, 21 atopic (but not asthmatic), 16 mild
asthmatic, and 24 moderate/severe asthmatic subjects were exposed to S02 concentrations between 0 and
0.6 ppm. While the moderate/severe asthmatics experienced greater decrements in lung function than
mild asthmatics following exposure to clean air during exercise, their increases in response to increasing
S02 concentrations were similar to those of the mild asthmatic group. Thus, it was concluded by the
authors that respiratory response to S02 was not strongly dependent on the clinical severity of asthma.
However, the apparent lack of correlation between S02 response and asthma severity should be
interpreted with caution, since the S02 response may have been attenuated by medication usage.
Classification of asthma severity in this study was based on medication use to control asthma. Individuals
who required regular medication to manage asthma were classified as "moderate/severe" asthmatics,
while asthmatic subjects who did not use medication between episodes were classified as "mild"
asthmatics. Three of the moderate/severe asthmatics were unable to withhold medication usage prior to
most exposures. Trenga et al. (1999) observed that 25 out of 47 adult asthmatics experienced a drop in
FEVi versus baseline of between 8 and 44% (mean = 17.2%) following a 10 min mouthpiece exposure to
0.5 ppm S02 during moderate exercise. However, severity of asthma, as defined by medication use, was
not shown to be a predictor of sensitivity to S02. In a study of medication-dependent moderate
asthmatics, Linn et al. (1990) found that normal treatment (typically regular use of a long-acting
bronchodilator) did not prevent the airway responses to S02 and exercise. However, S02-induced
bronchoconstriction was significantly reduced when normal medication was supplemented with
administration of a short-acting beta agonist immediately preceding exposure.
Quick-relief and long-term-control asthma medications have been shown to provide varying
degrees of protection against the bronchoconstrictive effect of S02 in mild and moderate asthmatics (see
Annex Table D-l; (U.S. EPA, 1994c; Gong et al., 1996; 2001; Lazarus et al., 1997; Linn et al., 1988;
1990; Myers et al., 1986). While no therapy has been shown to completely eliminate the respiratory
effects of S02 in asthmatics, some short- and long-acting asthma medications are capable of significantly
reducing S02-induced bronchoconstriction (Gong et al., 1996; 2001; Koenig et al., 1987; Linn et al.,
1990). However, asthma is often poorly controlled even among severe asthmatics due to inadequate drug
therapy or poor compliance among those who are on regular medication (Rabe et al., 2004). Mild
asthmatics, who constitute the majority of asthmatic individuals, are much less likely to use asthma
3-6
-------
medication than asthmatics with more severe disease (O'Byrne, 2007; Rabe et al., 2004). It is therefore
reasonable to conclude that all asthmatics, mild, moderate and severe, are at high risk of experiencing
adverse respiratory effects of S02 exposure.
One of the key studies discussed in the 1994 Supplement to the Second Addendum was by
Horstman et al. (1986). In this study, 27 asthmatic subjects were exposed to concentrations of S02
between 0 and 2 ppm S02 for 10 min on different days under exercising conditions (yE = 42 L/min). The
authors reported that for 22% of the subjects, the concentration of S02 needed to produce a doubling of
sRaw compared to clean air exposure [PC(S02)] was < 0.5 ppm, with 2 subjects (7.4%) experiencing
moderate decrements in lung function following exposure to concentrations of S02 at or below 0.3 ppm
(see Figure 3-1). For approximately 15% of the subjects, the PC(S02) was > 2 ppm, with approximately
35% of asthmatic subjects experiencing a doubling in sRaw versus clean air at < 0.6-ppm S02.
S02-induced decrements in lung function (increased sRaw and decreased FEVi) have frequently
been associated with increases in respiratory symptoms among asthmatics (Balmes et al., 1987; Gong et
al., 1995; Linn et al., 1983b; 1987; 1988; 1990). Linn et al. (1987) exposed 40 mild and moderate
asthmatics during 10 min periods of exercise to 0, 0.2, 0.4, and 0.6 ppm S02. The effect of S02 on lung
function and respiratory symptoms was assessed immediately following exposure, and the individual-
specific results have been made available to the U.S. EPA by the study authors (Smith, 1994). Following
exposure to 0.6 ppm S02 and after adjusting for effects of exercise in clean air, 21 of the 40 subjects
demonstrated moderate or greater decrements in lung function, defined as a >15% decrease in FEVi, a
>100% increase in sRaw, or both. Of these 21 responders, 14 (67%) also experienced mild to severe
respiratory symptoms (6 mild, 6 moderate, and 2 severe). In the same study, 14 asthmatics experienced
moderate or greater decrements in lung function at 0.4 ppm S02, 5 of whom (36%) also experienced mild
to moderate respiratory symptoms (2 mild, 3 moderate). Five asthmatics experienced moderate or greater
decrements in lung function at the lowest S02 concentration tested (0.2 ppm), with 1 of the 5 (20%) also
experiencing mild respiratory symptoms.
100-
— 75.
>»
u
c
m
>
3
£
3
o
50.
25 ¦
X
X
X
X
X*
//:
0.25
—, 1 r
0.5 0.75 1.0
2.0
PC(S02) (ppm)
5.0
10.0
Source: Horstman et al. (1986).
Figure 3-1. Distribution of individual airway sensitivity to SO2. Each data point represents the
value of PC(S02) for an individual subject. PC(S02) is defined as the provocative
concentration of SO2 causing a doubling of sRaw compared to clean air exposure.
3-7
-------
It has been proposed that, as in asthmatics, individuals with COPD may also be more susceptible to
S02-induced respiratory health effects. However, this group has not been extensively studied in human
clinical studies. Among a group of older adults with physician-diagnosed COPD, Linn et al. (1985a)
reported no significant effect on lung function following 15 min exposures to S02 at concentrations of 0.4
and 0.8 ppm. While it was concluded that older adults with COPD appear to be less sensitive to S02 when
compared with younger adult asthmatics, the authors suggested that the lack of response may have been
due in part to the very low levels of exercise used in the study (yE =18 L/min), which would result in a
lower dose of S02 reaching the lower airway. In contrast to studies with asthmatics, most of the subjects
in this study regularly used bronchodilators and were permitted their use up to 4 h prior to the study.
3.1.3.3. Airway Inflammation
A very limited number of human clinical studies have investigated the role of airway inflammation
in the asthmatic response following peak exposure to S02. Gong et al. (2001) observed an S02-induced
increase in sputum eosinophil counts in exercising asthmatics 2 h after a 10 min exposure to 0.75 ppm
S02. The results of this study provide some evidence that S02 may elicit an inflammatory response in the
airways of asthmatics which extends beyond the short time period typically associated with S02 effects.
3.1.3.4. Mixtures and Interactive Effects
The interaction of S02 with other common air pollutants or the sequential exposure of S02 after
prior exposure to another pollutant can potentially modify S02-induced respiratory effects. However, only
a few human clinical studies have looked at the interactive effects of coexisting ambient air pollutants. In
a human clinical study designed to simulate an ambient "acid summer haze," Linn et al. (1997) exposed
healthy and asthmatic children (9-12 years of age) for 4 h with intermittent exercise to a mixture of S02
(0.1 ppm), H2S04 (100 |ig/m3). and 03 (0.1 ppm). Compared with exposure to filtered air, exposure to the
air pollution mixture did not result in statistically significant changes in lung function or respiratory
symptoms.
In a human clinical study of asthmatic adolescents (12- to 16-years-old), Koenig et al. (1983)
evaluated changes in FEVi following a 10-min mouthpiece exposure during moderate exercise to
1 mg/m3 NaCl alone and in combination with 0.5 and 1.0 ppm S02. Statistically significant decreases of
15 and 23% were reported in FEVi following exposure to 1 mg/m3 NaCl in combination with 0.5 and
1.0 ppm S02, respectively. No significant changes in FEVi were observed between pre- and
post-exposure to 1 mg/m3 NaCl without S02. The effect observed in this study may be the result of the
presence of hygroscopic particles that can carry S02 deeper into the lung.
Koenig et al. (1990) also examined the effect of 15-min exposures to 0.1 ppm S02 via mouthpiece
in adolescent asthmatics engaged in moderate levels of exercise. Immediately preceding this exposure,
subjects were exposed for 45 min to 0.12 ppm 03 during intermittent moderate exercise. Subjects also
underwent two additional exposure sequences with the same exercise regimen: 15-min exposure to
0.1 ppm S02 following a 45-min exposure to clean air, and 15-min exposure to 0.12 ppm 03 following a
45-min exposure to 0.12 ppm 03. The authors found that the change in FEVi relative to baseline was
significantly different following the 03-S02 exposure (8% decrease) when compared to the change
following the air-S02 or 03-03 exposures (decreases of 3% and 2%, respectively). In a more recent study
using a mouthpiece exposure system, Trenga et al. (2001) reported that among adult asthmatics, exposure
to 03 (0.12 ppm for 45 min) resulted in a slight increase in lung function responses to S02 at a
concentration of 0.25 ppm (6.5% decrease in FEVi with pre-exposure to 03, compared with a 3.4%
decrease in FEVi with pre-exposure to filtered air). Hazucha and Bates (1975) demonstrated a synergistic
effect of concurrent exposure to S02 (0.37 ppm) and 03 (0.37 ppm) on lung function in healthy
asthmatics; however, no such effect was observed in a similar study conducted by Bedi et al. (1979).
3-8
-------
Jorres and Magnussen (1990) and Rubinstein et al. (1990) investigated the effects of a prior N02
exposure on S02-induced bronchoconstriction in asthmatic adults. While Jorres and Magnussen suggested
that prior exposure to N02 increased the responsiveness to S02, Rubinstein et al. did not find that N02
exacerbated the effects of S02. Linn et al. (1980) reported no difference in lung function or respiratory
symptoms among a group of exercising asthmatics exposed to both clean air and a combination of N02
(0.5 ppm) and S02 (0.3 ppm).
3.1.3.5. Summary of Evidence on the Effect of Peak Exposure on Respiratory Health
Collectively, evidence from earlier studies considered in the previous review, along with a limited
number of new human clinical studies, consistently indicates that with elevated ventilation rates a large
percentage of asthmatic individuals (up to 60%) experience moderate or greater decrements in lung
function, frequently accompanied by respiratory symptoms, following peak exposures to S02 at
concentrations of 0.4-0.6 ppm (Balmes et al., 1987; Gong et al., 1995; Horstman et al., 1986; Linn et al.,
1983b; 1987; 1988; 1990). S02-induced decrements in lung function have also been observed at lower
S02 concentrations (0.2-0.3 ppm) in a smaller fraction (-5-30%) of asthmatic subjects (Bethel et al.,
1985; Linn et al., 1987; 1988; 1990; Sheppard et al., 1981). At these concentrations, S02-induced
decrements in lung are less likely to be accompanied by respiratory symptoms (Linn et al., 1983b; 1987;
1988; 1990; Roger et al., 1985). These findings are consistent with the current understanding of the
potential modes of action for respiratory health as described in Section 3.1.2. Among asthmatics, both the
magnitude of S02-induced decrements in lung function and the percent of individuals affected have
consistently been shown to increase with increasing exposure to S02 concentrations between 0.2 and 1.0
ppm. This is summarized in Table 3-1 along with supporting evidence of S02-induced increases in
respiratory symptoms at various exposure concentrations. The table includes data from all studies where
individual data are presented or have been made available by the authors (Smith, 1994). This information
represents the response to S02 among groups of relatively healthy asthmatics and cannot necessarily be
extrapolated to the most sensitive asthmatics in the population who are likely more susceptible to the
respiratory effects of exposure to S02.
Although the vast majority of human clinical studies involving controlled exposure to S02 have
been conducted in adult asthmatics, there is a relatively strong body of evidence to suggest that
adolescents may experience many of the same respiratory effects at similar S02 exposure concentrations
(Koenig et al., 1981, 1983; 1987; 1988; 1990; 1992). It should be noted, however, that in all of these
studies involving adolescents, S02 was administered via inhalation through a mouthpiece rather than an
exposure chamber. This exposure technique bypasses nasal absorption of S02, likely resulting in a relative
increase of pulmonary S02 uptake (see Section 2.7.1) (Linn et al., 1983a).
3-9
-------
Table 3-1. Percentage of asthmatic adults in controlled human exposures experiencing SO2
induced decrements in lung function.
so2
Cone Exposure
(ppm) Duration
No. Ventilation
Subj (L/min)
Cumulative Percentage of
Responders (Number of Subjects)1
sRaw
> 100% > 200% > 300% t
FEV1
Lung >15o/o^ > 20% 4- >30%nU
Func
Study
Respiratory
Symptoms:
Supporting
Studies
0.2
0.25
0.3
10 min
10 min
5 min
5 min
0 min
0 min
0 min
0 min
0 min
40
40
-40
-40
sRaw
FEV1
5% (2)
13% (5)
0
5% (2)
0
3% (1)
Linn et al. (1987)2
Linn et al. (1987)
19
9
28
-50-60
-80-90
-40
sRaw
sRaw
sRaw
32% (6)
22% (2)
4% (1)
16% (3)
0
0
Bethel etal. (1985)
Bethel etal. (1985)
Roger et al. (1985)
20
21
20
21
-50
-50
-50
-50
sRaw
sRaw
FEVi
FEVi
10% (2)
33% (7)
15% (3)
24% (5)
5% (1)
10% (2)
0
14% (3)
5% (1)
0
0
10% (2)
Linn et al.
Linn et al.
Linn et al.
Linn et al.
990
990
Limited evidence of
S02-induced
increases in
respiratory symptoms
in some asthmatics:
Linn et al. (1983b;
1984; 1987; 1988;
1990), Schachteret al.
(1984)
0.4
0.5
0 min
0 min
i min
0 min
0 min
40
40
-40
-40
sRaw 23% (9)
FEVi 30% (12)
8% (3)
23% (9)
3% (1)
13% (5)
Linn et al.
Linn et al.
10
28
45
-50-60
-40
-30
sRaw 60% (6)
sRaw 18% (5)
sRaw 36% (16)
40% (4)
4% (1)
16% (7)
20% (2)
4% (1)
13% (6)
Bethel et al. (1983)
Roger et al. (1985)
Magnussen et al.
(1990)5
Stronger evidence
with some statistically
significant increases in
respiratory symptoms:
Balmes et al. (1987)5,
Gong et al. (1995),
Linn et al. (1983b;
1987), Roger et al.
(1985)
0.6
1.0
0 min
0 min
0 min
0 min
0 min
0 min
0 min
0 min
40
20
21
40
20
21
-40
-50
-50
-40
-50
-50
sRaw 35% (14)
sRaw 60% (12)
sRaw 62% (13)
FEVi
FEVi
FEVi
53% (21)
55% (11)
43% (9)
28% (11)
35% (7)
29% (6)
48% (19)
55% (11)
33% (7)
18% (7)
10% (2)
14% (3)
20% (8)
5% (1)
14% (3)
Linn et al.
Linn et al.
Linn et al.
Linn et al.
Linn et al.
Linn et al.
1987
1988
1990
1987
1988
1990
Clear and consistent
increases in SO2-
induced respiratory
symptoms: Linn et al.
(1983b; 1984; 1987;
1988; 1990), Gong et
al. (1995), Horstman
et al. (1988)
28
10
-40
-40
sRaw 50% (14)
sRaw 60% (6)
25% (7)
20% (2)
14% (4)
0
Roger et al. (1985)4
Kehrl et al. (1987)
1Data presented from all references from which individual data were available. Percentage of individuals who experienced greater than or equal to a 100, 200, or 300% increase in
specific airway resistance (sRaw), or a 15, 20, or 30% decrease in FEVi. Lung function decrements are adjusted for effects of exercise in clean air (calculated as the difference between
the percent change relative to baseline with exercise/S02 and the percent change relative to baseline with exercise/clean air). Quality control of data was performed separately by two
EPA staff scientists.
Responses of mild and moderate asthmatics reported in Linn et al. (1987) have been combined. Data reported only for the first 10 min period of exercise in the first round of exposures.
3Analysis includes data from only mild (Linn et al., 1988) and moderate (Linn et al., 1990) asthmatics who were not receiving supplemental medication.
4One subject was not exposed to 1.0 ppm due to excessive wheezing and chest tightness experienced at 0.5 ppm. For this subject, the values used for 0.5 ppm were also used for 1.0
ppm under the assumption that the response at 1.0 ppm would be equal to or greater than the response at 0.5 ppm.
indicates studies in which exposures were conducted using a mouthpiece rather than a chamber.
3-10
-------
3.1.4. Respiratory Effects Associated with Short-Term (> 1 h)
Exposure
3.1.4.1. Respiratory Symptoms
Epidemiologic studies have examined the association between ambient S02 concentrations and
respiratory symptoms in both adults and children. In air pollution field studies, respiratory symptoms are
usually assessed using questionnaire forms (or "daily diaries") completed by study subjects. Questions
address the daily experience of coughing, wheezing, shortness of breath (or difficulty breathing),
production of phlegm, and others.
Children
Epidemiologic studies on respiratory symptoms published since the last review are summarized in
Annex Table F-l; key studies are discussed in detail below.
Locations Pollutants
All 8 urban areas S02 only
S02 adjusting for 03
7 urban areas S02 only
S02 adjusting for 03 and N02
3 urban areas S02 only
S02 adjusting for 03 N02 and PM
i 1 1 r
0.90 1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.
Odds Ratio
Source: Mortimer et al. (2002).
Figure 3-2. Odds ratios (95% CI) for incidence of morning asthma symptoms of 846 asthmatic
children from the National Cooperative Inner-City Asthma Study. Effects associated
with a 20 ppb increase in 3-h avg SO2 with a lag of 1-2 day moving average are
presented. SO2 effect estimates from single- and multipollutant models are shown.
The strongest epidemiologic evidence for an association between respiratory symptoms and
exposure to ambient S02 comes from two large U.S. multicity studies (Mortimer et al., 2002; Schildcrout
et al., 2006). Mortimer et al. examined 846 asthmatic children from eight U.S. urban areas in the National
Cooperative Inner-City Asthma Study (NCICAS) for summertime air pollution-related respiratory
3-11
-------
symptoms. Median 3-h avg S02 (8 to 11 a.m.) levels ranged from 17 ppb in Detroit, MI to 37 ppb in East
Harlem, NY. Morning symptoms were found to be most strongly associated with an avg of a 1- to 2-day
lag of S02 concentrations. In multipollutant models with 03 and N02 (measured in seven cities), the S02
association remained robust (see Figure 3-2). When PMi0 was also included in the multipollutant models,
the S02 effect estimate decreased only slightly; however, it became nonsignificant, possibly due to
reduced statistical power (only three of eight cities were included in the analysis adjusting for PMi0) or
collinearity resulting from adjustment of multiple pollutants (in addition to PM10, 03 and N02 were also
adjusted for in this model).
In the Childhood Asthma Management Program (CAMP) study, the association between ambient
air pollution and asthma exacerbations in children (n = 990) from eight North American cities was
investigated (Schildcrout et al., 2006). S02 measurements were available in seven of the eight cities. The
median 24-h avg S02 concentrations ranged from 2.2 ppb (interquartile range [IQR]: 1.7, 3.1) in San
Diego, CAto 7.4 ppb (IQR: 5.3, 10.7) in St. Louis, MO. Results for the associations between asthma
symptoms and all pollutants are shown in Figure 3-3. Analyses indicate that although S02 was positively
related to increased risk of asthma symptoms at all lags, only the 3-day moving avg was statistically
significant. No associations were observed between S02 and rescue inhaler use. Stronger associations
were observed for CO and N02. The effect estimates appear to be slightly larger in joint-pollutant models
with CO or N02, particularly at a 2-day lag, but did not change much when PMi0 was jointly considered.
Pollutants
SO,
S0s and CO
SO, and NO,
SO, and PM
m
0
1
2
May moving sum
0
1
2
3-day moving sum
0
1
2
3-day moving sum
0
1
2
3-day moving sum
100 1.10 1.20
Odds Ratio
Source: Schildcrout et al. (2006).
Figure 3-3. Odds ratios (95% CI) for daily asthma symptoms of 990 asthmatic children from the
Childhood Asthma Management Program Study. Effects associated with a 10 ppb
increase in within-subject concentrations of 24-h avg SO2 are presented. Data
collected from November 1993 to September 1995 were used. Results from single- and
joint-pollutant models are shown.
3-12
-------
A longitudinal study of 1,844 schoolchildren during the summer from the Harvard Six Cities Study
suggested that the association between S02 and respiratory symptoms could be confounded by PMi0
(Schwartz et al., 1994). The median 24-h avg S02 concentration during this period was 4.1 ppb (10th-
90th percentile: 0.8, 17.9; max 81.9). S02 concentrations were found to be associated with cough
incidence and lower respiratory tract symptoms. Of the pollutants examined, PMi0 had the strongest
associations with respiratory symptoms. In two-pollutant models, the effect of PMi0 was found to be
robust to adjustment for other copollutants, while the effect of S02 was substantially reduced after
adjustment for PMi0. Because the PMi0 concentrations were correlated strongly to S02-derived sulfate
particles (r = 0.80), the diminution of the S02 effect estimate may indicate that for PMi0 dominated by
fine sulfate particles, PM10 has a slightly stronger association than S02. This study further investigated the
concentration-response function and observed a nonlinear relationship between S02 concentrations and
respiratory symptoms. Though an increasing trend was observed at concentrations as low as 10 ppb, no
statistically significant increase in the incidence of lower respiratory tract symptoms was seen until
concentration exceeded a 24-h avg S02 of 22 ppb.
In the Pollution Effects on Asthmatic Children in Europe (PEACE) study, a multicenter study of 14
cities across Europe, the effects of acute exposure to various pollutants including S02 on the respiratory
health of children with chronic respiratory symptoms (n = 2,010) was examined during the winter of
1993-1994 (Roemer et al., 1998). Mean 24-h avg S02 concentrations ranged from 1 ppb in the urban area
of Umea, Sweden, to 43 ppb in the urban area of Prague, Czech Republic. No associations were observed
between S02 and daily prevalence of respiratory symptoms or bronchodilator use at any of the single- and
multiday lags considered. In addition, no associations were observed for any of the other pollutants
examined. It should be noted that during the study period, there were only two major air pollution
episodes, at the beginning and end of the study period. In the epidemiologic model, the control for time
trend was accomplished through the use of linear and quadratic terms. Given the timing of the air
pollution episodes, the quadratic trend term would have removed most of the air pollution effect.
Other studies that participated in the PEACE study and analyzed results for longer periods of time
have observed statistically significant associations between S02 and respiratory symptoms in children.
Van der Zee et al. (1999) looked at the association between respiratory symptoms and S02 in 7- to 11-
year-old children (n = 633) with and without chronic respiratory symptoms in the Netherlands. Significant
associations with lower respiratory tract symptoms and increased bronchodilator use were observed for
S02, as well as PMi0, BS, and sulfate, in symptomatic children living in urban areas (n = 142). In a two-
pollutant model with PMi0, the results were robust for bronchodilator use, but slightly reduced for lower
respiratory tract symptoms. A subgroup analysis of this cohort was conducted by Boezen et al. (1999).
They examined 7- to 11-year-old children (n = 459) in the Netherlands and tested them for AHR and
atopy. It was hypothesized that children with AHR, as measured using a methacholine (MCh) challenge,
and atopy, indicated by raised serum total IgE (> 60 kU/L, the median value), may be susceptible to the
effects of air pollution. One of the strengths of this study was the use of AHR and serum IgE
concentration to indicate susceptibility; these measurements would be less prone to error than self-
reported chronic respiratory symptoms. A total of 121 children were found to have AHR and relatively
high serum total IgE; 67 had AHR and relatively low serum total IgE, 104 had no AHR but had a
relatively high serum total IgE concentration, and 167 were found to have neither AHR nor relatively high
serum total IgE. For the subset of children with relatively low serum total IgE with or without AHR, no
associations were observed between S02 and any respiratory symptoms. However, for children with
relatively high serum total IgE either with or without AHR, the prevalence of lower respiratory tract
symptoms increased with increasing S02 concentrations. For children with AHR and relatively high
serum total IgE, the OR for the prevalence of lower respiratory tract symptoms was 1.70 (95% CI: 1.26,
2.29) with a 5-day moving avg for every 10 ppb increase in S02. For children without AHR but with
relatively high serum total IgE, the OR was 1.82 (95% CI: 1.33, 2.50) with a 5-day moving avg.
Additional studies have examined the relationship between respiratory symptoms and ambient S02
concentrations and generally found positive associations, including two U.S. studies (Delfino et al.,
2003a; Neas et al., 1995) and several European studies (Hoek and Brunekreef, 1995; Peters et al., 1996a;
3-13
-------
Roemer et al., 1993; Segala et al., 1998; Timonen and Pekkanen, 1997). However, some did not find a
consistent association between respiratory symptoms and S02 concentrations (Hoek and Brunekreef,
1993; Romieu et al., 1996). None of these studies examined possible confounding of the S02 effect by
copollutants.
Reference
Location
Population
Schwartz et al. (1994) 6 U.S. Cities Children (n = 300)
Neas et al. (1995) Uniontown, PA Non-asthmatics (n = 98)
*Ward et al. (2002) Birmingham and Children (n = 162)
Sandwell, U.K.
*Van der Zee
etal (1999)
5 Urban areas in With chronic respiratory
the Netherlands symptoms (n = 142)
Lag
0
1-4
0
1
0-6
0
1
0-4
Hoek and The Netherlands Children (n = 1,078) 0
Brunekreef (1994)
1
0-6
*Ward et al. (2002) Birmingham and Children (n = 162)
Sandwell, U.K.
Segala et al. (1998) Paris, France Mild asthmatics (n = 43)
Romieu et al. (1996) N. Mexico City, Mild asthmatics (n = 71) Not
Mexico stated
| Summer |
All year |
I
0.4
I
0.6
I I I I I I I
0.8 1.0 1.2 1.4 1.6 2.0
Odds Ratio
*Note that van der Zee et al. (1999) and Ward et al. (2002a) presented results for prevalence of cough.
Figure 3-4. Odds ratios (95% CI) for incidence of cough among children, grouped by season. For
single-day lag models, current day and/or previous day SO2 effects are shown, except
for Segala et al. (1998b), which only presented results for a 3-day lag. Multiday lag
models represent the effect of the mean concentration from the range of days noted.
Risk estimates are standardized per 10 ppb increase in 24-h avg SO2 level.
Figure 3-4 and Figure 3-5 present the odds ratios for S02-related cough, and lower respiratory tract
or asthma symptoms, respectively, among children from epidemiologic studies published since the last
NAAQS review. All studies that reported quantitative results with relevant data are included in the figure.
The results for cough were somewhat variable with wide confidence intervals, as shown in Figure 3-4.
3-14
-------
The studies conducted in the summer generally indicate increased risk of cough from exposure to SO2. A
more consistent effect of SO2 is observed on lower respiratory tract or asthma symptoms (Figure 3-5).
Although there is some variability in the individual effect estimates, the majority of the odds ratios appear
to be greater than one. As was the case with cough, stronger associations with lower respiratory tract or
asthma symptoms were observed in the summer, as opposed to the winter. There was some variability
among the different lags of exposure; however, effects were generally observed with current day or
previous day exposure and, in some cases, with a distributed lag of 2 to 3 days.
Reference
Location
Population
Mortimer et al. (2002) 8 U.S. Cities
"Schildcrout et al. (2006) 7 U.S. Cities
Romieu et al. (1996) N. Mexico City,
Mexico
Lag
Asthmatics (n = 846) 1-2
Schwartz et al. (1994) 6 U.S. Cities Children (n = 300) 1
*Van der Zee et al. (1999) 5 Urban areas With chronic respiratory 0
in the Netherlands symptoms (n = 142)
Hoek and Brunekreef (1994) The Netherlands Children (n = 1,078) 0
1
Segala et al. (1998) Paris, France Mild asthmatics (n = 43) 0
Asthmatics (n = 881) 0
1
0-2
Mild asthmatics (n = 71) Not
stated
Summer |
All year ~|
~1—
0.8
i i i i i i
1.0 1.2 1.4 1.6 1.8 2.0 2.2
Odds Ratio
*Note that van der Zee et al. (1999) and Schildcrout et al. (2006) presented results for prevalence of lower respiratory tract or asthma symptoms.
Figure 3-5. Odds ratios (95% CI) for the incidence of lower respiratory tract or asthma symptoms
among children, grouped by season. Risk estimates are standardized per 10 ppb
increase in 24-h avg SO2 level. For single-day lag models, current day and/or previous
day SO2 effects are shown. Multiday lag models represent the effect of the moving
average from the range of days noted.
3-15
-------
Overall, recent epidemiologic studies provided evidence for an association between ambient S02
exposures and increased respiratory symptoms in children, particularly those with asthma or chronic
respiratory symptoms. Recent U.S. multicity studies observed significant associations between S02 and
respiratory symptoms at a median range of 17 to 37 ppb (75th percentile: -25 to 50) across cities for 3-h
avg S02 (NCICAS, Mortimer et al., 2002) and 2.2 to 7.4 ppb (90th percentile: 4.4 to 14.2) for 24-h avg
S02 (CAMP, Schildcrout et al., 2006). However, an earlier study that examined the concentration-
response function found that a statistically significant increase in the incidence of lower respiratory tract
symptoms was not observed until concentrations exceeded a 24-h avg S02 of 22 ppb, though an
increasing trend was observed at concentrations as low as 10 ppb (Harvard Six Cities Study, Schwartz et
al., 1994). In the limited number of studies that examined potential confounding by copollutants through
multipollutant models, the S02 effect was generally found to be robust after adjusting for PM and other
copollutants. More details of the literature published since the last review are found in Annex Table F-l.
Adults
Compared to the number of studies conducted with children, fewer epidemiologic studies were
performed that examined the effect of ambient S02 exposure on respiratory symptoms in adults. Most of
these studies focused on potentially susceptible populations, i.e., those with asthma or COPD. One of the
larger studies was conducted by van der Zee et al. (2000) in 50- to 70-year-old adults, with (n = 266) and
without (n = 223) chronic respiratory symptoms in the Netherlands. In adults both with and without
chronic respiratory symptoms, no consistent associations were observed between S02 levels and respira-
tory symptoms or medication use. A subgroup analysis of this cohort examining S02-related respiratory
symptoms in individuals with AHR and atopy was conducted by Boezen et al. (2005). The subgroup of
individuals with elevated serum total IgE, both with (n = 48) and without (n = 112) AHR, were found to
be more susceptible to air pollutants when contrasted with those who did not have elevated serum total
IgE (n = 167). Significant associations were observed between previous-day 24-h avg S02 concentrations
and the prevalence of upper respiratory tract symptoms in those with elevated serum total IgE. Stratified
analyses by gender indicated that, among those with AHR and elevated IgE, only males (n = 25) were at a
higher risk for respiratory symptoms. The OR for these males was 3.54 (95% CI: 1.79, 7.07) increase in
24-h avg S02 for a 5-day moving avg, compared with 1.05 (95% CI: 0.59, 1.91) for the females.
Studies by Desqueyroux et al. (2002a; 2002b) examined the association between air pollution and
respiratory symptoms in other potentially susceptible populations, i.e., those with severe asthma (n = 60,
mean age 55 years) and COPD (n = 39, mean age 67 years), in Paris, France. The mean 24-h avg S02
concentration was 3 ppb (range: 1, 10) in the summer and 7 ppb (range: 1, 31) in the winter. No
associations were observed between S02 concentrations and the incidence of asthma attacks or episodes
of symptom exacerbation in severe asthmatics or individuals with COPD. Among the pollutants
considered, 03 was found to have the strongest effect in these studies.
Several other European studies did observe an association between ambient S02 concentrations and
respiratory symptoms in adults with asthma or chronic bronchitis (Higgins et al., 1995; Neukirch et al.,
1998; Peters et al., 1996a). Only one of these studies examined possible confounding of the association by
copollutants. Higgins et al. (1995) examined the effect of summertime air pollutant exposure on respira-
tory symptoms in 62 adults with either asthma, COPD, or both. The max 24-h avg S02 level was 45 ppb.
An association was observed between S02 and symptoms of wheeze, and it remained robust after
adjustment for 03 and N02. The effects of PM were not examined in this study.
Results from the epidemiologic studies examining the association between S02 and respiratory
symptoms in adults were generally mixed, with some showing positive associations and others finding no
relationship at current ambient levels. There was limited epidemiologic evidence which suggested that
atopic adults may be at increased risk for S02-induced respiratory symptoms. The overall epidemiologic
evidence that 24-h avg S02 exposures at or near ambient concentrations has an effect on adults is incon-
clusive. However, as discussed in Section 3.1.3.1, human clinical studies have observed an effect of peak
exposures to S02 on respiratory symptoms, particularly among S02-sensitive asthmatics, with 10 min
3-16
-------
exposures to S02 concentrations as low as 0.2-0.6 ppm under exercise conditions. These effects in clinical
studies are at levels that have sometimes been measured in ambient air for similarly short-time durations.
3.1.4.2. Lung Function
The 1982 AQCD reported bronchoconstriction, indicated by increased pulmonary resistance, as the
most sensitive indicator of lung function effects of acute S02 exposure, based on the observations of
increased pulmonary resistance in guinea pigs that were acutely exposed to 0.16 ppm S02. Since then,
only a few animal toxicological studies have measured lung function at or near ambient levels of S02.
These recent studies, and those using higher concentrations of S02, are summarized in Annex Table E-l.
Increased pulmonary resistance and decreased dynamic compliance were observed in conscious guinea
pigs exposed to 1 ppm S02 for 1 h (Amdur et al., 1983). Effects were seen immediately after exposure
and were not present 1 h post-exposure. No changes in tidal volume, minute volume or breathing
frequency were found. These same investigators also exposed guinea pigs to 1 ppm S02 for 3 h/day for 6
days (Conner et al., 1985). No changes were observed in pulmonary function or respiratory parameters,
i.e., diffusing capacity for CO, functional reserve capacity, vital capacity, total lung capacity, respiratory
frequency, tidal volume, pulmonary resistance or pulmonary compliance. In another study, Barthelemy et
al. (1988) demonstrated a 16% increase in airway resistance following a 45-min exposure of anesthetized
rabbits to 0.5 ppm S02 via an endotracheal tube. This latter exposure is more relevant to oronasal than
nasal breathing.
Children
Most epidemiologic studies discussed in the previous section on respiratory symptoms also
examined lung function. In these studies self-administered PEF meters were primarily used to assess lung
function. PEF follows a circadian rhythm, with the highest values found during the afternoon and lowest
values during the night and early morning (Borsboom et al., 1999). Therefore, these studies generally
analyzed PEF data stratified by time of day. The epidemiologic studies on lung function are summarized
in Annex Table F-l.
Mortimer et al. (2002) examined 846 asthmatic children from eight U.S. urban areas in the
NCICAS for changes in PEF related to air pollution. The mean 3-h avg S02 was 22 ppb across the eight
cities during the study period of June through August, 1993. No associations were observed between S02
concentrations and morning or evening PEF. Of all the pollutants examined, including PMi0, 03, and N02,
only 03 was associated with changes in morning PEF.
In another U.S. study (Neas et al. 1995), 83 children from Uniontown, PA reported twice-daily PEF
measurements during the summer of 1990. The mean daytime 12-h avg S02 concentration was 14.5 ppb
(max 44.9). No associations were observed between daytime 12-h avg S02 concentrations and mean
deviation in evening PEF, even after concentrations were weighted by the proportion of hours spent
outdoors during the prior 12-h. Statistically significant associations were observed for 03, total sulfate
particles, and particle-strong acidity.
A study by van der Zee et al. (1999) observed associations between ambient S02 concentrations
and daily PEF measurements in 7- to 11-year-old children (n = 142) with chronic respiratory symptoms
living in urban areas of the Netherlands. The OR for a > 10% decrement in evening PEF per 10 ppb
increase in 24-h avg S02 was 1.20 (95% CI: 0.97, 1.47) with same-day exposure. A greater effect was
observed at a 2-day lag, OR = 1.40 (95% CI: 1.18, 1.67), and this effect remained robust in a two-
pollutant model with PMi0, OR = 1.34 (95% CI: 1.08, 1.64).
Multipollutant analyses also were conducted in a study by Chen et al. (1999), which examined the
effects of short-term exposure to air pollution on the pulmonary function of 895 children, ages 8 to 13
years, in three communities in Taiwan. The daytime 1-h max S02 the day before spirometry ranged from
0 to 72.4 ppb. In a single-pollutant model, 1-h max S02 concentration at a 2-day lag was significantly
3-17
-------
associated with FVC, -50.80 mL (95% CI: -97.06, -4.54), or a 2.6% decline, per 40 ppb 1-h max S02.
However, in multipollutant models, authors noted that only 03 remained significantly associated with
FVC and FEVi. Effect estimates for S02 in multipollutant models were not provided.
While additional studies have observed associations between ambient S02 concentrations and
changes in lung function in children (e.g., Hoek and Brunekreef, 1993; Peters et al., 1996a; Roemer et al.,
1993; Segala et al., 1998; Timonen and Pekkanen, 1997), several other studies did not find a significant
association between S02 and lung function parameters (e.g., Delfino et al., 2003a; Peacock et al., 2003;
Romieu et al., 1996). In addition, within studies that did observe an association, the correlations between
S02 and other pollutants, particularly PM indices, were high [for example, r = 0.8-0.9 with TSP] (Peters
et al., 1996a) making it difficult to separate the contributions of individual pollutants.
In conclusion, while some epidemiologic studies observed a positive association between short-
term S02 exposure and lung function in children, several others, including a large U.S. multicity study,
did not observe such an association. The limited evaluation of potential confounding by copollutants also
indicated mixed results. Overall, the evidence is insufficient to conclude that short-term exposure to
ambient S02 has an independent effect on lung function in children.
Adults
Only a limited number of epidemiologic studies have been conducted examining the association
between ambient S02 concentrations and lung function in adults, as in the case of respiratory symptoms.
In a cross-sectional survey, Xu et al. (1991) investigated the effects of indoor and outdoor air pollutants
on the respiratory health of 1,140 adults (aged 40 to 69 years) living in residential, industrial, and
suburban areas of Beijing, China. The annual mean concentrations of S02 in residential, industrial, and
suburban areas from 1981 to 1985 were 49 ppb, 22 ppb, and 7 ppb, respectively. Log-transformed S02
and TSP were significantly associated with reductions in FEVi and FVC. The authors cautioned that since
S02 and TSP concentrations were strongly correlated, the effect of S02 could not be separated.
Van der Zee et al. (2000) observed an association between S02 and morning PEF in 50- to 70-year-
old adults (n = 138) with chronic respiratory symptoms living in urban areas of the Netherlands. No
associations were observed with evening PEF. The OR for a > 20% decrement in PEF was 1.21 (95% CI:
0.76, 1.92) per 10 ppb increase in 24-h avg S02 with same-day exposure and 1.56 (95% CI: 1.02, 2.39) at
a 1-day lag. No associations were observed for a > 10% decrement in PEF. The authors hypothesized that
while S02 level did not have much effect on PEF in most subjects, a small subgroup of individuals
experienced fairly large PEF decrements when S02 levels were high. No multipollutant analyses were
conducted.
Higgins et al. (1995) examined the association between pulmonary function and air pollution in 75
adults with either asthma, COPD, or both. Exposure to S02 was associated with increased variation in
PEF, but not with mean or minimum PEF. The S02 effects on PEF variation were robust to adjustment for
03 and N02. Effects of PM were not considered. Neukirch et al. (1998) also observed associations
between lung function and S02 concentrations in a study of asthmatic adults in Paris, France; however,
significant associations were found for all pollutants examined, including BS, PMi3, and N02. Other
epidemiologic studies observed only weak relationships between ambient S02 concentrations and lung
function in adults (Peters et al., 1996a; Taggart et al., 1996).
Evidence from human clinical studies clearly indicates that asthmatic individuals experience
moderate or greater decrements in lung function, as well as increased respiratory symptoms, following
peak exposure (5-10 min) to S02 (Balmes et al., 1987; Gong et al., 1995; Horstman et al., 1986; Linn et
al., 1983b; 1987). These effects were seen at peak concentrations as low as 0.2-0.6 ppm. In a human
clinical study by Tunnicliffe et al. (2003) that evaluated the effect of 1-h exposures to 0.2 ppm S02 in
resting healthy and asthmatic subjects, no significant changes were observed in lung function as measured
by FEVi, FVC, and maximal midexpiratory flow (MMEF). However, these results are not unexpected
given that subjects were exposed while at rest.
3-18
-------
In summary, the epidemiologic studies examining adults do not provide strong evidence for an
association between short-term exposure to ambient S02 and lung function. While some studies did
observe associations between S02 exposure and decrements in lung function parameters, the strong
correlation between S02 and various copollutants in most studies, and the lack of evidence evaluating
potential confounding by copollutants, limit interpretation of independent effects of S02 on lung function.
3.1.4.3. Airway Inflammation
The animal toxicological studies on airway inflammation are summarized in Annex Table E-2. In
one study, guinea pigs were exposed to 1 ppm S02 for 3 h/day for 1-5 days and bronchoalveolar lavage
was performed (Conner et al., 1989). No change in numbers of total cells or neutrophils in lavage fluid
was observed over this time period. However, in two models of allergic sensitization, S02 exposure
increased airway inflammation. In one study Park et al. (2001a), guinea pigs were exposed to 0.1 ppm
S02 for 5 h/day for 5 days and sensitized with 0.1% ovalbumin aerosols for 45 min on days 3-5. One
week later, animals were subjected to bronchial challenge with 1.0% ovalbumin and bronchoalveolar
lavage and histopathologic examination were performed 24 h later. Results demonstrated increased
numbers of eosinophils in lavage fluid, and an infiltration of inflammatory cells, bronchiolar epithelial
cell damage and plugging of the airway lumen with mucus and cells in the bronchial tissues of animals
treated with both S02 and ovalbumin, but not in animals treated with ovalbumin or S02 alone. In a second
study, rats which were sensitized and challenged with ovalbumin and exposed to 2 ppm S02 for 1 h/day
for 7 days had an increased number of inflammatory cells in bronchoalveolar lavage fluid and an
enhanced histopathological response compared with those treated with ovalbumin or S02 alone (Li et al.,
2007). Similarly, ICAM-1, a protein involved in regulating inflammation, and MUC5AC, a mucin
protein, were upregulated in lungs and trachea to a greater extent in rats treated with ovalbumin and S02
than those treated with ovalbumin or S02 alone. Further experiments are required to determine whether
exposure to near ambient S02 also enhance inflammatory responses in non-allergic and allergic rats.
In a human clinical study, Tunnicliffe et al. (2003) measured levels of exhaled NO (eNO) in
asthmatic and healthy adult subjects, before and after 1-h exposure to 0.2 ppm S02 under resting
conditions. While eNO concentrations were higher in the asthmatic than in healthy subjects, no significant
difference was observed between pre- and postexposure in either group.
One epidemiologic study by Adamkiewicz et al. (2004) examined eNO as a biological marker for
inflammation in 29 older adults (median age 70.7 years) in Steubenville, OH. The mean 24-h avg S02
concentration was 12.5 ppb (IQR 11.5). The authors reported that, while significant and robust
associations were observed between increased daily levels of fine PM (PM2 5) and increased eNO, no
associations were observed with any of the other pollutants examined, including S02, N02, and 03.
Overall, the very limited human clinical and epidemiologic evidence is insufficient to conclude that
exposure to S02 at current ambient concentrations is associated with inflammation in the airway.
However, toxicological studies indicated that repeated exposures to S02> at concentrations as low as 0.1
ppm in guinea pigs, may exacerbate inflammatory responses in allergic animals.
3.1.4.4. Airway Hyperresponsiveness and Allergic Sensitization
The toxicological studies describing S02-induced effects on airway responsiveness and allergic
sensitization in guinea pigs, rabbits, dogs, and sheep are summarized in Annex Table E-3. In one study,
Amdur et al. (1988) exposed guinea pigs for 1 h to 1 ppm S02 and measured airway responsiveness to
acetylcholine 2 h later. No AHR was observed. In a second study, Douglas et al., (1994) found no AHR
following a histamine challenge 24 h after exposure of rabbits to 5 ppm S02 for 2 h. In a third study,
exposure of sheep for 4 h to 5 ppm S02 failed to result in AHR following carbachol (Abraham et al.,
1981). In a fourth study, a 5-min exposure to 30 ppm but not to 10 ppm S02 resulted in AHR in dogs
3-19
-------
challenged with methacholine (Lewis and Kirchner, 1984). Collectively, these results show that a single
exposure to S02 at a concentration of 10 ppm or less failed to induce AHR following challenge in 4
different animal models.
However, two other studies demonstrated increased airway responsiveness in guinea pigs exposed
repeatedly to S02 and allergen. Riedel et al. (1988) studied the effect of S02 exposure on local bronchial
sensitization to inhaled antigen. Guinea pigs were exposed by inhalation to 0.1, 4.3 and 16.6 ppm S02 for
8 h/d for 5 days. During the last 3 days, S02 exposure was followed by exposure to nebulized ovalbumin
for 45 min. Following bronchial provocation with inhaled ovalbumin (0.1%) one week later, airway
obstruction was measured by whole body plethysmography. In addition, specific antibodies against
ovalbumin were measured in serum and bronchaolveolar fluids. Results show significantly higher
bronchial obstruction in animals exposed to S02 (at all concentration levels) with ovalbumin compared
with animals exposed only to ovalbumin. In addition, significant increases in anti-ovalbumin IgG
antibodies were detected in bronchoalveolar lavage fluid of animals exposed to 0.1, 4.3 and 16.6 ppm S02
and in serum from animals exposed to 4.3 and 16.6 ppm S02 compared with controls exposed only to
ovalbumin. These results demonstrate that repeated exposure to S02 can enhance allergic sensitization in
the guinea pig at a concentration as low as 0.1 ppm. In a second study, guinea pigs were exposed to 0.1
ppm S02 for 5 h/day for 5 days and sensitized with 0.1% ovalbumin aerosols for 45 min on days 3 to 5
(Park et al., 2001a). One week later, animals were subjected to bronchial challenge with 1.0% ovalbumin
and lung function was evaluated 24 h later by whole body plethysmography. Results demonstrated a
significant increase in enhanced pause (Penh), a measure of airway obstruction, in animals exposed to S02
with ovalbumin but not in animals treated with ovalbumin or S02 alone. These experiments also indicate
that near ambient levels of S02 may play a role in exacerbating allergic responses in the guinea pig.
In a human clinical study evaluating S02-induced AHR to an inhaled allergen (house dust mite),
Devalia et al. (1994) found that neither S02 (0.2 ppm) nor N02 (0.4 ppm) enhanced airway response to
the allergen in asthmatic individuals. However, following concurrent exposure (6 h) to S02 and N02
while at rest, subjects did exhibit an increased response to the inhaled allergen. In a subsequent study,
Rusznak et al. (1996) confirmed these findings and observed that the combination of S02 and N02
enhanced airway response to house dust mite antigen up to 48 hours post-exposure.
A limited number of epidemiologic studies also examined the association between S02 and AHR.
These studies are summarized in Annex Table F-l. Soyseth et al. (1995a) investigated the effect of short-
term exposure to S02 and fluoride on the number of capillary blood eosinophils, and the prevalence of
AHR in schoolchildren, ages 7 to 13 yr, (n = 620) from two regions in Norway, a valley containing an
S02-emitting aluminum smelter (Ardal) and a similar but nonindustrialized valley (Laerdal). The median
24-h avg S02 concentration was 8 ppb (10th-90th percentile: 1, 33) in the exposed area and 1 ppb (10th-
90th percentile: 0, 4) in the nonindustrialized valley. The mean number of eosinophils was significantly
greater in children living near the aluminum smelter compared to the nonindustrialized area. However,
within children in the exposed area, a negative concentration-response relationship was observed between
mean eosinophils and previous-day 24-h avg S02. The observed association between S02 and eosinophils
was limited to atopic children. In children living in the exposed area, a statistically significant positive
association was observed between prevalence of AHR and previous-day 24-h avg S02 concentrations.
Similar associations were observed for fluoride. The authors hypothesized that recent exposure to S02
may have induced changes in the airway leading to AHR, in addition to recruitment of eosinophils to the
airways in atopic subjects. Exposure to PM was not assessed in this study.
A study by Taggart et al. (1996) examined the effect of summertime air pollution levels in
northwestern England on AHR in nonsmoking, asthmatic subjects (n = 38) aged 18 to 70 years who were
determined to be MCh reactors. Subjects were tested multiple times, for a total of 109 evaluable challenge
tests, with a range of two to four tests per subject. The max 24-h avg S02 concentration during the study
period was 40 ppb. This study reported that 24-h avg S02 levels were marginally associated with a
decreased dose of MCh required for a 20% drop in the postsaline FEVi (PD20FEVi).
In summary, the animal toxicological evidence suggests that repeated exposures to S02 at
concentrations as low as 0.1 ppm in guinea pigs can exacerbate AHR following allergic sensitization. Two
3-20
-------
recent human clinical studies have demonstrated an increase in airway response to an inhaled allergen in
asthmatic subjects following exposures to a combination of 0.2-ppm S02 and 0.4-ppm N02. These
findings are consistent with the very limited epidemiologic evidence that suggests that exposure to S02
may lead to AHR in atopic children and asthmatic adults.
3.1.4.5. Respiratory Illness-Related Absences
An additional concern has been the potential for S02 exposure to enhance susceptibility to, or the
severity of illness resulting from respiratory infections, especially in children. School absenteeism is used
as an indicator of morbidity in children caused by acute conditions. Respiratory conditions are the most
frequent cause, particularly influenza and the common childhood infectious diseases. Studies discussed in
this section are summarized in Annex Table F-1. Park et al. (2002) examined the association between air
pollution and school absenteeism in 1,264 first- to sixth-grade students attending school in Seoul, Korea.
The study period extended from March 1996 to December 1999, with a mean 24-h avg S02 concentration
of 9.19 ppb (SD 4.61). Note that analyses were performed using Poisson Generalized Additive Model
(GAM) with default convergence criteria. Same-day S02 concentrations were positively associated with
illness-related absences (16% excess risk [95% CI: 13, 22] per 10 ppb increase in 24-h avg S02), but
inversely associated with non-illness-related absences (9% decrease [95% CI: 2, 15]). PMi0 and 03
concentrations also were positively associated with illness-related absences. In two-pollutant models
containing S02 and either PM10 or 03, the S02 estimates were robust.
A study by Ponka (1990) observed results that were consistent with those from the Park et al.
(2002) study. Ponka found that absenteeism due to febrile illnesses among children in day care centers
and schools, and in adults was significantly higher on days of higher S02 concentrations (>8.1 ppb
weekly mean of 1-h avg), compared to days of lower S02 concentrations in Helsinki, Finland. In addition,
on days of higher S02 concentrations, the mean weekly number of cases of upper respiratory tract
infections and tonsillitis reported from health centers increased. Temperature, but not N02, was also found
to be associated with febrile illnesses and respiratory tract infections. From these epidemiologic studies, it
is unknown whether S02 increases susceptibility to infection or whether its presence exacerbates
morbidity following infection.
Pino et al. (2004) examined the association between air pollution and respiratory illnesses in a
cohort of 504 infants recruited at 4 months of age from primary health care units in southeastern Santiago,
Chile. The infants were followed through the first year of life. The mean 24-h avg S02 concentration was
11.6 ppb (5th-95th percentile: 3.0, 29.0). The most frequent diagnosis during follow-up was wheezing
bronchitis. No associations were observed between current-day or previous-day S02 and wheezing
bronchitis, but with a 7-day lag, a 21% (95% CI: 8, 39) excess risk in wheezing bronchitis was observed
per 10 ppb increase in 24-h avg S02. However, it should be noted that stronger associations were
observed with PM2 5, which was well-correlated with S02 (r = 0.73). These epidemiologic studies are
summarized in Annex Table F-l.
To summarize, very few studies have examined the association between ambient S02
concentrations and absences from school or work as a result of respiratory illnesses. The limited evidence
indicates a possible association between exposure to S02 concentrations and increased respiratory
illnesses, particularly among young children; however, this association was also seen with PM, which was
correlated with S02.
3.1.4.6. Emergency Department Visits and Hospitalizations for Respiratory Diseases
Total respiratory causes for ED visits and hospital admissions typically include asthma, bronchitis
and emphysema (collectively referred to as COPD), upper and lower respiratory tract infections,
pneumonia, and other minor categories. Temporal associations between ED visits or hospital admissions
3-21
-------
for respiratory diseases and the ambient concentrations of S02 have been the subject of more than fifty
peer-reviewed research publications since 1994. In addition to considerable statistical and analytical
refinements, recent studies have examined responses of morbidity in different age groups, the effect of
seasons on ED and hospital usage, and multipollutant models to characterize the effects of copollutant
mixtures. The epidemiologic studies of ED visits and hospital admissions for respiratory causes are
summarized in Annex Table F-2.
All Respiratory Diseases
There are relatively few studies of ED visits for all respiratory causes in contrast to the quantity of
studies that examine hospital admissions for all respiratory causes. Collectively, studies of ED visits and
hospitalizations provide evidence to support an association between ambient S02 levels and ED visits and
hospitalizations for all respiratory causes. When analyses were restricted by age, the results among
children (0-14 years) and older adults (65+ years) were mainly positive, though not all statistically
significant. The studies that examined the association of these outcomes and S02 levels among adults (15-
64 years) reported a mix of positive and negative results. When all age groups were combined, the results
of ED and hospitalization studies were mainly positive; however, the excess risk estimates were generally
smaller compared to the children and older adults groups (see additional discussion in Section 4.2.3). It is
possible that the effects observed in the combined age groups were driven by increases in the very young
or older adult subpopulations. S02-related relative risks from the hospitalization and ED studies,
separated by analyses among all ages and age-specific analyses, are shown in Figure 3-6. All studies that
reported quantitative data are included in the figure (studies using GAM with default convergence criteria
are not included). Figure 3-6, as well as Figure 3-7 are presented to assess the general consistency of the
findings. Overall, the effect estimates in this figure range from a -5% to 20% excess risk in ED visits or
hospital admissions for respiratory causes per 10 ppb increase in 24-h avg S02, with the large majority of
studies suggesting an increase in risk.
Wilson et al. (2005) examined ED visits for all respiratory causes in Portland, ME from 1996-2000
and in Manchester, NH from 1998-2000. The mean 1-h max S02 concentration in Portland was 11.1 ppb
(SD 9.1), and was higher during the winter months (mean 17.1 ppb [SD 12.0]) and lower in the summer
(mean 9.1 ppb [SD 8.0]). In Manchester, the mean 1-h max S02 concentration was 16.5 ppb (SD 14.7
ppb), and also was higher in the winter months (mean 25.7 ppb [SD 15.8]) than in the summer months
(mean 10.6 ppb [SD 15.1]). Though the authors reported the 1-h max S02 concentrations, they used the
24-h avg S02 concentrations in their analyses. When all ages where included in analyses, Wilson et al.
(2005) found positive associations between ED visits and S02, with a 7% (95% CI: 3.0, 12) and 1% (95%
CI: -3.0, 5.0) excess risk per 10 ppb increase in 24-h avg S02 at a 0-d lag in Portland, ME and
Manchester, NH, respectively.
Peel et al. (2005) investigated ED visits for all respiratory causes in Atlanta, GA from 1993-2000.
This study included 484,830 ED visits. The mean 1-h max S02 concentration was 16.5 ppb (SD 17.1).
The researchers found a weak positive relationship between ED visits and S02, though the increased risk
was not statistically significant (1.6% [95% CI: -0.6, 3.8] excess risk per 40 ppb increase in 1-h max
S02). Tolbert et al. (2007) recently reanalyzed these data with four additional years of data and found
similar results (0.8% [95% CI: -0.7, 2.3]). An analysis by Dab et al. (1996) examined the association
between S02 and hospital admissions for all respiratory causes in Paris, France, using both the 24-h avg
and 1-h max. It should be noted that these researchers observed similar effect estimates for both exposure
metrics; however, only the estimate using 24-h avg was statistically significant (1.1% [95% CI: 0.1, 2.0]
excess risk per 10 ppb increase in 24-h avg S02 versus 1.9% [95% CI: -1.3, 5.0]) per 40 ppb increase in
1-h max S02).
3-22
-------
Reference
Location
Lag
Tdbertetal. (2007)*
Wilson etal. (2005)
Atkinson et al. (1999a)
Wilson et al. (2005)
Atkinson etal, (1999a)
Wiisonetal. (2005)
Atkinson etal. (1999a)
Wiisonetal. (2005)
Atkinson etal. (1999a)
Atlanta, GA
Portland. ME
Manchester. NH
London, UK
Portland. ME
Manchester NH
London, UK
Portland, ME
Manchester. NH
London, UK
Portland. ME
Manchester NH
London, UK
0-2
0
0
1
0
0
2
0
0
2
0
0
1
Cakmaketal, (2006)
Burnett et al. (1997}*
Luginaahetal, (2005)
Atkinson etal. (1999b)
Ponce de Leon etal. (1996)
Waiters etal (1994)
I. (1996}
Lkxca etal. (2005}
Hag-en etal. <2000}
Oftedaletal. (2003)
Petroeschevskyetal. (2001)
Wong etal. (1999}
Luginaahetal. (2GQ5)
Yang etal (2003}
Atkinson etal. (1999b)
Ponce cfe Leon etal. (1996)
Barnettetal. (2005)*
Multeity. Canada
Toronto, ON
Windsor, ON
London, UK
London, UK
Birmingham, UK
Paris. France
Torrelavega. Spain
Drammen, Noway
Drammen, Norway
Brisbane. Australia
Hong Kong China
Windsor, ON
Vancouver, BC
London, UK
London, UK
Mulboty. Australia
Petroeschevsky et al. (2001) Brabane, Australia
Gouveia and Fletcher (2000)
Wong et al. (1999)
Luginaah el al. (2005)
Spixetal. (1998}
Atkinson et al. (1999b)
Ponce de Leon etal, (1996)
Vigotti et al. (1996)
Schoutenelal (1996)
Petroeschevsky el al. (2001)
Wong etal. (1999)
Schwartz (1995)
Schwartz et al. (1996)
Fung etal, (2006}
Yang etal. (2003)
Luginaah et al. (2005)
Spixetal. (1998)
Aftirvson et al. (1S$9b)
Ponce de Leon etal. (1996)
Vigotti etal. (1996)
Schouten et al (1996)
Petroeschevsky etal. (2001)
Wong etal. (1999)
S5o Paulo, Brazil
Hong Kong. China
Windsor, ON
Multicity. Europe
London, UK
London, UK
Milan, Italy
Amsterdam, The Netherlands
Rotterdam, The Netherlands
Brisbane, Australia
Hong Kong, China
New Haven, CT
Tacoma. WA
Cleveland, OH
Vancouver, BC
Vancouver, BC
Windsor, ON
Multicity, Europe
London, UK
London, UK
Milan, Italy
Amsterdam, The Netherlands
Brisbane, Australia
Hong Kong, China
3
0-3
1
1
0
0-2
NR.
NR
NR
0
0
2
1
1
0-1
04
1
0
NR
3
1
0
0-3
0-2
1
0
2
0
0-1
0-6
0
1
NR
3
2
0
0-3
0
0
EO visits
All Ages
0-14 yrs
15-64 yrs
65+ yrs
Hospital visits
iSummer -
All Ages
Males.
- Females
U Summer
Winter' -
. Females
0-14 yrs
5-14 yrs
1-4 ^rs-
i
CM yrs -
- 5-14 yis
Femafes .
¦ Males
15-64 yrs
65+ yrs
Females -
-Males
T
0.6 0.8
T
T
1.2 1.4
Relative Risk
1 r
1.6 1.8
2.0
Figure 3-6. Relative risks (95% CI) of S02-associated emergency department visits and
hospitalizations for all respiratory causes among all ages and separated by age group.
Risk estimates are standardized per 10 ppb increase in 24-h avg SO2 concentrations
or 40 ppb increase in 1-h max SO2 (*).
3-23
-------
When analyses were stratified to include only children (0-14 years), evidence of a modest
association between S02 and ED visits or hospitalizations for all respiratory causes in children was
reported in several Australian (Barnett et al., 2005; Petroeschevsky et al., 2001) and European (Anderson
et al., 2001 [using GAM default convergence criteria]; Atkinson et al., 1999a; 1999b) studies. Excess
risks ranging from 3% to 22% per 10 ppb increase in 24-h avg S02 were reported by these studies. In a
multicity study spanning Australia and New Zealand, Barnett et al. (2005) compared hospital admission
data collected from 1998-2001 with ambient S02 concentrations, where the mean 24-h avg S02
concentration ranged from 0.9 to 4.8 ppb. The authors found a 22% (95% CI: 5, 42) excess risk per 10
ppb increment in 24-h avg S02 among children (1-4 years) in these cities. Petroeschevsky et al. (2001)
found similar results for the 0-4 age group in their Brisbane, Australia study (22.4% increase [95% CI:
8.7, 37.7]). However, some additional U.S. (Wilson et al., 2005), European (Fusco et al., 2001 [using
GAM default convergence criteria]; Ponce de Leon et al., 1996), and Latin American (Braga et al., 1999;
2001) studies did not find statistically significant associations between ambient S02 concentrations and
hospitalizations for all respiratory causes among children.
Wilson et al. (2005) found a positive association between ED visits and S02, with a 16% (95% CI:
7.0, 22.0) excess risk per 10 ppb increase in 24-h avg S02 at a 0-d lag in Portland, and a 7% (95% CI: -
5.0, 21.0) in Manchester, NH when only older adults (65+ years) were considered. In another two-city
study, Schwartz (1995) compared 13,740 hospital admission among older adults in New Haven, CT and
Tacoma, WA from 1988-1990 with ambient S02 concentrations. The mean 24-h avg S02 concentration
was 29.8 ppb (90th percentile: 159) in New Haven and 16.8 ppb (90th percentile: 74) in Tacoma. Schwartz
found positive associations between hospitalizations and S02, with a 2% (95% CI: 1.0, 3.0) excess risk at
a 2-d lag in New Haven and 3% (95% CI: 1.0, 6.0) excess risk at a 0-d lag in Tacoma per 10 ppb increase
in 24-h avg S02. In two-pollutant models, the S02 effect estimate from New Haven, but not Tacoma, was
found to be robust to adjustment for PM10. Here, the term robust is used to indicate that there was little
change in the magnitude of the central estimate, though statistical significance may have been lost. In
Vancouver, BC, both Fung et al. (2006) and Yang et al. (2003b) also found positive associations between
hospitalizations among older adults and S02. In a multipollutant model including coefficient of haze
(CoH), N02, 03, and CO, the S02 effect estimate diminished slightly (Yang et al., 2003b).
Additional evidence of a positive association between ED visits or hospitalizations for all
respiratory causes among older adults and S02 comes from several European (Spix et al., 1998; Sunyer et
al., 2003a; Vigotti et al., 1996) and Australian (Petroeschevsky et al., 2001) studies. Excess risks ranging
from 1% to 12% per 10 ppb increase in 24-h avg S02 were reported by these studies. Petroeschevsky et
al. (2001) examined 33,710 hospital admissions in Brisbane, Australia from 1987-1994. The mean
24-h avg S02 concentration was 4.1 ppb, and was highest in the winter months (4.8 ppb) and lowest in the
spring (3.7 ppb). Petroeschevsky et al. found a 12% (95% CI: 2.0, 23.0) excess risk per 10 ppb increase in
24-h avg S02 at 0-d lag. Additional European studies did not find statistically significant associations
between ambient S02 concentrations and ED visits or hospitalizations for all respiratory causes among
older adults (Anderson et al., 2001 [using GAM with default convergence criteria]; Atkinson et al., 1999a;
1999b; Ponce de Leon et al., 1996; Schouten et al., 1996).
In summary, studies generally observed small, positive associations between ambient S02 concen-
trations and ED visits and hospitalizations, particularly among children and older adults (65+ years). The
positive evidence from these studies is supported by the results of panel, human clinical, and limited
toxicological studies that also found a positive relationship between S02 levels and adverse respiratory
outcomes.
Asthma
Studies of ED visits and hospitalizations provide evidence to support an association between
ambient S02 levels and ED visits and hospitalizations for asthma. The results from the hospitalization and
ED studies, separated by analyses among all ages and age-specific analyses, are shown in Figure 3-7.
Overall, central effect estimates in the figure range from a -10% to 40% excess risk in ED visits and
3-24
-------
hospitalizations for asthma per 10 ppb increase in 24-h avg S02. Most of the effect estimates are positive
(suggesting an association with S02 and ED visits and hospitalizations for asthma), though few are
statistically significant at the 95% confidence level.
When all ages were included in the analyses, Wilson et al. (2005) found a positive association
between ED visits and S02, with an 11% (95% CI: 2, 20) excess risk per 10 ppb increase in 24-h avg S02
at a 0-d lag in Portland, ME and a positive, though not statistically significant association in Manchester,
NH (6% increase [95% CI: -4, 17]). Ito et al. (2007b) found a 36% (95% CI: 22.2, 51.2) excess risk in
asthma ED visits per 10 ppb increase in 24-h avg S02 during warm months in New York City. This effect
was robust to the inclusion of PM2 5 in the model, though this association was diminished once N02 was
included in the model. Another study conducted in New York City (NY DOH, 2006) found a 10% (95%
CI: 5, 15) excess risk in asthma hospital admissions per 10 ppb increase in 24-h avg S02 for Bronx
residents, but a null association for the residents of Manhattan (-1% [95% CI: -11, 11]). A study
conducted in Atlanta (Peel et al., 2005) found a null relationship between asthma ED visits and 1-h max
S02 (0.2% increase [95% CI: -3.2, 3.4]). A study by Jaffe et al. (2003) examined the association between
S02 and ED visits for asthma in three cities in Ohio - Cincinnati, Cleveland, and Columbus - in
asthmatics aged 5 to 34 years. The mean 24-h avg S02 concentrations were 14 ppb (range: 1-50) in
Cincinnati, 15 ppb (range: 1-64) in Cleveland, and 4 ppb (range: 0-22) in Columbus. A positive
association was observed in the multicity analysis, with a 6.1% (95% CI: 0.5, 11.5) excess risk in asthma
visits observed per 10 ppb increase in 24-h avg S02. In the city-stratified analyses, significant
associations were observed only for Cincinnati (17.0% [95% CI: 4.6, 30.8]).
When analyses were stratified to include children (0-14 years) only, Wilson et al. (2005) found
positive, but not statistically significant associations between ED visits and S02 in Portland, ME (5%
[95% CI: -12, 25]) and Manchester, NH (20% [95% CI: -3, 49]). Similarly, Lin et al. (2003b) observed a
positive association between hospitalizations for asthma and S02 among girls (30% [95% CI: 6, 60]), and
a negative association for boys (-10% [95% CI: -23.4, 5.8] Toronto, ON; mean 24-h avg S02 of 5.36 ppb
[SD 5.90]). Stronger evidence comes from a study of childhood asthma hospitalizations conducted in
Bronx County, New York (Lin et al., 2004e). In this study, the authors conducted a case-control study of
children aged 0-14 years and examined the association of daily ambient S02 concentrations (categorized
into quartiles of both avg and max levels) and cases admitted to the hospital for asthma or controls who
were admitted for reasons other than asthma. The mean 24-h avg S02 was below 17 ppb for both cases
and controls across all lag days examined. The authors found that cases were exposed to higher 24-h avg
S02 than controls. When the highest exposure quartile was compared with the lowest, the ORs were
strongest when a 3-day lag was employed (OR 2.16 [95% CI: 1.77, 2.65] for 24-h avg S02; OR 1.86
[95% CI: 1.52, 2.27] for 1-h max S02). The results were positive and statistically significant for all lag
days examined. These results suggest a consistent positive association between S02 exposure and
hospitalizations for childhood asthma.
Additional evidence of a positive association between ED visits or hospitalizations for asthma
among children and S02 comes from several European (Anderson et al., 1998; Atkinson et al., 1999a;
1999b; Hajat et al., 1999; Sunyer et al., 1997; 2003a [using GAM with default convergence criteria];
Thompson et al., 2001) and Asian (Park et al., 2002 [using GAM with default convergence criteria])
studies. Excess risks ranging from 2% to 10% per 10 ppb increase in 24-h avg S02 were reported by these
studies. Several of these studies observed that the S02 effect estimate was robust to adjustment for BS and
N02 (Anderson et al., 1998; Sunyer et al., 1997), but one study observed that the S02 effect diminished
considerably with adjustment for PM10 and benzene (Thompson et al., 2001). Atkinson et al. (1999a)
compared 165,032 respiratory hospital admissions in London from 1992-1994 with ambient S02 levels
(mean 24-h avg of 7.2 ppb [SD 4.7]). They found a 10% (95% CI: 4.0, 16.0) excess risk per 10 ppb
increase in 24-h avg S02 at 1-d lag for children aged 0-14 years. Additional European (Fusco et al., 2001
[using GAM with default convergence criteria]), Australian (Barnett et al., 2005; Petroeschevsky et al.,
2001), Asian (Ko et al., 2007b; Lee et al., 2006a) and Latin American (Gouveia and Fletcher, 2000)
studies did not find statistically significant associations between ambient S02 concentrations and
hospitalizations for all respiratory causes among children.
3-25
-------
Ito el al (2007)
NY DOH (2006)
Peel et al. (2005)*
Wilson el al. {2005)
Atkinson el al. (1999a)
Hajatel al. (1999)
Galan el al. {2003)
Wilson el al. (2005)
Atkinson et al. (1999a)
Hajatetal. (1999)
Jaffe el al. (2003)
Wilson el al. (2005)
Atkinson et al. (1999a)
Hajatetal. (1999)
Boutin-Forzano el al. (2004)
Castellsague el al. (1995)
Tenias et al. (1998)
Wilson etal. (2005)
Hajatetal. (1999)
New York. NY
Bronx, NY
Manhattan. NY
Atlanta, GA
Portland. ME
Manchester. NH
London. UK
London. UK
Madrid, Spain
Portland, ME
Manchester. NH
London. UK
London. UK
Cincinnati. OH
Cleveland, OH
Columbus, OH
Mufticily, OH
Portland, ME
Manchester. NH
London. UK
London. UK
Marseille. France
Barcelona, Spain
Valencia, Spain
Portland, ME
Manchester. NH
London, UK
Lag
0-4
0*4
0-2
0
0
1
0-2
0
0
0
1
0-3
2
2
3
NR
0
0
1
0-3
0
2
1
0
0
0
0-1
EO visits
All Ages
I
Wipter-
. Summer
1
0-14 yrs
15-64 yrs
Winter -
- Summer
65+ yrs
Anderson el al. (1998)
Atkinson et al. (1999b)
Wallers el al. (1994)
Dab el al. (1996)
Schouten et al. (1996)
Petroeschevsky et al- (2001)
Tsai el al. (2006)
Wong et al. (1999)
Ko et al. (2007a)
Lin et al. (2004a)
Lin et al. (2003)
Sunyer et al. (1997)
Anderson el al. (1998)
Atkinson et al. (1999b)
Barnett el al. (2005)*
Petroeschevsky et al (2001)
Gouveia and Fletcher (2000)
Lee et al. (2006)
Sheppard et al. (1999)
Sunyer etai. (1997)
Anderson el al. (1998)
Atkinson et al. (1999b)
Petroeschevsky et al (2001 )*
Koetal. (2007a)
Anderson el al. (1998)
Atkinson et al, (1999b)
London. UK
London. UK
Birmingham, UK
0-3
1
0
[ Hospital visits j
Paris. France 2
Amsterdam. The Netherlands 0-3
Brisbane, Australia 2
Kaohsiung, Taiwan 0-2
Summer.
£25 "C -
~ Winter
Hong Kong. China
Hong Kong. China
Bronx, NY
T oronto. ON
Multicity, Europe
London. UK
London. UK
Multicity, Australia
Brisbane. Australia
Sao Paulo, Brazil
Hong Kong, China
Seattle. WA
Murticily, Europe
London. UK
London, UK
Brisbane, Australia
Hong Kong, China
London, UK
London. UK
0
0-3
NR
0-6
Best lag
0-3
1
0-1
0
2
0
0
0-3
0-2
3
0
0-3
0-3
2
Females -
I
All Ages
- <25*C
0-14 yrs
1-4 yrs_
5-14 yrs
15-64 yrs
65+ yrs
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Relative Risk
Figure 3-7. Relative risks (95% CI) of S02-associated emergency department visits and
hospitalizations for asthma among all ages and age-specific groups. Risk estimates
are standardized per 10 ppb increase in 24-h avg SO2 concentrations or 40 ppb
increase in 1-h max SO2 (*).
3-26
-------
In summary, small, positive associations were observed between ambient S02 concentrations and
ED visits and asthma hospitalizations. Evidence from these studies is further supported by the results of
panel and human clinical studies that have also found S02-related respiratory effects in asthmatics.
Chronic Obstructive Pulmonary Disease
There are relatively few studies that have examined the association of ED visits and
hospitalizations for COPD and ambient S02 levels, and very little evidence that an association exists. A
recent study (Ko et al., 2007a) found a significant association between hospital admissions for COPD (not
including asthma) in Hong Kong (1.8% [95% CI: 0.3, 3.8]) excess risk per 10 ppb increase in 24-h avg
S02 concentration). Three additional studies reported positive and statistically significant results for
COPD and S02; all three studies included asthma in their diagnostic definition of COPD (Anderson et al.,
2001; Moolgavkar, 2003a; Sunyer et al., 2003b). Anderson et al. (1999) reported a 12% (95% CI: 5.0,
20.0) excess risk per 10 ppb increase in 24-h avg S02 among children, while Moolgavkar (2003b) and
Sunyer et al. (2003b) found 5% and 2% excess risks per 10 ppb increase in 24-h avg S02 among older
adult populations, respectively. Other studies examining COPD did not find statistically significant results
(Atkinson et al., 1999a; Burnett et al., 1999 [using GAM with default convergence criteria]; Michaud et
al., 2004).
Overall, this limited and inconsistent evidence does not support a relationship between ED visits
and hospitalizations for COPD and ambient S02 levels.
Respiratory Diseases Other than Asthma or COPD
Studies of ED visits or hospital admissions for other respiratory diseases looked at several other
specific outcomes. There are limited studies with mixed results for upper respiratory tract infections
(Burnett et al., 1999 [using GAM with default convergence criteria]; Hajat et al., 2002 [using GAM with
default convergence criteria]; Lin et al., 2005; Peel et al., 2005), pneumonia (Barnett et al., 2005;
Moolgavkar et al., 1997 [using GAM with default convergence criteria]; Peel et al., 2005), bronchitis
(Barnett et al., 2005; Michaud et al., 2004), and allergic rhinitis (Hajat et al., 2002 [using GAM with
default convergence criteria]; Villeneuve et al., 2006b). The limited evidence is suggestive of an
association between S02 levels and ED visits or hospitalizations for lower respiratory tract diseases
(Atkinson et al., 1999a; Farhat et al., 2005 [using GAM with default convergence criteria]; Hajat et al.,
2002 [using GAM with default convergence criteria]; Lin et al., 1999; Martins et al., 2002 [using GAM
with default convergence criteria]). All of the studies that characterized this relationship found a positive
and statistically significant excess risk associated with increases in S02. Excess risks ranging from 3% to
33% per 10 ppb increase in 24-h avg S02 were reported by these studies.
In summary, only a few studies provide results for respiratory health outcomes other than asthma
and COPD, and these results are mixed. This makes it difficult to draw conclusions about the effects of
S02 on these diseases. Limited evidence is indicative of an association between ambient S02 levels and
ED visits for lower respiratory tract diseases.
Summary of Evidence on Emergency Department Visits and Hospitalizations for Respiratory
Diseases
Small, positive associations exist between ambient S02 concentrations and ED visits and
hospitalizations for all respiratory causes, particularly among children and older adults (65+ years).
Similar associations are found for asthma. The S02-related changes in ED visits or hospital admissions
for respiratory causes ranged from -5% to 20% excess risk, with the large majority of studies suggesting
an increase in risk. Mean 24-h avg S02 levels ranged from 1 to 30 ppb in these studies, with maximum
values ranging from 12 to 75 ppb. No association was observed between S02 levels and ED visits and
3-27
-------
hospitalizations for COPD. Given the limited number of studies with mixed results, it is difficult to draw
conclusions about the effect of S02 on other respiratory diseases, though studies of lower respiratory tract
diseases are somewhat indicative of an association.
The potential influence of copollutants has not been systematically considered in the epidemiologic
literature. A limited subset of the studies examined potential confounding by copollutants using
multipollutant regression models. Figure 3-8 presents S02 excess risk estimates with and without
adjustment for various copollutants. PM and N02 are the main foci, since these pollutants have been
found to be highly-correlated with S02 in epidemiologic studies and have known respiratory health
effects. Multipollutant regression analyses indicated that although copollutant adjustment had varying
degrees of influence on the S02 effect estimates, the effect of S02 on respiratory health outcomes
appeared to be generally robust and independent of the effects of gaseous copollutants, including N02
(Anderson et al., 1998; Lin et al., 2004c; Sunyer et al., 1997), and 03 (Anderson et al., 1998; Hajat et al.,
1999; Tsai et al., 2006; Yang et al., 2003b; 2005). The evidence for PMi0 was less consistent, with three
studies finding that positive S02 effect estimates became negative, though not statistically significant,
with the inclusion of PMi0 in the model (Galan et al., 2003; Schwartz, 1995 [in New Haven, CT]; Tsai et
al., 2006 [in Tacoma, WA]). Several other studies found the effects estimates for S02 to be generally
robust to the inclusion of PM10 in the model (Burnett et al., 1997b; Hagen et al., 2000; Hajat et al., 1999
[using GAM with default convergence criteria]; Schwartz, 1995 [in New Haven, CT]). The studies that
examined PM2 5 and PMi0.2.5 in copollutant models found that the S02 estimates were generally robust to
the adjustment for PM of these size fractions (Burnett et al., 1997b; Ito et al., 2007; Lin et al., 2003b; NY
DOH, 2006).
The results of several studies (Anderson et al., 1998; Hajat et al., 1999; Schouten et al., 1996; Spix
et al., 1998; Wong et al., 1999a) have demonstrated a greater increase in ED visits and hospitalizations for
respiratory illnesses during the summer months, despite the fact that the avg concentrations for S02 in
some of the areas studied were greatest in winter. In contrast, some studies found the associations between
ED visits and hospital admissions and respiratory disease with similar increases in S02 to be greater in
winter than summer (Vigotti et al., 1996; Walters et al., 1994). Other studies were unable to discern a
seasonal difference in ED visits and hospitalizations for respiratory causes (Castellsague et al., 1995;
Tenias et al., 1998; Wong et al., 2002c [using GAM with default convergence criteria]). These effects
were not consistent across age groups. Warmer months were more likely to show evidence of an
association with adverse respiratory outcomes in children, while older adults appeared more likely to be
affected during the cooler months. These seasonal associations remain somewhat uncertain and require
additional investigation.
In conclusion, a large number of epidemiologic studies provide evidence of an association between
ambient S02 concentrations and ED visits and hospitalizations for all respiratory causes, in particular
among children and older adults (65+ years), and for asthma. The findings are generally robust when
additional copollutants are included in the model. These associations are supported by panel studies that
observed S02-related increases in asthma and other respiratory symptoms in children, and human clinical
and animal toxicological studies that found a positive relationship between S02 exposure and various
respiratory outcomes.
3-28
-------
Reference
Location
Schwartz (1995)
New Haven, CT
Tacoma, WA
Burnett et al. (1997) Toronto, ON
- Summer
Hagen et al. (2000)
Yang et al. (2003)
Drammen, Norway
Vancouver, BC
Respiratory
f'Mic
< 3 Yrs -
ttPMio
— PMio
— PM2 5
— PMl 0-2.5
— PM10
0
¦ > 65 Yrs
I to et al. (2007)*
- Summer
NY DOH (2006)*
Lin et al. (2003)
- Females
New York, NY
Bronx, NY
Manhattan, NY
Toronto, ON
I to et al. (2007)*
- Summer
NY DOH (2006)*
Hajat et al. (1999)*
Sunyer etal. (1997)
New York, NY
Bronx, NY
Manhattan, NY
London, UK
Multicity - Europe
Sunyer et al. (1997) Multicity - Europe
Anderson et al. (1998) London, UK
Hajat et al. (1999)* London, UK
Galan et al. (2003)* Madrid, Spain
Tsai et al. (2006)1 Kaohsiung, Taiwan
Anderson et al. (1998) London. UK
Tsai et al. (2006) Kaohsiung, Taiwan
Ito et al. (2007)* New York, NY
- Summer
Anderson et al. (1998) London, UK
Hajat et al. (1999)* London. UK
Tsai et al. (2006) Kaohsiung, Taiwan
0.6
- PM2 5
* PM25
. PM2 5
-PM2.5+ PMlO-2.5
All ages
0-14 yrs
- BS
¦ BS
-BS
£25°C -
¦ PM10
-PM10
<25°C -
PM10
All ages
525°C
<25 C
All ages
0-14 yrs
•..v1:"- c
<25 C
no2
• = single pollutant
x= multipollutant
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Relative Risk
Figure 3-8. Relative risks (95% CI) of S02-associated emergency department visits (*) and
hospitalizations for all respiratory causes and asthma, with and without copollutant
adjustment. Risk estimates are standardized per 10 ppb increase in 24-h avg SO2
concentrations or 40 ppb increase in 1-h max SO2.
3-29
-------
3.1.4.7. SO2-PM Interactions and Other Mixture Effects
As discussed earlier, S02 is a component of complex air pollution mixtures that vary
geographically and temporally (e.g., by hour, week, and season). The 1982 AQCD addressed the question
of possible effects of PM on the response to S02. It was noted that sorption of S02 onto liquid or solid
particles, which may act as carriers, tended to increase its potency in animal toxicological experiments.
However, the mechanism for the effect was not known. Since then, additional animal studies have
demonstrated respiratory responses following inhalation of S02 which was adsorbed onto metal oxide or
carbon particles. These studies are summarized in Annex Table E-4 and both confirm and extend earlier
findings. In all of the recent studies, the resulting particles were submicron in size; they would be
expected to deposit in the lower respiratory tract. Acute and subacute exposures to S02 and PM resulted
in additive or more-than-additive effects on pulmonary resistance (Amdur et al., 1983; Chen et al., 1991),
diffusing capacity for CO (Amdur et al., 1988), AHR following an acetylcholine challenge (Amdur et al.,
1988; Chen et al., 1992), and decreased host defense responses (Clarke et al., 2000; Jakab et al., 1996).
Many of these studies reported transformation of S02 to sulfite, sulfate, sulfur trioxide and H2S04
depending on conditions of temperature and relative humidity (Amdur et al., 1983; 1988; Chen et al.,
1991; Clarke et al., 2000; Jakab et al., 1996). Respiratory responses observed in these experiments were
in some cases attributed to the formation of particular sulfur-containing species. For example, repeated
exposure to 20 (ig/m3 carbon black-associated sulfate resulted in impaired host defenses (Clarke et al.,
2000). However, the relevance of these animal toxicological studies has been called into question because
concentrations of both PM (1 mg/m3 and higher) and S02 (1 ppm and higher) utilized in these studies are
much higher than ambient levels. Furthermore, the S02-adsorbed PM utilized in some of these studies is
not representative of ambient PM. For example, some of the laboratory-generated aerosols contained
sulfite but atmospheric chemistry studies do not indicate significant amounts of sulfite ion in atmospheric
PM. In summary, animal toxicological studies conducted since the last review suggest that S02 effects
may be potentiated by coexposure to PM but the relevance of these results to ambient exposures is not
clear.
The 1982 AQCD also described an informative study of complex air pollutants which was
conducted in dogs (Stara et al., 1980). In dogs that were exposed to S02 + H2S04, with or without
irradiated or non-irradiated auto exhaust concentrations relevant to urban exposures, functional lung
changes were observed at 61 months of exposure and at 2 years after exposures ended. Morphological and
biochemical changes were observed at 2.5-3 years after exposure. Additional animal studies have been
conducted since the 1982 AQCD which involved binary mixtures, laboratory-generated complex mixtures
(e.g., simulation of regional air pollution), or actual ambient air mixtures. Annex Tables E-5 through E-7
summarize results from short-term studies on possible toxicity relationships between S02 and 03, S02 and
sulfates, as well as the effects of complex air pollution mixtures in healthy animals and disease models.
Possible interactions between S02 and cold air were also examined (Annex Table E-8). Generally, most
studies with ambient or laboratory-generated complex mixtures did not include a S02-only exposure
group, making it difficult to determine the contribution of SOx. No definitive conclusions can be made
from these studies.
3.1.4.8. Summary of Evidence on the Effect of Short-Term (> 1 h) Exposure on
Respiratory Health
Numerous epidemiologic studies have observed associations between short-term (> 1-h, generally
24-h avg) exposure to S02 and respiratory health effects, ranging from respiratory symptoms to ED visits
and hospital admissions for respiratory causes. The associations between ambient S02 concentrations and
several respiratory outcomes were generally consistent, with the large majority of studies showing
positive associations, and multicity studies, as well as several single-city studies, indicating statistically
significant findings. The effects on lung inflammation and AHR related to short-term exposure to S02 at
3-30
-------
levels as low as 0.1 ppm found in animal toxicological studies provide a degree of coherence and
biological plausibility for the observed epidemiologic associations. In addition, the causal respiratory
effects of peak exposures (5-10 min) of S02 at levels as low as 0.2 ppm found in the human clinical
studies of asthmatics (see Section 3.1.3.) provide further evidence of biological plausibility for the effects
associated with short-term exposure to S02.
Two recent multicity studies (Mortimer et al., 2002; Schildcrout et al., 2006) and several other
studies (Delfino et al., 2003a; Neas et al., 1995; van der Zee et al., 1999) have found an association
between short-term ambient S02 concentrations and respiratory symptoms in children. In the limited
number of studies that assessed potential confounding by copollutants using multipollutant models, the
S02 effect on respiratory symptoms was generally found to be robust to adjustment for copollutants.
These findings indicate an association between short-term exposure to ambient S02 exposure and
respiratory symptoms in children, particularly those with asthma. Several recent studies (Desqueyroux et
al., 2002a; 2002b; van der Zee et al., 2000) found no association between ambient S02 levels and
respiratory symptoms in adults, though there was limited epidemiologic evidence which suggested that
atopic adults as well as children may be at increased risk for S02-induced respiratory symptoms (Boezen
et al., 1999; 2005).
Epidemiologic studies do not provide strong evidence for an association between short-term
ambient S02 exposure and lung function in either children (Delfino et al., 2003a; Mortimer et al., 2002;
Roemer et al., 1998) or adults (e.g., Peters et al., 1996a; Taggart et al., 1996). Several other studies
reported positive results; however, the generally mixed findings, as well as the relative lack of evidence
available to evaluate potential confounding by copollutants, limits the causal interpretation of ambient
S02 on lung function.
Only one epidemiologic study (Adamkiewicz et al., 2004) evaluated inflammation, as indexed by
eNO, and found no association with S02 exposure. Animal toxicological studies found that repeated
exposure to S02 leads to increased airway inflammation in two models involving animals which were
sensitized to an antigen (Li et al., 2007; Park et al., 2001a). Studies of other ambient pollutants indicate
that influx of macrophages and other inflammatory cells, with the related release of cytokines, is a
common response to injury.
Effects of short-term exposure to S02 on AHR have been observed. In two animal toxicological
studies, repeated exposure to 0.1 ppm S02 led to AHR in guinea pigs sensitized to an antigen (Park et al.,
2001a; Riedel et al., 1988). Human clinical studies by Devalia et al. (1994) and Rusznak et al. (1996)
demonstrated enhanced airway responses to an inhaled allergen in asthmatic subjects following exposure
to a combination of S02 (0.2 ppm) and N02 (0.4 ppm). This effect was not observed following exposure
to either S02 or N02 alone. These findings of increased airway resistance are in concordance with the
limited epidemiologic study results that showed S02-induced increases in AHR among atopic children
and asthmatic adults (Soyseth et al., 1995a; Taggart et al., 1996).
Epidemiologic studies provide evidence for an association between ambient S02 levels and ED
visits and hospitalizations for all respiratory diseases in two susceptible populations: children (Dab et al.,
1996; Petroeschevsky et al., 2001; Walters et al., 1994) and older adults (65+ years) (Fung et al., 2006;
Schwartz, 1995; Spix et al., 1998; Wong et al., 1999a). Evidence for an association between ambient S02
levels and these outcomes in non-elderly adults was weaker. A modest association between ambient S02
and ED visits and hospitalizations for asthma was also observed. S02 effect estimates were generally
robust to the inclusion of copollutants, including PM, 03, CO and N02j indicating that the observed effects
of S02 on respiratory endpoints is independent of the effects of other ambient air pollutants.
In summary, recent epidemiologic studies, supported by a limited number of animal toxicological
studies conducted at near ambient concentrations, indicate an association between short-term (> 1-h,
generally 24-h avg) exposure to S02 and several measures of respiratory health, including respiratory
symptoms, inflammation, airway hyperreponsiveness, and ED visits and hospitalizations for respiratory
causes. The epidemiologic evidence further observed that the S02-related respiratory effects were more
pronounced in asthmatic children and older adults (65+ years).
3-31
-------
3.1.5. Evidence of the Effects of SO2 on Respiratory Morbidity from
Intervention Studies
Many epidemiologic studies have examined the association of short-term S02 concentrations and
various respiratory morbidity outcomes. These studies collectively suggest that increased ambient S02
concentrations are associated with increased risk of respiratory outcomes, ranging from respiratory
symptoms to ED visits and hospitalizations. Further contributing to the evidence base are intervention
studies that reported decreases in respiratory morbidity following improvements in air quality, particularly
reductions in S02 concentrations.
In Hong Kong, a sudden change in regulation in July 1990 required all power plants and road
vehicles to use fuel oil with a sulfur content of < 0.5% by weight. These regulations were enforced
quickly, and provided opportunities to observe changes in morbidity before and after the intervention.
Peters et al. (1996b) followed 3,521 children (mean age 9.5 years) residing in two districts with good and
poor air quality before the intervention from 1989 to 1991. The intervention resulted in large reductions in
S02 (up to 80% in polluted district), along with a modest reduction in sulfate (38% in polluted district).
Only a small change in TSP levels was observed after the intervention (15% decline in polluted district).
In 1989 and 1990, an excess risk of respiratory symptoms was observed in the polluted district. After the
intervention, there was a greater decline in reported symptoms of cough, sore throat, phlegm, and
wheezing in the polluted compared with the unpolluted district. For example, the OR for cough,
comparing the polluted to the unpolluted district, was 1.22 (95% CI: 1.05, 1.42) in 1989 and 1990, and
decreased to 0.92 (95% CI: 0.73, 1.15) in 1991.
A study by Keles et al. (1999) evaluated the prevalence of chronic rhinitis among high school
students before and after installation of a natural gas network for domestic heating and industrial works,
in a polluted area of Istanbul, Turkey. Concentrations of CO, N02, and hydrocarbons were relatively low
compared to S02 and TSP in this area. After the intervention, the annual mean TSP concentration declined
by 23% from 89.7 (ig/m3 to 68.8 |_ig/nr\ An even greater decline (46%) was observed for S02, from an
annual mean of 70.8 ppb to 38.2 ppb. The prevalence of rhinitis decreased significantly from 62.5% to
51% of the student population (p < 0.05) following the installation of the natural gas network. Symptoms
of rhinitis were associated with air pollution levels, but not with any of the other factors considered,
including sex, household crowding, heating source, and smoking status. Although the effects from TSP
could not be separated from S02 effects, this study demonstrated that reductions in both pollutants (with
greater declines in S02) resulted in significant reductions in the prevalence of chronic rhinitis in a highly
polluted area.
Another study in Germany observed that reductions in air pollutant levels were associated with
improvement in reported respiratory symptoms. Heinrich et al. (2002) examined the influence of reduced
air pollution levels on respiratory symptoms in children aged 5 to 14 years (n = 7,632). Questionnaires
were collected from the children during 1992-1993, 1995-1996, and 1998-1999 in three study areas.
During the study period, S02 concentrations decreased by more than 90% and TSP concentrations
decreased by approximately 60%. Concentrations of nucleation-mode particles (10-30 nm) increased
during this time period. For most respiratory outcomes, the prevalence continued to decline in each of the
three surveys. The temporal changes followed similar trends in all three study areas. Stronger effects
between S02 and prevalence of respiratory symptoms were observed among children without indoor
exposures. For those without indoor exposures, ORs of 1.21 (95% CI: 1.11, 1.32) were observed for
prevalence of bronchitis and 1.11 (95% CI: 1.02, 1.22) for frequent colds per 5-ppb increase in the annual
mean of S02. Frye et al. (2003) reported changes in lung function parameters associated with declines in
S02 concentrations in 2,493 children during this period as well. The researchers observed a 0.6% (95%
CI: 0.1, 1.2) increase in FVC and a 0.4% (95% CI: -0.1, 0.9) increase in FEVi per 5-ppb decrease in the
annual mean of S02. They concluded that the decreasing prevalence of respiratory symptoms and the
increase in lung function following decreases in air pollution levels might indicate the reversibility of
adverse health effects in children.
3-32
-------
In summary, these studies observed that improvements in air quality, in particular large decreases in
S02 concentrations, were associated with improvements in respiratory symptoms and lung function.
However, the decreased respiratory morbidity following large reductions in ambient S02 concentrations
does not preclude the possibility that other constituents of the pollution mixture that share the same
source as S02 are also responsible for adverse effects. In the German and Turkey studies, both S02 and
TSP concentrations decreased dramatically. Although PMi0 levels before and after the intervention were
stable in Hong Kong, large reductions in ambient nickel and vanadium were observed concomitantly with
reductions of sulfur after the intervention (Hedley et al., 2006). Animal toxicological studies conducted at
higher concentrations (> 1 mg/m3 PM and < 1 ppm S02) suggest that S02 effects may be potentiated by
coexposure to PM, but the relevance of these results to ambient exposures is not clear. The improvements
in respiratory health may be jointly attributable to declines in both S02 and PM. Considered collectively
with the larger body of evidence from epidemiologic, human clinical, and animal toxicological studies,
these intervention studies are supportive of S02-related effects on respiratory morbidity.
3.1.6. Summary of Evidence of the Effect of Short-Term SO2 Exposure
on Respiratory Health
Evaluation of the health evidence led to the conclusion that there is a causal relationship between
respiratory morbidity and short-term exposure to S02. This conclusion is supported by the consistency,
coherence, and plausibility of findings observed in human clinical studies with 5-10 min exposures,
epidemiologic studies using largely 24-h avg exposures, and animal toxicological studies using exposures
of minutes to hours.
The strongest evidence for this causal relationship comes from human clinical studies reporting
respiratory symptoms and decreased lung function following peak exposures of 5-10 min duration to S02.
These effects have been observed consistently across studies involving exercising mild to moderate
asthmatics. Statistically significant decrements in lung function accompanied by respiratory symptoms
including wheeze and chest tightness have been clearly demonstrated following exposure to 0.4-0.6 ppm
S02. Although studies have not reported statistically significant respiratory effects following exposure to
0.2-0.3 ppm S02, some asthmatic subjects (5-30%) have been shown to experience moderate to large
decrements in lung function at these exposure concentrations.
A larger body of evidence supporting this determination of causality comes from numerous
epidemiologic studies reporting associations with respiratory symptoms, ED visits, and hospital
admissions with short-term S02 exposures, generally of 24-h avg. Important new multicity studies and
several other studies have found an association between 24-h avg ambient S02 concentrations and
respiratory symptoms in children, particularly those with asthma. Furthermore, limited epidemiologic
evidence indicates that atopic children and adults may be at increased risk for S02-induced respiratory
symptoms. Generally consistent and robust associations also were observed between ambient S02
concentrations and ED visits and hospitalizations for all respiratory causes, particularly among children
and older adults (65+ years), and for asthma. Results of experiments in laboratory animals support these
observations; studies in animals sensitized with antigen demonstrate that repeated exposure to near
ambient S02 levels (as low as 0.1 ppm in guinea pigs) can exacerbate allergic responses including airway
inflammation and AHR.
Mean 24-h avg S02 levels ranged from 1 to 30 ppb in the epidemiologic studies, with maximum
values ranging from 12 to 75 ppb. In the human clinical studies, respiratory effects were observed in
exercising asthmatics following 5-10 min exposure to S02 at levels as low as 0.2 ppm. 5-min S02 data
acquired from a limited number of ambient monitoring sites across the U.S. during the years 1997 to 2006
indicated that 0.2% of the hourly maximum 5-min avg were at or above a concentration of 0.2 ppm. It is
difficult to unequivocally relate the 24-h avg S02 concentrations typically assessed in epidemiologic
studies with the peak exposures in the human clinical studies. The apparent gap between the S02
3-33
-------
concentrations at which respiratory health effects are observed in the epidemiologic studies and the
human clinical studies may be partially attributable to the differences in the study type (e.g., sample size,
study subject selection, exposure conditions). Collectively, the findings from both human clinical and
epidemiologic studies provide a strong basis for concluding a causal relationship between respiratory
morbidity and short-term exposure to S02.
3.2. Systemic Morbidity Associated with Short-Term SO2
Exposure
3.2.1. Summary of Findings from the Previous Review
The studies reviewed in the 1982 AQCD primarily investigated respiratory health outcomes. There
were no key animal toxicological or human clinical studies available at the last review to address effects
of S02 exposure on the cardiovascular system. The only report was a study in dogs exposed to air
pollutant mixtures (S02 + H2S04 with or without nonirradiated or irradiated auto exhaust) (Stara et al.,
1980). No changes were observed in cardiovascular function at the end of 3 years of exposure and 3 years
after exposure. No epidemiologic studies linking exposure to S02 with cardiovascular physiological
endpoints or ED visits or hospital admissions for cardiovascular causes were examined in the last review.
Furthermore, no studies of S02 effects on other organ systems were addressed in the 1982 AQCD.
3.2.2. Cardiovascular Effects Associated with Short-Term Exposure
The biological basis for S02-related cardiovascular health effects may lie in the stimulation of
chemosensitive receptors found in the respiratory tract which respond to irritants like S02 Vagally-
mediated responses may affect the cardiovascular system by inducing bradycardia and either hypotension
or hypertension, as discussed in Section 3.1.2. Alternatively oxidation reactions mediated by the S02
metabolites sulfite and bisulfite which have been absorbed into the systemic circulation may potentially
alter cardiovascular function. In general, vagally-mediated responses have been observed at lower
concentrations of S02 than oxidative injury.
Since 1982, several animal toxicological studies have addressed the effects of S02 on
cardiovascular endpoints. These are summarized below and in Annex Table E-9. In addition, there is one
noteworthy study examining the hematological effects of short-term S02 exposure (Annex Table E-10).
Acute exposure of rats to 0.87 ppm S02 for 24 h resulted in increased hematocrit, sulfhemoglobin and
osmotic fragility as well as decreased whole blood and packed cell viscosities (Baskurt, 1988). These
results indicate a systemic effect of inhaled S02 and are consistent with an oxidative injury to red blood
cells. Only one study since 1982 measured systemic levels of sulfite or bisulfite following S02 inhalation
(Gunnison et al., 1987; Annex Table E-ll). Further studies are required to confirm that inhalation
exposures of S02 at or near ambient levels increase blood sulfite and bisulfite levels sufficiently for
oxidative injury to blood cells or other tissues.
Recent epidemiologic studies have examined the association between air pollution and
cardiovascular effects, including increased heart rate (HR), reduced heart rate variability (HRV),
incidence of ventricular arrhythmias, changes in blood pressure, incidence of myocardial infarctions (MI),
and ED visits and hospitalizations due to cardiovascular causes. The results of these cardiovascular
studies are summarized in Annex Tables F-3 and F-4.
3-34
-------
3.2.2.1. Heart Rate and Heart Rate Variability
Heart rate variability (HRV) is generally determined by analyses of time (e.g., standard deviation of
normal R-R intervals [SDNN]) and frequency domains (e.g., low frequency [LF] / high frequency [HF]
ratio by power spectral analysis, reflecting autonomic balance) measured during 24 h of
electrocardiography (ECG). Brook et al. (2004) stated that HRV, resting heart rate, and blood pressure are
modulated by a balance between the two determinants of autonomic tone (the sympathetic and
parasympathetic nervous systems). An imbalance of cardiac autonomic control may predispose
susceptible people to greater risk of ventricular arrhythmias and mortality from cardiovascular causes
(Brook et al., 2004; Liao et al., 2004).
A limited number of human clinical studies examined the effect of S02 on HRV. During a
controlled exposure of 12 healthy subjects and 12 subjects with asthma to 0.2 ppm S02 for 1 h under
resting conditions, Tunnicliffe et al. (2001) reported that HF power, LF power, and total power were
higher with S02 exposures compared to air exposure in the healthy subjects, but that these indices were
reduced during S02 exposure in the subjects with asthma. The LF/HF ratios were unchanged in both
groups. The authors postulated two autonomic pathways for S02-mediated bronchoconstriction. In
healthy subjects, the dominant pathway was proposed to be the rapidly adapting receptor/C-fiber route,
which results in activation of a central nervous system reflex with an increase in vagal tone. In the
asthmatic subjects, proximal airway narrowing was proposed as the dominant response, possibly through
neurogenic inflammation. This likely causes a compensatory central nervous system-mediated reduction
in vagal tone, resulting in bronchodilation of the distal airway. While there were no detectable changes in
symptoms or lung function in either of the groups, this study provides some evidence that exposure to
S02 may elicit systemic responses at these low levels (0.2 ppm).
In a similar study, Routledge et al. (2006) exposed patients with stable angina as well as healthy
subjects to 50 (ig/m3 carbon particles and to 0.2 ppm S02, each alone and in combination, for 1 h under
resting conditions. HRV, C-reactive protein, and coagulation markers were measured. The authors
reported that for the healthy subjects, S02 exposure was associated with a decrease in HRV markers of
cardiac vagal control 4 h after exposure. However, it should be noted that there was no apparent
difference in the absolute value of the root mean square of successive RR interval differences (r-MSSD)
at 4 h postexposure between the control, S02, carbon, and carbon/S02 groups. The significant difference
reported in the change in r-MSSD from baseline to 4 h postexposure with S02 appears to be due to a
higher baseline value of r-MSSD preceding the S02 exposure compared to the baseline value of r-MSSD
preceding the air exposure. There were no changes in HRV among the patients with stable angina. The
authors noted that this lack of response in the heart patients may be due to a drug treatment effect rather
than decreased susceptibility; a large portion of the angina patients were taking beta blockers, which are
known to increase indices of cardiac vagal control.
In an epidemiologic study, Liao et al. (2004) investigated short-term associations between ambient
pollutants and cardiac autonomic control from the fourth cohort examination (1996 through 1998) of the
population-based Atherosclerosis Risk in Communities (ARIC) study using a cross-sectional study
design. Men and women aged 45 to 64 years (n = 6,784) from three U.S. study centers in North Carolina,
Minnesota, and Mississippi were examined. Resting, supine, and 5-min beat-to-beat R-R interval data
were collected. The mean 24-h avg S02 level measured 1 day prior to the HRV measurement was 4 ppb
(SD 4). In addition to S02, the potential effects of PMi0, 03, CO, and N02 were evaluated. Previous-day
S02 concentrations were positively associated with HR and inversely associated with SDNN and LF
power. Consistently more pronounced associations were suggested between S02 and HRV among persons
with a history of coronary heart disease. Significant associations with HRV indices also were observed for
PMio and the other gaseous pollutants. When the regression coefficients for each individual pollutant
model were compared, the effects of PMi0 on HRV were considerably larger than the effects for the
gaseous pollutants, including S02. No multipollutant analyses were conducted.
Gold et al. (2000; 2003) examined the effect of short-term changes in air pollution on HRV in a
panel study of 21 older adults (aged 53 to 87 years) in Boston, MA. The study participants were observed
3-35
-------
up to 12 times from June to September 1997. The mean 24-h avg S02 concentration was 3.2 ppb (range:
0, 12.6). The 24-h avg S02 concentration was associated with decreased HR in the first 5-min rest period,
but not in the overall 25-min study protocol. The effect estimate for S02 slightly diminished but remained
marginally significant in a two-pollutant model with PM25. The inverse association between S02 and HR
observed in this study are in contrast to the S02-related increases in HR observed by Liao et al. (2004)
and Peters et al. (1999). No associations were observed between HRV and S02. The strongest associations
with HRV were observed for PM2 5 and 03.
Another study of air pollutants and HRV was conducted in Boston, MA on 497 men from the
Normative Aging Study (Park et al., 2005b). The best 4-consecutive-min interval from a 7-min sample
was used for the HRV calculations. For the exposure variable, 4, 24, and 48 h moving averages matched
on the time of the ECG measurement for each subject were considered. The mean 24-h avg S02
concentration was 4.9 ppb (range: 0.95, 24.7). Associations with measures of HRV were reported for
PM2 5 and 03, but not with S02 for any of the averaging periods. In another study conducted in Boston,
MA, Schwartz et al. (2005) found significant effects of increases in PM2 5 on measures of HRV, while
no associations with S02 were observed. Other studies examined the relationship of S02 with HRV (Chan
et al., 2005; de Paula Santos et al., 2005; Holguin et al., 2003; Luttmann-Gibson et al., 2006). Most of
these studies, with the exception of de Paula Santos et al., did not observe associations with S02.
In the limited number of epidemiologic studies that examined a possible effect of S02 on HRV,
there were some positive findings; however, results reported from the human clinical studies were
inconsistent. The overall evidence is insufficient to conclude that S02 has an effect on cardiac autonomic
control.
3.2.2.2. Repolarization Changes
In addition to the role played by the autonomic nervous system in arrhythmogenic conditions,
myocardial vulnerability and repolarization abnormalities are believed to be key factors contributing to
the mechanism of such diseases.
Two in vitro studies (Nie and Meng, 2005, 2006) conducted with a 1:3 molarmolar mixture of the
S02 derivatives bisulfite and sulfite demonstrated effects of a 10-|am bisulfite: sulfite mixture on sodium
and L-type calcium currents (which included changes in inactivation and/or activation, recovery from
inactivation, and inactivation/activation time constants) in ventricular myocytes. These in vitro
observations suggested a potential role for L-type calcium current in cardiac injury following S02
exposure. Additional toxicological studies are necessary to evaluate repolarization changes at ambient
levels of S02.
In an epidemiologic study, Henneberger et al. (2005) examined the association of repolarization
parameters (QT duration, T-wave complexity, variability of T-wave complexity, and T-wave amplitude)
with air pollutants in patients with preexisting coronary heart disease (n = 56, all males) in East Germany.
The patients were examined repeatedly once every 2 weeks for 6 months, for a total of 12 ECG
recordings. The mean 24-h avg S02 concentration was 2 ppb (range: 1, 4). Ambient S02 concentrations
during the 24-h preceding the ECG were associated with the QT interval duration, but not with any other
repolarization parameters. Stronger associations were observed between PM indices and QT interval
duration, T-wave amplitude, and T-wave complexity.
Evidence from the limited number of in vitro toxicological studies indicates that L-type calcium
current may have a role in cardiac injury following S02 exposure at higher than ambient concentrations.
In the single epidemiologic study of S02 and repolarization changes, an association between S02 and one
of several repolarization parameters examined (QT duration) was observed; however, stronger
associations were reported for PM.
3-36
-------
3.2.2.3. Cardiac Arrhythmias
One toxicological study examined the effects of PM, ultrafine carbon, and S02 on spontaneous
arrhythmia frequency in 18-month-old rats (Nadziejko et al., 2004). The rats were exposed to 1 ppm S02
for 4 h. No significant change in the frequency of spontaneous arrhythmias was found with S02 and
ultrafine carbon exposure. However, rats exposed to concentrated ambient PM had a significantly greater
increase in the frequency of delayed beats than rats exposed to air.
In a panel study of 100 patients with implanted cardioverter defibrillators (ICDs) in Eastern
Massachusetts, Peters et al. (2000a) tested the hypothesis that patients with ICDs would experience life-
threatening arrhythmias after an air pollution episode. The mean 24-h avg S02 concentration measured at
two sites in Boston during the study period was 7 ppb (5th-95th percentile: 1, 19). ICDs monitor ECG
abnormalities, and treat ventricular fibrillation or ventricular tachycardias by administering shock therapy
to restore the normal cardiac rhythm. The ICD device also stores information on each tachyarrhythmia
and shock. There was no association between S02 and defibrillator discharges in the 33 subjects who had
any defibrillator discharges during the follow-up period or in the 6 subjects who had at least 10
discharges. There was some evidence that N02 was associated with increased defibrillatory interventions
in the subjects with any defibrillator discharges. Among the patients with at least 10 events, N02, CO, and
PM2 5 were found to be associated with defibrillator discharges.
In a follow-up study designed to confirm the findings of Peters et al. (2000a), Dockery et al. (2005)
used a larger sample of ICD patients in Boston (n = 203) with a longer follow-up period. The median
concentration of 48-h avg S02 averaged across multiple sites in Boston was 4.9 ppb (IQR 4.1). No
significant associations were found between ventricular arrhythmic episode days and any of the air
pollutants. However, when the analysis was stratified by recent arrhythmias (i.e., within 3 days), there
was evidence of an excess risk of ventricular arrhythmia with S02, PM2 5, black carbon, N02, and CO.
Since PM2 5, black carbon, N02, and CO were correlated with each other and with S02, the authors noted
that differentiating the independent effects of the pollutants would be difficult. A case-crossover analysis
of the same data by Rich et al. (2005) also observed associations with 48-h avg S02, but the S02 effect
was not found to be robust to adjustment by PM2 5. In a similar study conducted in St. Louis, MO, an
excess risk was associated with S02 concentrations in the 24 h prior to an arrhythmia, but not with PM2 5
and 03 (Rich et al., 2006a). In this study, none of the other measured pollutants (PM, elemental carbon,
03, CO, N02) were correlated with S02. The authors suggested that the different effects observed in St.
Louis and Boston may be due to differences in the source or mix of air pollutants in these cities. Finally,
findings from a time series study of tachyarrythmic events among 518 patients over a 10 year period in
Atlanta do not indicate an association with S02, nor with the other pollutants studied including PM2 5 and
its components (Metzger et al., 2007).
Additional studies have examined the relationship of S02 with arrhythmias in Vancouver, and
observed associations at very low ambient S02 concentrations (mean 24-h avg S02 of -2.5 ppb with a
max of 8.1 ppb). Vedal et al. (2004) stated that of all pollutants examined, the only one with somewhat
consistent positive associations with arrhythmia events was S02. In season-stratified analyses, S02 was
positively associated with arrhythmias in the winter, while in the summer the association was negative.
On the other hand, in the Rich et al. (2004) study, positive associations were observed in the summer but
not in the winter. The authors stated that it was difficult to interpret these findings.
Collectively, the epidemiologic evidence for an association between short-term exposure to S02
and arrhythmias is inconsistent. The limited toxicological evidence does not provide biological plausiblity
for an effect.
3.2.2.4. Blood Pressure
Two animal toxicological studies examined the effect of S02 on blood pressure Halinen et al.
(2000a) examined blood pressure changes in guinea pigs which were hyperventilated to simulate exercise,
3-37
-------
and exposed to 1, 2.5, and 5 ppm S02 in cold, dry air. After 10-min exposures to each S02 concentration,
separated by 15-min exposures to clean, warm, humid air, a transient increase in blood pressure was
observed during 5 ppm S02 exposure in cold, dry air. In a second study (Halinen et al., 2000b),
hyperventilated guinea pigs were exposed to cold, dry air alone or to 1 ppm S02 in cold, dry air for 60
min. The study reported similar increases in blood pressure and HR with exposure to cold, dry air or to
S02 in cold, dry air. The increase in HR was gradual, while increases in blood pressure generally occurred
during the first 10 to 20 min of exposure. Similar effects were observed with exposure to cold, dry air or
to S02 in cold, dry air, suggesting that effects were associated with cold, dry air rather than with S02.
Ibald-Mulli et al. (2001) examined the association between blood pressure and S02 using survey
data from the MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) Project. Blood
pressure measurements were taken from 2,607 men and women. The mean 24-h avg S02 concentration
was 23 ppb (range: 5, 91). An increase in systolic blood pressure was associated with 24-h avg S02 and
TSR However, in a two-pollutant model with TSP, the effect of S02 on blood pressure was substantially
reduced and became nonsignificant while the effect of TSP was robust.
In a study by de Paula Santos et al. (2005), changes in blood pressure in association with S02 were
investigated in vehicular traffic controllers (n = 48) aged 31 to 55 years living in Sao Paulo, Brazil, where
vehicles are the primary source of air pollution. The mean 24-h avg S02 level, measured at six different
stations around the city, was 7 ppb (SD 3). Blood pressure was measured every 10 min when subjects
were awake (6 a.m. to 11 p.m.) and every 20 min during sleep (11 p.m. to 6 a.m.). Results indicated that
S02, as well as CO, were associated with increases in systolic and diastolic blood pressure. However,
when a two-pollutant model was used to test the robustness of the associations, only the CO effect
remained statistically significant.
Very few studies have examined the effects of short-term S02 exposure on blood pressure.
Collectively, the limited toxicological and epidemiologic evidence that exposure to S02 has effects on
blood pressure is inconclusive.
3.2.2.5. Blood Markers of Cardiovascular Risk
Folsom et al. (2001) demonstrated that elevated levels of fibrinogen, white blood cell count, factor
VIII coagulant activity (factor VIII-C), and von Willebrand factor were associated with risk of
cardiovascular disease. Schwartz (2001) investigated the association between various blood markers of
cardiovascular risk and air pollution among subjects in the Third National Health and Nutrition
Examination Survey (NHANES III) in the U.S. conducted between 1989 and 1994 across 44 counties.
The NHANES III is a random sample of the U.S. population with oversampling for minorities (30% of
NHANES sample) and the elderly (20% of the sample). The mean S02 concentration was 17.2 ppb (IQR
17) across the 25 counties where data were available. This study looked at fibrinogen levels, platelet
counts, and white blood cell counts. After controlling for age, ethnicity, gender, body mass index, and
smoking status and number of cigarettes per day, S02 was found to be positively associated with white
blood cell counts. PMi0 was associated with all blood markers. In two-pollutant models, PMi0 remained a
significant predictor of white blood cell counts after controlling for S02, but not vice versa.
A recent cross-sectional study (Liao et al., 2005) investigated the effects of air pollution on plasma
hemostatic and inflammatory markers in the ARIC study (n = 10,208). The authors hypothesized that
short-term exposure to air pollutants was associated with increased levels of inflammatory markers and
lower levels of albumin, as serum albumin is inversely associated with inflammation. The mean 24-h avg
S02 concentration was 5 ppb (SD 4). Significant curvilinear relationships were observed between S02 and
factor VIII-C, white blood cell counts, and serum albumin. The authors noted that since no biological
explanation could be offered for the "U"-shaped curve between S02 and factor VIII-C and the "inverse
U"-shape between S02 and albumin, generalization of the association should be exercised with caution.
No associations were observed between S02 and fibrinogen or von Willebrand factor.
3-38
-------
In another large cross-sectional study of 7,205 office workers in London, Pekkanen et al. (2000)
examined the association between plasma fibrinogen and ambient air pollutants. The mean 24-h avg S02
was 9 ppb (10th-90th percentile: 5, 19). Associations with fibrinogen were observed for all pollutants
examined, either in all-year or summer-only analyses, except for S02 and 03.
Taken together, results from the limited number of studies is insufficient to determine the effect of
S02 on various blood markers of cardiovascular risk.
3.2.2.6. Acute Myocardial Infarction
The association between air pollution and the incidence of MI was examined in a small number of
studies. As part of the Determinants of Myocardial Infarction Onset Study, Peters et al. (2001) examined
772 patients with MI living in greater Boston, MA. A case-crossover design was used to assess changes in
the risk of acute MI after exposure to potential triggers. The mean 24-h avg S02 was 7 ppb (range: 1, 20)
during the study period. Similarly, the mean 1-h avg S02 was 7 ppb (range: 0, 23). In an analysis that
considered both the 2-h avg (between 60 and 180 min before the onset of symptoms) and 24-h avg
(between 24 and 48 h before the onset) concentrations jointly, the study found no significant association
between risk of MI and S02. Of all the pollutants considered, only PM2 5 and PMi0 were found to be
associated with an excess risk of MI. In a study of 5,793 confirmed cases of acute MI in King County
Washington, Sullivan et al. (2005) also used a case-crossover design to investigate the association of
ambient levels of several air pollutants 1, 2, 4 and 24 h before the MI onset. No association with S02 (or
with PM2 5) was observed. The mean S02 level was 9 ppb (range: 0-39 ppb) at the time of the study.
In the MONICA Project, the effect of air pollution on acute MI was studied in Toulouse, France,
using a case-crossover study design (Ruidavets et al., 2005b). The mean 24-h avg S02 level was 3 ppb
(5th-95th percentile: 1, 5). A total of 399 cases of acute MI were recorded during the study period. 03, but
not S02, was found to be associated with the incidence of acute MI. Exposure to PM was not considered
in this study.
Only a limited number of studies examined the association between ambient S02 concentrations
and incidence of acute MI. These studies provide no evidence that exposure to S02 increases the risk of
MI.
3.2.2.7. Emergency Department Visits and Hospitalizations for Cardiovascular
Diseases
The current review includes more than 30 peer-reviewed studies that address the effect of S02
exposure on ED visits or hospitalizations for cardiovascular diseases. These studies are discussed briefly
in this section and further summarized in Annex Table F-4.
All Cardiovascular Diseases
The disease grouping of all cardiovascular diseases typically includes all diseases of the circulatory
system (e.g., heart diseases and cerebrovascular diseases, ICD9 Codes 390-459). A summary of the
associations reported for ambient S02 concentrations with all cardiovascular diseases are presented in
Figure 3-9.
In a study of 11 cities in Spain, an excess risk of 3.6% (95% CI: 0.6, 6.7) per 10 ppb increase in
24-h avg S02 at a 0-1 day lag was observed for all cardiovascular disease admissions (Ballester et al.,
2006). The mean 24-h avg S02 level in the cities studied was 6.6 ppb. In addition, time-series data linking
S02 with hospital admissions for cardiovascular diseases in three metropolitan areas in the U.S. (i.e.,
Cook, Maricopa, Los Angeles Counties) was conducted (Moolgavkar 2000; reanalysis 2003). Among
older adults (65+ years) in Los Angeles County, a 13.7% (95% CI: 11.3, 16.1) excess risk in admissions
3-39
-------
was observed per 10 ppb increase in 24-h avg S02 at lag 0 day, in the reanalysis using a Generalized
Linear Model (GLM) and natural splines to adjust for temporal trends rather than GAM. The median
24-h avg S02 level for Los Angeles County was 2 ppb during the study period. Results for Maricopa and
Cook counties were not presented in the reanalysis.
Reference
Location
Lag
| All Ages |
Metzger et al. (2004)*
Atlanta, GA
0-2
i •
Ballester et al. (2006)
14 Spanish cities
0-1
i
i—®—
Atkinson et al. (1999)
London, England
0
i
Poloniecki et al. (1997)
London, England
1
i
i
Ballester et al. (2001)
Valencia, Spain
2
i
i
i
Llorca et al. (2005)
Torrelavega, Spain
0
i
i
i
Petroeschevsky et al. (2001)
Brisbane, Australia
0
i
i
Chang et al. (2005)
Taipei, Taiwan
0-2
< o 2>£0°C
i
1 ®—
i
i
i
i
<20°C
Atkinson et al. (1999)
London, England
0
i
1 0
1
| 15-64yrs |
Petroeschevsky et al. (2001)
Brisbane, Australia
0
1
1
1
1
1
Moolgavkar (2003)
Los Angeles, CA
0
1
1
1
0 | 65+ yrs
Atkinson et al. (1999)
London, England
0
1
1—®—
Jalaludin et al. (2006)*
Sydney, Australia
0
1
1
1
Petroeschevsky et al. (2001)
Brisbane, Australia
1
1
1 o
1
0.8 1.0 1.2 1.4
Relative Risk
Figure 3-9. Relative risks (95% CI) of S02-associated emergency department visits (*) and
hospitalizations for all cardiovascular causes, arranged by age group. Risk estimates
are standardized per 10 ppb increase in 24-h avg SO2 concentrations or 40 ppb
increase in 1-h max SO2.
In a large single city study Metzger et al. (2004) examined approximately 4.4 million hospital visits
to 31 hospitals from 1993 to 2000 in Atlanta, GAand reported a 1.4% (95% CI: -1.5, 4.4) excess risk in
ED visits for cardiovascular causes per 40-ppb increase in 1-h max S02. Peel et al. (2006) conducted
3-40
-------
analyses using the same dataset to compare results obtained with a case-crossover design to the Metzer et
al. (2007) results, which were obtained using a time series approach. Peel et al. (2006) and Metzger et al.
(2004) report similar findings. The median 1-h max S02 level in Atlanta during the study period was 11
ppb (10th-90th percentile: 2-39). Results from several single-city studies in Europe, Australia, and
Taiwan indicated positive associations with S02 (Atkinson et al., 1999a; Ballester et al., 2001; Jalaludin et
al., 2006; Petroeschevsky et al., 2001; Poloniecki et al., 1997), though others observed negative
associations (Chang et al., 2005; Llorca et al., 2005) (see Figure 3-9).
Specific Cardiovascular Diseases
Several studies examined the effect of ambient S02 on hospital admissions for cardiac disease
(ICD9 Codes 390-429), ischemic heart disease (IHD, ICD9 Codes 410-414), dysrhythmia (ICD9 Code
427), congestive heart failure (CHF, ICD9 Code 428), MI (ICD9 Code 410) or cerebrovascular diseases
(ICD9 Codes 430-438). In a study of the six cities of Metropolitan Toronto, Burnett et al. (1997b)
reported an association between cardiac hospitalizations and ambient S02 (23.8% [95% CI: 5.5, 45.2])
excess risk per 40 ppb increase in 1-h max S02) that was robust to the inclusion of other gaseous
pollutants in the model. PM was not identified as an independent risk factor for hospital admissions in
this study. A European multicity study reported statistically significant positive associations with cardiac
disease admissions (Ballester et al., 2006). However, adjustment for PMi0 and CO in two-pollutant
models diminished the association reported by Ballester et al. by approximately 50%. Findings for cardiac
disease admissions reported in several additional single city studies conducted in the U.S., Canada,
Australia and Europe were inconsistent (Fung et al., 2005b; Jalaludin et al., 2006; Llorca et al., 2005;
Michaud et al., 2004).
Analyses restricted to diagnoses of IHD (Jalaludin et al., 2006; Lee et al., 2003b; Lin et al., 2003a;
Metzger et al., 2004; Peel et al., 2007), CHF (Burnett et al., 1997c; Koken et al., 2003; Metzger et al.,
2004; Morris et al., 1995; Peel et al., 2007; Wellenius et al., 2005a) dysrhythmia (Koken et al., 2003;
Metzger et al., 2004; Peel et al., 2007), MI (Koken et al., 2003; Lin et al., 2003a), and angina pectoris
(Hosseinpoor et al., 2005) were also conducted. Metzger et al. (2004) observed weak nonsignificant or
negative associations of 1-h max S02 with IHD, CHF, and dysrhythmia. Using the same dataset, Peel et
al. (2007) investigated effect modification of cardiovascular disease outcomes across comorbid disease
status categories, including hypertension, diabetes, COPD, dysrhythmia, and CHF. Authors observed only
weak nonsignificant or negative associations for IHD, CHF, and dysrhythmia with ambient 1-h max S02
level in any comorbid disease category. Both increases in admissions or ED visits (Jalaludin et al., 2006;
Koken et al., 2003; Wellenius et al., 2005a) and weak or negative associations (Burnett et al., 1997;
Hosseinpoor et al., 2005; Lee et al., 2003b; Lin et al., 2003a) were reported in other studies.
Studies of the effect of S02 on cerebrovascular admissions were also considered. Positive
associations were reported for ischemic stroke (Villeneuve et al., 2006a; Wellenius et al., 2005a, 2005b).
However, Wellenius et al. (2005b) reported stronger associations for N02 and CO than S02, and the
association reported by Villenueve et al. (2006a) was diminished in two-pollutant models. No meaningful
positive associations of ambient S02 with cerebrovascular diseases were observed in several other studies
(Henrotin et al., 2007a; Jalaludin et al., 2006; Metzger et al., 2004; Peel et al., 2007; Tsai et al., 2003a).
Summary of Evidence on Emergency Department Visits and Hospitalizations from
Cardiovascular Diseases
Several studies have observed positive associations between ambient S02 concentrations and ED
visits or hospital admissions for cardiovascular diseases (e.g., all cardiovascular diseases, cardiac
diseases, cerebrovascular diseases) particularly among individuals 65+ years of age, but results are not
consistent across studies. The strongest evidence comes from a large multicity study conducted in Spain
Ballester et al. (2006) that observed statistically significant positive associations between ambient S02
3-41
-------
and cardiovascular disease admissions; however, the S02 effect was found to diminish by half with PMi0
and CO adjustment. Only a limited number of studies assessed potential confounding by copollutants
despite the moderate to strong correlation between S02 and various copollutants in most studies. While
some studies indicated that the association of S02 with cardiovascular hospitalizations were generally
robust to adjustment for BS and PMi0 (Ballester et al., 2001; Fung et al., 2005b), several other studies,
including that by Balleseter et al. (2006), observed that the effect of S02 on cardiovascular ED visits and
hospitalizations may be confounded by copollutant exposures. Jalaludin et al. (2006) reported a 3%
excess risk in cardiovascular disease hospital admissions per 0.75 ppb incremental change in 24-h avg
S02 in single-pollutant models, which was reduced to null when CO was included. Chang et al. (2005)
noted that the observed negative association of S02 with all cardiovascular disease hospitalizations was
strengthened after adjusting for N02, PMi0, and CO in two-pollutant models. The authors attributed this
finding to possible collinearity problems between S02 and copollutants. None of the epidemiologic
studies examined effects of possible interactions among copollutants.
3.2.2.8. Summary of Evidence on the Effects of Short-Term SO2 Exposure on
Cardiovascular Health
Biologically plausible modes of action (e.g., vagally-mediated irritant responses and oxidative
injury) that could explain short-term S02 effects on the cardiovascular system were summarized in a
previous section of this chapter (Section 3.1.2). However, consideration of these modes of action in light
of findings from additional animal toxicological, human clinical, and epidemiologic studies has led to the
conclusion that the evidence as a whole is inadequate to infer a causal relationship.
Specifically, evidence from human clinical and epidemiologic studies of HRV in healthy persons as
well as persons with asthma or cardiovascular disease was inconsistent and did not support an effect of
S02 on the autonomic nervous system, despite some positive findings. In the single epidemiologic study
of S02 and repolarization changes, an association with QT interval duration was observed. While in vitro
studies suggested a potential role for L-type calcium current in cardiac injury, the relevance of these
studies to ambient exposures is unknown. Epidemiologic evidence from studies of the effect of S02 on
ICD-recorded arrhythmias was inconsistent. Furthermore, studies of blood pressure and blood markers of
cardiovascular risk failed to provide consistent evidence to suggest a role for S02 in cardiovascular
disease development. Finally, although some studies of hospital admissions and ED visits for
cardiovascular diseases reported positive and statistically significant associations with S02, findings were
inconsistent across this body of literature as a whole. Many researchers were unable to distinguish the
effect of S02 from correlated copollutants while others reported a reduction in the S02 effect in two-
pollutant models. The inconsistency of the evidence, lack of coherence across and within disciplines, as
well as limitations inherent to the observational studies (e.g., inadequate control of copollutant exposures)
contributed to this causal determination.
3.2.3. Other Effects Associated with Short-Term SO2 Exposure
The short-term effects of S02 on other organ systems were not examined in the previous review. A
review of animal toxicological studies published since the 1982 AQCD indicates a limited number of
research inquiries addressing the systemic effects of short-term S02 exposure in various other organs. The
most recent studies on these are summarized in Annex Tables E-12 through E-16.
Of note are three ex vivo acute exposure studies using S02 derivatives (sulfite and bisulfite) on
hippocampal or dorsal root ganglion neurons isolated from Wistar rats (Du and Meng, 2004a, 2004b,
2006). Perturbations were observed in potassium-, sodium-, and calcium-gated channels at concentrations
of 0.01-100 |_iM. These authors speculated that such effects might correlate with the neurotoxicity that has
3-42
-------
been associated with S02 inhalation. However effects on the nervous system have generally been studied
using chronic exposures > 5 ppm S02. Effects observed at these levels are of questionable significance in
evaluating the health effects at ambient levels. These studies are summarized in Annex Table E-12.
3.3. Mortality Associated with Short-Term SO2 Exposure
3.3.1. Summary of Findings from the Previous Review
The studies available to review in the 1982 AQCD were mostly from historical data including
London, England, and New York City air pollution episodes. Effects of SOx (mainly S02) were
investigated along with PM indices because they shared a common source, coal burning, and separating
their associations with mortality was a challenge that many of the earlier episodic studies could not
resolve. The S02 levels observed in these air pollution episodes were several orders of magnitude higher
than the current avg levels observed in U.S. cities (e.g., in the 1962 New York City episode, S02 in
Manhattan peaked at 400 to 500 ppb). Some of these London and New York City studies suggested that
PM, not S02, was associated with observed mortality, but the 1982 AQCD could not resolve the relative
roles of these two pollutants and suggested that the clearest mortality associations were seen when both
pollutants were at high levels (24-h avg values of both BS and S02 exceeding 1000 |_ig/m3 [-400 ppb for
S02]) and less so at lower ranges although the review of the studies and reanalyses found no clear
evidence of a threshold for S02.
The 1986 Second Addendum to the 1982 AQCD reviewed more reanalyses of the London data and
analyses of New York City, Pittsburgh, and Athens data. While these reanalyses and some new analyses
confirmed earlier findings (and suggested stronger evidence of BS effects than of the S02 effects), given
the remaining uncertainties with exposure error and statistical modeling, there was not sufficient
information to quantitatively determine concentration-response relationships at lower concentrations of
either PM or S02.
A series of short-term mortality effects studies in the late 1980s and early 1990s (Dockery et al.,
1992; Fairley, 1990; Pope, 1989; Pope et al., 1992; Schwartz and Dockery, 1992a, 1992b) showed
associations between mortality and PM indices at relatively low levels. Since then, a large number of
epidemiologic studies have investigated the adverse health effects of air pollution with hypotheses mainly
focused on PM, and S02 was often analyzed as one of the potential confounders in these studies.
3.3.2. Mortality and Short-Term SO2 Exposure in Multicity Studies and
Meta-Analyses
In reviewing the range of S02 mortality effect estimates, multicity studies provide especially useful
information because they analyze data from multiple cities using a consistent method, avoiding potential
publication bias. There have been several multicity studies from the U.S., Canada, and Europe, some of
which will be discussed in the sections below. Meta-analysis studies also provide useful information on
describing heterogeneity of effect estimates across studies; however, in contrast to multicity studies, the
observed heterogeneity may reflect the variation in analytical approaches across studies. In addition, the
effect estimate from a meta-analysis may be subject to publication bias, unless the analysis specifically
examines such bias and adjusts for it. These studies, as well as many other single-city studies, are
summarized in Annex Table F-5.
3-43
-------
3.3.2.1. Multicity Studies
National Morbidity, Mortality, and Air Pollution Study
The time-series analysis of the largest 90 U.S. cities (Samet et al., 2000a; reanalysis Dominici et
al., 2003) in the National Morbidity, Mortality, and Air Pollution Study (NMMAPS) is by far the largest
multicity study conducted to date to investigate the mortality effects of air pollution, but its primary focus
was PM10. It should also be noted that, according to the table of mean pollution levels in the original
report (Samet et al., 2000a), S02 data were missing in 28 of 90 cities. Annual 24-h avg mean S02 levels
ranged from 0.4 ppb (Riverside, CA) to 14.2 ppb (Pittsburgh, PA), with a mean of 5.9 ppb during the
study period of 1987 to 1994. The analysis in the original report used GAM models with default
convergence criteria. Dominici et al. (2003) reanalyzed the data using GAM with stringent convergence
criteria as well as using GLM. It should be noted that this model's adjustment for weather effects employs
more terms than other time-series studies in the literature, suggesting that the model adjusts for potential
confounders more aggressively than the models in other studies.
A = SO; alone
B = SOJ + PM,0
c = soj + pmI0+o3
d = so2 + pm,0 + no,
E = SO, + PM,„ + CO
BCD
Models
Source: Dominici et al. (2003).
Figure 3-10. All cause mortality excess risk estimates for SO2 from the National Morbidity,
Mortality, and Air Pollution Study. Posterior means and 95% posterior intervals of
national average estimates of SO2 effects on all-cause mortality from non-external
causes per 10 ppb increase in 24-h avg SO2 at 0,1, and 2-day lags within sets of the 62
cities with pollutant data available.
Figure 3-10 shows the all-cause mortality excess risk estimates for S02 from Dominici et al.
(2003). The mortality excess risk estimate with a 1-day lag was 0.60% (95% CI: 0.26, 0.95) per 10 ppb
increase in 24-h avg S02. PMi0 and O , (in summer) appeared to be more strongly associated with
3-44
-------
mortality compared to the other gaseous pollutants. The model with PMi0 and N02 resulted in an
appreciably reduced S02 excess risk estimate, 0.38% (95% CI: -0.62, 1.38) per 10 ppb increase in
24-h avg S02. These results suggest that the observed S02-mortality association could be confounded by
PMio and N02. The authors stated that the results did not indicate associations of S02, N02, and CO with
all-cause mortality.
Canadian Multicity Studies
There have been three Canadian multicity studies conducted by the same group of investigators
examining the association between mortality and short-term exposure to air pollutants (Burnett et al.,
1998a; 2000; 2004). This section focuses on Burnett et al. (2004) as this study is the most extensive
Canadian multicity study, both in terms of the length and coverage of cities. The discussion in this study
focused on N02, because N02 was the best predictor of short-term mortality fluctuations among the
pollutants. This was also the case in the Burnett et al. (1998a) study of the gaseous pollutants in 11
Canadian cities. The mean 24-h avg S02 levels across the 12 cities was 5.8 ppb, with city means ranging
from 1 ppb in Winnipeg to 10 ppb in Halifax. The population-weighted avg was 5 ppb. The mean S02
levels in this study were similar to those in the NMMAPS (mean 24-h avg S02 levels across the 62
NMMAPS cities was 5.9 ppb).
All-cause (nonaccidental), cardiovascular, and respiratory mortality were analyzed in Burnett et al.
(2004). For S02, PM2 5, PMio_2 5, PMi0 (arithmetic addition of PM2 5 and PMio_2 5), CoH, and CO, the
strongest mortality association was found at a 1-day lag, whereas for N02, it was the 3-day moving avg
(i.e., avg of 0, 1, and 2-day lags), and for 03, it was the 2-day moving avg. The daily 24-h avg values
showed stronger associations than the daily 1-h max values for all the gaseous pollutants and CoH except
for 03. The S02 all-cause mortality excess risk estimate was 0.74% (95% CI: 0.29, 1.19) per 10 ppb
increase in the 24-h avg S02 with a 1-day lag. After adjusting for N02, the S02 effect estimate was
reduced to 0.42% (95% CI: 0.01, 0.84), while the N02 effect estimate was only slightly affected. In this
analysis, no regression analysis using both S02 and PM was conducted. The Burnett et al. (2000) analysis
observed that the simultaneous inclusion of S02 and PM2 5 in the model reduced the S02 effect estimate
by half, whereas the PM2 5 estimate was only slightly reduced. Overall, these results suggest that S02 was
not an important predictor of daily mortality in the Canadian cities and that its mortality associations
could be confounded with N02 or PM.
Air Pollution and Health: A European Approach
Several Air Pollution and Health: a European Approach (APHEA) analyses have reported S02
mortality excess risk estimates. Katsouyanni et al. (2006) examined the association of PMi0, BS, and S02
with all-cause mortality in 12 European cities using the standard APHEA (GLM) approach. The same data
set was reanalyzed to adjust for the seasonal cycles (Samoli et al., 2001; 2003). The reanalysis by Samoli
et al. (2003) produced results that were similar to those in the original analysis by Katsouyanni et al.
(2006). Since the original analysis presented more results, including multipollutant model results,
discussion will focus on this analysis.
The study by Katsouyanni et al. (1997) includes seven western European cities (Athens, Barcelona,
Cologne, London, Lyon, Milan, and Paris) and five central eastern European cities (Bratislava, Kracow,
Lodz, Poznan, and Wroclaw). The data covered at least 5 consecutive years for each city within the years
1980 through 1992. The S02 levels in these cities were generally higher than in the U.S. or Canada, with
the median 24-h avg S02 ranging from 5 ppb in Bratislava to 28 ppb in Kracow. Analysis was restricted to
days when PM and S02 concentrations did not exceed 200 (.ig/ni1 (76 ppb for S02) to evaluate the effects
of moderate to low exposures. The data were analyzed by each center separately following a standardized
method, but the lag for the "best" model was allowed to vary in these cities from 0 to 3 days. The city-
specific effect estimates were then examined in the second stage for source of heterogeneity using city-
3-45
-------
specific variables such as mean pollution and weather variables, accuracy of the air pollution
measurements, health of the population, smoking prevalence, and geographical differences.
The city-specific estimates were found to be heterogeneous and, among the explanatory variables,
only the separation between western and central eastern European cities resulted in more homogeneous
groups. The all-cause mortality excess risk estimates were 1.14% (95% CI: 0.88, 1.39), 1.99% (95% CI:
1.15, 2.83), and 0.46% (95% CI: -0.23, 1.15) for all the 12 cities combined, western cities, and central
eastern cities, respectively, per 10 ppb increase in the 24-h avg S02 at variable single-day lags. Zmirou et
al. (1998) analyzed cardiovascular and respiratory mortality in 10 of the 12 APHEA cities and observed
that the cause-specific mortality excess risk estimates were higher than those for all-cause mortality. As in
the analyses of all-cause mortality, S02 effect estimates for these cause specific deaths were higher in
western European cities than in central eastern European cities.
Seasonal analyses indicated that the summer estimate was slightly higher than the winter estimate
in the western cities, but the difference was not statistically significant. The results for the two-pollutant
model with S02 and BS were presented for the western cities, with a similar extent (-30%) of reductions
in the estimates of both pollutants (1.31% [95% CI: 0.40, 2.23] for S02). Furthermore, for western cities,
they estimated effects for S02 for days with high or low BS levels and the corresponding BS effects for
days with high or low S02 levels and found that the effects were similar in the stratified data. From these
results, Katsuoyanni et al. (2006) suggested that the effects of the two pollutants were independent.
Overall, the APHEA studies provide some suggestive evidence that the effect of short-term
exposure to S02 on mortality is independent of PM. This is somewhat in contrast to the U.S. and
Canadian studies. The S02 levels were much higher in the European cities, but the type of PM
constituents also might be different.
The Netherlands Study
In the Netherlands studies by Hoek et al. (2000; 2001; renalysis by Hoek, 2003) the association
between air pollutants and mortality was examined in a large population (14.8 million for the entire
country) over the period of 1986 through 1994. The Netherlands were not part of the APHEA analysis.
The median 24-h avg S02 level in the Netherlands was 4 ppb (6 ppb for the four major cities). All the
pollutants examined, including PMi0, BS, 03, N02, S02, CO, sulfate, and nitrate, were associated with all-
cause mortality, and for single-day models, a 1-day lag showed the strongest associations for all the
pollutants. The following effect estimates are all from the GLM models with natural splines for
smoothing functions. The S02 excess risk estimate in a single-pollutant model was 1.31% (95% CI: 0.69,
1.93) per 10 ppb increase in 24-h avg S02 at a 1-day lag and 1.78% (95% CI: 0.86, 2.70) at an avg of 0-
to 6-day lag. Seasonal analyses showed slightly greater effect estimates during the summer compared to
the winter. S02 was most highly correlated with BS (r = 0.70). The simultaneous inclusion of S02 and BS
reduced the effect estimates for both pollutants (S02 effect estimate was 1.07% [95% CI: -0.27, 2.42] per
10 ppb increase with an avg of 0- to 6-day lag of 24-h avg S02). PMi0 was less correlated with S02 (r =
0.65), and the simultaneous inclusion of these pollutants resulted in an increase in the S02 effect estimate.
These results from the analysis of the Netherlands data suggested some indication of confounding
between S02 and BS.
Cause specific analysis showed larger excess risk estimates for COPD (3.61% [95% CI: -0.29,
7.67] per 10 ppb increase in the avg of 0- through 6-day lags of 24-h avg S02) and pneumonia (6.56%
[95% CI: 1.16, 12.24]) deaths compared to that from all causes, but because essentially all of the
pollutants showed larger effect estimates for these sub-categories, it is difficult to interpret these estimates
as effects of S02 alone. Similarly, the effect estimates for heart failure (7.1% [95% CI: 2.6, 11.7]) and
thrombosis-related deaths (9.6% [95% CI: 3.1, 16.6]) were larger than that for total cardiovascular (2.7%
[95% CI: 1.3,4.1]) causes. Since the same pattern was seen for other pollutants as well, it is difficult to
interpret these cause-specific effect estimates due to S02 alone or any one of the pollutants analyzed.
3-46
-------
Other European Multicity Studies
Other European multicity studies were conducted in 8 Italian cities (Biggeri et al., 2005), 9 French
cities (Le Tertre et al., 2002), and 13 Spanish cities (Ballester et al., 2002). The studies by Le Tertre et al.
and Ballester et al. were conducted using GAM methods with the default convergence setting.
Biggeri et al. (2005) analyzed eight Italian cities (Turin, Milan, Verona, Ravenna, Bologna,
Florence, Rome, and Palermo) for mortality and hospital admissions (mortality data were not available for
Ravenna and Verona). The study period varied from city to city between 1990 and 1999. Only single-
pollutant models were examined in this study. The S02 excess risk estimates were 4.14% (95% CI: 1.05,
7.33), 4.94% (95% CI: 0.41, 9.67), and 7.37% (95% CI: -3.58, 19.57) per 10 ppb increase with an avg of
0- and 1-day lag of 24-h avg S02 for all-cause, cardiovascular, and respiratory deaths, respectively. Since
all the pollutants showed positive associations with these mortality categories and the correlations among
the pollutants were not presented, it is not clear how much of the observed associations are shared or
confounded. The mortality excess risk estimates were not heterogeneous across cities for all the gaseous
pollutants. It should be noted that in Turin, Milan, and Rome, the mean S02 values declined by 50% from
the first half to the second half of the study period, while the levels of other pollutants declined by smaller
fractions. This also complicates the interpretation of S02 effect estimates in this study, which are much
higher than those from the APHEA studies.
The Le Tertre et al. (2002) study of nine French cities examined BS, S02, N02, and 03 by generally
following the APHEA protocol, but using GAM with default convergence criteria and using the avg of
lags 0 and 1 day for combined estimates. S02 data were not available in one of the nine cities (Toulouse).
All four pollutants were positively associated with mortality outcomes. The study did not report
descriptions of correlation among the pollutants or conduct multipollutant models, and therefore, it is
difficult to assess the potential extent of confounding among these pollutants. The S02 effect estimates
were homogeneous across cities, with the exception of Bordeaux, which was the only city that used strong
acidity as a proxy for S02.
The Spanish Multicentre Study on Air Pollution and Mortality (EMECAM) examined the
association of PM indices (i.e., PMi0, TSP, BS) and S02 with mortality in 13 cities (Ballester et al., 2002).
These studies followed the APHEA protocol, but using GAM with default convergence criteria. The daily
mean 24-h avg S02 concentrations ranged from 3 to 17 ppb. In the seven cities where 1-h max S02 data
were also available, mean concentrations ranged from 21 to 43 ppb. The combined effect estimates for
all-cause and respiratory mortality were statistically significant for both 24-h avg S02 and 1-h max S02.
Controlling for PM indices substantially diminished the effect estimates for 24-h avg S02, but not for
1-h max S02. The authors reported that these results could indicate an independent impact of peak values
of S02 more than an effect due to a longer exposure.
3.3.2.2. Meta-Analyses of Air Pollution-Related Mortality Studies
Meta-Analysis of All Criteria Pollutants
Stieb et al. (2002) reviewed time-series mortality studies published between 1985 and 2000, and
conducted a meta-analysis to estimate combined effects for PM10, CO, N02, 03, and S02. Since many of
the studies reviewed in that analysis used GAM with default convergence parameters, Stieb et al. (2003)
updated the estimates by separating the GAM versus non-GAM studies. In addition, separate combined
estimates were presented for single- and multipollutant models. There were more GAM estimates than
non-GAM estimates for all the pollutants except for S02. For S02, there were 29 non-GAM estimates
from single-pollutant models and 10 estimates from multipollutant models. The lags and multiday
averaging used in these estimates varied. The combined estimate for all-cause mortality was 0.95% (95%
CI: 0.64, 1.27) per 10 ppb increase in 24-h avg S02 from the single-pollutant models and 0.85% (95% CI:
0.32, 1.39) from the multipollutant models. Because these estimates are not from an identical set of
3-47
-------
studies, the difference (or lack of a difference, as in this case) between the two estimates may not
necessarily be due to the effect of adding a copollutant in the model. Note that the data extraction
procedure of this meta-analysis for the multipollutant models was to include from each study the
multipollutant model that resulted in the greatest reduction in effect estimates compared with that
observed in single-pollutant models. It should also be noted that all the multicity studies whose combined
estimates have been discussed in the previous section were published after this meta-analysis.
Health Effects Institute Review of Air Pollution Studies in Asia
The Health Effects Institute (HEI) conducted a comprehensive review of air pollution health effects
studies in Asia (HEI, 2004) that summarized the results from mortality and hospital admission studies
published in the peer-reviewed scientific literature from 1980 through 2003. Of the 138 papers the report
identified, most were studies conducted in East Asia (mainland China, Taipei, Hong Kong, South Korea,
and Japan). The levels of S02 in these Asian cities were generally higher than in U.S. or Canadian cities,
with more than half of these studies reporting mean 24-h avg S02 levels of > 10 ppb. Based on a
comparison of the reported mean S02 levels from the same cities in different time periods, it is clear that
the S02 levels declined significantly in the 1990s. The meta-analysis used the most recent estimate for
each city to reflect recent pollution levels. Based on the criteria of having at least one year of data, model
adjustment for major time-varying confounders, and reporting effect estimates per unit increase in air
pollution, the meta-analysis included 28 time-series studies (11 from South Korea, 6 from mainland
China, 6 from Hong Kong, and 1 each from Taipei, India, Singapore, Thailand, and Japan). The lags
selected to compute combined estimates were inevitably variable; a systematic approach was used to
favor the a priori lag stated in the study, followed by the most significant lag, and then the largest effect
estimate.
Among the health outcomes examined in the meta-analysis, all-cause mortality was addressed in
the largest number of studies (13 studies) and S02 was the most frequently studied pollutant (11 studies).
The report generally focused on the results of single-pollutant models, as there were too few studies with
results of comparable multipollutant models to allow meaningful analysis. The S02 mortality effect
estimates showed evidence of heterogeneity. The combined estimate for all-cause mortality was 1.49%
(95% CI: 0.86, 2.13) per 10 ppb increase in 24-h avg S02. One of the limitations noted in the report was
that some degree of publication bias was present in these studies.
3.3.3. Evidence of the Effect of SO2 on Mortality from an Intervention
Study
Many time-series studies provide estimates of excess risk of mortality, but a question remains as to
the likelihood of a reduction in deaths when S02 levels are actually reduced. A sudden change in
regulation in Hong Kong in July 1990 required the conversion to fuel oil with low sulfur content. The
reduction in respiratory symptoms among children living in the polluted district in Hong Kong after the
intervention was previously discussed in Section 3.1.5. Hedley et al. (2002) examined changes in
mortality rates following the intervention. The S02 levels after the intervention declined about 50% (from
about 17 ppb to 8 ppb), but the levels for PMi0, N02, and sulfate did not change and 03 levels slightly
increased. The seasonal mortality analysis results showed that the apparent reduction in seasonal death
rate occurred only during the first winter, and this was followed by a rebound (i.e., higher than expected
death rate) in the following winter, then returned to the expected pattern three to five years after the
intervention. Using Poisson regression of the monthly deaths, the avg annual trend in death rate
significantly declined after the intervention for all causes (2.1%), respiratory causes (3.9%), and
cardiovascular causes (2.0%), but not from other causes. These results seem to suggest that a reduction in
S02 leads to an immediate reduction in deaths and a continuing decline in the annual trend in death rates.
3-48
-------
Hedley et al. (2002) estimated that the expected average gain in life expectancy per year due to the lower
S02 levels was 20 days for females and 41 days for males.
Interpreting these results is somewhat complicated by an upward trend in mortality across the
intervention point, which the authors noted was due to increased population size and aging. The results
suggest that such an upward trend is less steep after the introduction of low sulfur fuel. While the Poisson
regression model of monthly deaths does adjust for trend and seasonal cycles, the regression model does
not specifically address the influence of influenza epidemics, which can vary from year to year. This issue
also applies to the analysis of warm to cool season change in death rates. The most prominent feature of
the time-series plot (or the fitted annual cycle of monthly deaths) presented in this study is the lack of a
winter peak for respiratory and all-cause mortality during the year immediately following the
intervention. Much could be made of this lack of a winter peak, but no discussion of the potential impact
of (or a lack of) influenza epidemics is provided. These issues complicate the interpretation of the
estimated decline in upward trend of mortality rate or the apparent lack of winter peak.
The decline in mortality following the intervention does not preclude the possibility that other
constituents of the pollution mixture that share the same source as S02 are responsible for the adverse
effects. Even though PMi0 levels before and after the intervention were stable in Hong Kong, it is possible
that constituents that do not explain a major fraction of PM may have declined. As also noted previously
in Section 3.1.5, Hedley et al. (2006) noted large reductions in ambient nickel and vanadium
concomitantly with reductions of sulfur after the intervention. S02 also may be serving as a modifier of
the effect of respirable particles. Thus, while the Hong Kong data are supportive of S02-mortality effects,
the possibility remains that mortality effects may be caused by constituents of S02-associated sources.
3.3.4. Summary of Evidence on the Effects of Short-Term SO2
Exposure on Mortality
The epidemiologic evidence on the effect of short-term exposure to S02 on all-cause
(nonaccidental) and cardiopulmonary mortality is suggestive of a causal relationship at ambient
concentrations. The epidemiologic studies are generally consistent in reporting positive associations
between S02 and mortality; however, there was a lack of robustness of the observed associations to
adjustment for copollutants.
Figure 3-11 presents all-cause S02 mortality excess risk estimates from the multicity studies and
meta-analyses. The mortality effect estimates from single-pollutant models range from 0.6% (NMMAPS)
to 4.1% (Italian 8-cities study) per 10 ppb increase in 24-h avg S02 concentrations, but given the large
confidence band in the Italian study, a more stable range may be 0.6 to 2%. It is noteworthy that the S02
effect estimates for the NMMAPS and Canadian 12-city studies are quite comparable (0.6 and 0.7%,
respectively), considering the differences in the modeling approach. The heterogeneity of estimates in the
multicity studies and meta-analyses may be due to several factors, including the differences in model
specifications, averaging/lag time, S02 levels, and effect-modifying factors. Only the APHEA study
examined possible sources of heterogeneity for S02-related mortality. They examined several potential
effect modifiers such as the mean levels of pollution and weather variables, accuracy of the air pollution
measurements, health of the population, smoking prevalence, and geographical differences. The only
variable that could explain the heterogeneity of city-specific effect estimates was the geographic
separation (western versus central eastern European cities) for both S02 and BS, but heterogeneity in the
S02 effect estimates remained within the western cities.
3-49
-------
Reference
Study
Lag
Dominici et al. (2003) NMMAPS 90 cities study
Burnett et al. (2004) Canadian 12 cities study 0-2
Katsouyanni et al. (1997) APHEA1 (12 European cities) Variable
7 W. European cities
5 Central E. European cities
Biggeri et al. (2005) Italian 8 cities study
Hoek (2003)
The Netherlands study
0-1
1
0-6
Stieb et al. (2003) Meta-analysis, international Variable
Health Effects Institute (2004) Meta-analysis, Asian cities Variable
Relative Risk
1.00 1.01 1.02 1.03 1.04
I I I I I
• Single pollutant
O Multipollutant
With PM10 and N02
With NO,
-O With BS
With BS
With various copollutants
Figure 3-11. Relative risks (95% CI) of S02-associated all-cause (nonaccidental) mortality, with and
without copollutant adjustment, from multicity and meta-analysis studies. Effect
estimates are standardized per 10 ppb increase in 24-h avg SO2 concentrations. For
multipollutant models, results from the models that resulted in the greatest reduction
in the SO2 effects are shown. (NMMAPS = National Morbidity, Mortality, and Air
Pollution Study; APHEA = Air Pollution and Health: a European Approach.)
3-50
-------
Relative Risk
Reference
Study
Katsouyanni et al. (1997) APHEA1 (7 W. European cities)
Zmirou et al. (1998) APHEA1 (5 W. European cities)
Biggeri et al. (2005)
Hoek (2003)
Italian 8 cities study
*Le Tertre et al. (2002) French 9 cities study
*Ballester et al. (2002) Spanish 13 cities study
The Netherlands study
Lag
0.95 1.00 1.05 1.10 1.15 1.20
J I I L
Variable
Variable
0-1
0-1
0-1
0-6
X All cause
• Respiratory
o Cardiovascular
COPD
-• Pneumonia
*Note: Le Tertre et al. (2002) and Ballester et al. (2002) performed analyses using Poisson GAM with default convergence criteria.
Figure 3-12. Relative risks (95% CI) of S02-associated mortality for all (nonaccidental), respiratory,
and cardiovascular causes from multicity studies. Effect estimates are standardized
per 10 ppb increase in 24-h avg SO2 concentrations. (APHEA = Air Pollution and
Health: a European Approach.)
Several multicity studies provided effect estimates for broad cause-specific categories, typically
respiratory and cardiovascular mortality. A summary of these effect estimates, along with the all-cause
mortality estimates for comparison, are presented in Figure 3-12. These results from multicity studies
suggest that the mortality effect estimates for cardiovascular and respiratory causes were generally larger
than that for all-cause mortality, though in some cases the effects were not statistically significant,
possibly because of reduced statistical power by which to examine cause-specific associations. In these
3-51
-------
studies, the effect estimates for respiratory mortality were also found to be larger than the cardiovascular
mortality effect estimates, suggesting a stronger association of S02 with respiratory mortality compared to
cardiovascular mortality. This finding is consistent with the observed greater effects of S02 on respiratory
morbidity compared to cardiovascular morbidity.
As shown previously in Figure 3-11, the mortality effect estimates from the multipollutant models
in the multicity studies suggest some extent of confounding between S02 and PM and/or N02, as
indicated by the instability of the effect estimates in multipollutant models. It should be noted, however,
that interpretation of the single- versus multipollutant model results are complicated by potential
interaction among copollutants and differing degrees of measurement error for correlated pollutants.
Very few studies specifically examined possible interactions among the copollutants. Katsouyanni
et al. (2006) examined the effect estimates for S02 and BS in seven western European cities for subsets
stratified by high and low levels of the other pollutant and found that the estimates were similar for days
with low or high levels of the other pollutant. From these results, Katsouyanni et al. suggested that the
effects of S02 and BS were independent.
In summary, recent epidemiologic studies have reported associations between mortality and S02,
often at mean 24-h avg levels of < 10 ppb. The range of the excess risk estimates for S02 on all-cause
mortality is 0.4 to 2% per 10 ppb increase in 24-h avg S02 in several multicity studies and meta-analyses.
The effect estimates for more specific categories may be larger. The larger European study suggests that
the observed heterogeneity in S02 effect estimates is at least in part regional. The intervention study from
Hong Kong supports the idea that a reduction in S02 levels results in a reduction in deaths, but this does
not preclude the possibility that the causal agent is not S02 but rather something else that is associated
with S02 sources. Results from the multicity studies suggest that S02-mortality excess risk estimates may
be confounded to some extent by copollutants, making a definitive distribution of effects among the
pollutants difficult. However, the interpretation of multipollutant model results also requires caution
because of possible interaction among the copollutants and influence of varying measurement error. Very
limited information was available to determine possible interaction effects between S02 and PM or other
copollutants. Overall, the evidence that S02 is causally related to mortality at current ambient levels is
suggestive, but limited by potential confounding and lack of understanding regarding the interaction of
S02 with copollutants in the epidemiologic data.
3.4. Morbidity Associated with Long-Term SO2 Exposure
3.4.1. Summary of Findings from the Previous Review
The 1982 AQCD addressed some effects of long-term S02 exposure. It was reported that
bronchoconstriction resulted from chronic exposure to 5.1 ppm S02 in dogs but not in monkeys. This
increased pulmonary resistance was thought to occur as a result of morphological changes in the airway
or hypersecretion of mucus leading to airway narrowing. However, there were no remarkable pulmonary
pathological findings in monkeys and dogs in these studies. This could have been due to the conventional
light microscopic examination applied, which could not detect alterations in surface membranes or subtle
changes in cilia.
It was also noted that repeated exposures of rats > 50 ppm S02 produced chronic bronchitis similar
to that seen in humans although there was no evidence to suggest that bronchitis developed in humans at
ambient levels of S02 Furthermore, nasal mucosal alterations were observed in mice exposed to 10 ppm
S02 for 72 h by inhalation. Lack of data on morphological effects of S02 at near ambient concentrations
was noted. In addition, some alterations in lung host defenses were discussed with chronic exposure to
S02 at doses exceeding ambient concentrations.
3-52
-------
In the 1982 AQCD, only a few epidemiologic studies provided sufficient quantitative evidence
relating respiratory symptoms or pulmonary functions changes to long-term exposure to S02. Briefly, a
study by Lunn et al. (1967) in Sheffield, England, provided the strongest evidence of an association
between pulmonary function decrements and increased frequency of lower respiratory tract symptoms in
5- to 6-year-old children chronically exposed to ambient BS (annual level of 230 to 301 |_ig/m3) and S02
levels (69 to 105 ppb). A follow-up study by Lunn et al. (1970) found no effect with much lower levels of
BS (range: 48, 169 |_ig/m3) and S02 (range: 36, 97 ppb); it was suggested that this might be due to
insufficient power to detect small health effect changes.
The 1986 Second Addendum presented three additional studies that examined the effects of long-
term exposure on respiratory health. A study by Ware et al. (1986) reported that respiratory symptoms
were associated with annual avg TSP in the range of -30 to 150 (ig/m3 in children (n = 8,380) from six
U.S. studies. Only cough was found to be significantly associated with S02. Although the increase in
symptoms did not appear concomitantly with any decrements in lung function, this may indicate different
mechanisms of effect. Other studies by Chapman et al. (1995) and Dodge et al. (1985) also observed
increased prevalence of cough among children and young adults living in areas of higher S02
concentrations; however, it was noted that the observed effects might have been due to intermittent high
S02 peak concentrations.
In addition to respiratory effects from long-term exposure to S02, the potential carcinogenicity of
S02 or other SOx was also examined in the previous review. The 1982 AQCD concluded that little or no
clear epidemiologic evidence substantiated the hypothesized links between S02 or other SOx and cancer,
though there was some animal toxicological evidence that led to the conclusion that S02 may be
considered a suspect carcinogen/cocarcinogen. There was very limited consideration of the effects of
long-term exposure to S02 on other organ systems.
Since the 1982 AQCD and the 1986 Second Addendum, a number of animal toxicological and
epidemiologic studies have investigated the effect of long-term exposure to S02 on respiratory morbidity,
including asthma, bronchitis and respiratory symptoms, lung function, morphological effects, and lung
host defense. Additional studies have examined the effect of long-term S02 exposure on genotoxic and
carcinogenic effects, cardiovascular effects, and prenatal and neonatal outcomes, which are also briefly
discussed in this section.
3.4.2. Respiratory Effects Associated with Long-Term Exposure to SO2
3.4.2.1. Asthma, Bronchitis, and Respiratory Symptoms
Several epidemiologic studies have examined the association between long-term exposure to S02
and other air pollutants on asthma, bronchitis, and a variety of respiratory symptoms. These studies are
summarized in Annex Table F-6. In the Six Cities Study of Air Pollution and Health, cross-sectional
associations between air pollutants and respiratory symptoms were examined in 5,422 white children aged
10 to 12 years old from Watertown, MA, St. Louis, MO, Portage, WI, Kingston-Harriman, TN,
Steubenville, OH, and Topeka, KS (Dockery et al., 1989). Annual means of 24-h avg S02 concentrations
ranged from 3.5 ppb in Topeka to 27.8 ppb in Steubenville. Except for 03, the correlations among pairs of
pollution measures varied between 0.53 and 0.98. No associations were observed between S02 and a
variety of respiratory symptoms, including bronchitis, chronic cough, chest illness, persistent wheeze, and
asthma. Stronger associations were observed for PM indices.
Dockery et al. (1996) examined the respiratory health effects of acid aerosols in 13,369 white
children aged 8 to 12 years old from 24 communities in the U.S. and Canada between 1988 and 1991. The
city-specific annual mean S02 concentration was 4.8 ppb, with a range of 0.2 to 12.9 ppb. With the
exception of the gaseous acids, nitrous and nitric acid, none of the particulate or gaseous pollutants,
including S02, were associated with increased asthma or any asthmatic symptoms. Stronger associations
3-53
-------
with particulate pollutants were observed for bronchitis and bronchitic symptoms. For S02, the only
significant association found was with chronic phlegm, with an OR of 1.19 (95% CI: 1.00, 1.40) per
5 ppb increase in S02.
Herbarth et al. (2001) performed a meta-analysis of three cross-sectional surveys conducted in East
Germany investigating the relationship between lifetime exposure (from birth to completion of
questionnaire survey) to S02 and TSP in children and the prevalence of chronic bronchitis. Using a
logistic model that included variables on parental predisposition (mother or father with bronchitis) and
environmental tobacco smoke exposure, the authors reported that the OR for bronchitis due to a lifetime
exposure to S02 was 3.51 (95% CI: 2.56, 4.82) (the concentration change for which the OR was based
was not presented). No associations were found between TSP and the prevalence of bronchitis in children.
As part of the international SAVIAH (Small-Area Variation in Air Pollution and Health) study,
Pikhart et al. (2001) examined the respiratory health effects from long-term exposure to S02 in children (n
= 6,959) from two central European cities with high pollution levels (Prague, Czech Republic, and
Poznan, Poland). A novel technique was used to estimate the outdoor concentrations of S02 at a small-
area level. Outdoor S02 was measured by passive samplers at 130 sites in the two cities during 2-week
periods. Concentrations of S02 at each location in the study areas were estimated from these data by
modeling using a geographic information system (GIS). The estimated mean exposure to outdoor S02 was
32 ppb (range: 25, 37) in Prague and 31 ppb (range: 17, 53) in Poznan. The prevalence of wheezing or
whistling in the past 12 months was associated with S02 (OR of 1.08 [95% CI: 1.03, 1.13] per 5 ppb
increase in S02). Moreover, the lifetime prevalence of wheezing or whistling (OR 1.03 [95% CI: 1.00,
1.07]) and lifetime prevalence of physician-diagnosed asthma (OR 1.09 [95% CI: 1.00, 1.19]) also were
associated with S02 levels. In the SAVIAH study, the only other pollutant considered in relation to health
outcomes was N02. An earlier publication by Pikhart et al. (2000) presented preliminary results of the
Prague data and indicated that the observed associations between N02 and respiratory symptoms were
generally similar to that of S02.
The International Study of Asthma and Allergies in Children (ISAAC) included thousands of
children in several European countries and Taiwan (Hirsch et al., 1999; Hwang et al., 2005; Penard-
Morand et al., 2005; Ramadour et al., 2000; Studnicka et al., 1997). Penard-Morand et al. examined the
effect of long-term exposures to air pollution and prevalence of exercise-induced bronchial reactivity
(EIB), flexural dermatitis, asthma, allergic rhinitis, and atopic dermatitis in 9,615 children aged 9 to 11
years in six French communities. Using 3-year averaged concentrations of S02, the investigators reported
that the prevalence of EIB reactivity, lifetime asthma, and allergic rhinitis were significantly associated
with increases in S02 exposure. The estimated 3-year averaged concentration of S02 was 2 ppb in the
low-exposure schools and 4 ppb in the high-exposure schools. In a single-pollutant model, the ORs were
2.37 (95% CI: 1.44, 3.77) for EIB and 1.58 (95% CI: 1.00, 2.46) for lifetime asthma per 5 ppb increase in
S02. In this study, S02 was correlated with PMi0 (r = 0.76) but not with 03 (r = -0.02). Using a two-
pollutant model that included PM10, the associations of S02 with EIB and lifetime asthma were fairly
robust (<5% change).
In a German study of 5,421 children, the annual mean S02 concentration was associated with
morning cough reported in the last 12 months, but not bronchitis (Hirsch et al., 1999). This study further
observed that the association of S02 and other air pollutants with respiratory symptoms were stronger in
nonatopic than in atopic children. The authors noted that these findings were in line with the hypothesis
that these air pollutants induce nonspecific irritative rather than allergic inflammatory changes in the
airway mucosa, as irritative effects would affect the clinical course in nonatopic children more strongly
than in atopics whose symptoms are also determined by allergen exposure.
In contrast to the studies noted above, other studies using the ISAAC protocol did not observe an
association between long-term exposure to S02 and respiratory symptoms. In France, Ramadour et al.
(2000) performed an epidemiologic survey of 2,445 children aged 13 to 14 years living in communities
with contrasting levels of air pollution to determine the relationship between long-term exposure to
gaseous air pollutants and prevalence rate of rhinitis, asthma, and asthma symptoms. The avg S02
concentrations during the 2-month survey period ranged from 7 ppb to 22 ppb across the seven
3-54
-------
communities. This study found no relationship between the mean levels of S02, N02, or 03 and the
above-mentioned symptoms. Another study of 843 children from eight nonurban communities in Austria
did not observe consistent associations between S02 and prevalence of asthma and symptoms (Studnicka
et al., 1997). Compared to the lowest S02 concentration category, the ORs in the higher S02
concentration categories (third and fourth quartiles) did not exceed one for any of the symptoms
examined (wheeze, cough, bronchitis, and asthma).
A cohort study was conducted by Goss et al. (2004) to examine the effect of air pollutants on a
potentially susceptible population, patients with cystic fibrosis. Study participants included 11,484
patients (mean age 18.4 years) enrolled in the Cystic Fibrosis Foundation National Patient Registry in
1999-2000. Exposure was assessed by linking air pollution values from ambient monitors with the
patient's home ZIP code. During the study period, the mean S02 concentration was 4.9 ppb (SD 2.6, IQR:
2.7, 5.9). This study found no association between S02 and the odds of having two or more pulmonary
exacerbations. One of the limitations addressed by the authors was the lack of information regarding
tobacco use or environmental tobacco smoke, an important risk factor for pulmonary exacerbations.
Several studies examined the effects of long-term exposure to S02 on asthma, bronchitis, and
respiratory symptoms. The studies reported positive associations in children; the notable exception was
the Harvard Six Cities Study. However, there were inconsistencies in the results observed: some found
effects on bronchitic but not asthmatic symptoms; others found the converse. A major limitation was that
some subjects were asked to recall prevalence of symptoms in the last 12 months or in a lifetime; such
long recall periods may have caused significant recall bias. Another concern is the high correlation of
long-term avg S02 and copollutant concentrations, particularly PM, and the very limited evaluation of
potential confounding in these studies. Overall, the epidemiologic studies do not provide sufficient
evidence to conclude that long-term exposure to S02 has an effect on asthma, bronchitis, or respiratory
symptoms.
3.4.2.2. Lung Function
Only a few new animal toxicological studies involving longer-term inhalation exposures to S02
were conducted since the last review. These studies are summarized here and in Annex Table E-l. Rabbits
that were neonatally immunized to Alternaria tenuis and exposed to 5 ppm S02 for 13 weeks beginning in
the neonatal period (Douglas et al., 1994) did not demonstrate alterations in lung resistance, dynamic
compliance, trans-pulmonary pressure, tidal volume, respiration rate or minute volume. Similarly, no
changes in physiological function were noted in dogs exposed to 15 ppm S02 for 2 h/day and 4-5
days/week for 5 months (Scanlon et al., 1987), although changes were noted at 50 ppm. However, Smith
et al. (1989) found decreased residual volume and quasistatic compliance in rats at 4 months of exposure
to 1 ppm S02 for 5 h/day and 5 days/week.
Only a limited number of epidemiologic studies examined the association between long-term
exposure to S02 and changes in lung function. The Harvard Six Cities Study by Dockery et al. (1989)
reported that no associations were observed between lung function and long-term exposure to air
pollution, including S02, in a cohort of more than 5,000 children. An analysis of NHANES II data by
Schwartz (1989), which included information on children and youths from 44 cities but was limited by a
cross-sectional study design, also did not observe an association with S02, though inverse associations of
FVC and FEVi with annual concentrations of TSP, N02 and 03 were found. Additional studies conducted
in Europe observed mixed results.
In a longitudinal cohort study of 1,150 children in nine communities in Austria, Frischer et al.
(1999) examined the effect of long-term exposure to air pollutants on lung function. Lung function was
measured in the spring and fall over a 3-year period from 1994 through 1996. Annual mean S02
concentrations ranged from 2 to 6 ppb across the nine communities. The authors reported no consistent
associations between S02, PMi0, or N02 and lung function. For S02, a negative parameter estimate was
observed during the summer, but a positive estimate was found during the winter period. Horak et al.
3-55
-------
(2002a, 2002b) extended the study of Frischer et al. (1999) with an additional year of data. The mean S02
concentration was 6 ppb in the winter and 3 ppb in the summer. This study found a positive association
between wintertime S02 concentrations and changes in FVC, which became null with PMi0 in a two-
pollutant model.
Jedrychowski et al. (1999) conducted a prospective cohort study of 1,001 preadolescent children
from two areas of Krakow, Poland, that differed in ambient air pollutants. In the city center, which had
higher air pollution, the mean annual level of S02 was 16.7 ppb (SD 12.5). In comparison, the mean
annual S02 level in the control area was 12.1 ppb (SD 8.4). A similar difference in TSP levels was
observed between the city center and control area. The adjusted ORs comparing the city center to the
control area for the occurrence of slower lung function growth over a two-year period were 2.10 (95% CI:
1.27, 3.46) for FVC and 2.10 (95% CI: 1.27, 3.48) for FEVi in boys. The adjusted ORs for girls were 1.54
(95% CI: 0.89, 2.64) for FVC and 1.51 (95% CI: 0.90, 2.53) for FEVj. However, as both TSP and S02
levels were higher in the city center, the observed effects on lung function growth cannot be specifically
attributable to S02.
One notable study examined the potential effect of long-term exposure to air pollution on lung
function in adults. The study by Ackermann-Liebrich et al. (1997) included 9,651 adults aged 18 to 60
years old residing in eight different areas in Switzerland (Study on Air Pollution and Lung Diseases in
Adults [SAPALDIA]). They observed a 0.1% decrease in FEVi per 5 ppb increase in S02 for adults.
Significant associations also were observed for PMi0 and N02. The limited number of study areas and
high intercorrelation between the pollutants made it difficult to assess the effect of an individual pollutant.
The authors concluded that air pollution from fossil fuel combustion, which was the main source of air
pollution for S02, N02, and PMi0 in Switzerland, was associated with decrements in lung function
parameters in this study.
Collectively, the results from the limited number of animal toxicological and epidemiologic studies
on the effect of long-term exposure to ambient S02 on lung function is inconclusive.
3.4.2.3. Morphological Effects
Several animal toxicological studies of morphological effects resulting from subacute to chronic
S02 exposures have been published since the 1982 AQCD. These studies are summarized in Annex Table
E-17. No alveolar lesions (including electron microscopic evaluation) or changes in numbers of tracheal
secretory cells were observed in guinea pigs exposed to 1 ppm S02 for 3 h/day for 6 days (Conner et al.,
1985). No pulmonary or nasal lesions were observed in rats exposed to 5 ppm S02 for 2 h/day and
5 days/week for 4 weeks (Wolff et al., 1989). A weakness of the latter study is that histopathological
methods were not reported. However, a third study reported histopathological changes in the respiratory
system involving lesions in the bronchioles. Smith et al. (1989) exposed rats for 4 to 8 months to 1 ppm
S02 for 5 h/day and 5 days/week and observed increased incidence of bronchiolar epithelial hyperplasia
and a small increase (12%) in numbers of nonciliated epithelial cells in terminal respiratory bronchioles at
4 but not 8 months of exposure. A limitation of the study was the examination of a single concentration,
which does not allow for concentration-response assessment or identification of a no-effect-level.
In summary, evidence from these animal toxicological studies is insufficient to conclude that long-
term exposure to ambient S02 causes prolonged effects on lung morphology.
3.4.2.4. Lung Host Defense
The 1982 AQCD reported some detrimental effects of S02 on lung host defenses that generally
occurred at concentrations exceeding ambient exposure concentrations. In rats exposed to 0.1 ppm S02
for ~2 to 3 weeks, clearance of labeled particles from the lung was accelerated at 10 and 23 days
following exposure. In rats exposed to 1 ppm for ~2 to 3 weeks, clearance was accelerated at 10 days and
3-56
-------
slowed down at 25 days. Tracheal mucus flow was decreased with a 1-year exposure of dogs to 1 ppm
S02, but was unaffected by a 30-min exposure of donkeys to 25 ppm S02. Studies in mice suggested no
effect on susceptibility to bacterial infection with exposure to S02 concentrations of < 5 ppm for 3
months. Antiviral defenses were impaired in mice exposed to 7-10 ppm S02 for 7 days. No alterations in
pulmonary immune system were reported with chronic exposure of mice to 2 ppm S02.
Several studies on lung host defense have been conducted since the last review and are summarized
in Annex Table El 8. Only one study published after the last review evaluated mucociliary clearance in
rats after exposure to S02. In this subchronic study, no effect on clearance of radiolabeled particles from
the lung was observed in rats exposed to 5 ppm S02 for 2 h/day for 4 weeks (Wolff et al., 1989). These
findings are in contrast to the altered clearance reported in the 1982 AQCD. Three other recent studies
were conducted evaluating the effects of 10 ppm S02 on immune responses. Impairment of host defense
responses was seen following continuous exposure to S02 for 1-3 weeks (Azoulay-Dupuis et al., 1982),
but not in response to a single 4 h exposure (Clarke et al., 2000; Jakab et al., 1996).
In summary, animal toxicological studies do not provide sufficient evidence to assess the effects of
long-term exposure to ambient S02 on lung host defense.
3.4.2.5. Summary of Evidence on the Effects of Long-Term Exposure on Respiratory
Health
The overall epidemiologic evidence on the respiratory effects of long-term exposure to S02 is
inadequate to infer a causal relationship. Studies that examined the effects of long-term exposure to S02
on asthma, bronchitis, and respiratory symptoms observed positive associations in children. However, the
variety of outcomes examined and the inconsistencies in the observed results make it difficult to assess
the impact of long-term exposure of S02 on respiratory symptoms. In the limited number of studies
examining the S02 associations with lung function, results were generally mixed. A major consideration
in evaluating S02-related health effects in these epidemiologic studies of long-term exposure is the high
correlation among the pollutant levels observed, particularly between long-term avg S02 and PM
concentrations. The lack of evidence available to evaluate potential confounding by copollutants limits
the ability to make a causal determination based on these studies.
A limited number of animal toxicological have examined the effect of long-term exposure to S02
on lung function. Results from these studies do not provide strong biological plausibility for effects of
long-term ambient exposure to S02 on respiratory morbidity. These studies observed no effects on
physiological lung function at S02 concentrations < 5 ppm in rabbits and dogs; however, one study found
decreased residual volume and quasistatic compliance at 1 ppm S02 in rats. In addition, no morphological
changes were found in guinea pigs exposed subacutely to 1 ppm S02, or in rats exposed subchronically to
5 ppm S02 While mild, bronchiolar epithelial hyperplasia was observed in rats exposed to 1 ppm for 4
months, this change was not apparent at 8 months. Furthermore, animal toxicological studies provide no
evidence for decrements in lung host defense at or near ambient levels of S02.
Overall, the available evidence from the generally limited number of epidemiologic and animal
toxicological studies is inadequate to infer that respiratory effects occur from long-term exposure to S02
at ambient concentrations.
3.4.3. Carcinogenic Effects Associated with Long-Term Exposure
The 1982 AQCD concluded that little or no clear epidemiologic evidence substantiated the
hypothesized links between S02 or other SOx and cancer. From the toxicological studies, it was noted that
while there were some indications of carcinogenicity for both S02 and S02 + benzo[a]pyrene (B[a]P),
3-57
-------
complex exposure regimens, problematic dose determinations, and/or inadequately reported experimental
details led to the conclusion that S02 could only be considered a suspect carcinogen/cocarcinogen.
Since the last review, numerous studies have examined the genotoxic effects of S02. These are
summarized in Annex Table E-19. S02 and its metabolite sulfite were found not to be mutagenic or to
induce DNA damage in vitro (Pool-Zobel et al., 1990; Pool et al., 1988b). However, inhalation studies
demonstrated increased mouse bone marrow micronucleated polychromatic erythrocytes and DNA
damage in cells isolated from various organs when mice were exposed for 4-6 h/day for 7 days to 5-30
ppm S02 (Meng et al., 2002; 2005b; Ruan et al., 2003). These in vivo studies suggest that inhaled S02
may have systemic effects at high concentrations, but they are of questionable significance in evaluating
the effects of S02 at ambient levels.
The carcinogenic potential of S02 was examined in animal toxicological studies which are
summarized in Annex Table E-ll. Gunnison et al. (1988) conducted a two-part study in which rats were
exposed either for 21 weeks (6 h/day, 5 days/week) by inhalation to 0, 10, or 30 ppm S02, or for 21 weeks
to two tungsten-supplemented, molybdenum-deficient diets. This latter regimen induced a condition of
sulfite oxidase deficiency, resulting in elevated systemic levels of sulfite:bisulfite relative to control
values (e.g., in plasma, from 0 to 44 (.iM: and in tracheal tissue, from 33 to 69 or 550 nmol/g wet weight).
Beginning with week 4, some groups from each regimen received weekly tracheal installations of 1-mg
B[a]P for 15 weeks. Overall results indicated that squamous cell carcinoma was not induced, or in the
B[a]P groups coinduced or promoted, by S02 inhalation or elevated systemic sulfite:bisulfite. Researchers
found a very high incidence of animals with tumors in the groups exposed to only B[a]P (128 ofl44
animals). As a result, carcinogenicity or cocarcinogenicity of S02 or sulfite:bisulfite could only have been
detected as a shortening of tumor induction time or an increase in rate of tumor appearance, and neither
was observed. As noted by the authors, these findings do not support the conclusion that S02 exposure
enhances the carcinogenicity of B[a]P . It was proposed that S02 exposure, by elevating systemic
sulfite:bisulfite, would generate g 1 utathione- V-su 1 fonates. which in turn could inhibit glutathione S-
transferase (GST) and reduce intracellular GSH and, thus, interfere with a major detoxication pathway for
B[a]P. See Annex Table E-20 for discussion (Menzel et al. 1986).
Two similar studies were published that investigated the ability of 10 to 11 months of exposure (16
h/day) to 4 ppm S02, 6 ppm N02, or their combination to affect the carcinogenicity of either urban
suspended PM (SPM) (Ito et al., 1997) or diesel exhaust particle (DEP) extract-coated carbon particles
(Ohyama et al., 1999). The former study found that, while exposure to SPM extract-coated carbon
particles significantly increased pulmonary endocrine cell (PEC) hyperplasia, coexposure to S02, N02, or
their combination was without additional affect. Also, irrespective of gas coexposure, SPM extract-coated
carbon particles demonstrated a few PEC papillomas versus control frequencies of zero.
Using Syrian golden hamsters, Heinrich et al. (1989) investigated whether coexposure to 10 ppm
S02 and 5 ppm N02 for 6 to 8 months (5 days/week, 19 hours/day) could enhance tumorigenesis induced
by a single subcutaneous injection of diethylnitrosamine (DEN) during week 2. The combined gas
exposure did not affect body weight gain and only minimally shortened survival times. Compared to the
DEN groups, serial sacrifices of gas-exposed animals demonstrated progressively increasing numbers of
tracheal mucosal cells and aberrant tracheal cell cilia. In the lung, effects related to gas mixtures were
largely limited to a progressive type of alveolar lesion that involved the lining of bronchiolar epithelium
and the appearance of pigment-containing AM and to a mild, diffuse thickening of the alveolar septa.
Exposure to the combined gases by itself did not induce tumors of the upper respiratory tract, nor did it
enhance the induction of such tumors by DEN.
In addition to the animal toxicological studies that examined the genotoxic and carcinogenic
potential of S02, a limited number of recent epidemiologic studies have investigated the relationship
between long-term exposure to S02 and lung cancer incidence and mortality. These studies are
summarized in Annex Table F-7. Nyberg et al. (2000) conducted a case-control study of men aged 40 to
75 years with (n = 1,042) and without (n = 2,364) lung cancer in Stockholm County, Sweden. They
mapped residence addresses to a GIS database to assign individual exposures to S02 from defined
emission sources (mainly local oil-fueled residential heating). Available S02 measurement data were used
3-58
-------
to calibrate the model. In this study, S02 was considered an indicator of air pollution from residential
heating. Exposure to N02, considered to be a marker of traffic pollution, also was evaluated in this study.
The 90th percentile 30-year avg S02 level was 30 ppb. After adjusting for potential confounders (e.g.,
smoking, occupational exposures), long-term avg heating-related S02 exposure was not associated with
an increase in risk of lung cancer incidence. A weak association for the 30-year avg traffic-related N02
exposure was observed.
Very similar results were reported in a Norwegian study by Nafstad et al. (2003). The study
population is a cohort of 16,209 men who enrolled in a study of cardiovascular diseases in 1972. The
Norwegian cancer registry identified 422 incident cases of lung cancer. S02 exposure data were modeled
based on residence using data for observed concentrations and emission from point sources (e.g., industry
and heating of buildings and private homes) and traffic. Once again, no association was observed between
long-term exposure to S02 and lung cancer incidence.
Three additional European cohort studies examined the associations between long-term exposure to
air pollution and lung cancer mortality (Beelen et al., 2008; Filleul et al., 2005; Nafstad et al., 2004) in
cohorts ranging in size from 14,284 to 120,852 subjects, who were followed for 9 to > 20 years.
Consistent with the results for lung cancer incidence, none of these studies observed an association
between long-term S02 exposure and lung cancer mortality. These studies are discussed in further detail
in Section 3.5.2.2.
Similar to the European cohort studies, studies conducted in the U.S. generally did not observe an
association between long-term exposure to S02 and lung cancer mortality. In the reanalysis of the Harvard
Six Cities Study, Krewski et al. (2000) estimated a RR of 1.03 (95% CI: 0.91, 1.16) per 5 ppb increase in
avg S02 over the study period, while Pope et al. observed a positive but not statistically significant (RR
-1.04 per 5 ppb increase in avg S02 from 1982 to 1998) association in the extended analysis of the
American Cancer Society (ACS) cohort. The California Seventh-day Adventists study by Abbey et al.
(1999) did observe a statistically significant association between lung cancer mortality and S02 (and most
of the pollutants examined including PMi0, sulfate, 03, and N02), but the number of lung cancer deaths in
this cohort was very small (12 for female, 18 for male) and, therefore, it is difficult to interpret these
estimates. More detailed discussions of these studies are provided in Section 3.5.2.1.
In conclusion, the toxicological studies indicate that S02 at high concentrations may cause DNA
damage but fails to induce carcinogenesis, cocarcinogenesis, or tumor promotion. Furthermore, results
from the limited number of epidemiologic studies examining the association between long-term exposure
to ambient S02 and excess risk of lung cancer incidence and mortality are inconclusive.
3.4.4. Cardiovascular Effects Associated with Long-Term Exposure
The effects of S02 on the cardiovascular system were not addressed in the 1982 AQCD. Since then,
animal toxicological studies have reported oxidation and glutathione (GSH) depletion (Langley-Evans et
al., 1996; Meng et al., 2003b; Wu and Meng, 2003) in the hearts of rodents which were exposed by
inhalation to S02. However, as concentrations of S02used in these studies were 5 ppm and above, the
oxidative injury observed is probably not relevant to cardiovascular effects seen at ambient levels of S02.
These and other animal toxicology studies measuring cardiovascular endpoints are summarized in Annex
Table E-9.
A recent epidemiologic study examined the association between long-term exposure to air
pollution, including S02, and one or more fatal or nonfatal cardiovascular events. In the Women's Health
Initiative cohort study, Miller et al. (2007) studied 65,893 postmenopausal women between the ages of 50
and 79 years without previous cardiovascular disease in 36 U.S. metropolitan areas from 1994 to 1998.
Subjects' exposures to air pollution were estimated using residents' five-digit ZIP code, assigning the
annual mean levels of air pollutants measured at the nearest monitor. A total of 1,816 women had one or
more fatal or nonfatal cardiovascular events, including 261 deaths from cardiovascular causes. Hazard
3-59
-------
ratios for the first cardiovascular event were estimated. The results for models that only included subjects
with non-missing exposure data for all pollutants (n = 28,402 subjects, resulting in 879 cardiovascular
events) are described here. In the single-pollutant models, PM2 5 showed the strongest associations with
cardiovascular events among the pollutants (Hazard Ratios = 1.24 [95% CI: 1.04, 1.48] per 10 (ig/m3
increase in annual avg), followed by S02 (1.07 [95% CI: 0.95, 1.20] per 5 ppb increase in the annual avg).
In the multipollutant model where all the pollutants (i.e., PM2 5, PM10-2.5, CO, S02, N02, 03) were
included in the model, the PM25 association with overall cardiovascular events was even stronger (1.53
[95% CI: 1.21, 1.94]). The association with S02 also became stronger (1.13 [95% CI: 0.98, 1.30]).
Correlations among these pollutants were not described and, therefore, the extent of confounding among
these pollutants in these associations could not be examined, but among all the air pollutants considered,
PM2 5 was clearly the best predictor of cardiovascular events.
The available toxicological and epidemiologic evidence to assess the effect of long-term exposure
to S02 on cardiovascular health is too limited to make any conclusions at this time.
3.4.5. Prenatal and Neonatal Outcomes Associated with Long-Term
Exposure
Several animal toxicological studies examined developmental effects of S02 and are summarized in
Annex Table E-13. No changes in birth weight or neurobehavioral development were noted in mouse
pups prenatally exposed to 5-30 ppm S02 (Petruzzi et al., 1996), while some behavioral modifications
were seen in adults exposed prenatally to these same levels (Fiore et al., 1998). However, effects observed
at such high concentrations of S02 are of questionable relevance.
In recent years, the effects of prenatal and neonatal exposure to air pollution have been examined in
epidemiologic studies by several investigators (see Annex Table F-8). The most common endpoints
studied are low birth weight, preterm delivery, and measures of intrauterine growth. Preterm birth and low
birth weight may result in serious long-term health outcomes for the infant. Preterm birth is the leading
cause of infant mortality and is a major determinant of a variety of adverse neurodevelopmental outcomes
and chronic adverse respiratory effects (Berkowitz and Papiernik, 1993). Low birth weight has also been
linked with increased risk of infant mortality and morbidity. Other studies have examined associations
between maternal exposure to ambient air pollution and sudden infant death syndrome (SIDS) and
neonatal hospitalizations.
These studies analyzed air pollution data and birth certificates from a given area. In evaluating the
results of these studies, it is important to consider the limitations of these data. For example, the reliability
and validity of birth certificate data have been reviewed (Buescher et al., 1993; Piper et al., 1993) and
have been found to vary in degrees of reliability by specific variables. The variables considered the most
reliable include birth weight, maternal age, race, and insurance status. Whereas gestational age, parity and
delivery type (vaginal vs. cesarean) were reasonably reliable, obstetrical complications and maternal
lifestyle factors such as smoking and alcohol consumption were not reliable. Another concern in these
studies regards adequate control for potential confounders. While most of these studies adequately
controlled for maternal education, parity, age, and sex of child, many did not adjust for SES, occupational
exposures, indoor pollution levels, maternal smoking, alcohol use, prenatal care, or concurrent
temperature exposures as fetal growth is associated with all of these factors. This makes overall
comparisons across studies a difficult task.
While most studies analyzed avg S02 exposure for the whole pregnancy, many also considered
exposure during specific trimesters, or other time periods (e.g., first and last months of gestation).
Different exposure periods have been examined because the biological mechanisms and timing of critical
exposures that link air pollution to adverse birth outcomes are yet to be determined. For example, fetal
growth is much more variable during the third trimester; therefore, exposure during the third trimester
would have the greatest likelihood of an association. However, insufficient placentation during the first
3-60
-------
trimester may be associated with early environmental insult, w hereby subsequent fetal growth is hindered.
Similarly, it is possible that preterm delivery is associated with insufficient placentation resulting from
early exposure. Furthermore, preterm delivery may be the result of acute exposures just prior to delivery.
Reference
Location
i
Maisonet et al. (2001)
6 Northeastern cities, U.S.
1
| 1st trimester |
'White •
—* African American
i
Hispanic
i
i •
Liu et al. (2003)
Vancouver, BC
i
i *
Dugandzic et al. (2006)
Nova Scotia, Canada
i
" f
Bobak (2000)
Czech Republic
i
i
i
Maisonet et al. (2001)
6 Northeastern cities, U.S.
All J*
| 2nd trimester |
i
White i
• 1 African American
i
Hispanic-
i
i
Dugandzic et al. (2006)
Nova Scotia, Canada
i
*T
Bobak(2000)
Czech Republic
1
1
1
Maisonet et al. (2001)
6 Northeastern cities, U.S.
1
All 1,
1
| 3rd trimester |
Wtjiite .
* i African American
•—1 Hispanic
Liu et al. (2003)
Vancouver, BC
i
i
Dugandzic et al. (2006)
Nova Scotia, Canada
i
" I
Bobak (2000)
Czech Republic
i
Wang etal. (1997)
Beijing. China
i
i
i
0.8 1.0 1.2 1.4
Relative Risk
Figure 3-13. Relative risks (95% CI) for low birth weight, grouped by trimester of SO2 exposure.
Risk estimates are standardized per 5 ppb increase in SO2 concentrations.
Epidemiologic studies examining the effects of air pollutants on low birth weight are summarized
in Figure 3-13. Maisonet et al. (2001) examined the association between air pollution and low birth
weight in six northeastern cities: Boston, MA; Hartford, CT; Philadelphia, PA; Pittsburgh, PA;
Springfield, MA; and Washington, DC. The study population consisted of 89,557 singleton, full-term, live
3-61
-------
births (37-44 weeks of gestation) born between January 1994 and December 1996. Low birth weight was
classified as < 2,500 g (5.5 lbs.). This study observed an association between low birth weight and S02
concentrations during each trimester among Caucasians; however, the association was not consistent in
other races and ethnicities.
An excess risk for low birth weight associated with ambient S02 concentrations was reported by
Dugandzic et al. (2006) in a large cohort study of 74,284 women with full-term, singleton births from
1988-2000 in Nova Scotia, Canada. The mean 24-h avg S02 concentration over the study period was 10
ppb (IQR 7). These investigators found that exposure only during the first trimester was associated with
increased risk of low birth weight. The RR was 1.14 (95% CI: 1.04, 1.26) per 5 ppb increase in S02 level.
Liu and Krewski (2003) found similar results in a study of pregnancy outcomes and air pollution in
Vancouver, Canada. The mean 24-h avg S02 concentration was 4.9 ppb (IQR 7.7) from 1985 to 1998.
Maternal exposure during the first month was associated with an increased risk of low birth weight (OR
1.11 [95% CI: 1.01, 1.22]). Additional studies from the U.S., Europe, Latin America and Asia have
reported positive associations between low birth weight and maternal exposure to S02 during the first
(Bell et al., 2007; Bobak, 2000; Ha et al., 2001; Mohorovic, 2004; Yang et al., 2003a), second (Bobak,
2000; Gouveia et al., 2003; Lee et al., 2003a) and third (Bobak, 2000; Lin et al., 2004b; Wang et al.,
1997) trimesters.
Preterm delivery, intrauterine growth retardation (IUGR), and birth defects are additional adverse
birth outcomes that have been associated with ambient S02 levels. In a time-series analysis using data
from four Pennsylvania counties, Sagiv et al. (2005) reported that the mean 6-week S02 exposure prior to
birth was associated with increased risk of preterm birth, with a RR of 1.05 (95% CI: 1.00, 1.10) per
5 ppb increase in S02. A 5 ppb increase in S02 concentrations three days before birth was associated with
a RR of 1.02 (95% CI: 0.99, 1.05). The authors discussed two plausible mechanisms for the effects of air
pollution on preterm birth: (1) changes in blood viscosity due to inflammation as a result of air pollution
(Peters et al., 1997); and (2) maternal infection during pregnancy as a consequence of impaired immunity
from air pollution exposure. Liu and Krewski (2003) reported that S02 exposure during the last month of
pregnancy was associated with preterm birth, with an OR of 1.09 (95% CI: 1.01, 1.19) for a 5 ppb
increase in S02, in Vancouver, Canada. Similar results were found for studies conducted in the Czech
Republic (Bobak, 2000), Korea (Leem et al., 2006), and Beijing (Xu et al., 1995).
Liu and Krewski (2003) further reported that S02 exposure during the last month of pregnancy was
associated with IUGR (OR 1.07 [95% CI: 1.01, 1.13]). However, in a later study in the Canadian cities of
Calgary, Edmonton and Montreal, Liu et al. (2007) did not observe associations between maternal
exposure to S02 and increased risk of IUGR.
Two Brazilian studies examined exposure to S02 and neonatal deaths. Pereira et al. (1998) found a
positive association between S02 and intrauterine mortality in Sao Paulo during a 2-year period, though
the effect was sensitive to model specifications and did not support a concentration-response relationship.
The most robust association with intrauterine mortality was observed for an index of three gaseous
pollutants (N02, S02, CO). Lin et al. (2004b) found that a 5 ppb increase in S02 was associated with an
increase of 8.8% (95% CI: 5.8, 11.8). A similar relationship was found for PMi0. The creation of an index
containing both PMi0 and S02 allowed the observation of their cumulative effects on daily death counts.
The result of this analysis was similar in magnitude to the effect of S02 alone. An ecologic cohort study
of infant mortality in the U.S. found no association with annual averages of S02 concentration (Lipfert et
al., 2000a).
Gilboa et al. (2005) conducted a population-based case-control study to investigate the association
between maternal exposure and air pollutant exposure during weeks 3-8 of pregnancy, the risk of selected
cardiac birth defects and oral clefts in live births, and fetal deaths between 1997 and 2000 in seven Texas
counties. When the highest quartile of exposure was compared to the lowest, the authors observed a
positive association between S02 and isolated ventricular septal defects (OR 2.16 [95% CI: 1.51, 3.09]).
Although this is the only study to have examined the effect of maternal exposure to S02 on birth defects,
it supports the notion that the developing embryo and growing fetus are susceptible to maternal air
pollution exposure.
3-62
-------
Several studies examined adverse health outcomes in relation to S02 concentrations during the
neonatal period. Dales et al. (2006) evaluated hospitalizations for respiratory disorders in neonates < 4
weeks of age from hospitals in 11 large Canadian cities during a 15-year study period (population-
weighted avg 24-h avg S02 of 4.3 ppb). The researchers observed a 5.5% (95% CI: 2.8, 8.3) excess risk in
respiratory hospitalizations associated with a 10 ppb increase in 24-h avg S02 concentrations with a 2-d
lag. This effect was slightly attenuated after adjusting for PMi0 and gaseous copollutants. To investigate
the influence of ambient S02 concentrations on SIDS, Dales et al. (2004) conducted a time-series analysis
comparing daily rates of SIDS and daily S02 concentrations from 12 large, Canadian cities during a 16-
year period. The mean 24-h avg S02 level across the 12 cities was 5.51 ppb (IQR 4.92). There was an
18.0% (95% CI: 4.4, 33.4) excess risk in SIDS incidence for a 10 ppb increase in 24-h avg S02 levels.
The authors concluded that the effect of S02 was independent of sociodemographic factors, temporal
trends, and weather.
In summary, epidemiologic studies on birth outcomes have observed positive associations between
S02 exposure and low birth weight; however, toxicological studies provide very little biological
plausibility for reproductive outcomes related to S02 exposure. The inconsistent results across trimesters
of pregnancy and the lack of evidence regarding confounding by copollutants further limit the
interpretation of these studies. The limited number of studies addressing preterm delivery, IUGR, birth
defects, neonatal hospitalizations, and infant mortality make it difficult to draw conclusions regarding the
effect of S02 on these outcomes.
3.4.6. Other Organ System Effects Associated with Long-Term
Exposure
The 1982 AQCD presented only one chronic exposure study which was relevant to nervous system
effects. Dogs were exposed for 68 months to a mixture of S02 and H2S04 (Stara et al., 1980). No effects
on visual-evoked brain potentials during or immediately after exposure to the SOx mixture were
observed. Since then, numerous studies have examined brain lipid content, lipid peroxidation and
glutathione content and antioxidant enzymes following inhalation exposure of rodents to S02 at
concentrations of 10 ppm or higher. Concentrations of 5 ppm or higher S02 were used in studies
examining neurobehavior and neurodevelopment in mice. These studies are summarized in Annex Table
E-12.
In the past 25 years, numerous animal toxicological studies have evaluated the effects of long-term
S02 exposure on other organ systems such as reproductive, hematological, gastrointestinal, renal,
lymphatic, and endocrine systems. Most of these studies used concentrations of S02 of 5 ppm or higher.
Many of these studies examined alteration profiles of lipid peroxidation and antioxidant levels (Langley-
Evans et al., 1996; Meng et al., 2003c; 2004) and are summarized in Annex Table E-10, E-13 through E-
16, and E-21.
3.5. Mortality Associated with Long-Term SO2 Exposure
3.5.1. Summary of Findings from the Previous Review
At the time of the 1982 AQCD, the available studies on the effects of long-term exposure to S02 on
mortality were all ecological cross-sectional studies. This study design could not take into consideration
such confounders as cigarette smoking, occupational exposures, and social status. In addition, there were
3-63
-------
questions regarding how representative the aerometric data used were for community exposure.
Therefore, it was concluded that the epidemiologic studies did not provide valid quantitative data relating
respiratory disease or other types of mortality to long-term (annual avg) exposures to S02 or PM.
The 1986 Secondary Addendum reviewed more studies of this type, with information on more
detailed components of PM (inhalable and fine particles, and particulate sulfate). While some studies
suggested importance of the size of PM, the fundamental problem of the study design made it difficult to
interpret the effect estimates. The 1986 Secondary Addendum also reviewed a Japanese study in which
the death rates from asthma and chronic bronchitis in a highly polluted section of Yokkaichi, an industrial
city with large S02 emissions from the largest oil-fired power plant in Japan, were compared with those in
a less polluted area of the same city (Imai et al., 1986). SOx levels (measured using the lead peroxide
method) averaged across several monitoring sites in the polluted harbor area ranged from around 1.0 to
2.0 mg/day (annual avg) during 1964 through 1972 and then steadily declined to less than 0.5 mg/day in
1982. This is in contrast to levels consistently < 0.3 mg/day in the low pollution areas throughout 1967
through 1982. Annual avg levels for other pollutants (i.e., N02, TSP, oxidants) monitored in the high
pollution area were consistently low from 1974 through 1982. The results indicated elevated rates of
chronic bronchitis mortality in the highly polluted area compared to the less polluted area, but the 1986
Secondary Addendum could not conclude that this was due to S02 alone, because sulfate or other
particulate SOx such as H2S04 could have been responsible.
Several, more recent studies have examined long-term exposure effects of air pollution, including
S02, on mortality. These studies are summarized in Annex Table F-9. As with short-term exposure
studies, the focus of most of these studies was mainly on PM though some focused on traffic-related air
pollution. They all used Cox proportional hazards regression models with adjustment for potential
confounders. The designs of these studies were better than earlier cross-sectional studies as the outcome
and most of the potential confounders (e.g., smoking history, occupational exposure) were measured on
an individual basis. However, the geographic scale and method for exposure estimates varied across these
studies.
3.5.2. Associations of Mortality and Long-Term Exposure in Key
Studies
3.5.2.1. U.S. Cohort Studies
Harvard Six Cities Studies
Dockery et al. (1993) conducted a prospective cohort study to study the effects of air pollution with
the main focus on PM components in six U.S. cities. These cities were chosen based on levels of air
pollution, with Portage, WI and Topeka, KS representing the least polluted cities and Steubenville, OH
representing the most polluted city. Mean S02 levels ranged from 1.6 ppb in Topeka to 24.0 ppb in
Steubenville from 1977 to 1985. Cox proportional hazards regression was conducted with data from a 14-
to 16-year follow-up study of 8,111 adults in the six cities. Dockery et al. (1993) reported that lung cancer
and cardiopulmonary mortality were more strongly associated with the concentrations of inhalable and
fine PM, and sulfate particles, than with the levels of TSP, S02, N02, or acidity of the aerosol.
Krewski et al. (2000) conducted a sensitivity analysis of the Harvard Six Cities Study and
examined associations between gaseous pollutants (i.e., 03, N02, S02, and CO) and mortality. S02
showed positive associations with total mortality (RR = 1.05 [95% CI: 1.02, 1.09] per 5 ppb increase in
avg S02 over the study period) and cardiopulmonary deaths (1.05 [95% CI: 1.00, 1.10]), but in this
dataset S02 was highly correlated with PM2 5 (r = 0.85), sulfate (r = 0.85), and N02 (r = 0.84).
3-64
-------
American Cancer Society Cohort Studies
Pope et al. (1993) investigated associations between long-term exposure to PM and the mortality
outcomes in the ACS cohort. Ambient air pollution data from 151 U.S. metropolitan areas in 1981 were
linked with individual risk factors in 552,138 adults who resided in these areas when enrolled in the
prospective study in 1982. Death outcomes were ascertained through 1989. PM2 5 and sulfate were
associated with total, cardiopulmonary, and lung cancer mortality, but not with mortality for all other
causes. Gaseous pollutants were not analyzed in this study.
Krewski and co-investigators (Jerrett et al., 2003a; Krewski et al., 2000) conducted an extensive
sensitivity analysis of the Pope et al. (1993) ACS data, augmented with additional gaseous pollutants data.
The mean S02 concentrations were 7.18 ppb in the warm season (April to September) and 11.24 ppb in
the cool season (October to March). Among the gaseous pollutants examined, only S02 showed positive
associations with mortality. The RR for total mortality was 1.06 (95% CI: 1.05, 1.07) per 5 ppb increase
in the annual avg S02. Analysis using S02 measured in different seasons produced a somewhat higher
estimate for the warm season than that for the cool season (7% compared to 5% excess risk per 5 ppb
increase). Although the subjects in the ACS cohort came from all regions of the U.S., the majority of the
151 cities were located in the eastern U.S., where both S02 and sulfate tend to be higher. PM2 5 levels are
also higher in the east. To address the influence of these spatial patterns, which may confound
associations between mortality and these pollutants, Krewski et al. (2000) conducted extensive two-stage
regression modeling. In these models, the association between S02 and mortality was diminished but
persisted after adjusting for sulfate, PM2 5, and other variables. For example, in the spatial filtering model
(which resulted in the largest reduction of the S02 effect estimate when sulfate was included), the S02
total mortality RR estimate was 1.07 (95% CI: 1.03, 1.11) in the single-pollutant model and 1.04 (95%
CI: 1.02, 1.06) with sulfate in the two-pollutant model. The effect estimates for PM2 5 and sulfate also
were diminished when S02 was included in the models. The result further showed that S02 effect
estimates were generally insensitive to adjustment for spatial correlation. Thus, these results suggest that
the association between S02 and mortality may be confounded with PM, but the association cannot be
accounted for by PM2 5 or sulfate alone. Krewski et al. (2000) noted that their reanalysis of the ACS and
Harvard Six Cities studies suggested that mortality might be attributed to more than one component of the
complex mixture of ambient air pollutants in urban areas in the U.S..
The original Pope et al. (1993) study and the Krewski et al. (2000) reanalysis both used the air
pollution exposure estimates that are based on the average over the Metropolitan Statistical Area (MSA),
which consists of multiple counties. To investigate the effects of geographic scale over which the air
pollution exposures are averaged, (Willis et al., 2003), reanalyzed the ACS cohort data using the exposure
estimates averaged over the county scale, and compared the results with those based on the MSA-scale
avg exposure. Less than half of the cohort used in the MSA-based study was used in the county-scale
based analysis, because of the limited availability of sulfate monitors and the reduced sample size due to
the loss of subjects when using the five-digit ZIP codes. The mean (9.3 ppb versus 10.7 ppb) and range
(0.0 to 29.3 ppb versus 0.0 to 27.2 ppb) of the MSA- and county-level S02 data sets were similar. In the
analysis comparing the two-pollutant model with sulfate and S02, they found that the inclusion of S02
reduced sulfate effect estimates substantially (> 25%) in the MSA-scale model but not substantially
(< 25%) in the county-scale model. In the MSA-level analysis (with 113 MSAs), the S02 RR estimate
was 1.04 (95% CI: 1.02, 1.06) per 5 ppb increase, with sulfate in the model. In the county-level analysis
(91 counties) with sulfate in the model, the corresponding estimate was smaller (1.02 [95% CI: 1.00,
1.05]). It should also be noted that the correlation between covariates were different between the MSA-
level data and county-level data. The correlation between S02 and sulfate was 0.48 in the MSA-level data,
but it was 0.56 in the county-level data. The correlation between poverty rate and S02 was -0.16 in the
MSA-level data, but it was 0.15 in the county-level data. Thus, the extent of confounding between S02
and PM components as well as among other covariates in the model can be affected by the geographic
scale of aggregation of exposure estimates. It is not clear, however, if the smaller geographic scale
increases or decreases exposure characterization error for S02, because a certain extent of smoothing
3-65
-------
(averaging) over distance may reduce very local concentration peaks that are not relevant to the city-wide
population.
Pope et al. (2002) extended analysis of the ACS cohort with double the follow-up time (to 1998)
and triple the number of deaths compared to the original Pope et al. (1993) study. In addition to PM2 5, all
the gaseous pollutant data were retrieved for the extended period and analyzed for their associations with
death outcomes. As in the 1995 analysis, the air pollution exposure estimates were based on the MSA-
level averages. PM2 5 was associated with total, cardiopulmonary, and lung cancer mortality but not with
deaths for all other causes. S02 was associated with all the mortality outcomes, including all other causes
of deaths. The S02 RR estimate for total mortality was 1.03 (95% CI: 1.02, 1.05) per 5 ppb increase (1982
to 1998 avg). The association of S02 with mortality for all other causes (sulfate also showed this pattern)
makes it difficult to interpret the effect estimates. This lack of specificity for S02 (in contrast to PM) is
not consistent with causal inference.
The EPRI-Washington University Veterans' Cohort Mortality Studies
Lipfert et al. (2000b) conducted an analysis of a national cohort of -70,000 male U.S. military
veterans who were diagnosed as hypertensive in the mid 1970s and were followed up for about 21 years
(up to 1996). This cohort was 35% black and 57% were current smokers (81% of the cohort had been
smokers at one time). PM2 5, PMi0, PMi0.2 5, TSP, sulfate, CO, 03, N02, S02, and lead were examined in
this analysis. No mean or median level of S02 was reported. The county of residence at the time of entry
to the study was used to estimate exposures. Four exposure periods (from 1960 to 1996) were defined,
and deaths during each of the three most recent exposure periods were considered. The results for S02
were presented only qualitatively as part of their preliminary screening regression results. Lipfert et al.
(2000b) noted that lead and S02 were not found to be associated with mortality, thus were not considered
further. They also noted that the pollution effect estimates were sensitive to the regression model
specification, exposure periods, and the inclusion of ecological and individual variables. The authors
reported that indications of concurrent mortality risks were found for N02 and peak 03.
Lipfert et al. (2006b) examined associations between traffic density and mortality in the same
cohort, whose follow-up period was extended to 2001. As in their 2000 study, four exposure periods were
considered but included more recent years. The 95th percentiles of daily avg in each of the exposure
periods were considered for S02. For the 1997-2001 data period, the estimated mortality RR for S02 was
0.99 (95% CI: 0.97, 1.01) per 5 ppb increase in a single-pollutant model. They reported that traffic density
was a better predictor of mortality than ambient air pollution variables with the possible exception of 03.
The log-transformed traffic density variable was only weakly correlated with S02 (r = 0.32) and PM2 5 (r
= 0.50) in this data set.
Lipfert et a. (2006a) further extended analysis of the veterans' cohort data to include the EPA's
Speciation Trends Network (STN) data, which collected chemical components of PM2 5. They analyzed
the STN data for year 2002, again using county-level averages. PM2 5 and gaseous pollutants data for
1999 through 2001 were also analyzed. As in the previous Lipfert et al. (2006b) study, traffic density was
the most important predictor of mortality, but associations were also seen for elemental carbon, vanadium,
nickel, and nitrate. 03, N02, and PM10 also showed positive but weaker associations. Once again, no
associations were observed between long-term exposure to S02 and mortality.
Seventh-day Adventist Study
Abbey et al. (1999) investigated associations between long-term ambient concentrations of PMi0,
sulfate, S02, 03, and N02 (1973 through 1992) and mortality (1977 through 1992) in a cohort of 6,338
nonsmoking California Seventh-day Adventists. Monthly indices of ambient air pollutant concentrations
at 348 monitoring stations throughout California were interpolated to ZIP codes according to home or
work location of study participants, cumulated, and then averaged over time. They reported associations
between PM10 and total mortality for males and nonmalignant respiratory mortality for both sexes. S02
3-66
-------
was not associated with total mortality (RR 1.07 [95% CI: 0.92, 1.24] for males and 1.00 [95% CI: 0.88,
1.14] for females per 5 ppb increase in multiyear avg S02), cardiopulmonary deaths, or respiratory
mortality for either gender.
3.5.2.2. European Cohort Studies
A study by Beelen et al. (2008) examined associations between traffic-related air pollution and
mortality. They analyzed data from the Netherlands Cohort Study on Diet and Cancer with 120,852
subjects who were followed from 1987 to 1996. BS, N02, S02, PM25, and four types of traffic-exposure
estimates were analyzed. While the local traffic component was estimated for BS, N02, and PM2 5, no
such attempt was made for S02, because there was "virtually no traffic contributions to this pollutant."
Thus, only "background" S02 levels were reflected in the exposure estimates. Traffic intensity on the
nearest road was associated with all-cause mortality and a larger RR was observed for respiratory
mortality. Results were similar for BS, N02 and PM2 5, but no associations were found for S02 (RR = 0.98
[95% CI: 0.93, 1.03] per 5 ppb increase in multiyear avgS02).
Nafstad et al. (2004) investigated the association between mortality and long-term exposure to air
pollution exposure in a cohort of Norwegian men followed from 1972-1973 through 1998. Data from
16,209 males (aged 0 to 49 years) living in Oslo, Norway, in 1972-1973 were linked with data from the
Norwegian Death Register and with estimates of the avg annual air pollution levels at the participants'
home addresses. PM was not considered in this study because measurement methods changed during the
study period. Exposure estimates for NOx and S02 were constructed using models based on subject
addresses, emission data for industry, heating, and traffic, and measured concentrations. While NOx was
associated with total, respiratory, lung cancer, and ischemic heart disease deaths, S02 did not show any
associations with mortality. The authors noted that the S02 levels were reduced by a factor of 7 during the
study period (from 5.6 ppb in 1974 to 0.8 ppb in 1995), whereas NOx did not show any clear downward
trend.
Filleul et al. (2005) investigated long-term effects of air pollution on mortality in 14,284 adults
who resided in 24 areas from seven French cities when enrolled in the Air Pollution and Chronic
Respiratory Diseases (PAARC) survey in 1974. Daily measurements of S02, TSP, BS, N02, and NO were
made in the 24 areas for 3 years (1974 through 1976). Models were run before and after exclusion of six
area monitors influenced by local traffic as determined by a NO:N02 ratio of > 3. Before exclusion of the
six areas, none of the air pollutants was associated with mortality outcomes. After exclusion of these
areas, analyses showed associations between total mortality and TSP, BS, N02, and NO but not S02 (RR
= 1.01 [95% CI: 0.97, 1.06] per 5 ppb multiyear average) or acidimetric measurements. It should be noted
that S02 levels in these French cities declined markedly between the 1974 through 1976 period and the
1990 through 1997 period by a factor of 2 to 3, depending on the city. The changes in air pollution levels
over the study period complicate interpretation of reported effect estimates.
3.5.2.3. Cross-Sectional Analysis Using Small Geographic Scale
Elliott et al. (2007) examined associations of BS and S02 with mortality in Great Britain using a
cross-sectional analysis. However, unlike the earlier ecological cross-sectional mortality analyses in the
U.S. in which mortality rates and air pollution levels were compared using large geographic boundaries
(i.e., MSAs or counties), in the Elliot et al. analysis, the mortality rates and air pollution were compared
using a much smaller geographic unit, the electoral ward, with a mean area of 7.4 km2 and a mean
population of 5,301 per electoral ward. Death rates were computed for four successive 4-year periods
from 1982 to 1994 and associated with 4-year exposure periods from 1966 to 1994. The number of deaths
from all causes in the 10,520 wards was 420,776. Of note, S02 levels declined from 41.4 ppb in the 1966
to 1970 period to 12.2 ppb in 1990 to 1994. This type of analysis does not allow adjustments for
3-67
-------
individual risk factors, but the study did adjust for SES data available for each ward from the 1991
census. Social deprivation and air pollution were more highly correlated in the earlier exposure windows.
They observed associations for both BS and S02 and mortality outcomes. The estimated effects were
stronger for respiratory illness than other causes of mortality for the most recent exposure period and most
recent mortality period (when pollution levels were lower). The adjustment for social deprivation reduced
the effect estimates for both pollutants. The adjusted mortality RRs for S02 for the pooled mortality
periods using the most recent exposure windows were 1.021 (95% CI: 1.018, 1.024) for all causes, 1.015
(95% CI: 1.011, 1.019) for cardiovascular, and 1.064 (95% CI: 1.056, 1.072) for respiratory causes per
5 ppb increase in S02. The effect estimates for the most recent mortality period using the most recent
exposure windows were larger. Simultaneous inclusion of BS and S02 reduced effect estimates for BS but
not S02. Elliott et al. (2007) noted that the results were consistent with those reported in the Krewski et al.
(2000) reanalysis of the ACS study. This analysis was ecological, but the exposure estimates in the
smaller area compared to that in the U.S. cohort studies may have resulted in less exposure
misclassification error, and the large underlying population appears to be reflected in the narrow
confidence bands of effect estimates. The results from this study suggest an association between long-
term exposures (especially in recent years) to S02 and mortality.
3.5.3. Summary of Evidence on the Effect of Long-Term Exposure on
Mortality
The available epidemiologic evidence on the effect of long-term exposure to S02 on mortality is
inadequate to infer a causal relationship at this time. The ecological cross-sectional studies examined in
the 1982 AQCD and 1986 Secondary Addendum found suggestive relationships between long-term
exposure to S02 and mortality. However, there were concerns as to whether the observed association was
due to S02 alone, because sulfate or other particulate SOx such as H2S04 could have been responsible. In
the more recent longitudinal cohort studies, once again, positive associations have been observed between
long-term exposure to S02 and mortality; however, several issues affect the interpretation of these results.
Figure 3-14 presents all-cause mortality RR estimates associated with long-term exposure to S02
from the U.S. and European cohort studies. The overall range of RRs spans 0.97 to 1.07 per 5 ppb
increase in the annual (or longer period) avgS02. The analyses of the Harvard Six Cities and the ACS
cohort data, which likely provide effect estimates that are most useful for evaluating possible health
effects in the U.S., observed RRs of 1.02 to 1.07. Note that each of the U.S. cohort data has its own
advantages and limitations. The Harvard Six Cities data have a small number of exposure estimates, but
the study cities were carefully chosen to represent a range of air pollutant exposures. The ACS cohort had
far more subjects, but the population was more highly educated than the representative U.S. population.
Since educational status appeared to be an important effect modifier of air pollution effects in both
studies, the overall effect estimate for the ACS cohort may underestimate that for the more general
population. However, it should also be noted that several other U.S. and European studies did not observe
an association between long-term exposure to S02 and mortality.
The geographic scale of analysis appears to influence S02 effect estimates and exposure error. In a
reanalysis of the ACS data, the county-level analysis showed a smaller S02 effect estimate than MSA-
level analysis. For sulfate, the opposite pattern was found. Thus, the impact of the geographic scale of
analysis may also depend on the spatial distribution of air pollutants. The cross-sectional analysis in Great
Britain using small-scale electoral wards observed an effect estimate similar to the lower end of the range
of effect estimates for all-cause mortality from U.S. cohort studies, though it is not clear if the effect
estimates from this cross-sectional study are directly comparable to those from cohort studies.
3-68
-------
Relative Risk
Reference
Study
0.8
Analysis Method
Krewski et al. (2000) Harvard Six Cities study MSA-level
Krewski et al. (2000); ACS study
Jerrett et al. (2003) ACS study
Willis et al. (2003) ACS study
Pope et al. (2002) ACS study
(extended follow-up)
MSA-level
Spatial filtering
Spatial filtering
MSA-level
County-level
MSA-level
Lipfert et al. (2006a) Veterans cohort study County-level
Abbey et al. (1999) Seventh-day Adventist study
Beelen et al. (2008) The Netherlands cohort study
Nafstad et al. (2004) Norwegian study
Filleul et al. (2005) French PAAC survey
0.9
l
Female
1.0
l
1.1
l
Male
Male-
1.2
l
1.3
Male
S02 only
o S02 with sulfate
Figure 3-14. Relative risks (95% CI) of S02-associated all-cause (nonaccidental) mortality, with and
without adjustment for sulfate, from longitudinal cohort studies. Effect estimates are
standardized per 5 ppb increase in SO2 concentrations. The exposure estimates for
Krewski et al. (2000) and Pope et al. (2002) are based on MSA (Metropolitan Statistical
Area)-level averaging; Lipfert et al. (2006b) used county-level averaging.
3-69
-------
Another important issue that these studies could not resolve was the possible confounding and/or
interaction among PM indices and S02. The possibility that the observed effects may not be due to S02,
but other constituents that come from the same source as S02, or that PM may be more toxic in the
presence of S02 or other components associated with S02, cannot be ruled out. For example, the ACS
cohort came from all regions of the U.S., but a major fraction of the ACS cities were located in the eastern
U.S., where both S02 and sulfate levels tend to be higher. Therefore, even with sophisticated spatial
modeling, separating possible confounding of S02 effects by PM is challenging. Future and on-going
studies that take into consideration within- versus between-city variation of these pollutants may help
elucidate this issue.
Overall, the results from two major U.S. epidemiologic studies observe an association between
long-term exposure to S02 or sulfur-containing particulate air pollution and mortality. However, several
other U.S. and European cohort studies did not observe an association. The lack of consistency across
studies, inability to distinguish potential confounding by copollutants, and uncertainties regarding the
geographic scale of analysis limit the interpretation of a causal relationship.
3-70
-------
Chapter 4. Public Health Impact
This chapter addresses several issues relating to the broader public health impact from exposure to
ambient S02. The first section discusses the shape of the concentration-response relationship for S02,
with consideration of interindividual variability in responses and evaluation of the limited evidence
available to assess threshold values for health effects. The next section identifies characteristics of
subpopulations which may experience increased risks from S02 exposures, through either enhanced
susceptibility (e.g., as a result of pre-existing disease, genetic factors, age) and/or differential vulnerability
associated with increased exposure (e.g., close proximity to sources, activities).
4.1. Assessment of Concentration-Response Function and
Potential Thresholds
An important consideration in characterizing the public health impacts associated with S02
exposure is whether the concentration-response relationship is linear across the full concentration range,
or if there are concentration ranges where there are departures from linearity (i.e., nonlinearity). Of
particular interest is the shape of the concentration-response curve at and below the level of the current
S02 NAAQS level of a 24-h avg level of 0.14 ppm or the annual avg of 0.03 ppm, and whether a
threshold of effect may exist among these lower concentrations. The assumption of a threshold indicates
an ambient S02 concentration below which adverse health outcomes are not elicited. Lack of a threshold
implies that exposure to even the lowest measured ambient S02 concentrations has the potential to cause
toxicity.
In the context of risk assessment for human health, it is important to consider the distinction
between individual thresholds and population thresholds. There is wide variability in the human
population, and a threshold for a population is defined by the threshold for the most sensitive individual
in that population (Gaylor et al., 1988). Some human clinical studies provide individual-level response
data in relation to different levels of S02 exposure; this allows evaluation of both the percentage of
individuals showing responses across the range of exposures as well as the concentration at which an
individual begins to indicate a response. Very few epidemiologic studies, and no human clinical studies,
evaluate whether there is a population-level threshold, which is the concentration of S02 that must be
exceeded to elicit a health response in the study population.
Human clinical and epidemiologic studies that examined the shape of the concentration-response
function are presented below. The discussion focuses on respiratory morbidity effects associated with
short-term exposure to S02, for which the strongest causal evidence exists.
4.1.1. Evidence from Human Clinical Studies
In human clinical studies of exercising asthmatics, moderate S02-induced decrements in lung
function have been observed at the lowest levels tested (i.e., 0.2 to 0.3 ppm, 5 to 10 min exposures) in
some individuals (approximately 5-30% of subjects). Statistically significant respiratory effects have been
consistently observed at concentrations of 0.4-0.6 ppm, with 20-60% of asthmatics experiencing moderate
to large decrements in lung function following 5-10 min exposures (see Table 3-1). Smaller, yet
statistically significant decrements in lung function have also been demonstrated at S02 concentrations <
0.2 ppm when preceded by exposure to 03 (see Section 3.1.3). Human clinical studies are valuable in
characterizing the concentration-response relationship in relatively healthy asthmatics, but cannot be used
4-1
-------
to evaluate potential population threshold directly, as controlled human exposure studies have not been
conducted at SO2 concentrations below 0.1 ppm and do not include the most sensitive asthmatics.
100
A sRaw
¦D
47%,
40
!3%
22%'
5%-
0
0.2
0.4
0.6
0.8
1
SO2 Concentration (ppm)
100
B FEV.
O 80
T2
q. 70
60
51%
40
20%'
20
0
0.2
0.4
0.6
0.8
1
SO2 Concentration (ppm)
Source: Linn et al. (1987; 1988; 1990); Smith (1994)
Figure 4-1. Percent of mild and moderate asthmatics (vE = 40-50 L/min) experiencing an SO2-
induced increase in (a) sRaw of > 100% or a decrease in (b) FEV1 of > 15%, adjusted
for effects of moderate to heavy exercise in clean air. The data represents lung
function measurements from 40,41,40, and 81 subjects at concentrations of 0.2, 0.3,
0.4, and 0.6 ppm, respectively. A two parameter logistic model was fit to the data using
Bayesian estimation with noninformative priors (Lunn et al., 2000).1
1 The form of the logistic model was P 1 / (1 + 6Xp( Ct f3 1 Og( ))) , where p is the fraction of asthmatics who
experienced a moderate or greater decrement in lung function as defined above. Median estimates are presented along with upper and lower 95%
confidence intervals.
4-2
-------
With increasing exposure concentration between 0.2 and 1.0 ppm, there is a clear increase both in
magnitude of respiratory effect and percent of asthmatics affected (Table 3-1). A subset of the data
presented in this table was taken from a series of studies conducted by Linn et al. (1987; 1988; 1990) and
is presented graphically in Figures 4-1 through 4-3. In these studies, mild and moderate asthmatics were
exposed for 10 min to S02 concentrations between 0 and 0.6 ppm during moderate to heavy exercise.
These particular studies were selected for inclusion in this meta-analysis owing to similarities between
exposure protocols, with all subjects being exposed to multiple concentrations of S02. In the 1987 study,
subjects were exposed to S02 concentrations of 0, 0.2, 0.4, and 0.6 ppm, while in the 1988 and 1990
studies, subjects were exposed to concentrations of 0, 0.3, and 0.6 ppm. The percent of asthmatics
experiencing moderate or greater S02-induced decrements in lung function (increase in sRaw > 100% or
decrease in FEVi > 15%) is shown in Figure 4-1. At 0.2 ppm, between 5 and 13% of subjects are affected,
and this fraction increases with increasing concentration, with approximately 50% of subjects
experiencing respiratory effects at a concentration of 0.6 ppm.
Figures 4-2 and 4-3 present the concentration-response relationship between S02 and decrements in
lung function among asthmatics in the Linn et al. (1987; 1988; 1990) studies. Concentration response
relationships are presented for all asthmatics, as well as for S02-sensitive asthmatics, i.e., those
asthmatics experiencing significant decrements in lung function at the highest exposure concentration
used (0.6 ppm). This analysis demonstrates a clear increase in the magnitude of respiratory effects with
increasing exposure concentration, with more marked effects observed at lower concentrations among the
S02-sensitive asthmatics. The results of a study by Gong et al. (1995) support this conclusion: the authors
observed a linear relationship between S02 concentration (0, 0.5, and 1.0 ppm) and both lung function
(decrease in FEVi, and increase in sRaw) and respiratory symptoms.
450
• All Asthmatics
-•-Sensitive Asthmatics
400
350
300
250
200
150
"S 100
-50
-100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
S02 Concentration (ppm)
Figure 4-2. S02-induced increase in sRaw among mild and moderate asthmatics following 10 min
exposures with moderate to heavy exercise (Ve = 40-50 L/min). Responses presented
for all asthmatics as well as S02-sensitive asthmatics, defined here as asthmatics
experiencing a >100% SO2 induced increase in sRaw at 0.6 ppm. The analysis includes
data from 40 subjects exposed to concentrations of 0.0, 0.2, 0.4, and 0.6 ppm, 14 of
whom were S02-sensitive (Linn et al., 1987), as well as 41 subjects exposed to 0.0, 0.3,
and 0.6,25 of whom were S02-sensitive (Linn et al., 1988; 1990). Error bars = 1 SE.
4-3
-------
-10
-20
-25
All Asthmatics
Sensitive Asthmatics
-30
-35
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
S02 Concentration (ppm)
Figure 4-3. SCMnduced decrease in FEVi among mild and moderate asthmatics following 10 min
exposures with moderate to heavy exercise (Ve = 40-50 L/min). Responses presented
for all asthmatics as well as SO2 sensitive asthmatics, defined here as asthmatics
experiencing a >15% SO2 induced decrease in FEV1 at 0.6 ppm. The analysis includes
data from 40 subjects exposed to concentrations of 0.0, 0.2, 0.4, and 0.6 ppm, 21 of
whom were SO2 sensitive (Linn et al., 1987), as well as 41 subjects exposed to 0.0, 0.3,
and 0.6,20 of whom were SO2 sensitive (Linn et al., 1988; 1990). Error bars = 1 SE.
4.1.2. Evidence from Epidemiologic Studies
Although there are numerous epidemiologic studies that examined the association between SO2 and
various health effects, only a few of these studies attempted to evaluate the concentration-response
function. Most studies assumed a linear or log-linear relationship between ambient S02 concentrations
and the health outcome in their evaluations.
Epidemiologic studies have examined the concentration-response relationship for SO2 using
various statistical methods, including the comparison of effect estimates in increasing quartiles or
quintiles, plotting the risk observed against increasing SO2 concentrations, and using nonparametric
smoothed curves to assess the nonlinearity of the SCVeffect relationship. Most of the epidemiologic
studies that examined the concentration-response function between SO2 exposure and respiratory
morbidity observed that the relationship could not be distinguished from linear across the entire
concentration range.
The association between asthma hospitalizations and ambient 24-h avg SO2 concentrations was
examined in a case-control study of children in Bronx County, NY (Lin et al., 2004d). The 24-h avg
concentration ranged from 2.9 to 66.4 ppb. The authors categorized 24-h avg SO2 concentrations and
estimated ORs for each category using the lowest exposure group as the reference (2.9 to 9.2 ppb). They
observed an increasing linear trend across the range of concentrations, with more marked effects observed
at 24-h avg S02 concentrations greater than 40 ppb (Figure 4-4). A similar concentration-response
relationship for daily 1-h max SO2 levels was also observed. During the years 2003-2005, the 24-h avg
SO2 concentrations for the 90th and 95th percentiles were 10 and 13 ppb, respectively.
4-4
-------
4.0
Same day
¦ ¦ 0-1 day
b -a 0-2 day
o— o 0-3 day
3.5
3.0
¦I 2.5
re
a.
¦g 2.0
¦a
O
0.5
'/
¦C?,
O
Q.
SO2 Concentration (ppb)
Source: Lin et al. (2004d)
Figure 4-4. Adjusted odds ratios of asthma hospitalizations by groupings of 24-h avg SO2
concentrations in Bronx County, New York. All groups were compared with the lowest
exposure group (2.9-9.2 ppb). ORs for 24-h avg SO2 concentrations on the same day,
as well as from a 2-day, 3-day, and 4-day moving average lag are presented.
The Harvard Six Cities Study by Schwartz et al. (1994) investigated the concentration-response
function and observed a nonlinear relationship between SO2 concentrations and respiratory symptoms. A
figure plotting the relative odds of incidence of lower respiratory tract symptoms against SO2
concentrations lagged 1 day indicated that no statistically significant increase in the incidence of lower
respiratory tract symptoms was seen until SO2 concentrations exceeded a 24-h avg of 22 ppb though an
increasing trend was observed at concentrations as low as 10 ppb (see Figure 4-5). In a study of
respiratory hospitalizations, Ponce de Leon et al. (1996) found that a weak relationship with SO2 was only
observable at 24-h avg S02 concentrations above 23 ppb. In both the Schwartz et al. (1994) and Ponce de
Leon et al. (1996) studies, a statistically significant increased risk was observable only at 24-h avg SO2
concentrations that were above the 90th percentile. The nonlinearity observed in these concentration-
response functions is dependent on only a few influential observations; thus, should be interpreted with
caution.
4-5
-------
1.4 -
o
0
10
20
30
40
24-h avg SOz (ppb)
Source: Schwartz et al. (1994).
Figure 4-5. Relative odds ratio of incidence of lower respiratory tract symptoms smoothed
against 24-h avg SO2 concentrations on the previous day, controlling for temperature,
city, and day of week.
A study by Jaffe et al. (2003) examined the association between S02 and ED visits for asthma in
three cities in Ohio, and found significant associations only in Cincinnati using Poisson regression
analysis. To examine the concentration-response function, they also conducted quintile analyses. In
Cincinnati, an increasing linear trend in risk was observed across the range of concentrations. Wong et al.
(2002a [using GAM with default convergence criteria]) constructed a plot of risk against 24-h avg S02
concentrations to examine the concentration-response relationship in Hong Kong and London. In general,
a linear relationship between risk of respiratory hospitalizations and S02 was observed across the range of
S02 concentrations in Hong Kong, but not in London. Several other studies that examined the
concentration-response relationship found that the association between respiratory hospitalizations and
S02 did not deviate from linearity (Atkinson et al., 1999b; Burnett et al., 1997b; Hajat et al., 1999; Hajat
et al., 2002).
Discerning a possible population-level threshold for air pollution-related effects in epidemiologic
studies is quite challenging. Using PM2 5 as an example, Brauer et al. (2002a) examined the relationship
between ambient concentrations and mortality risk in a simulated population with specified common
individual threshold levels. They found that no population threshold was detectable when a low threshold
level was specified for individuals. Other factors that may make it difficult to detect a threshold if one
exists include wide interindividual variability in sensitivity to S02 exposure, and the inadequacy of
currently deployed ambient monitors for accurate and precise measurements at lower 24-h avg S02 levels.
Ambient concentrations of S02 have been declining since the 1980s and are now at or very near the limit
of detection (~3 ppb) of the ambient monitors in the regulatory network. The mean 24-h avg S02
concentration across the metropolitan statistical areas (MSAs) from 2003 through 2005 was 4 ppb (5th-
95th percentile: 1-13). Thus, there is greater uncertainty at the lower concentration range compared to the
higher concentrations, which likely limits the ability to detect any potential threshold that may exist.
4-6
-------
The overall evidence from epidemiologic studies that evaluate the concentration-response
relationship is not sufficient to conclude that the relationship deviated from linearity; however, it should
be noted that these studies generally lack power to distinguish between linear and non-linear response
forms.
4.1.3. Summary of Evidence on Concentration-Response Functions
and Thresholds
In the previous two sections, evidence from human clinical and epidemiologic studies on the
concentration-response function that may inform identification of any potential population threshold was
presented. Results from human clinical studies indicate wide interindividual variability in response to S02
exposures, with peak (5 to 10 min) exposures at levels as low as 0.2-0.3 ppm eliciting respiratory
responses in some asthmatic individuals. A clear increase in the magnitude of respiratory effects was
observed with increasing exposure concentrations between 0.2 and 1.0 ppm during 5-10 min S02
exposures.
Several epidemiologic studies that examined the concentration-response function between short-
term (24-h avg or 1-h max) exposure to S02 and respiratory morbidity observed that the relationship
could not be distinguished from linear across the entire concentration range. Given the various limitations
in observing a possible threshold in population studies, the lack of evidence does not necessarily indicate
that there is indeed no threshold in S02 health effects. Some epidemiologic studies did report that though
there was generally an increasing trend at the lower S02 concentrations, a marked increase in S02-related
respiratory health effects was observed at higher concentrations. However, as these observations were
based on a few potentially influential data points (24-h avg S02 concentrations above the 90th percentile),
the results should be interpreted with caution. The overall limited evidence from epidemiologic studies
examining the concentration-response function of S02 health effects is inconclusive regarding the
presence of an effect threshold at current ambient levels.
4.2. Susceptible and Vulnerable Populations
Not all individuals exposed to pollutants respond similarly. Some subpopulations are at increased
risk to the detrimental effects of pollutant exposure; additionally, considerable interindividual variability
exists within sensitive subpopulations. The NAAQS are intended to provide an adequate margin of safety
for both general populations and sensitive subpopulations, or those subgroups potentially at increased risk
for ambient air pollution health effects.
In general, a sensitive population might exhibit an adverse health effect to a pollutant at
concentrations lower than those needed to elicit the same response in the general population, or exhibit a
more severe adverse effect than the general population when exposed to the same pollutant
concentrations. The term susceptibility generally encompasses innate or acquired factors that make
individuals more likely to experience effects with exposure to pollutants. Genetic or developmental
factors can lead to innate susceptibility, while acquired susceptibility may result from age, from disease,
or personal risk factors such as smoking, diet, or exercise; personal risk factors such as smoking, diet, or
exercise habits are also associated with the development of heart and lung diseases. In addition, new
attention has been paid to the concept of some population groups having increased vulnerability to
pollution-related effects due to factors including SES (e.g., reduced access to health care) or particularly
elevated exposure levels. Factors potentially contributing to susceptibility or vulnerability to air pollution
are included in Table 4-1.
4-7
-------
It is important to recognize that there is some overlap between the general concepts of
susceptibility and vulnerability. For example, life stage is an important factor potentially determining both
susceptibility and vulnerability. Children may be particularly vulnerable because of differences in
exposure arising from their behavior and susceptible due to absorption or metabolism rates. Aging may
also increase susceptibility to adverse effects from a pollutant because as individuals age, their bodies'
ability to defend against and respond to injury may diminish.
The previous review of the S02 NAAQS identified certain groups within the population that may
be more susceptible to the effects of S02 exposure, including asthmatics, individuals not diagnosed as
asthmatic but with atopic disorders (e.g., allergies), and individuals with COPD or cardiovascular disease.
Recent information on potentially susceptible or vulnerable subpopulations is assessed below.
Table 4-1. Factors Potentially Contributing to Susceptibility or Vulnerability to Air Pollution
Susceptibility Factors
Vulnerability Factors
Respiratory diseases (e.g., asthma)
Increased activity patterns
Cardiovascular diseases
Decreased air conditioning use
Genetic factors
Increased level of exertion
Age, Gender
Work environment (e.g., outdoor workers)
Race/ethnicity
Lower SES
Pro-inflammatory conditions, e.g., diabetes
Lower education level
Obesity
Residential location (e.g., proximity to roadways)
Adverse birth outcomes (e.g., low birth weight)
Geographic location (West vs. East)
4.2.1. Pre-existing Disease
A recent report of the NRC (2004) emphasized the need to evaluate the effect of air pollution on
susceptible groups, including those with respiratory illnesses and cardiovascular diseases. Generally,
asthma, COPD, conduction disorders, CHF, diabetes, and MI are conditions believed to put persons at
greater risk of adverse events associated with air pollution. Asthmatics are known to be one of the most
S02-responsive subgroups in the population; the evidence related to respiratory illness, including asthma
and other factors, is discussed in further detail below.
4.2.1.1. Pre-existing Respiratory Diseases
The 1982 AQCD concluded that asthmatics are more susceptible to respiratory effects from S02
exposures than the general public. This conclusion was primarily drawn from the strong human clinical
evidence. Recent epidemiologic studies have strengthened this conclusion, reporting associations between
a range of health outcomes with both short-term and long-term S02 exposures in subjects with respiratory
disease.
In human clinical studies, asthmatics have been shown to be more responsive to respiratory effects
of S02 exposures than healthy non-asthmatics. While S02-attributable decrements in lung function
generally have not been demonstrated at concentrations <1.0 ppm in non-asthmatics (Lawther et al.,
1975; Linn et al., 1987; Schachter et al., 1984), statistically significant increases in respiratory symptoms
and decreases in lung function have consistently been observed in exercising asthmatics following peak (5
to 10 min) S02 exposures to concentrations of 0.4-0.6 ppm (Gong et al., 1995; Horstman et al., 1986;
4-8
-------
Linn et al., 1983b). Moderate or greater S02-induced decrements in lung function have also consistently
been observed at lower S02 concentrations (0.2-0.3 ppm) in some asthmatics (Bethel et al., 1985; Linn et
al., 1987; 1988; 1990; Sheppard et al., 1981). It is important to note that the most severe asthmatics in the
population are excluded from participating in controlled human exposures to S02. Therefore, it is not
unreasonable to presume that the most sensitive asthmatics would respond at lower levels than those used
in human clinical studies. There is no evidence from human clinical studies that individuals with COPD
have increased susceptibility to S02-induced respiratory effects.
A number of epidemiologic studies reported increased respiratory morbidity associated with S02
exposures in asthmatics. Notably, two U.S. multicity studies observed associations between ambient S02
concentrations and respiratory symptoms in asthmatic children (Mortimer et al., 2002; Schildcrout et al.,
2006). Additional studies also have indicated generally positive associations for asthma among children
and included a U.S. study (Delfino et al., 2003a) and several European studies (Higgins et al., 1995;
Neukirch et al., 1998; Peters et al., 1996a; Roemer et al., 1993; Segala et al., 1998; Taggart et al., 1996;
Timonen and Pekkanen, 1997; van der Zee et al., 1999). Studies of adults found no consistent association
between respiratory symptoms among asthmatics and S02 concentrations (Desqueyroux et al., 2002a;
2002b; Romieu et al., 1996; van der Zee et al., 2000).
A positive association between ambient S02 concentrations and ED visits and hospitalizations
provides further evidence that asthmatics are susceptible to the effects of S02. The associations between
ambient concentrations of 24-h avg S02 and ED visits and hospitalizations for asthma in the U.S. are
generally positive (Jaffe et al., 2003; Lin et al., 2004d; Michaud et al., 2004; Wilson et al., 2005), though
a large time-series study conducted in Atlanta, GA did not find an association between ambient 1-h max
S02 levels and asthma ED visits (Peel et al., 2005). Studies conducted outside the U.S. (Atkinson et al.,
1999a; 1999b; Hajat et al., 1999; Sunyer et al., 1997; Thompson et al., 2001) also generally found
positive results.
In summary, substantial evidence from epidemiologic studies suggests that individuals with
preexisting respiratory diseases, particularly asthma, are more susceptible to respiratory health effects,
though not mortality, from S02 exposures than the general public. The observations from human clinical
studies indicating increased sensitivity to S02 exposures in asthmatic subjects compared to healthy
subjects provide coherence and biological plausibility for these observations in epidemiologic studies.
4.2.1.2. Pre-existing Cardiovascular Diseases
The evidence available to evaluate the susceptibility of populations with cardiovascular disease for
S02-related health effects is very limited. One human clinical study observed no evidence to suggest that
patients with stable angina were more susceptible to S02-related health effects compared with healthy
subjects (Routledge et al., 2006). The authors noted that this lack of response in the heart patients may be
due to a drug treatment effect rather than decreased susceptibility. Liao et al. (2004) investigated short-
term associations between ambient pollutants and cardiac autonomic control and observed that
consistently more pronounced associations were found between S02 and HRV among persons with a
history of coronary heart disease. In another epidemiologic study, Henneberger et al. (1996) examined the
association of repolarization parameters with air pollutants in East German men with preexisting coronary
heart disease. Ambient S02 concentrations during the 24-h preceding the ECG were associated with the
QT interval duration, but not with any other repolarization parameters.
Evidence is inconsistent in studies analyzing the associations between ambient levels of air
pollutants and ED visits or hospitalizations for cardiovascular diseases. A recent epidemiologic study
investigated the association of S02 with cardiac-related hospital admissions among persons with
preexisting cardiopulmonary conditions and observed no associations with ambient 1-h max S02 level for
any cardiac disease investigated (i.e., ischemic heart disease [IHD], CHF, and dysrhythmia) across strata
of comorbid disease status, including hypertension, diabetes, and COPD (Peel et al., 2007).
4-9
-------
Goldberg et al. (2003) compared the risk estimates for death with the underlying cause of CHF and
those deaths classified as having CHF one year before death and did not find associations between air
pollution and those with CHF as an underlying cause of death. The authors found associations between
some of the air pollutants examined (coefficient of haze [CoH], S02, and N02) and the deaths that were
classified as having CHF one year before death, but the association with the specific cause of death was
not unique to S02. This pattern of association, including but not specific to S02, with specific causes of
death also was observed in an additional cohort of patients with CHF (Kwon et al., 2001).
In conclusion, the very limited evidence examining the susceptibility of individuals with
preexisting cardiovascular disease to adverse health effects from ambient S02 exposures is inconclusive.
4.2.2. Genetic Factors for Oxidant and Inflammatory Damage from Air
Pollutants
A consensus now exists among scientists that genetic factors related to health outcomes and
ambient pollutant exposures merit serious consideration (Gilliland et al., 1999; Kauffmann, 2004).
Several criteria must be satisfied in selecting and establishing useful links between polymorphisms in
candidate genes and adverse respiratory effects. First, the product of the candidate gene must be
significantly involved in the pathogenesis of the effect of interest, which is often a complex trait with
many determinants. Second, polymorphisms in the gene must produce a functional change in either the
protein product or in the level of expression of the protein. Third, in epidemiologic studies, the issue of
confounding by other genes or environmental exposures must be carefully considered.
Several glutathione S-transferase (GST) families have common, functionally important
polymorphic alleles (e.g., homozygosity for the null allele at the GSTM1 and GSTT1 loci, homozygosity
for the A105G allele at the GSTP1 locus) that significantly reduce expression of enzyme function in the
lung. Exposure to radicals and oxidants from air pollution induces decreases in GSH that increase GST
transcription. Individuals with genotypes that result in enzymes with reduced or absent glutathione
peroxidase activity are likely to have reduced oxidant defenses and increased susceptibility to inhaled
oxidants and radicals.
Gilliland et al. (2002) examined effects of GSTM1, GSTT1, and GSTP1 genotypes and acute
respiratory illness, specifically respiratory illness-related absences from school. The goal was to examine
potential susceptibilities on this basis, but not specifically to air pollutants. They concluded that fourth
grade schoolchildren who inherited a GSTP1 Val-105 variant allele had a decreased risk of respiratory
illness-related school absences, indicating that GSTP1 genotype influences the risk and/or severity of
acute respiratory infections in school-aged children.
Lee et al. (2004) studied ninth grade schoolchildren with asthma in Taiwan for a gene-
environmental interaction between GSTP1-105 genotypes and outdoor pollution. They examined general
district air pollution levels of low (mean S02 level of 3.6 ppb from 1994 to 2001), moderate (mean S02 of
6.2 ppb), and high (mean S02 of 8.6 ppb) and found that compared with individuals with any Val-105
allele in the low air pollution district, lie-105 homozygotes in the high air pollution district had a
significantly increased risk of asthma.
Gauderman et al. (2002) describe a study method that uses principal components analysis
computed on single nucleotide polymorphism (SNP) markers to test for an association between a disease
and a candidate gene. For example, they evaluated the association between respiratory symptoms in
children and four SNPs in the GSTP1 locus, using data from the Southern California Children's Health
Study (CHS). The authors observed stronger evidence of an association using the principal components
approach (p = 0.044) than using either a genotype-based (p = 0.13) or haplotype-based (p = 0.052)
approach. This method may be applied to relationships in this and other databases to evaluate aspects of
air pollutants such as S02.
4-10
-------
Reference
Location
Wilson et al. (2005)*
Luginaah el al, (2005)
- Females
Atkinson et al. (1999a)*
Atkinson et al, (1999b)
Wong et al. (1999)
Wilson et al. (2005)*
Atkinson et al. (1999a)'
Atkinson et al. (1999b)
Hajatetal. (1999)*
Anderson et al. (1998)
Portland, ME
Manchester. NH
Windsor, ON
London, UK
London, UK
Ponce de Leon et al. (1996) London, UK
Petroeschevsky et al. (2001) Brisbane, Australia
Hong Kong. China
Portland, ME
Manchester, NH
London, UK
London, UK
London. UK
London, UK
Petroeschevsky et al. (2001) Brisbane, Australia
t
I
-~-r-
I —~-
5-14 yrs-
Respiratory
0-4 yrs
• = all ages
~ = children
- = adults
* = older adults
0.6 0.8 1.0 1.2 1.4
Relative Risk
1.6
1.8
Figure 4-6. Relative risks (95% CI) of age-specific associations between short-term exposure to
SO2 and respiratory ED visits* and hospitalizations. Risk estimates are standardized
per 10 ppb increase in 24-h avg SO2 concentrations or 40 ppb increase in 1-h max SO2.
Winterton et al. (2001) attempted to identify a genetic biomarker for susceptibility to S02. They
screened 62 asthmatic subjects for S02 responsiveness using an inhalation challenge and collected genetic
material via buccal swabs to test for associations between S02 sensitivity and specific gene
4-11
-------
polymorphisms. Subjects inhaled 0.5 ppm S02 by mouthpiece for 10 min while wearing noseclips during
moderate exercise on a treadmill. Subjects were defined as S02-sensitive if FEVi was decreased 12%.
Genetic polymorphisms as biomarkers of susceptibility were evaluated in five regions coding for the (32-
adrenergic receptor, the a subunit of the interleukin-4 (IL-4) receptor, the Clara cell secretory protein
(CC16), tumor necrosis factor-a (TNF-a), and lymphotoxin-a (also known as TNF-(3). The authors found
a significant association between response to S02 and the homozygous wild-type allele of TNF-a position
-308. All of the S02-sensitive subjects had the homozygous wild-type allele for TNF-a position -308,
while 61% of the nonresponders had this genotype. Homozygosity for the TNF-1 allele was associated
with a 5-fold increased risk of physician-diagnosed asthma relative to other genotypes. None of the other
polymorphisms showed significant trends.
In summary, the differential effects of air pollution in general among genetically diverse
subpopulations have been examined for a number of GST genes and other genotypes. The limited number
of studies may provide some insight into susceptible groups and a potential genetic role in such. Only one
of these studies specifically examined S02 as the exposure of interest, and it found a significant
association with the homozygous wild-type allele for TNF-a. Khoury et al. (2000a) states that while
genomics is still in its infancy, opportunities exist for developing, testing, and applying its tools to public
health research of outcomes with possible environmental causes. At this time, there are insufficient data
on which to base a conclusion regarding the effect of S02 exposure on genetically distinct subpopulations.
4.2.3. Age-Related Susceptibility
The American Academy of Pediatrics (2004) notes that children and infants are among the most
susceptible to many air pollutants, including S02. Eighty percent of alveoli are formed postnatally and
changes in the lung continue through adolescence; furthermore, the developing lung is highly susceptible
to damage from exposure to environmental toxicants (Dietert et al., 2000). Children also have increased
vulnerability as they spend more time outdoors, are highly active, and have high minute ventilation,
which collectively increase the dose they receive (Plunkett et al., 1992; Wiley et al. 1991a, 1991b).
A number of epidemiologic studies have observed increased respiratory symptoms in children
associated with increasing S02 exposures (Mortimer et al., 2002; Schildcrout et al., 2006; Schwartz et al.
1994), though there is no evidence from a limited number of studies suggesting this same effect in adults
(Desqueyroux et al. 2002a, 2002b; van der Zee et al., 2000). Similarly, adverse respiratory effects have
been observed in adolescents following S02 exposure in a laboratory setting (Koenig et al., 1981; 1983;
1987; 1988; 1990). However, there is no evidence from human clinical studies to suggest that the
respiratory effects in adolescents are more severe than those observed in adults.
Older adults are frequently classified as being particularly susceptible to air pollution. The basis of
the increased sensitivity in the older adults is not known, but one possibility is that it may be related to
changes in the respiratory tract lining fluid antioxidant defense network (Kelly and Mudway, 2003) or a
general reduction in immune competence.
A number of studies, investigating the association between ambient S02 levels and ED visits or
hospital admissions for all respiratory causes or asthma, stratified their analyses by age group. Figure 4-6
summarizes the evidence of age-specific associations between S02 and acute respiratory ED visits and
hospitalizations. Several studies demonstrated that the excess risk of ED visits or hospitalizations for all
respiratory causes or asthma was higher for children (e.g. Atkinson et al., 1999a; 1999b; Petroeschevsky
et al., 2001) and older adults (e.g. Petroeschevsky et al., 2001; Wilson et al., 2005; Wong et al. 1999)
when compared to the risk for adults or all ages together. This is more clearly depicted in the summary
density curves in Figure 4-7 and Figure 4-8, created using the effect estimates presented in Figure 4-6. As
shown in these two figures, the effect estimates for children and older adults are slightly larger than that
for adults or all ages for both all respiratory diseases and asthma ED visits and hospitalizations.
4-12
-------
50
40
>*
w 30
c
o
Q
20
10
° 0.90 0.95 1.00 1.05 1.10 1.15 1.20
Relative Risk
Figure 4-7. Summary density curves of the relative risks of age-specific associations between
short-term exposure to SO2 and ED visits and hospitalizations for all respiratory
causes. Risk estimates are standardized per ppb increase in 24-h avg SO2
concentrations or 40 ppb increase in 1-h max SO2. Density curves drawn from effect
estimates in Figure 4-6.
Age Groups
— All Ages
Children
- Adults
— Older Adults
30
25
20
(/)
i 15
10
5
0 0.90 0.95 1.00 1.05 1.10 1.15 1.20
Relative Risk
Figure 4-8. Summary density curves of the relative risks of age-specific associations between
short-term exposure to SO2 and ED visits and hospitalizations for asthma. Risk
estimates are standardized per ppb increase in 24-h avg SO2 concentrations or 40 ppb
increase in 1-h max SO2. Density curves drawn from effect estimates in Figure 4-6.
Cakmak et al. (2007b) reported that among seven Chilean urban centers, the percent increase in
nonaccidental mortality associated with a 10 ppb increase in 24-h avg S02 was 3.4% (95% CI: 0.7, 6.1)
for those < 65 years of age and 5.6% (95% CI: 2.2, 9.1) for those > 85 years of age. The authors
concluded that older adults are particularly susceptible to dying from air pollution, and suggested that
concentrations deemed acceptable for the general population may not adequately protect those aged > 85
years.
Age Groups
All Ages
— Children
Adults
— Older Adults
4-13
-------
There is limited epidemiologic evidence to suggest that children and older adults (65+ years) are
more susceptible to the adverse respiratory effects associated with ambient S02 concentrations when
compared to the general population.
4.2.4. Other Potentially Susceptible Populations
Although data specific to S02 exposures is lacking for the susceptibility factors listed below,
several other potentially susceptible groups deserve specific mention in this document. These include
individuals in a chronic pro-inflammatory state (e.g., diabetics), obese individuals, and children born
prematurely or with low birth weight.
Chronic inflammation appears to enhance susceptibility for air pollution-related cardiovascular
events in older individuals, persons with diabetes, coronary artery disease, obesity, and past Mis (Bateson
and Schwartz, 2004; Goldberg et al., 2001; Peel et al., 2007; Zanobetti and Schwartz, 2002). Dubowsky et
al. (2006) reported that individuals with conditions associated with both chronic inflammation and
increased cardiac risk were more vulnerable to the short-term pro-inflammatory effects of air pollution.
This included individuals with diabetes; obesity; and concurrent diabetes, obesity and hypertension.
Zanobetti and Schwartz (2001) reported more than twice the risk for hospital admissions for heart disease
in persons with diabetes than in persons without diabetes associated with exposure to ambient air
pollution, indicating that persons with diabetes are an important at-risk group. Data from the Third
National Health and Nutrition Examination Survey indicated that 5.1% of the U.S. population older than
20 years of age has diagnosed diabetes and an additional 2.7% has undiagnosed diabetes (Harris et al.,
1998). Moreover, another study found that subjects with impaired glucose tolerance without type II
diabetes also had reduced HRV (Schwartz, 2001). This may indicate that the at-risk population may be
even larger.
Mortimer et al. (2000) reported that among asthmatic children, birth characteristics continue to be
associated with increased susceptibility to air pollution later in life, demonstrating that air pollution-
induced asthma symptoms were more severe in children born prematurely or of low birth weight.
Specifically, the authors revealed asthmatic children born more than three weeks prematurely or weighing
less than 2,500 grams (5.5 pounds) had a six-fold decrease in breathing capacity associated with air
pollution compared to full-weight, full-term children. The low birth weight and premature children also
reported a five-fold greater incidence of symptoms like wheezing, coughing and tightness in the chest.
4.2.5. Factors that Potentially Increase Vulnerability to SO2
A limited amount of information exists on exposures to S02 among vulnerable populations. As
noted above, vulnerability is characterized by extrinsic factors that may increase a population's risk from
air pollution, such as being more highly exposed than the general population, or reduced SES. Because
indoor and personal S02 concentrations are generally much lower than outdoor or ambient measurements,
individuals that spend most of their time indoors, such as older adults, are not anticipated to be vulnerable
to high S02 exposures, though in some cases they may be more susceptible to the effects of these
exposures than the general population due to preexisting health factors. Another factor that potentially
alters vulnerability to S02 is air conditioning use due to the reduced penetration of S02 into buildings
when windows are closed.
Other individuals with increased vulnerability include those who spend a lot of time outdoors at
increased exertion levels, for example outdoor workers and individuals who exercise or play outdoor
sports. Exercise may cause an increase in uptake of S02 resulting from an increase in ventilation rate and
accompanying shift from nasal to oronasal breathing. Children, who generally spend more time playing
outdoors, may qualify as both a susceptible population (due to their developing physiology) and as a
4-14
-------
vulnerable population since ambient S02 concentrations are several-fold higher than indoor
concentrations.
SES is a known determinant of health and there is evidence that SES modifies the effects of air
pollution (Makri and Stilianakis, 2007; O'Neill et al., 2003). Both higher exposures to air pollution and
greater susceptibility to its effects may contribute to a complex pattern of risk among those with lower
SES. Conceptual frameworks have been proposed to explain the relationship between SES, susceptibility
and exposure to air pollution. Common to these frameworks is the consideration of the broader social
context in which people live and its effect on health in general (Gee and Payne-Sturges, 2004; O'Neill et
al., 2003), as well as on maternal and child health (Morello-Frosch and Shenassa, 2006), and asthma
(Wright et al., 2007) specifically. Multilevel modeling approaches that allow parameterization of
community level stressors such as increased life stress, as well as individual risk factors, are considered
by these authors. In addition, statistical methods that allow for temporal and spatial variability in exposure
and susceptibility, have been discussed in the recent literature (Jerrett and Finkelstein, 2005; Kunzli,
2005).
Several studies have examined modification by SES indicators on the association between
mortality and PM (Finkelstein et al., 2003; Jerrett et al., 2004; Martins et al., 2004; O'Neill et al., 2003;
Romieu et al., 2004) or other indices such as traffic density, distance to roadway or a general air pollution
index (Finkelstein et al., 2005; Ponce et al., 2005; Woodruff et al., 2003). However, modification of S02
associations has been examined in a few studies. For example, in a study conducted in 10 large Canadian
cities, living in communities in which individuals have lower household education and income levels
increased the individual's vulnerability to air pollution (Cakmak et al., 2006). These effects were
statistically significant for several gaseous criteria pollutants, but not for S02. In addition, Finkelstein et
al. (2003) evaluated neighborhood levels of income and air pollution in southern Ontario, Canada. They
found that both income and S02 levels were associated with mortality differences. Specifically, among
people with below-median income, the relative risk for those with above-median exposure to S02 was
1.18 (95% CI: 1.11, 1.26); the corresponding relative risk among subjects with above-median income was
1.03 (95% CI: 0.83, 1.28). Overall, there is very limited evidence available from which conclusions on the
human health effects from the interaction between SES and S02 can be drawn.
Other factors that may potentially increase vulnerability to S02 are residential or geographic
location. However, residential location is not as strong of a predictor of exposure vulnerability for S02 as
for traffic-related pollutants, because meteorological conditions have a greater impact on pollutant plume
direction from primary point sources such as coal-fired power plants.
4.2.6. Summary of Potentially Susceptible and Vulnerable Populations
In summary, subgroups considered to be potentially susceptible and/or vulnerable include children
and older adults; people with other respiratory disease; genetic factors; SES; and populations
experiencing heightened exposure levels (e.g., those living near roadways or other "hot spots" or engaged
in outdoor work or exercise). Also of concern are individuals who generally may not be inherently
susceptible to S02-related health effects but may experience transient increases in airways sensitivity to
SOx induced by other respiratory irritants such as recent viral respiratory infection (Stempel and Boucher,
1981). These groups comprise a large fraction of the U.S. population. Given the heterogeneity of
individual responses to air pollution, the severity of health effects experienced by a susceptible subgroup
may be much greater than that experienced by the population at large (Zanobetti et al., 2000).
4-15
-------
Chapter 5. Summary and Conclusions
Previous chapters present the most policy-relevant information related to the review of the NAAQS
for SOx. This chapter integrates key findings from atmospheric sciences, ambient air data analyses,
exposure assessment, dosimetry, and health evidence. The EPA framework for causal determinations
described in Chapter 1 has been applied to the body of evidence in order to judge the scientific data about
exposure to SOx and health effects in a two-step process. The first step is to determine the weight of
evidence in support of causation at relevant pollutant exposures and characterize the strength of any
resulting causal classification. The EPA framework applied here employs a five-level hierarchy for causal
determination:
¦ Causal relationship
¦ Likely to be a causal relationship
¦ Suggestive of a causal relationship
¦ Inadequate to infer a causal relationship
¦ Suggestive of no causal relationship
The second step evaluates the entirety of policy-relevant quantitative evidence regarding the
concentration-response relationships including levels and exposure durations at which effects are
observed, and subpopulations that experience effects that differ from the general population. This
integration of evidence results in identification of a study or set of studies that best estimates the
concentration-response relationships for the U.S. population, given the current state of knowledge.
Together the two steps in the framework lead to: 1) causal determinations for a range of health outcomes,
and 2) characterization of the magnitude of these responses, including susceptible or vulnerable
subpopulations, over a range of relevant exposures.
This chapter summarizes and integrates the newly available scientific evidence that best informs
consideration of the policy-relevant questions that frame this review, presented in Chapter 1. Section 5.1
presents trends in emissions of S02 and provides a brief summary of the ambient air quality at short- and
long-term exposures ranging from 5 min to one year. Section 5.2 discusses the evidence for the
occurrence and plausibility of health effects following short- and long-term exposure to ambient S02, and
the levels at which these health effects occur. Section 5.3 integrates the evidence and discusses important
uncertainties identified in the interpretation of the scientific evidence. Section 5.4 presents the evidence
for potentially susceptible and vulnerable populations to health effects from S02 exposure. Finally,
conclusions based on the available scientific evidence for human health effects associated with S02
exposure are presented in Section 5.5.
5.1. Emissions and Ambient Concentrations of SO2
Anthropogenic S02 is emitted mainly by fossil fuel combustion (chiefly coal and oil) and metal
smelting. The largest source of emissions is from elevated point sources such as the stacks of power
plants and industrial facilities. Since 1990, in response to controls applied under the Acid Rain Program
(U.S. EPA, 2006a), S02 emissions from these sources have declined substantially. Emissions demonstrate
a strong gradient increasing from west to east, owing to the high concentration of S02-emitting electric
generating utilities in the Ohio River Valley, the Southeast, Texas, Illinois, and Missouri. Policy Relevant
Background (PRB) levels of S02 are estimated to be in the range of a few hundredths of a ppb (< 1% of
typical ambient levels) across most of the U.S., though much higher values are found in areas affected by
volcanic or geothermal activity.
5-1
-------
The levels of the current primary NAAQS for SOx are 0.14 ppm for 24-h avg S02 concentrations
and 0.03 ppm for annual avg S02 concentrations, not to be exceeded more than once per year.
Exceedances of the current primary NAAQS for SOx have become rare in recent years, as both the mean
24-h and annual avg S02 concentrations in the U.S. for the years 2003 to 2005 were ~4 ppb. For the 24-h
avg, 99th percentile and maximum values were -25 ppb and -145 ppb, respectively. The 99th percentile
and maximum values of the annual avg were -13 ppb and -15 ppb, respectively.
Mean 1-h max concentrations in these years were -13 ppb, with a 99th percentile value of -95 ppb
and maximum value of -700 ppb. The large differences between 99th percentile and maximum values for
the shorter term averages suggest that the maxima are strongly limited spatially and temporally and are
not a major determinant of the mean values.
5-min avg S02 data are collected without a specific regulatory mandate. Hourly maximum 5-min
avg data were voluntarily supplied from 98 monitors located in certain regions of the U.S. during the
period 1997 to 2006, with only 16 monitors providing all twelve 5-min averages in each hour. The median
value of the hourly maximum 5-min avg in this limited data set ranged from 1 to 8 ppb, while the 99th
percentile value ranged from 21 to 184 ppb, depending on location. Because of the nonuniform
spatiotemporal distribution of the existing 5-min data set across the U.S., these values are likely not
representative of nationwide 5-min distributions.
5.2. Health Effects of SO2
Evaluation of the health evidence, with consideration of issues related to atmospheric sciences,
exposure assessment, and dosimetry, led to the conclusion that there is a causal relationship between
respiratory morbidity and short-term exposure to S02. This conclusion is supported by the consistency,
coherence, and plausibility of findings observed in the human clinical, epidemiologic, and animal
toxicological studies. In human clinical studies, respiratory effects were observed following 5-10 min
exposures to S02 at concentrations >0.2 ppm in asthmatics engaged in moderate to heavy levels of
exercise. In the epidemiologic studies, respiratory effects were observed in areas where the maximum
ambient 24-h avg S02 concentration was below the current 24-h avg NAAQS level of 0.14 ppm. Mean
24-h avg S02 levels ranged from 1 to 30 ppb in these epidemiologic studies conducted in the U.S. and
abroad, with maximum 24-h avg S02 values ranging from 12 to 75 ppb. Animal toxicological studies
indicate that repeated exposures to S02, at concentrations as low as 0.1 ppm in guinea pigs, may
exacerbate airway inflammation and hyperresponsiveness in allergic animals.
The respiratory health effects of S02 are consistent with the mode of action of S02 as it is currently
understood. The immediate effect of S02 on the respiratory system is bronchoconstriction. This response
is mediated by chemosensitive receptors in the tracheobronchial tree. These receptors trigger reflexes at
the central nervous system level resulting in bronchoconstriction, mucus secretion, mucosal vasodilation,
cough, and apnea followed by rapid shallow breathing. In some cases, local nervous system reflexes also
may be involved. Asthmatics are more sensitive to the effects of S02 likely resulting from preexisting
inflammation associated with this disease. This inflammation may lead to enhanced release of mediators,
alterations in the autonomic nervous system and/or sensitization of the chemosensitive receptors. These
biological processes are likely to underlie decreased lung function and increased hyperresponsiveness
observed in response to S02 exposure.
The definitive evidence for the causal relationship comes from human clinical studies reporting
respiratory symptoms and decreased lung function in exercising asthmatics following 5-10 min exposures
to S02 at concentrations which have sometimes been measured in ambient air for similarly short-time
durations. With increased ventilatory rates during exercise, the pattern of S02 absorption shifts from the
upper airways to the tracheobronchial airways in conjunction with a shift from nasal to oronasal
breathing. Mode of breathing is an important determinant of the severity of S02-induced
bronchoconstriction, with the greatest responses occurring during oral breathing followed by oronasal
5-2
-------
breathing and the smallest responses observed during nasal breathing. In the human clinical studies,
5-30% of relatively healthy exercising asthmatics experienced moderate or greater decrements in lung
function (> 100% increase in sRaw or > 15% decrease in FEVi) with peak exposures to S02
concentrations of 0.2-0.3 ppm (see Table 5-1). At concentrations >0.4 ppm, a greater percentage (20-
60%) of asthmatics experience S02-induced decrements in lung function, which are frequently
accompanied by respiratory symptoms. A clear concentration-response relationship has been
demonstrated following exposures to S02 at concentrations between 0.2 and 1.0 ppm, both in terms of
increasing severity of effect and percentage of asthmatics adversely affected. Animal toxicological studies
have also reported bronchoconstriction with short-term exposures of 1 ppm S02 (see Table 5-2). The
limited 5-min S02 data acquired from ambient monitors located in certain regions of the U.S. during the
years 1997 to 2006 indicated that 0.2% of the hourly maximum 5-min avg were greater than 0.2 ppm.
Table 5-1. Key health effects of short-term exposure to SO2 observed in human clinical studies.
Concentration Exposure Effects Study
0.2-0.3 ppm 5-10 min Moderate to large reductions in FEVi and increases in sRaw observed among some Bethel et al. (1985); Horstman et al.
asthmatic adults (5-30%) during moderate to heavy exercise. Bronchial (1986);Koenig et al. (1990); Linn et al.
responsiveness to SO2 may be enhanced when preceded by exposure to O3. Limited (1983b; 1987; 1988; 1990); Schachter
evidence of S02-induced increases in respiratory symptoms. et al. (1984); Sheppard et al. (1981);
Trenga et al. (2001)
1-6 h Enhanced airway responses to an inhaled allergen following exposure to SO2 with Devalia et al. (1994); Routledge et al.
NO2 in resting asthmatics. No evidence of respiratory symptoms or decrements in (2006); Rusznak et al. (1996);
lung function in resting asthmatics or healthy adults. The evidence that SO2 exposure Tunnicliffe et al. (2001; 2003)
may lead to changes in heart rate variability is weak and inconsistent.
0.4-0.5 ppm 1-10 min Moderate or greater decrements in lung function clearly demonstrated in asthmatics
during exercise, with significant interindividual variability in response (-20-35% of
asthmatics experiencing moderate or greater decrements in lung function).
Respiratory effects observed following 5-10 min of exposure are generally not
enhanced by repeat exposures. Respiratory symptoms (e.g., wheezing, chest tight-
ness) are observed at concentrations as low as 0.4 ppm and have been shown to
increase with increasing exposure concentrations.
Balmes et al. (1987); Gong et al.
(1995); Horstman et al. (1986); Koenig
et al. (1983); Linnet al. (1983b; 1987);
Magnussen et al. (1990); Schachter
et al. (1984); Sheppard et al. (1981);
Trenga et al. (1999)
~1-h Decrements in lung function among asthmatics following 10 min of exercise at the end Linn et al. (1987); Roger et al. (1985)
of a 60-75 min exposure are statistically significant, but less severe than effects
observed following a 10 min period of exercise at the start of the exposure.
0.6-1.0 ppm 1-10 min Clear and consistent S02-induced increases in respiratory symptoms observed
among exercising asthmatics. Moderate to large decrements in lung function
demonstrated in 35-60% of asthmatics. Respiratory effects attributed to SO2 among
asthmatics during exercise may be diminished after cessation of exercise, even with
continued SO2 exposure. No respiratory effects reported in healthy, non-asthmatics.
Balmes et al. (1987); Gong et al.
(1995); Hackney et al. (1984); Hors-
tman et al. (1986; 1988); Koenig et al.
(1983); Linnet al. (1987; 1988; 1990);
Roger et al.(1985); Schachter et al.
(1984)
1-6 h Decrements in lung function among asthmatics following 5-10 min of exercise at the Linn et al. (1984; 1987); Hackney et al.
end of a 1-6 h exposure are statistically significant, but less severe than effects (1984); Roger et al. (1985)
observed following a 5-10 min period of exercise at the start of the exposure.
5-3
-------
Table 5-2. Key respiratory health effects of exposure to SO2 in animal toxicological studies.
Study
Exposure
Species
Effects
LUNG FUNCTION
Amdur et al.
(1983)
1 ppm (2.62 mg/m3); 1 h;
head only
Hartley guinea pigs,
male, age not reported,
200-300 g,n = 18-23/
group
An 11% increase in pulmonary resistance and 12% decrease in dynamic compliance
were observed. Neither effect persisted into the 1 h period following exposure. No
effects were observed for breathing frequency, tidal volume, or min volume.
Barthelemy
et al. (1988)
0.5 or 5 ppm
(1.3 or 13.1 mg/m3); 45 min;
intratracheal
Rabbit, sex not reported,
adult, mean 2.0 kg,
n = 5-9/group; rabbits
were mechanically
ventilated
Lung resistance increased by 16% and 50% in response to 0.5 and 5 ppm SO2,
respectively.
Conner etal.
(1985)
1 ppm (2.62 mg/m3); 3-h/day
for 6 days; nose only; animals
evaluated for up to 48-h
following exposure
Hartley guinea pig, male,
age not reported,
250-320 g, n = <18/
group/time point
No effect was observed on residual volume, functional reserve capacity, vital capacity,
total lung capacity, respiratory frequency, tidal volume, pulmonary resistance,
pulmonary compliance, diffusing capacity for carbon monoxide or alveolar volume at 1
or 48 h after last exposure.
INFLAMMATION AND MORPHOLOGY
Li et al.
(2007)
2 ppm (5.24 mg/m3); 1 h/day
for 7 days; with and without
exposure to ovalbumin
Wistar rats, male, age not Increased number of inflammatory cells in BAL fluid, increased levels of MUC5AC and
reported ICAM-1 and an enhanced histopathological response compared with those treated with
ovalbumin or SO2 alone
Park et al.
(2001a)
0.1 ppm (0.26 mg/m3); 5 h/day Dunkin-Hartley guinea
for 5 days; whole body; with pig, male, age not re-
and without exposure to ported, 250-350 g,
ovalbumin n = 7-12/group
After bronchial challenge, the ovalbumin/S02-exposed group had significantly
increased eosinophil counts in BAL fluids compared with all other groups, including the
S02-only group. The bronchial and lung tissue of the ovalbumin/S02-exposed group
showed infiltration of inflammatory cells, bronchiolar epithelial damage, and mucus and
cell plug in the lumen.
Conner etal.
(1989)
1 ppm (2.62 mg/m3); 3-h/day
for 1-5 days; nose only; bron-
choalveolar lavage performed
each day
Hartley guinea pig, male,
age not reported,
250-320 g, n = 4
No change in numbers of total cells and neutrophils, protein levels or enzyme activity in
lavage fluid following SO2 exposure.
Conner etal.
(1985)
1 ppm (2.62 mg/m3); 3 h/day/6 Hartley guinea pigs,
day; evaluated up to 72 h male, age not reported,
postexposure 250-320 g, n = 18/
group/time point
No alveolar lesions.
Smith et al.
(1989)
1 ppm (2.62 mg/m3); 5-h/day,
5 day/wk up to 4 and 8 mos
Sprague-Dawley rats,
male, young adult, initial
weight not reported, n =
12-15 data point
Increased bronchial epithelial hyperplasia and number of nonciliated epithelial cells
observed at 4 mos.
AIRWAY HYPERRESPONSIVENESS AND ALLERGIC SENSITIZATION
Park et al.
(2001a)
0.1 ppm (0.26 mg/m3); 5-
h/day for 5 days; whole body;
with and without exposure to
ovalbumin
Dunkin-Hartley guinea
pig, male, age not re-
ported, 250-350 g,
n = 7-12/group
After bronchial challenge, the ovalbumin/S02-exposed group had significantly
increased enhanced pause (indicator of airway obstruction) compared with all other
groups, including the SO2 group. Study authors concluded that low level SO2 may
enhance the development of ovalbumin-induced asthmatic reactions in guinea pigs.
Riedel et al.
0.1, 4.3, or 16.6 ppm (0, 0.26,
11.3, or 43.5 mg/m3); 8-h/day
for 5 days; whole body;
animals were sensitized to
ovalbumin on the last 3 days
of exposure
Perlbright-White guinea
pig, female, age not
reported, 300-350 g,
n = 5 or 6/g roup (14
controls)
Bronchial provocation with ovalbumin was conducted every other day for 2 wks,
starting at 1 wk after the last exposure. Numbers of animals displaying symptoms of
bronchial obstruction after ovalbumin provocation was increased in all SO2 groups
compared to air-exposed groups. Anti-ovalbumin antibodies (IgG total and lgG1) were
increased in BAL fluid and serum of S02-exposed compared to air-exposed controls,
with statistical significance attained for IgG total in BAL fluid at a4.3 ppm SO2 and in
serum at all SO2 concentrations. Results indicate that in this model, repeated exposure
to even low concentrations of SO2 can potentiate allergic sensitization of the airway.
A larger body of evidence supporting this determination of causality comes from numerous
epidemiologic studies published since the previous NAAQS review reporting associations with
5-4
-------
respiratory symptoms, ED visits, and hospital admissions with short-term S02 exposures, generally of 24-
h avg. These studies were conducted in areas where the maximum ambient 24-h avg S02 concentration
was consistently below the current 24-h avg NAAQS level of 0.14 ppm. Mean 24-h avg S02 levels
ranged from 1 to 30 ppb, with maximum values ranging from 12 to 75 ppb, in the epidemiologic studies,.
Important new multicity studies and several other studies have found an association between 24-h avg
ambient S02 concentrations and respiratory symptoms in children, particularly those with asthma.
Furthermore, limited epidemiologic evidence indicates that atopic children and adults may be at increased
risk for S02-induced respiratory symptoms. Generally consistent associations also were observed between
ambient S02 concentrations and ED visits and hospitalizations for all respiratory causes, particularly
among children and older adults (> 65 years), and for asthma. The S02-related changes in ED visits or
hospital admissions for respiratory causes ranged from -5% to 20% excess risk.
The consistency and coherence of the epidemiologic evidence for respiratory effects associated
with short-term exposure to S02 are illustrated in Figure 5-1 and Figure 5-2, which present effect
estimates for respiratory symptoms, ED visits, and hospitalizations in children. Associations between
short-term ambient S02 concentrations and respiratory symptoms, ED visits, and hospitalizations are
largely positive, with several of the more precise effect estimates (suggestive of greater study power)
indicating statistical significance. The epidemiologic findings of asthma symptoms with 24-h avg S02
exposures are generally coherent with increases in symptoms reported in asthmatics in human clinical
studies with 5-10 min exposures; it is possible that these epidemiologic associations are determined in
large part by peak exposures within a 24-h period. The effects of S02 on respiratory symptoms, lung
function, and airway inflammation observed in the human clinical studies using peak exposures further
provides a basis for a progression of respiratory morbidity resulting in increased ED visits and hospital
admissions. Collectively, these findings provide biological plausibility for the observed associations
between ambient S02 levels and ED visits and hospitalizations for all respiratory diseases and asthma,
notably in children and older adults (> 65 years).
Overall, the epidemiologic evidence for respiratory morbidity is consistent, with associations
reported in studies conducted in numerous locations using a variety of methodological approaches. The
potential influence of copollutants has not been systematically considered in the epidemiologic literature.
A limited subset of the studies examined potential confounding by copollutants using multipollutant
regression models. These analyses indicated that although copollutant adjustment had varying degrees of
influence on the S02 effect estimates, the effect of S02 on respiratory health outcomes appeared to be
generally robust and independent of the effects of gaseous copollutants, including N02 and 03. The
evidence for PMi0 was less consistent, though it was noted that in the limited number of studies that
examined PM2 5 and PM10_2 5 the S02 estimates were generally robust to the inclusion of these copollutants
in the regression model. These findings suggest that the observed effects of S02 on respiratory endpoints
occur independent of the effects of other ambient air pollutants.
Intervention studies provide additional evidence that supports a causal relationship between S02
exposure and respiratory health effects. Hill (1965) emphasized that intervention studies can provide
strong support for causal inferences. Two notable studies conducted in several cities in Germany and in
Hong Kong reported that decreases in S02 concentrations were associated with improvements in
respiratory symptoms. In eastern Germany, a decrease in the prevalence of respiratory symptoms was
correlated with a steep decline in ambient S02 concentrations of more than 90% from 1992-1993 to 1998-
1999 (Heinrich et al., 2002). During this study period, decreases in other ambient air pollutants, including
-60% lower TSP concentrations, also occurred in these cities. In Hong Kong (Peters et al., 1996b),
respiratory health improved with similarly large reductions in S02 of up to 80% in the polluted district but
with much smaller reductions in TSP (less than 20%) compared with those in the cities in eastern
Germany. The possibility remains that these health improvements may be partially attributable to
declining concentrations of air pollutants other than S02, most notably PM or constituents of PM. Animal
toxicological studies conducted at higher concentrations (> 1 mg/m3 PM and > 1 ppm S02) suggest that
S02 effects may be potentiated by coexposure to PM but the relevance of these results to ambient
5-5
-------
exposures is not clear. Hence, the improvements in respiratory health may be jointly attributable to
declines in both S02 and PM.
Reference
Location
Schildcrout et al. (2006) 7 US Cities
Romieu et al. (1996) Mexico City.
Mexico
Mortimer et al, (2002) 8 US Cities
Schwartz et al. (1994) 6 US Cities
van der Zee et al. (1999) The Netherlands
(urban area)
Neas et al. (1995) Uniontown, PA
Hoek & Brunekreef (1994) The Netherlands
Roemeret al. (1993)
Ward et al. (2002a)
Pikhart et al. (2000)
Segala et al. (1998)
The Netherlands
Birmingham and
Sandwell, UK
Czech Republic
Paris, France
IRS
URS ¦
W-
RN
W-
. AS
LRS _
-AS
AS Asthma symptoms
I Inhaler use
C Cough
LRS Lower respiratory symptoms
URS Upper respiratory symptoms
W Wheeze
SOB Shortness of breath
P Phlegm
RN Runny nose
Rl Respiratory infection
LRS
LRS
-URS
-SOB
0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.4 2.8
Odds Ratios for Respiratory Symptoms
Figure 5-1. Odds ratios (95% CI) for the association between short-term exposures to ambient
SO2 and respiratory symptoms in children. Odds ratios are standardized per 10-ppb
increase in 24-h avg SO2 level. Studies are generally presented in the order of
increasing width of the 95% CI.
5-6
-------
Reference
Location
Ponce de Leon et al. (1996)
London, UK
Wong et al. (1999)
Hong Kong, China
Atkinson et al. (1999b)
London, UK
Atkinson et al, (1999a)*
London, UK
Gouveia and Fletcher (2000)
Sao Paulo, Brazil
Wilson et al. (2005)*
Portland, ME
Manchester, NH
Yang et al, (2003)
Vancouver, BC
Luginaah et al. (2005)
Windsor, ON
Petroeschevsky et al. (2001)
Brisbane, Australia
Barnett et al (2005)
Multicity, Australia
Lee et al. (2006)
Hong Kong, China
Sunyeretal. (1997)
Multicity, Europe
Anderson et al. (1998)
London, UK
Atkinson et al. (1999b)
London, UK
Hajat et al. (1999)*
London, UK
Atkinson et al. (1999a)*
London, UK
Lin et al. (2004a)
Bronx, NY
Petroeschevsky et al. (2001)
Brisbane. Australia
Gouveia and Fletcher (2000)
Sao Paulo, Brazil
Wilson et al. (2005)*
Portland, ME
Manchester, NH
Lin et al. (2003)
Toronto, ON
Barnett et al, (2005)
Multicity, Australia
Respiratory
Males -
5-14 yrs
Males-
1-4 yrs-
5-14 yrs
, Females
yrs.
-5-14 yrs
.1-4 yrs
Asthma
. Females
~~II~"
0.6
i"
-r~
1.2
—T~
1.8
0.8 1.0 1.2 1.4 1.6 1.8 2.0
Relative Risk for Respiratory ED Visits arid Hospitalizations
Figure 5-2. Relative risks (95% CI) for the association between short-term exposures to ambient
SO2 and emergency department (ED) visits (*) and hospitalizations for all respiratory
diseases and asthma in children. Relative risks are standardized per 10-ppb increase
in 24-h avg SO2 level. The studies are generally presented in the order of increasing
width of the 95% CI.
The ISA also evaluates the evidence of other health outcomes and exposure durations. For short-
term exposure to S02 and mortality, the evidence was suggestive of a causal relationship. Recent
epidemiologic studies have consistently reported positive associations between mortality and SO2, with
slightly larger effect estimates observed for respiratory mortality compared to cardiovascular mortality.
5-7
-------
However, the S02 effect estimates were generally reduced after adjusting for copollutants in the
regression models, indicating some extent of confounding among these pollutants. The evidence between
short-term S02 exposure and cardiovascular effects, and morbidity and mortality with long-term S02
exposures is inadequate to infer a causal relationship. The key conclusions on the health effects of S02
exposure are briefly summarized in Section 5.5.
5.3. Integration of the Evidence
This section highlights some key considerations for the evaluation of the evidence in this ISA. As
discussed above, clinical studies provide the definitive evidence that short-term S02 exposure is
associated with respiratory morbidity. Numerous epidemiologic studies report associations for a broader
range of respiratory health outcomes, at lower concentrations than the clinical studies and at levels below
the current 24-h avg standard. There is, however, uncertainty about the magnitude of the epidemiologic
effects estimates. Several sources of uncertainty and the implications for risk assessment are discussed
below.
Although the numerous epidemiologic studies provide supportive evidence in making a causal
determination for the effect of S02 on respiratory health, uncertainty remains in quantifying the
concentration-response relationship. Exposure measurement error is a key source of this uncertainty as
there are questions about the extent to which concentrations measured by the regulatory ambient
monitoring network typically used in epidemiologic studies can accurately represent an individual's
exposure to S02 of ambient origin. S02 monitors currently deployed in the regulatory monitoring
networks are adequate to determine compliance with current standards, since both the 24-h and annual
avg standards are substantially above the operating limit of detection of these monitors. However, these
monitors are inadequate for accurate and precise measurements at or near the current ambient mean 24-h
avg S02 level of ~4 ppb. Other factors contributing to exposure measurement error include the siting of
ambient monitors, spatial variation in ambient S02, variation in time-activity patterns, infiltration
characteristics of microenvironments, and instrument error.
Only a limited number of studies have focused on the relationship between personal exposure and
ambient concentrations of S02, in part because ambient levels have declined markedly over the past few
decades. Current indoor and outdoor concentrations are often beneath detection limits for passive personal
monitors used in S02 exposure studies. In such situations, associations between ambient concentrations
and personal exposures are inadequately characterized. However, in two studies with personal
measurements above detection limits (Brauer et al., 1989; Sarnat et al., 2006a), a reasonably strong
association (statistically significant regression slopes, with R2 = 0.15-0.43) was observed between
personal S02 exposure and ambient concentrations.
No studies have characterized the relationship between community avg exposure and ambient
concentrations, which is more directly relevant to community time-series, short-term panel, and long-term
cohort epidemiologic studies. Variations across a community in the fractional contribution of ambient S02
to exposure generally will not influence the magnitude of the observed health effect estimate in the
epidemiologic studies, though the standard error of the estimate would tend to be increased by
intracommunity variations. However, for community time-series and short-term panel studies, exposure
and analytical measurement errors would tend to bias the effect estimate towards the null, leading to
uncertainty in accurately quantifying the magnitude of the effect. In long-term exposure studies, the
variable ambient measurement and exposure error could also result in bias, but the extent and direction of
this bias is unclear.
Another factor that contributes to uncertainty in estimating the S02-related effect from
epidemiologic studies is that S02 is one component of a complex air pollution mixture including various
other components known to affect respiratory health. Several of these copollutants have been found to be
5-8
-------
correlated with S02 in epidemiologic studies (e.g., PM, N02, CO). Of particular interest is the correlation
between S02 and sulfate, the principal atmospheric oxidation product of S02. Short-term, mostly time-
series epidemiologic studies generally use intracity ambient concentration data which show very little or
no correlation between emitted S02 and transformed sulfate. In contrast, long-term epidemiologic studies
using intercity data can show correlations between S02 and sulfate on the order of 0.8 or higher. In these
studies, the fine-scale spatiotemporal variations in the intracity data are significantly reduced since sulfate
has sufficient time for production from S02, is dispersed over a wide spatial area, and can be mixed down
to ground level. Layered over these spatial and fine-scale temporal differences are seasonal and regional
dissimilarities driven by cities' various S02 emissions profiles and differing available time and sunlight
conditions for oxidation. Therefore, the relationship between S02 and sulfate indicates temporal and
spatial discongruities that can influence exposures and related health effects.
As a consequence of these uncertainties, the epidemiologic observations of S02 health effects can
be interpreted in different ways which are not mutually exclusive. First, the reported S02 effect estimates
in epidemiologic studies may reflect independent S02 effects on respiratory health. This is supported by
evidence from human clinical studies which indicate that peak exposures (5-10 min) to S02 at levels as
low as 0.2-0.3 ppm are capable of eliciting respiratory responses in asthmatics. In the epidemiologic
studies, S02-related effects on respiratory morbidity were observed in areas where the mean 24-h avg S02
levels ranged from 1 to 30 ppb, with maximum values ranging from 12 to 75 ppb. There are several
factors that may explain the difference between the concentration-response relationships in the human
clinical and epidemiologic studies. First, human clinical studies examine effects in groups of relatively
healthy individuals who do not represent the full range of susceptibilities present in the general
population. In addition, clinical studies include very small numbers of subjects as compared to the
epidemiologic studies that consider large populations exposed to ambient concentrations. These two
factors limit the power of human clinical studies to characterize exposure-response relationships at
ambient-relevant concentrations for the general population and sensitive subpopulations. Further,
comparisons are being made between 5-10 min exposures in the human clinical studies and 24-h avg
exposures values reported in the epidemiologic studies. It is difficult to know if high short-term S02
concentrations are driving the observed associations among the general population given the limited
number of ambient monitoring sites reporting short-term concentration data across the country. Among
the limited number of epidemiologic studies evaluating the concentration-response function, a few studies
found that a marked increase in S02-related respiratory health effects was only observed at higher
concentrations (above 90th percentile values). However, several others reported a linear relationship
across the entire range of concentrations.
Another interpretation is that ambient S02 may be serving as an indicator of complex ambient air
pollution mixtures sharing the same source as S02 (i.e., combustion of sulfur-containing fuels or metal
smelting). Other components of mixed emissions from these sources include trace elements such as
vanadium, nickel, selenium, and arsenic. Distinguishing effects of individual pollutants in multipollutant
regression models is made difficult by the possibility that a given air pollutant may be acting as a
surrogate for a less-well-measured or unmeasured pollutant, or that several pollutants may all be acting as
surrogates for the same mixtures of pollutants. Therefore, reported S02-related effects may represent
those of the overall mixture or other chemical components within the mixture. Although these issues
complicate the interpretation of effect estimates from multipollutant regression models, the limited
available evidence indicates that the effect of S02 on respiratory health outcomes appeared to be generally
robust and independent of the effects of gaseous copollutants, including N02 and 03, as well as particulate
copollutants, particularly PM2 5.
5-9
-------
5.4. Susceptible and Vulnerable Populations
Evidence from epidemiologic and human clinical studies, supported by animal studies, has
indicated that certain subgroups within the population are more susceptible and/or vulnerable to the
effects of S02 exposure. There is substantial evidence from epidemiologic and human clinical studies
indicating that asthmatics are more susceptible to respiratory health effects from S02 exposures than the
general public. Limited epidemiologic evidence further indicates that children and older adults (> 65
years) are more susceptible to the adverse respiratory effects associated with ambient S02 concentrations
when compared to the general population. A number of potentially susceptible groups, including obese
individuals, individuals having a chronic pro-inflammatory state like diabetics, and children born
prematurely or with low birth weight (< 2,500 grams), may experience increased adverse effects
associated with exposure to air pollution, but these relationships have not been examined specifically in
relation to S02. The differential effects of air pollution among genetically diverse subpopulations have
been examined for a number of glutathione-S-transferase (GST) genes and other genotypes. While limited
in number, these studies provide some insight into a potential genetic role in the determination of
susceptibility to air pollution.
Human clinical studies have clearly shown that exercising asthmatics are at greatest risk of
experiencing adverse respiratory effects related to S02 exposure. Oronasal breathing during exercise
increases vulnerability as it allows a larger fraction of inhaled S02 to reach the lower airways. Therefore,
individuals with increased vulnerability for S02-related respiratory health effects include those who spend
time outdoors at increased exertion levels, for example children, outdoor workers, and individuals who
exercise or play sports.
5.5. Conclusions
The important findings of this ISA on the health effects of S02 exposure, including the levels at
which effects are observed, are briefly summarized in Table 5-3. Also summarized are conclusions drawn
in the 1996 NAAQS review for comparison.
Collectively, the human clinical, epidemiologic, and animal toxicological data are sufficient to
conclude that there is a causal relationship between respiratory morbidity and short-term exposure to S02.
Observed associations between S02 exposure and an array of respiratory outcomes, including respiratory
symptoms, lung function, airway inflammation, AHR, and ED visits and hospitalizations from the human
clinical, animal toxicological, and epidemiologic studies, in combination, provide clear and convincing
evidence of consistency, specificity, temporal and biologic gradients, biological plausibility, and
coherence.
Human clinical studies consistently demonstrate respiratory morbidity among exercising asthmatics
following peak exposures (5-10 min) to S02 concentrations > 0.4 ppm, with respiratory effects occurring
at concentrations as low as 0.2 ppm in some asthmatics. In the epidemiologic studies, the S02-related
respiratory effects were consistently observed in areas where the maximum ambient 24-h avg S02
concentration was below the current 24-h avg NAAQS level of 0.14 ppm (see Tables 5-4 and 5-5).
Potentially susceptible and vulnerable subgroups include asthmatics, children, older adults, and
individuals who spend a lot of time outdoors at increased exertion levels.
In addition to respiratory morbidity related to short-term exposure to S02, studies of other health
outcomes and exposure durations were also evaluated in this ISA. The evidence is suggestive of a causal
relationship between short-term exposure to S02 and mortality. The evidence linking short-term S02
exposure and cardiovascular effects, and morbidity and mortality with long-term exposures to S02 is
inadequate to infer a causal relationship.
5-10
-------
Table 5-3. Key findings on the health effects of SO2 exposure
Short-Term Exposure: RESPIRATORY MORBIDITY
Causal relationship
RESPIRATORY SYMPTOMS
Conclusions from 1996 NAAQS Review: Among exercising asthmatics, there is a clear, statistically
significant increase in respiratory symptoms following peak exposures (5-10 min) to 0.6-1.0 ppm SO2.
Significant, but less severe symptoms are associated with peak SO2 exposures at concentrations of
0.4-0.5 ppm in human clinical studies.
In the epidemiologic studies, an association with aggravation of bronchitis is consistently observed at
24-h avg SO2 levels of 0.19 to 0.23 ppm and in some cases at levels below 0.19 ppm.
Conclusions from Current Review: Severity and incidence of respiratory symptoms has been shown to
increase with increasing concentrations between 0.2 and 0.6 ppm SO2 in exercising asthmatic adults
following peak exposures (5-10 min). Statistically significant increases in symptoms are observed at
SO2 concentrations > 0.4 ppm.
Epidemiologic studies provide consistent evidence of an association between ambient SO2 exposure
and increased respiratory symptoms in children, particularly those with asthma or chronic respiratory
symptoms. Multicity studies have observed these associations at a median range of 17 to 37 ppb
(75th percentile: -25 to 50) across cities for 3-h avg SO2 and 2.2 to 7.4 ppb (90th percentile: 4.4 to
14.2) for 24-h avg S02.
In contrast, the epidemiologic evidence on the association between S02 and respiratory symptoms in
adults is inconsistent at current short-term avg ambient SO2 concentrations.
LUNG FUNCTION
Conclusions from 1996 NAAQS Review: Bronchoconstriction has been found to be the most sensitive
indicator of lung function effects following acute exposure to SO2. Guinea pigs were found to be the
most sensitive species, with bronchoconstriction observed using 0.16 ppm S02. In human clinical
studies, < 10-20% of exercising asthmatic individuals experience large decrements in lung function
(i.e., sRaw increases > 200% or FEV1 decreases > 20%) following 5-10 min exposures to SO2
concentrations of 0.2-0.5 ppm. At 0.6-1.0 ppm SO2, s 20-25% of exercising asthmatics are similarly
affected.
Small, reversible declines in lung function in children are observed in epidemiologic studies at levels of
0.10 to 0.18 ppm but not at levels of 0.04 to 0.08 ppm.
Conclusions from Current Review: SC>2-induces moderate or greater decrements in lung function, i.e.
increases in sRaw (> 100%) or decreases in FEV-i (> 15%) in 5-30% of exercising asthmatics at
0.2-0.3 ppm and 20-60% of exercising asthmatics at 0.4-1.0 ppm with 5-10 min exposures.
Epidemiology results are inconsistent for the association between short-term avg S02 and lung
function in children and adults.
AIRWAYS INFLAMMATION
Conclusions from 1996 NAAQS Review: No conclusions in the previous review.
Conclusions from Current Review: A limited number of health studies have evaluated the effect of SO2
on airway inflammation. One human clinical study observed an SC>2-induced increase in sputum
eosinophil counts in exercising asthmatics 2 h after a 10 min exposure to 0.75 ppm SO2. The results
of this study provide some evidence that SO2 may elicit an inflammatory response in the airways of
asthmatics which extends beyond the short time period typically associated with SO2 effects.
Animal toxicological studies indicate that repeated exposures to S02, at concentrations as low as 0.1
ppm in guinea pigs, may exacerbate inflammatory responses in allergic animals.
5-11
-------
AIRWAYS HYPERRESPONSIVENESS
Conclusions from 1996 NAAQS Review: No conclusions in the previous review.
Conclusions from Current Review: Animal toxicological evidence indicates that repeated exposures to
S02, at concentrations as low as 0.1 ppm in guinea pigs, can exacerbate AHR following allergic
sensitization. In a human clinical study, concurrent exposure (6 h) to 0.2 ppm SO2 and 0.4 ppm N02
has been observed to enhance AHR to an inhaled antigen among resting asthmatics. These findings
are consistent with the very limited epidemiologic evidence that observes an association between
exposure to SO2 and AHR in atopic individuals.
ED VISITS/HOSPITALIZATIONS
Conclusions from 1996 NAAQS Review: No conclusions in the previous review.
Conclusions from Current Review: Epidemiologic studies provide evidence of an association between
ambient SO2 concentrations and ED visits and hospitalizations for all respiratory causes, particularly
among children and older adults (age 65+ years), and for asthma. This finding is coherent and
plausible with the increases in bronchoconstriction, decreases in lung functions, increases in
respiratory symptoms, and potential increases in airway inflammation and hyperresponsiveness
demonstrated in other epidemiologic, human clinical, and animal toxicological studies. The SO2 effect
estimates ranged from a 5% decreased risk to a 20% excess risk per 10-ppb increase in 24-h avg
SO2, with the large majority of studies observing an increase in risk. These effects were observed in
studies with mean 24-h avg concentrations as low as 4 ppb, but two studies evaluating the
concentration-response function observed that a marked increase in SC>2-related effects was only
observed higher concentrations (above 90th percentile values).
Short-Term Exposure: CARDIOVASCULAR MORBIDITY
Inadequate to infer a causal relationship
CARDIOVASCULAR EFFECTS; ED VISITS/HOSPITALIZATIONS
Conclusions from 1996 NAAQS Review: No conclusions in the previous review.
Conclusions from Current Review: There was some positive evidence of an association between 24-h
avg SO2 exposure and heart rate variability in the epidemiologic studies, but the evidence from two
human clinical studies were weak and inconsistent. Some epidemiologic studies have observed
positive associations between ambient SO2 concentrations and ED visits or hospital admissions for
cardiovascular diseases, but results are not consistent across studies and the SO2 effect estimate was
generally not robust to copollutant adjustment.
Short-Term Exposure: MORTALITY
Suggestive of a causal relationship
NONACCIDENTAL AND CARDIOPULMONARY MORTALITY
Conclusions from 1996 NAAQS Review: Epidemiologic studies based on historical air pollution episodes
observed the clearest mortality associations when both BS and SO2 concentrations were at high levels
(24-h avg values exceeding 1,000 |jg/m3 [-400 ppb for SO2]). Later studies observed that an in-
creased risk of mortality was associated with exposure to BS and SO2 levels in the range of 0.19 to
0.38 ppm. Because of the high colinearity between BS and SO2 levels, it is difficult to readily separate
the effects of these pollutants on mortality.
Conclusions from Current Review: Recent epidemiologic studies have consistently reported positive
associations between mortality and SO2, often at mean 24-h avg levels < 10 ppb. The range of SO2
excess risk estimates for nonaccidental mortality is 0.4 to 2% per 10 ppb increase in 24-h avg SO2 in
several multicity studies and meta-analyses. SO2 effect estimates for respiratory mortality were
generally larger than the cardiovascular mortality estimates, suggesting a stronger association of S02
with respiratory mortality compared to cardiovascular mortality. The SO2 effect estimates were
generally reduced when copollutants were added in the model, indicating some extent of confounding
among these pollutants.
5-12
-------
Long-Term Exposure: RESPIRATORY MORBIDITY
Inadequate to infer a causal relationship
RESPIRATORY SYMPTOMS AND LUNG FUNCTION
Conclusions from 1996 NAAQS Review: The limited available epidemiologic data indicated associations
between respiratory illnesses and symptoms and persistent exposures to SO2 in areas with long-term
averages exceeding 0.04 ppm.
Conclusions from Current Review: Several epidemiologic studies that examined the effects of long-term
exposure to SO2 on asthma, bronchitis, and respiratory symptoms observed positive associations in
children. However, the variety of outcomes examined and the inconsistencies in the observed results
make it difficult to assess the direct impact of long-term exposure of SO2 on respiratory symptoms.
The epidemiologic and animal toxicological evidence is also inadequate to infer that long-term
exposure to SO2 has a detrimental effect on lung function.
Long-Term Exposure: OTHER MORBIDITY
Inadequate to infer a causal relationship
CARCINOGENIC EFFECTS
Conclusions from 1996 NAAQS Review: Epidemiologic evidence did not substantiate the hypothesized
links between SO2 or other SOx and cancer, though there was some animal toxicological evidence
that led to the conclusion that SO2 may be considered a suspect carcinogen/cocarcinogen.
Conclusions from Current Review: Animal toxicological studies indicate that SO2 at high concentrations
may cause DNA damage but fails to induce carcinogenesis, cocarcinogenesis, or tumor promotion.
Epidemiologic studies did not provide evidence that long-term exposure to SO2 is associated with an
increased incidence of or mortality from lung cancer.
PRENATAL AND NEONATAL OUTCOMES
Conclusions from 1996 NAAQS Review: No conclusions in the previous review.
Conclusions from Current Review: Epidemiologic studies on birth outcomes have observed positive
associations between SO2 exposure and low birth weight. However, the inconsistent results across tri-
mesters of pregnancy and the lack of evidence to evaluate confounding by copollutants limit the
interpretation of these studies.
Long-Term Exposure: MORTALITY
Inadequate to infer a causal relationship
NONACCIDENTAL AND CARDIOPULMONARY MORTALITY
Conclusions from 1996 NAAQS Review: The available studies on the effects of long-term exposure to
SO2 on mortality were all ecological cross-sectional studies which did not take into consideration
potential confounders. In addition, it was concluded that effects from PM and S02 could not be
distinguished in these studies.
Conclusions from Current Review: Two major U.S. epidemiologic studies observed associations
between long-term exposure to SO2 and mortality, but several other U.S. and European cohort studies
did not observe an association. The relative risks ranged from 0.97 to 1.07 per 5-ppb increase in the
long-term avgS02. Evaluation of these studies is further limited by the inability to distinguish potential
confounding by copollutants and uncertainties regarding the geographic scale of analysis.
5-13
-------
Table 5-4. Effects of short-term exposure to SO2 on respiratory symptoms among children.
Study
n . .. Mean Concentration
P°Pulatlon (ppb)
SO2 Range
(PPb)
SO2 Upper
Percentile (ppb)
Standardized Odds
Ratio (95% Cl)a
UNITED STATES
Schildcrout et al. (2006)
Multicity, North America
Seattle, WA; Baltimore, MD; St.
Louis, MO (Nov 1993—Aug
1995); Denver, CO; San Diego,
CA (Nov 1993—Jul 1995);
Toronto, ON (Dec 1993—Jul
1995); Boston, MA (Jan 1994—
Sep 1995)
No SO2 data available in
Albuquerque, NM
Asthmatic children
(n = 990)
24-h avg: 2.2-7.4
(range of city-specific
medians)
NR
75th: 3.1-10.7
90th: 4.4-14.2
(range of city-
specific estimates)
Children grades 2-5 24-h avg: 4.1 (median) 6.8 (IQR)
(n = 1,844)
Asthma symptoms (3-day sum
lag):
S02 alone: 1.04(1.00,1.08)
SO2 + PM10:1.04 (0.99,1.08)
SO2 + NO2:1.04(1.00,1.09)
SO2 + CO: 1.05 (1.00,1.09)
Schwartz et al. (1994)
Multicity, U.S.
Watertown, MA (Apr-Aug 1985);
Kingston-Harriman, TN; St.
Louis, MO (Apr-Aug 1986);
Steubenville, OH; Portage, Wl
(Apr-Aug 1987); Topeka, KS
(Apr-Aug 1988)
75th: 8.2 Cough incidence (4-day avg lag):
90th: 17.9 SO2 alone: 1.15 (1.02,1.31)
Max: 81.9 PM10 adjusted: 1.08 (0.93,1.25)
N02 adjusted: 1.09 (0.94,1.30)
03 adjusted: 1.15(1.01,1.31)
Neas et al. (1995)
Uniontown, PA
Summer 1990
Children grades 4-5 12-h avg: 10.2
(n = 83) Overnight: 5.9
Daytime: 14.5
11.1 (IQR) Max: 44.9 Evening cough (lag 12-hV
1.19 (1.00,1.42)
Delfino et al. (2003a)
Hispanic children
1-h max: 7.0
2-26
90th: 11.0
Asthma symptom score >1:
Los Angeles, CA
w/asthma, age 10-
16 yrs
1.31 (1.10,1.55), lagO
Nov 1999-Jan 2000
Asthma symptom score >2:
(n = 22)
1.37 (0.87,2.18), lagO
Mortimer etal. (2002)
Asthmatic children,
3-h avg: 22 (shown in
0-75 (shown in
NR
Asthma symptoms (lag 1-21:
Multicity, U.S.
age 4-9 yrs
figure)
graph)
SO2 alone (8 cities): 1.19 (1.06,
Bronx, NY; East Harlem, NY;
(n = 846)
1.35)
Baltimore, MD; Washington, DC;
O3 adjusted (8 cities): 1.18 (1.05,
Detroit, Ml; Cleveland, OH;
1.33)
Chicago, IL; St. Louis, MO
O3 & NO2 adjusted (7 cities):
Jun-Aug 1993
1.19 (1.04,1.37)
O3, NO2 & PM10 adjusted
(3 cities): 1.23 (0.94,1.62)
Timonen and Pekkanen (1997)
Kuopio (urban and suburban)
Finland
Winter 1994
Children with
asthma or cough
symptoms, age
7-12 yrs
(n = 169)
24-h avg: 2.3
1.7 (IQR)
75th: 2.7
Max: 12.3
Upper respiratory symptoms
fURSV
2.70(1.19,6.15), lag 1
3.15 (1.22,8.27), lag 1-4
5-14
-------
Study
Population
Mean Concentration
(PPb)
SO2 Range
(PPb)
SO2 Upper
Percentile (ppb)
Standardized Odds
Ratio (95% Cl)a
Wardet al. (2002a, 2002b)
Birmingham and Sandwell, UK
Jan-Mar 1997
May-Jul 1997
Children, age at en-
rollment 9 yrs
(n = 162)
24-h avg:
Winter: 5.4 (median)
Summer: 4.7 (median)
Winter: 2.0-18.0
Summer: 2.0-10.0
NR
Cough (lag 01:
Winter: 0.81 (0.59,1.13)
Summer: 1.21 (1.05, 1.42)
Shortness of breath (lag 01:
Winter: 1.05 (0.83,1.36)
Summer: 0.95 (0.71, 1.27)
Wheeze (lag 01:
Winter: 0.90 (0.67,1.18)
Summer: 1.13 (0.81, 1.54)
Segala et al. (1998)
Paris, France
Nov 1992-May 1993
Children with
physician-diagnosed
asthma, age
7-15 yrs
(n = 84)
24-h avg: 8.3 (SD 5.2)
1.7-32.2
NR
Asthma:
1.73 (1.15,2.60), lagO
1.60 (1.01,2.53), lag 1
Wheeze:
1.22 (0.95,1.58), lagO
1.13 (0.68,1.88), lag 1
Boezen et al. (1999)
Amsterdam and Meppel
(urban and rural),
the Netherlands
3 winters, 1992-1995
Children with and
without BHR and
high serum concen-
trations of total IgE,
age 7-11 yrs,
(n = 632)
24-h avg: 1.7-8.7
Medians: 1.4-8.3
(range of city-specific
estimates)
1.9-23.6
NR
Among children with BHR and
relatively high serum total IgE:
Lower resciratorv svmDtoms
(LRS):
1.27 (1.09,1.49), lagO
1.25 (1.06,1.48), lag 1
1.69 (1.26, 2.28), 5-day avg
Van der Zee et al. (1999)
Urban and nonurban areas, The
Netherlands
3 winters, 1992-1995
Children with and
without chronic
respiratory
symptoms, age 7-11
yrs,
(n = 633)
24-h avg: 1.4-8.8
(range of city-specific
medians)
NR
Max: 6.5-58.5
(range of city-
specific estimates)
LRS. urban:
SO2 alone:
1.22 (1.01,1.46),
1.14(0.95,1.38),
1.34 (0.98,1.82),
PM10 adjusted:
1.18 (0.96,1.45),
1.03 (0.83,1.27),
1.08 (0.72,1.63),
LRS. nonurban:
0.94 (0.79,1.12),
0.94 (0.78,1.13),
1.10 (0.75,1.63),
Cough, urban:
0.93 (0.84,1.03),
1.08 (0.98,1.19),
1.08 (0.89,1.30) 5-day avg
Cough, nonurban:
1.05 (0.96,1.15), lagO
0.98 (0.90,1.08), lag 1
1.04 (0.83,1.30), 5-day avg
lag 0
lag 1
5-day avg
lag 0
lag 1
5-day avg
lag 0
lag 1
5-day avg
lag 0
lag 1
Hoek and Brunekreef (1994)
The Netherlands
3 winters, 1987-1990
Children age
7-11 yrs from 1
industrial and 3
nonindustrial
communities
(n = 1,078)
24-h avg: 5.7 (SD 5.5)
0.2-36.0
NR
Cough:
1.05 (0.88,1.26), lagO
LRS:
1.33 (1.06,1.69), lag 1
URS:
1.14(0.98,1.32), lagO
Pikhart et al. (2000)
Czech Republic
1993-1994
Children age
7-11 yrs from 1
industrial and 3
nonindustrial
communities
(n = 1,078)
24-h avg: 28.2
NR
75th: 36.5
Wheeze:
1.23 (0.95,1.62), lag NR
5-15
-------
Table 5-5. Effects of short-term SO2 exposure on emergency department visits and hospital
admissions for respiratory outcomes.
Study
Population
Mean Concentration
(PPb)
SO2 Range (ppb)
SO2 Upper
Percentile (ppb)
Standardized
Excess Risk
(95% Cl)a
EMERGENCY DEPARTMENT VISITS - ALL RESPIRATORY
UNITED STATES
Wilson et al. (2005)
Portland, ME
Jan 1998-Dec 2000
Manchester, NH
Jan 1996-Dec 2000
= 54,000 ED visits
in Portland and =
30,000 ED visits in
Manchester for all
respiratory causes
1-h max:
Portland: 11.1 (SD9.1)
Manchester: 16.5 (SD
14.7)
NR
NR
Portland (lag 0):
All ages: /% (3,12).
0-14: -4% (-11, 4). 15-64: 9% (5,14)
65+: 16% (7, 26)
Manchester (lag 0):
All ages: 1% (-3, 5)
0-14: 0% (8, 8). 15-64:1% (-3, 5)
65+: 7% (-5, 21)
Tolbert et al. (2007)
Atlanta, GA
Jan 1993-Dec 2004
> 1,000,000 ED
visits for all
respiratory causes
from 41 hospitals
1-h max: 14.9
1.0-149.0
75th: 20.0
90th: 35.0
0.8% (-0.7, 2.3), lag 0-2
Peel et al. (2005)
Atlanta, GA
Jan 1993-Aug 2000
= 480,000 ED visits
for all respiratory
causes from 31
hospitals
1-h max: 16.5 (SD 17.1)
NR
90th: 39.0
1.6% (-0.6, 3.8), lag 0-2
EUROPE
Atkinson et al. (1999a)
London, UK
Jan1992-Dec1994
98,685 ED visits
from 12 hospitals
24-h avg: 8.0 (SD 2.9)
2.8-30.9
90th: 11.7
All ages: 4% (1, 7), lag 1
0-14: 9% (4,14), lag 2
15-64: 4% (0,8), lag 2
65+: -3% (-8, 3), lag 1
EMERGENCY DEPARTMENT VISITS - ASTHMA
UNITED STAT
Ito et al. (2007)
New York, NY
Jan 1999-Dec 2002
Asthma ED visits, all 24-havg:
ages from 11
hospitals
All year: 7.8 (SD 4.6)
Warm: 5.4 (SD 2.2)
Cold: 10.2 (SD 5.1)
IQR:
75th:
All year: 5
All year: 10
Warm: 3
Warm: 7
Cold: 7
Cold: 13
95th:
All year: 17
Warm: 10
Cold: 19
All year (lag 0-1): 8.9% (4.9,13.0)
Warm (lag 0-1):
S02 alone: 35.9% (22.2,51.2)
PM2.5 adjusted: 29.6% (14.3, 46.8)
O3 adjusted: 26.8% (13.7, 41.5)
NO2 adjusted:-1.6% (-16.7,16.1)
CO adjusted: 31.1% (16.7, 47.2)
Cold (lag 0-1): 8.5% (4.8,12.4)
NY Department of
Health (2006)
Bronx & Manhattan
Jan 1999-Dec 2000
= 31,000 asthma
ED visits from 8
hospitals in the
Bronx and <* 5,000
asthma ED visits
from 14 hospitals in
Manhattan
24-havg: 11 (SD7.2)
NR
NR
Bronx (lag 0-4):
SO2 alone: 10% (5,15)
PM2.5 adjusted: 10% (4,16)
NO2 adjusted: 10% (4,15)
Manhattan (lag 0-4):
S02 alone:-1% (-11,11)
PM2.5 adjusted:-3% (-14,10)
N02 adjusted: 1% (-12,14)
Jaffe et al. (2003)
Cincinnati, OH
Cleveland, OH
Columbus, OH
Jul 1991-Jun 1996,
summer months only
(Jun-Aug)
4,416 ED visits for
asthma, age 5-34
yrs
24-h avg:
Cincinnati
Cleveland
Columbus
14 (SD 10)
15 (SD 10)
4 (SD 3)
Cincinnati: 1-50
Cleveland: 1-64
Columbus: 0-22
NR
Cincinnati: 17% (5, 31), lag 2
Cleveland: 3% (-4, 11), lag 2
Columbus: 13% (-14, 49), lag 3
All cities: 6% (1,11), best lags
5-16
-------
Study
Population
Mean Concentration
(PPb)
SO2 Range (ppb)
SO2 Upper
Percentile (ppb)
Standardized
Excess Risk
(95% Cl)a
Wilson et al. (2005)
Portland, ME
Jan 1998-Dec 2000
Manchester, NH
Jan 1996-Dec 2000
= 8,100 asthma ED
visits in Portland
and = 4,700 asthma
ED visits in
Manchester
1-h max:
Portland: 11.1 (SD9.1)
Manchester: 16.5 (SD
14.7)
NR
NR
Portland (lag 0):
All ages: 11% (2, 20). 0-14: 5% (-12, 25)
15-64:12% (1,23). 65+: 12% (-15, 47)
Manchester (lag 0):
All ages: 6% (-4, 17). 0-14: 20% (-3, 49)
15-64: 3% (-8,16). 65+: 12% (-29, 75)
Peel et al. (2005)
Atlanta, GA
Jan 1993-Aug 2000
= 110,000 asthma
ED visits, all ages
from 31 hospitals
1-h max: 16.5 (SD 17.1)
NR
90th: 39.0
0.2% (-3.2, 3.4), lag 0-2
EUROPE
Atkinson et al. (1999a)
London, UK
Jan1992-Dec1994
28,435 asthma ED
visits from 12
hospitals
24-h avg: 8.0 (SD 2.9)
2.8-30.9
50th: 7.4
90th: 11.6
All ages: 7% (2,13), lag 1
0-14:15% (7, 23), lag 1
15-64: 6% (-1,14), lag 1
Hajat et al. (1999)
London, UK
Jan1992-Dec1994
General practitioner
visits for asthma
24-h avg:
All year: 8.0 (SD 2.9)
Warm: 7.7 (SD 2.4)
Cool: 8.3 (SD 3.4)
NR
90th:
All year: 11.6
Warm: 10.7
Cool: 12.4
All ages: 6.6% (1.3,11.9), lag 0-2
0-14: 6.6% (-1.0,14.7), lag 0-3
15-64: 5.2% (-1.5,12.3), lag 0-3
65+: 7.2% (-4.3, 20.1), lag 0-1
Boutin-Forzano et al.
(2004)
549 ED visits for
asthma
24-h avg: 8.5
0.0-35.3
NR
3-49 yrs: 0.6% (-1.4, 2.7), lag 0
Marseille, France
Apr 1997-Mar 1998
Galan et al. (2003)
Madrid, Spain
4,827 ED visits for
asthma
24-h avg: 8.9 (SD 5.8)
1.9-45.6
75th: 11.8
90th: 16.5
All ages: 4.9% (-4.2,15.0), lag 0
Jan1995-Dec1998
Tenias et al. (1998)
Valencia, Spain
Jan1993-Dec1995
734 ED visits for
asthma
24-h avg: 10.0
Cold: 11.9
Warm: 8.2
1-h max: 21.2
Cold: 24.3
Warm: 18.1
NR
24-h avg:
75th: 12.9
95th: 16.0
1-h max:
75th: 27.1
95th: 35.8
> 14 yrs:
13.9% (-7.0, 39.4), lag 0
Castellsague et al.
(1995)
Barcelona, Spain
Jan1986-Dec1989
ED visits for asthma
from 4 hospitals
24-h avg:
Summer: 15.3
Winter: 19.5
NR
Summer:
75th: 20.3
95th: 30.8
Winter:
75th: 25.2
95th: 35.3
15-64 yrs:
Summer: 5.5% (-2.1,13.8), lag 2
Winter: 2.1% (-4.2, 9.0), lag 1
HOSPITAL ADMISSIONS - ALL RESPIRATORY
UNITED STATES
Schwartz (1995)
New Haven, CT
Tacoma, WA
Jan1988-Dec1990
= 8,800 admissions
in New Have and =
4,000 admissions in
Tacoma for all
respiratory causes,
ages 65+ yrs
24-h avg:
New Haven: 29.8
Tacoma: 16.8
IQR:
New Haven: 24.8
Tacoma: 11.5
New Haven:
75th: 38.2
90th: 60.7
Tacoma:
75th: 21.4
90th: 28.2
New Haven:
S02 alone: 2% (1,3), lag 2
PM10 adjusted: 2% (1, 3), lag 2
Tacoma:
SO2 alone: 3% (1, 6), lag 0
PM10 adjusted: -1% (-4, 3), lag 0
Schwartz et al. (1996)
Cleveland, OH
1988-1990
Hospital
admissions, ages
65+ yrs
24-h avg: 35.0
IQR: 25
75th: 45.0
90th: 61.0
0.8% (-0.3,1.5), lag 0-1
5-17
-------
Study
Population
Mean Concentration
(PPb)
SO2 Range (ppb)
SO2 Upper
Percentile (ppb)
Standardized
Excess Risk
(95% Cl)a
CANADA
Fung et al. (2006)
Vancouver, BC
Jun 1995-Mar 1999
= 41,000 respiratory 24-h avg: 3.46 (SD 1.82) 0.0-12.5
admissions forages
65+ yrs
NR
13% (1, 26), 7-day avg
Cakmak et al. (2007a)
Multicity, Canada
Calgary, Edmonton,
Halifax, London,
Ottawa, Saint John,
Toronto, Vancouver,
Windsor, Winnipeg
> 200,000 hospital
admissions for all
respiratory causes
24-h avg: 4.6 (all cities)
2.8-10.2 (range of city-
specific means)
0-75
Max: 14-75
(range of city-
specific estimates)
SO2 alone:
2.4% (1.1, 4.0)
O3, NO2 & CO adjusted:
1.1% (0.2,2.0)
SO2 + O3 + NO2:
14.9% (7.8,22.3)
(best lags selected for each city)
Apr 1993-Mar 2000
Yang et al. (2003b)
Vancouver, BC
Jan1986-Dec1998
Respiratory hospital
admissions among
young children
(< 3 yrs) and elderly
(65+ yrs)
24-h avg: 4.84 (SD 2.84)
IQR: 3.5
75th: 6.25
Max: 24.00
< 3 yrs (lag 2):
SO2 alone: 3% (-6,15)
O3 adjusted: 3% (-8, 12)
65+yrs (lag 0):
S02 alone: 6% (0,12)
O3 adjusted: 6% (0, 12)
Burnett et al. (1997b)
Toronto, ON
Jun 1992-Sep 1994,
summer months only
(May-Sep)
All respiratory
hospital admissions
1-h max: 7.9
0-26
75th: 11
95th: 18
S02 alone: 38.4% (13.1, 69.2), lag 0-3
PMisadj: 24.3% (-0.5, 55.2), lag 0-3
PM10-2.5 adj: 25.5% (0.8, 56.4), lag 0-3
PM10 adj: 23.1% (-2.9, 56.0), lag 0-3
Luginaah et al. (2005)
Windsor, ON
Apr 1995-Dec 2000
All respiratory
admissions from
4 hospitals
1-h max: 27.5 (SD 16.5)
0-129
NR
All ages (lag 1): Female: 8.7% (-2.7, 21.4)
Male:-9.5% (-19.7,1.9)
0-14 (lag 1): Female: 24.4% (2.3, 51.4)
Male: -9.7% (-24.4, 7.8)
15-64 (lag 1): Female: 6.5% (-14.0, 32.3)
Male: -5.9% (-29.5, 25.4)
65+ (lag 1): Female: 6.3% (-9.9, 25.4)
Male:-11.9% (-26.9, 6.1)
EUROPE
Oftedal et al. (2003)
Drammen, Norway
All respiratory
hospital admissions
24-havg: 1.1 (SD 0.8)
NR
NR
All ages:
71.8% (15.5,152.7), lag NR
Jan 1994-Dec 2000
Llorca et al. (2005)
Torrelavega, Spain
Hospital admissions
from one hospital
24-h avg: 5.0 (SD 6.3)
NR
NR
All ages:
1.0% (-2.8, 4.7), lag NR
Jan1992-Dec1995
Atkinson et al. (1999b)
London, UK
Jan1992-Dec1994
165,032 respiratory
hospital admissions
24-h avg: 8.0 (SD 2.9)
2.8-30.9
90th: 11.7
All ages: 3% (0, 6), lag 1
0-14: 8% (4,12), lag 1
15-64: 3% (-1,7), lag 3
65+: 3% (-0, 7), lag 3
Schouten et al. (1996) All respiratory 24-havg:
hospital admissions Amsterdam: 10.5
Rotterdam: 15.0
NR
NR
Multicity, The
Netherlands
Amsterdam, Rotterdam
Apr 1977-Sep 1989
1-h max:
Amsterdam: 24.4
Rotterdam: 37.2
Amsterdam (lag 0-3):
15-64 yrs: -2.3% (-5.5, 0.9)
65+: 0.2% (-2.8, 3.3)
Rotterdam (lag 0-2):
15-64: -2.9% (-6.2,0.5)
5-18
-------
Study
Population
Mean Concentration
(PPb)
SO2 Range (ppb)
SO2 Upper
Percentile (ppb)
Standardized
Excess Risk
(95% Cl)a
Spix et al. (1998)
Multicity, Europe
London, UK;
Amsterdam &
Rotterdam, the
Netherlands; Paris,
France; Milan, Italy
Jan 1977-Dec 1991
All respiratory
hospital admissions
24-h avg:
London: 10.9
Amsterdam: 7.9
Rotterdam: 9.4
Paris: 8.6
Milan: 24.8
NR
NR
15-64 yrs: 0.5% (-0.4,1.3), lag NR
65+: 1.1% (0.3, 2.4), lag NR
Dab et al. (1996)
Paris, France
Jan 1987-Sep 1992
Respiratory hospital
admissions from 27
hospitals
All year:
24-h avg: 11.2
1-h max: 22.5
Warm season
24-h avg: 7.6
1-h max: 16.1
Cold season
24-h avg: 15.1
1-h max: 29.4
NR
99th:
All year:
24-h avg: 50.0
1-h max: 87.5
Warm:
24-h avg: 18.5
1-h max: 50.3
Cold:
24-h avg: 56.0
1-h max: 100.9
All ages:
1.1% (0.1, 2.1), lag 0-2
Ponce de Leon et al.
(1996)
London, UK
1987-1988; 1991-1992
All respiratory 24-h avg: 12.1 (SD 4.7) NR
hospital admissions
75th: 14.7
90th: 17.7
95th: 20.3
All ages: 0.8% (-0.7, 2.4)
0-14: 0.9% (-1.5, 3.3), lag 1
15-64: 2.0% (-0.5, 4.7), lag 1
65+: 2.0% (-0.3, 4.4), lag 2
Walters et al. (1994)
Birmingham, UK
Jan1988-Dec1990
All respiratory
hospital admissions
24-h avg: All year: 14.7 NR
Spring: 16.1
Summer: 14.2
Autumn: 15.4
Winter: 12.9
Max: 47.5 All ages:
Summer: 1.5% (0.3, 2.7), lag 0
Winter: 4.5% (2.3, 6.5), lag 0
Hagen et al. (2000)
Drammen, Norway
Jan1994-Dec1997
All respiratory
hospital admissions
24-h avg:
Winter: 21. Spring: 18
Summer:15. Autumn:19
Number of monitors: 1
Winter: 11-33
Spring: 13-29
Summer: 5-24
Autumn: 16-23
NR
All ages:
92.8% (16.8, 218.8), lag NR
Vigotti et al. (1996)
Milan, Italy
Jan1980-Dec1989
All respiratory 24-h avg:
hospital admissions A|| yegr; 44 g
Winter: 94.8
Summer: 11.6
1.1-315.7
75th: All year: 62.0
Winter: 125.0
Summer: 15.0
95th: All year: 143.5
Winter: 201.0
Summer: 23.9
15-64 yrs: 1.3% (0.0, 2.5), lag 0
65+: 1.0% (0.0, 2.3), lag 0
AUSTRALIA
Barnett et al. (2005)
Multicity, Australia/New
Zealand
Auckland, Brisbane,
Canberra, Christchurch,
Melbourne, Perth,
Sydney
Jan 1998-Dec 2001
All respiratory
hospital admissions
24-h avg:
Auckland: 4.3
Brisbane: 1.8
Christchurch: 2.8
Sydney: 0.9
NA in Canberra,
Melbourne, and Perth
1-h max:
Brisbane: 7.6
Christchurch: 10.1
Sydney: 3.7
NA in Auckland,
Canberra, Melbourne,
and Perth
24-h avg:
Auckland: 0-24.3
Brisbane: 0-8.2
Christchurch: 0-11.9
Sydney: 0-3.9
1-h max:
Brisbane: 0-46.5
Christchurch: 0.1-
42.1
Sydney: 0.1-20.2
NR
1-4 yrs: 21.8% (4.5, 41.5), lag 0-1
5-14:15.8% (-34.2,104.0), lag 0-1
5-19
-------
Study
Population
Mean Concentration
(PPb)
SO2 Range (ppb)
SO2 Upper
Percentile (ppb)
Standardized
Excess Risk
(95% Cl)a
Petroeschevsky (2001)
Brisbane, Australia
Jan1987-Dec1994
33,710 respiratory
hospital admissions
24-h avg: 4.1
1-h max: 9.2
NR
NR
All ages: 8.0% (3.0, 13.1), lag 0
0-4: 22.4% (8.7, 37.7), lag 0-4
5-14: 21.1% (-5.5, 55.1), lag 0-4
15-64: 3.3% (-10.5,11.8), lag 1
65+: 12.1% (1.9, 23.4), lag 0
LATIN AMERICA
Gouveia and Fletcher
(2000)
Sao Paulo, Brazil
All respiratory
hospital admissions
24-h avg: 6.9 (SD 3.4)
1.2-22.9
75th: 8.3
95th: 13.5
<5 yrs: 3.7% (-1.7, 9.4), lag 1
Nov 1992-Sep 1994
ASIA
Wong et al. (1999b)
Hong Kong
Jan1994-Dec1995
All respiratory
admissions from 12
hospitals
24-h avg: 6.4
1.0-25.7
75th: 9.4
All ages: 3.5% (1.1, 5.7), lag 0
0-4:1.3% (-2.4, 4.9), lag 0
5-64: 2.1% (-1.1, 5.7), lag 0
65+: 6.2% (3.2, 9.9), lag 0
HOSPITAL ADMISSIONS - ASTHMA
UNITED STATES
Lin et al. (2004d)
New York, NY
Jun 1991-Dec 1993
2,629 cases; 2,236
controls, aged 0-14
yrs
24-h avg:
Cases: 16.78
Controls: 15.57
2 88-66 35
NR
19% (11, 29), lag NR
Sheppard et al. (1999;
reanalysis 2003)
Seattle, WA
7,837 asthma
hospital admissions
for patients <65 yrs
24-h avg: 8.0
NR
75th: 10.0
90th: 13.0
2.1% (-4.0, 6.2), lag 0
Jan1987-Dec1994
CANADA
Toronto, ON
Jan 1981-Dec 1993
7,319 asthma
hospital admissions
among 6-12 yr olds
24-h avg: 5.36 (SD 5.90)
Girls (7-day avg):
S02 alone: 29.8% (5.8, 60.1)
PM2.5 & PM10-2.5 adj: 42.3% (11.6, 80.2)
Boys (7-day avg):
S02 alone:-9.8% (-23.4, 5.8)
PM2.5 & PM10-2.5 adj: -12.6% (-27.3, 5.8)
EUROPE
Atkinson et al. (1999b)
London, UK
Jan1992-Dec1994
= 42,000 hospital
admissions for
asthma
24-h avg: 8.0 (SD 2.9)
2.8-30.9
90th: 11.7
All ages: 5% (1,10), lag 1
0-14:10% (4,16), lag 1
15-64: 7% (0,14), lag 3
65+: 9% (-2, 23), lag 2y
Schouten et al. (1996)
Multicity, The
Netherlands
All hospital
admissions for
asthma
24-h avg:
Amsterdam: 10.5
Rotterdam: 15.0
NR
NR
Amsterdam:
All ages:
-6.0% (-10.7,-1.1), lag 0-3
Amsterdam, Rotterdam
Apr 1977-Sep 1989
1-h max:
Amsterdam: 24.4
Rotterdam: 37.2
Sunyeretal. (1997)
Multicity, Europe:
Barcelona, Spain;
Helsinki, Finland; Paris,
France; London, UK
All hospital
admissions for
asthma
24-h avg:
Barcelona: 15.4
Helsinki: 6.0
London: 11.6
Barcelona: 0.8-60.2
Helsinki: 1.1-35.7
London: 3.4-37.6
Paris: 0.4-82.3
NR
0-14 yrs:
3.2% (-0.2, 6.8), best cumulative lag
15-64:
0.2% (-2.2, 2.6), best cumulative lag
Jan1986-Dec1992
Paris: 8.6
5-20
-------
Study
Population
Mean Concentration
(PPb)
SO2 Range (ppb)
SO2 Upper
Percentile (ppb)
Standardized
Excess Risk
(95% Cl)a
Dab et al. (1996)
Paris, France
Jan 1987-Sep 1992
Hospital admissions
for asthma from 27
hospitals
All year: NR
99th:
24-h avg: 11.2
All year:
1-h max: 22.5
24-h avg: 50.0
Warm season
1-h max: 87.5
24-h avg: 7.6
Warm:
1-h max: 16.1
24-h avg: 18.5
Cold season:
1-h max: 50.3
24-h avg: 15.1
Cold:
1-h max: 29.4
24-h avg: 56.0
1-h max: 100.9
All ages:
1.8% (0.1,3.6), lag 2
Anderson et al. (1998) All hospital 24-h avg: 12.0 (SD 4.4) 3.4-37.6 75th: 14.3 All ages: 7.4% (3.2,11.7), lag 0-3
, „ , „ , I,/ admissions for an«,-i7Q 0-14:5.4% (0.8,10.3), lag 0-3
London, UK 90th. 17.3 .. .. , .) ' n M
asthma 15-64: -1.8% (-6.9, 3.4), lag 0-2
Apr 1987-Feb 1992 95th: 19.5 65+: 8.2% (-1.9.19.3), lag 0-3
Walters et al. (1994)
Birmingham, UK
Jan1988-Dec1990
All hospital
admissions for
asthma
24-h avg:
All year: 14.7
Spring: 16.1
Summer: 14.2
Autumn: 15.4
Winter: 12.9
NR
Max: 47.5 All ages:
Summer: 0.4% (-2.8, 9.2), lag 0
Winter: 0.7% (-2.2,1.6), lag 0
AUSTRALIA
Barnett et al. (2005)
Multicity, Australia/New
Zealand
Auckland, Brisbane,
Canberra, Christchurch,
Melbourne, Perth,
Sydney
Jan 1998-Dec 2001
All hospital
admissions for
asthma
24-h avg:
Auckland: 4.3
Brisbane: 1.8
Christchurch: 2.8
Sydney: 0.9
NA in Canberra,
Melbourne, and Perth
1-h max:
Brisbane: 7.6
Christchurch: 10.1
Sydney: 3.7
NA in Auckland,
Canberra, Melbourne,
and Perth
24-h avg:
Auckland: 0-24.3
Brisbane: 0-8.2
Christchurch: 0-11.9
Sydney: 0-3.9
1-h max:
Brisbane: 0-46.5
Christchurch: 0.1-
42.1
Sydney: 0.1-20.2
NR
1-4yrs: 28.1% (-27.8,125.5), lag 0-1
5-14: 27.2% (-34.7,147.3), lag 0-1
Petroeschevsky et al.
(2001)
Brisbane, Australia
Jan1987-Dec1994
13,246 hospital
admissions for
asthma
24-h avg: 4.1
1-h max: 9.2
NR
NR
All ages: -5.9% (-12.4,1.1), lag 2
0-14: 8.0% (-2.9, 20.1), lag 0
15-64:-21.6% (-34.4,-6.2), lag 0
LATIN AMERICA
Gouveia and Fletcher
(2000)
Sao Paulo, Brazil
All hospital
admissions for
asthma
24-h avg: 6.9 (SD 3.4)
1.2-22.9
75th: 8.3
95th: 13.5
<5 yrs: 10.4% (-1.9, 24.2), lag 2
ASIA
Wong et al. (1999b)
Hong Kong
Jan1994-Dec1995
Hospital admissions
for asthma from 12
hospitals
24-h avg: 6.4
1.0-25.7
75th: 9.4
All ages: 4.6% (-0.5, 9.9), lag 0
Lee et al. (2006b) 26,663 hospital 24-h avg: 6.6 (SD 4.0) NR 75th: 8.2 <18 yrs:-3.7% (-6.7,-0.6), lag 0
Hong Kong admissions for
a a asthma
Jan 1997-Dec 2002
5-21
-------
Study
Population
Mean Concentration
(PPb)
SO2 Range (ppb)
SO2 Upper
Percentile (ppb)
Standardized
Excess Risk
(95% Cl)a
Ko et al. (2007b)
Hong Kong
Jan 2000-Dec 2005
69,716 hospital
admissions for
asthma from 15
hospitals
24-h avg:
All year: 7.2 (SD 4.9)
<20°C: 6.9 (SD 3.8)
>20°C: 7.3 (SD 5.4)
All year: 1.1-45.6
<20°C: 1.3-28.8
>20°C: 1.1-45.6
75th:
All year: 8.6
<20°C: 21.3
>20°C: 22.5
All ages: 1.1% (-1.6, 3.7), lag 0-3
0-14: Not statistically significant
(estimates NR)
15-65: 4.8% (0.3, 9.4), lag 0-3
65+: Not statistically significant (estimates
NR)
Tsai et al. (2006)
Kaohsiung, Taiwan
Jan 1996-Dec 2003
17,682 hospital
admissions for
asthma from 63
hospitals
24-h avg: 9.49
0.92-31.33
75th: 12.16
SO2 alone (lag 0-2):
>25°C: 3.1% (-7.5,14.8)
<25°C: 34.5% (12.9, 60.3)
PM10 adjusted (lag 0-2):
>25°C:-1.2% (-11.5,10.2)
<25°C: 4.7% (-13.2, 26.5)
NO2 adjusted (lag 0-2):
>25°C:-5.6% (-16.2.6.7)
<25°C: -41.2% (-53.0,-26.8)
CO adjusted (lag 0-2):
>25°C:-15.0% (-24.9,-3.8)
<25°C: 6.3% (4.7,8.1)
O3 adjusted (lag 0-2):
>25°C: 9.7% (-1.7, 22.2)
<25°C: 36.0% (14.2, 62.2)
5-22
-------
ANNEXES
-------
Annex A. Literature Selection
This Annex includes detailed information on the methods used to identify and select studies, and on
frameworks for evaluating scientific evidence relative to causality determination. While the overarching
framework is outlined in the introduction to Chapter 1, this Annex provides supporting information for
that framework, including excerpts from decision frameworks or criteria developed by other
organizations.
A.1. Literature Search and Retrieval
Literature searches are conducted continuously, to identify studies published since the last review.
The current review includes studies published subsequent to the 1982 AQCD for SOx (EPA, 1982).
Search strategies are iteratively modified in an effort to optimize the identification of pertinent
publications. Additional publications are identified for inclusion in several ways: review of pre-
publication tables of contents for journals in which relevant papers may be published; independent
identification of relevant literature by expert authors; and identification by the public and CASAC during
the external review process. Generally, only information that has undergone scientific peer review and has
been published, or accepted for publication, in the open literature is considered. Studies identified are
further evaluated by EPA staff and outside experts to determine if they merit inclusion. Criteria used for
study selection are summarized below.
A.2. General Criteria for Study Selection
In assessing the scientific quality and relevance of epidemiologic and animal or human
toxicological studies, the following considerations have been taken into account.
¦ Were the study populations adequately selected and are they sufficiently well defined to
allow for meaningful comparisons between study groups?
¦ Are the statistical analyses appropriate, properly performed, and properly interpreted?
¦ Are likely covariates (i.e., potential confounders or effect modifiers) adequately controlled
or taken into account in the study design and statistical analysis?
¦ Are the reported findings internally consistent, biologically plausible, and coherent with
other known facts?
¦ To what extent are the aerometric data, exposure, or dose metrics of adequate quality and
sufficiently representative to serve as indicators of exposure to ambient S02?
Consideration of these issues informs our judgments on the relative quality of individual studies and
allows us to focus the assessment on the most pertinent studies. The following two sections describe
criteria for selecting specific types of studies.
A.2.1. Criteria for Selecting Epidemiologic Studies
In selecting epidemiologic studies for this assessment, EPA considered whether a given study
contains information on (1) associations with measured SOx concentrations using short- or long-term
A-1
-------
exposures at or near ambient levels of SOx, (2) health effects of specific SOx species or indicators related
to SOx sources (e.g., combustion-related particles), (3) health endpoints and populations not previously
extensively researched, (4) multiple pollutant analyses and other approaches to address issues related to
potential confounding and modification of effects, and/or (5) important methodological issues (e.g., lag of
effects, model specifications, thresholds, mortality displacement) related to interpretation of the health
evidence. Among the epidemiologic studies, particular emphasis has been placed on those most relevant
to reviews of the NAAQS. Specifically, studies conducted in the U.S. or Canada may be discussed in
more detail than those from other geographic regions. Particular emphasis has been placed on: (1) recent
multicity studies that employed standardized methodological analyses for evaluating effects of SOx and
that provide overall estimates for effects based on combined analyses of information pooled across
multiple cities, (2) recent studies that provide quantitative effect estimates for populations of interest, and
(3) studies that consider SOx as a component of a complex mixture of air pollutants.
Not all studies were accorded equal weight in the overall interpretive assessment of evidence
regarding S02-associated health effects. Among studies with adequate control for confounding, increasing
scientific weight is accorded in proportion to the precision of their effect estimates. Small-scale studies
without a wide range of exposures generally produce less precise estimates compared to larger studies
with a broad exposure gradient. For time-series studies, the size of the study, as indicated by the duration
of the study period and total number of events, and the variability of S02 exposures are important
components that help to determine the precision of the health effect estimates. In evaluating the
epidemiologic evidence in this chapter, more weight is accorded to estimates from studies with narrow
confidence bands.
The goal of what was a balanced and objective evaluation that summarizes, interprets, and
synthesizes the most important studies and issues in the epidemiologic database pertaining to SOx
exposure, illustrated by using newly created or previously published summary tables and figures. For each
study presented, the quality of the exposure and outcome data, as well as the quality of the statistical
analysis methodology, are discussed. The discussion incorporates the magnitude and statistical strengths
of observed associations between S02 exposure and health outcomes.
A.2.2. Criteria for Selecting Animal and Human Toxicological Studies
Criteria for the selection of research evaluating animal toxicological or controlled human exposure
studies included a focus on those studies conducted at levels within about an order of a magnitude of
ambient S02 concentrations and those studies that approximated expected human exposure conditions in
terms of concentration and duration. Studies that elucidate mechanisms of action and/or susceptibility,
particularly if the studies were conducted under atmospherically relevant conditions, were emphasized
whenever possible. For controlled human exposure studies, emphasis was placed on studies that (1)
investigated potentially susceptible populations such as asthmatics, particularly studies that compared
responses in susceptible individuals with those in age-matched healthy controls; (2) addressed issues such
as concentration-response or time-course of responses; (3) investigated exposure to S02 separately and in
combination with other pollutants such as 03 and N02; (4) included control exposures to filtered air; and
(5) had sufficient statistical power to assess findings.
A-2
-------
KEY DEFINITIONS
Continuous,
comprehensive
literature review
of peer-reviewed
journal articles
Studies that do
not address
exposure and/or
effects of air
pollutant(s) under
review are
excluded.
Studies added
to the docket
during public
comment period.
Studies identified
during EPA
sponsored kickoff
meeting (including
studies in
preparation}.
Informative
studies
are identified
Studies are
evaluated for inclusion
in the ISA and I
or Annexes.
INFORMATIVE studies are well-designed,
properly implemented, thoroughly described.
HIGHLY INFORMATIVE studies reduce
uncertainty on critical issues, may include
analyses of confounding or effect modification
by copollutants or other variables, analyses of
concentration-response or dose-response
relationships, analyses related to time
between exposure and response, and offer
innovation in method or design.
POLICY-RELEVANT studies may include
those conducted at or near ambient concen-
trations and studies conducted in U.S. and
Canadian airsheds.
Selection of
studies
d t
discussed and
additional studies
identified during
CASAC peer
review of draft
document.
Policy relevant and highly informative studies discussed in the ISA text include
those that provide a basis for or describe the association between the criteria
pollutant and effects. Studies summarized in tables and figures are included
because they are sufficiently comparable to be displayed together. A study
highlighted in the ISA text does not necessarily appear in a summary table or
figure.
ANNEXES
Figure A-1.
All newly identified informative studies are included in the Annexes. Older, key
studies included in previous assessments may be included as well.
Selection process for studies included in the ISA. Studies are categorized into:
informative, highly informative, and policy-relevant.
A-3
-------
A.3. Other Approaches to the Causal Determination
The following sections include excerpts from several reports that document approaches for the
causality, or related decision-making processes. These sections provide supplementary documentation of
approaches that are similar in nature to EPA's causal decision framework.
A.3.1. Surgeon General's Report: The Health Consequences of
Smoking
The Surgeon General's Report (CDC, 2004) evaluated the health effects of smoking; building upon
the first Surgeon General's report published in 1964 (PHS, 1964). It also updated the methodology for
evaluating evidence that was first presented in the 1964 report. The 2004 report acknowledged the
effectiveness of the previous methodology, but standardizes the language surrounding causality of
associations.
The Surgeon General's Reports on Smoking played a central role in the translation of scientific
evidence into policy, conveying succinctly the link between smoking and a health effect. Specifically, the
report stated:
The statement that an exposure "causes" a disease in humans represents a
serious claim, but one that carries with it the possibility of prevention.
Causal determinations may also carry substantial economic implications
for society and for those who might be held responsible for the exposure
or for achieving its prevention.
To address the issue of identifying causality, the 2004 report provided the following summary of
the earlier 1964 report:
When a relationship or an association between smoking... and some
condition in the host was noted, the significance of the association was
assessed.
The characterization of the assessment called for a specific term. ... The
word cause is the one in general usage in connection with matters
considered in this study, and it is capable of conveying the notion of a
significant, effectual relationship between an agent and an associated
disorder or disease in the host.
No member was so naive as to insist upon mono-etiology in pathological
processes or in vital phenomena. All were thoroughly aware... that the
end results are the net effect of many actions and counteractions.
Granted that these complexities were recognized, it is to be noted clearly
that the Committee's considered decision to use the words "a cause," or
"a major cause," or "a significant cause," or "a causal association" in
certain conclusions about smoking and health affirms their conviction
(CDC, 2004)
A-4
-------
The 2004 report created uniformly labeled conclusions that were used throughout the document. The
following excerpts from the report also include a description of the methodology and the judgments used
to reach a conclusion:
Terminology of Conclusions and Causal Claims
Evidence is sufficient to infer a causal relationship.
Evidence is suggestive but not sufficient to infer a causal relationship.
Evidence is inadequate to infer the presence or absence of a causal
relationship (which encompasses evidence that is sparse, of poor quality,
or conflicting).
Evidence is suggestive of no causal relationship.
For this report, the summary conclusions regarding causality are
expressed in this four-level classification. Use of these classifications
should not constrain the process of causal inference, but rather bring
consistency across chapters and reports, and greater clarity as to what the
final conclusions are actually saying. As shown in Table 1.1 [see original
document], without a uniform classification the precise nature of the
final judgment may not always be obvious, particularly when the
judgment is that the evidence falls below the "sufficient" category.
Experience has shown that the "suggestive" category is often an
uncomfortable one for scientists, since scientific culture is such that any
evidence that falls short of causal proof is typically deemed inadequate to
make a causal determination. However, it is very useful to distinguish
between evidence that is truly inadequate versus that which just falls
short of sufficiency.
There is no category beyond "suggestive of no causal relationship" as it
is extraordinarily difficult to prove the complete absence of a causal
association. At best, "negative" evidence is suggestive, either strongly or
weakly. In instances where this category is used, the strength of evidence
for no relationship will be indicated in the body of the text. In this new
framework, conclusions regarding causality will be followed by a section
on implications. This section will separate the issue of causal inference
from recommendations for research, policies, or other actions that might
arise from the causal conclusions. This section will assume a public
health perspective, focusing on the population consequences of using or
not using tobacco and also a scientific perspective, proposing further
research directions. The proportion of cases in the population as a result
of exposure (the population attributable risk), along with the total
prevalence and seriousness of a disease, are more relevant for deciding
on actions than the relative risk estimates typically used for etiologic
determinations. In past reports, the failure to sharply separate issues of
inference from policy issues resulted in inferential statements that were
sometimes qualified with terms for action. For example, based on the
evidence available in 1964, the first Surgeon General's report on
A-5
-------
smoking and health contained the following statement about the
relationship between cardiovascular diseases and smoking:
It is established that male cigarette smokers have a higher death rate from
coronary artery disease than non-smoking males. Although the causative
role of cigarette smoking in deaths from coronary disease is not proven,
the Committee considers it more prudent from the public health
viewpoint to assume that the established association has causative
meaning, than to suspend judgment until no uncertainty remains (CDC,
2004)
Using this framework, this conclusion would now be expressed
differently, probably placing it in the "suggestive" category and making
it clear that although it falls short of proving causation, this evidence still
makes causation more likely than not. The original statement makes it
clear that the 1964 committee judged that the evidence fell short of
proving causality but was sufficient to justify public health action. In this
report, the rationale and recommendations for action will be placed in the
implications section, separate from the causal conclusions. This
separation of inferential from action-related statements clarifies the
degree to which policy recommendations are driven by the strength of
the evidence and by the public health consequences acting to reduce
exposure. In addition, this separation appropriately reflects the
differences between the processes and goals of causal inference and
decision making.
A.3.2. EPA: Guidelines for Carcinogen Risk Assessment
The EPA Guidelines for Carcinogen Risk Assessment, published in 2005 (EPA, 2005), was an
update to the previous risk assessment document published in 1986. This document served to guide EPA
staff and public about the Agency's risk assessment development and methodology. In the 1986
Guidelines, a step-wise approach was used to evaluate the scientific findings. However, this newer
document was similar to the Surgeon General's Report on Smoking in that it used single integrative step
after assessing all of the individual lines of evidence. Five standard descriptors were used to evaluate the
weight of evidence:
1. Carcinogenic to Humans
2. Likely to Be Carcinogenic to Humans
3. Suggestive Evidence of Carcinogenic Potential
4. Inadequate Information to Assess Carcinogenic Potential
5. Not Likely to Be Carcinogenic to Humans.
The 2005 Guidelines recommend that a separate narrative be prepared on the weight of evidence
and the descriptor. The Guidelines further recommend that the descriptors should only be used in the
context of a weight-of-evidence discussion.
A-6
-------
The following excerpt describes how a weight of evidence narrative should be developed and how
a descriptor should be selected (EPA, 2005):
The weight of the evidence should be presented as a narrative laying out
the complexity of information that is essential to understanding the
hazard and its dependence on the quality, quantity, and type(s) of data
available, as well as the circumstances of exposure or the traits of an
exposed population that may be required for expression of cancer. For
example, the narrative can clearly state to what extent the determination
was based on data from human exposure, from animal experiments, from
some combination of the two, or from other data. Similarly, information
on mode of action can specify to what extent the data are from in vivo or
in vitro exposures or based on similarities to other chemicals. The extent
to which an agent's mode of action occurs only on reaching a minimum
dose or a minimum duration should also be presented. A hazard might
also be expressed disproportionately in individuals possessing a specific
gene; such characterizations may follow from a better understanding of
the human genome. Furthermore, route of exposure should be used to
qualify a hazard if, for example, an agent is not absorbed by some routes.
Similarly, a hazard can be attributable to exposures during a susceptible
lifestage on the basis of our understanding of human development.
The weight of evidence narrative should highlight:
¦ the quality and quantity of the data;
¦ all key decisions and the basis for these major decisions; and
¦ any data, analyses, or assumptions that are unusual for or new to EPA.
To capture this complexity, a weight of evidence narrative generally
includes
¦ conclusions about human carcinogenic potential (choice of descriptor(s), described
below)
¦ a summary of the key evidence supporting these conclusions (for each descriptor
used), including information on the type(s) of data (human and/or animal, in vivo
and/or in vitro) used to support the conclusion(s)
¦ available information on the epidemiologic or experimental conditions that
characterize expression of carcinogenicity (e.g., if carcinogenicity is possible only by
one exposure route or only above a certain human exposure level),
¦ a summary of potential modes of action and how they reinforce the conclusions,
¦ indications of any susceptible populations or lifestages, when available, and
¦ a summary of the key default options invoked when the available information is
inconclusive.
To provide some measure of clarity and consistency in an otherwise free-
form narrative, the weight of evidence descriptors are included in the
first sentence of the narrative. Choosing a descriptor is a matter of
judgment and cannot be reduced to a formula. Each descriptor may be
applicable to a wide variety of potential data sets and weights of
evidence. These descriptors and narratives are intended to permit
sufficient flexibility to accommodate new scientific understanding and
A-7
-------
new testing methods as they are developed and accepted by the scientific
community and the public. Descriptors represent points along a
continuum of evidence; consequently, there are gradations and borderline
cases that are clarified by the full narrative. Descriptors, as well as an
introductory paragraph, are a short summary of the complete narrative
that preserves the complexity that is an essential part of the hazard
characterization. Users of these cancer guidelines and of the risk
assessments that result from the use of these cancer guidelines
should consider the entire range of information included in the
narrative rather than focusing simply on the descriptor.
In borderline cases, the narrative explains the case for choosing one
descriptor and discusses the arguments for considering but not choosing
another. For example, between "suggestive" and "likely" or between
"suggestive" and "inadequate," the explanation clearly communicates the
information needed to consider appropriately the agent's carcinogenic
potential in subsequent decisions.
Multiple descriptors can be used for a single agent, for example, when
carcinogenesis is dose-or route-dependent. For example, if an agent
causes point-of-contact tumors by one exposure route but adequate
testing is negative by another route, then the agent could be described as
likely to be carcinogenic by the first route but not likely to be
carcinogenic by the second. Another example is when the mode of action
is sufficiently understood to conclude that a key event in tumor
development would not occur below a certain dose range. In this case,
the agent could be described as likely to be carcinogenic above a certain
dose range but not likely to be carcinogenic below that range.
Descriptors can be selected for an agent that has not been tested in a
cancer bioassay if sufficient other information, e.g., toxicokinetic and
mode of action information, is available to make a strong, convincing,
and logical case through scientific inference. For example, if an agent is
one of a well-defined class of agents that are understood to operate
through a common mode of action and if that agent has the same mode of
action, then in the narrative the untested agent would have the same
descriptor as the class. Another example is when an untested agent's
effects are understood to be caused by a human metabolite, in which case
in the narrative the untested agent could have the same descriptor as the
metabolite. As new testing methods are developed and used, assessments
may increasingly be based on inferences from toxicokinetic and mode of
action information in the absence of tumor studies in animals or humans.
When a well-studied agent produces tumors only at a point of initial
contact, the descriptor generally applies only to the exposure route
producing tumors unless the mode of action is relevant to other routes.
The rationale for this conclusion would be explained in the narrative.
When tumors occur at a site other than the point of initial contact, the
descriptor generally applies to all exposure routes that have not been
adequately tested at sufficient doses. An exception occurs when there is
A-8
-------
convincing information, e.g., toxicokinetic data that absorption does not
occur by another route.
When the response differs qualitatively as well as quantitatively with
dose, this information should be part of the characterization of the
hazard. In some cases reaching a certain dose range can be a
precondition for effects to occur, as when cancer is secondary to another
toxic effect that appears only above a certain dose. In other cases
exposure duration can be a precondition for hazard if effects occur only
after exposure is sustained for a certain duration. These considerations
differ from the issues of relative absorption or potency at different dose
levels because they may represent a discontinuity in a dose-response
function.
When multiple bioassays are inconclusive, mode of action data are likely
to hold the key to resolution of the more appropriate descriptor. When
bioassays are few, further bioassays to replicate a study's results or to
investigate the potential for effects in another sex, strain, or species may
be useful.
When there are few pertinent data, the descriptor makes a statement
about the database, for example, "Inadequate Information to Assess
Carcinogenic Potential," or a database that provides "Suggestive
Evidence of Carcinogenic Potential." With more information, the
descriptor expresses a conclusion about the agent's carcinogenic
potential to humans. If the conclusion is positive, the agent could be
described as "Likely to Be Carcinogenic to Humans" or, with strong
evidence, "Carcinogenic to Humans." If the conclusion is negative, the
agent could be described as "Not Likely to Be Carcinogenic to Humans."
Although the term "likely" can have a probabilistic connotation in other
contexts, its use as a weight of evidence descriptor does not correspond
to a quantifiable probability of whether the chemical is carcinogenic.
This is because the data that support cancer assessments generally are not
suitable for numerical calculations of the probability that an agent is a
carcinogen. Other health agencies have expressed a comparable weight
of evidence using terms such as "Reasonably Anticipated to Be a Human
Carcinogen" (NTP) or "Probably Carcinogenic to Humans" (IARC).
1989).
A.3.3. Improving the NAS/IOM Presumptive Disability Decision-Making
Process for Veterans Report
A recent publication by the Institute of Medicine also provided foundation for the causality
framework adapted in this ISA (IOM, 2007). The Committee on Evaluation of the Presumptive Disability
Decision-Making Process for Veterans was charged by the Veterans Association to describe how
presumptive decisions are made for veterans with health conditions arising from military service
currently, as well as recommendations for how such decisions could be made in the future. The committee
proposed a multiple-element approach that includes a quantification of the extent of disease attributable to
A-9
-------
an exposure. This process involved a review of all relevant data to decide the strength of evidence for
causation, using one of four categories:
¦ Sufficient: the evidence is sufficient to conclude that a causal relationship exists.
¦ Equipoise and Above: the evidence is sufficient to conclude that a causal relationship is at
least as likely as not, but not sufficient to conclude that a causal relationship exists.
¦ Below Equipoise: the evidence is not sufficient to conclude that a causal relationship is at
least as likely as not, or is not sufficient to make a scientifically informed judgment.
¦ Against: the evidence suggests the lack of a causal relationship.
The following is an excerpt from the report and describes these four categories in detail:
In light of the categorizations used by other health organizations and
agencies as well as considering the particular challenges of the
presumptive disability decision-making process, we propose a four-level
categorization of the strength of the overall evidence for or against a
causal relationship from exposure to disease.
We use the term "equipoise" to refer to the point at which the evidence is
in balance between favoring and not favoring causation. The term
"equipoise" is widely used in the biomedical literature, is a concept
familiar to those concerned with evidence-based decision-making and is
used in VA processes for rating purposes as well as being a familiar term
in the veterans' community.
Below we elaborate on the four-level categorization which the
Committee recommends.
Sufficient
If the overall evidence for a causal relationship is categorized as
Sufficient, then it should be scientifically compelling. It might include:
¦ replicated and consistent evidence of a causal association: that is, evidence of an
association from several high-quality epidemiologic studies that cannot be explained
by plausible noncausal alternatives (e.g., chance, bias, or confounding)
¦ evidence of causation from animal studies and mechanistic knowledge
¦ compelling evidence from animal studies and strong mechanistic evidence from
studies in exposed humans, consistent with (i.e., not contradicted by) the
epidemiologic evidence.
Using the Bayesian framework to illustrate the evidential support and the
resulting state of communal scientific opinion needed for reaching the
Sufficient category (and the lower categories that follow), consider again
the causal diagram in Figure A-2. In this model, used to help clarify
matters conceptually, the observed association between exposure and
health is the result of: (1) measured confounding, parameterized by a; (2)
the causal relation, parameterized by (3; and (3) other, unmeasured
sources such as bias or unmeasured confounding, parameterized by y.
The belief of interest, after all the evidence has been weighed, is in the
size of the causal parameter (3. Thus, for decision making, what matters is
how strongly the evidence supports the proposition that |3 is above 0. As
A-10
-------
it is extremely unlikely that the types of exposures considered for
presumptions reduce the risk of developing disease, we exclude values of
(3 below 0. If we consider the evidence as supporting degrees of belief
about the size of (3, and we have a posterior distribution over the possible
size of (3, then a posterior like Figure A-2 illustrates a belief state that
might result when the evidence for causation is considered Sufficient.
As the "mass" over a positive effect (the area under the curve to the right
of the zero) vastly "outweighs" the small mass over no effect (zero), the
evidence is considered sufficient to conclude that the association is
causal. Put another way, even though the scientific community might be
uncertain as to the size of (3, after weighing all the evidence, it is highly
confident that the probability that (3 is greater than zero is substantial;
that is, that exposure causes disease.
Equipoise and Above
To be categorized as Equipoise and Above, the scientific community
should categorize the overall evidence as making it more confident in the
existence of a causal relationship than in the non-existence of a causal
relationship, but not sufficient to conclude causation.
For example, if there are several high-quality epidemiologic studies, the
preponderance of which show evidence of an association that cannot
readily be explained by plausible noncausal alternatives (e.g., chance,
bias, or confounding), and the causal relationship is consistent with the
animal evidence and biological knowledge, then the overall evidence
might be categorized as Equipoise and Above. Alternatively, if there is
strong evidence from animal studies or mechanistic evidence, not
contradicted by human or other evidence, then the overall evidence
might be categorized as Equipoise and Above. Equipoise is a common
term employed by VA and the courts in deciding disability claims (see
Appendix D [see original document]).
A-11
-------
Measured
Confounders/Covariates
1 ° 1
Exposure
to Substance
Health
Outcome
P
x^z
Unmeasured Confounders/Covariates, or
Other Sources of Spurious Association from Bias
Source: IOM (2008).
Figure A-2. Focusing on unmeasured confounders/covariates, or other sources of spurious
association from bias.
Posterior Over p
Posterior Mass
Over an Effect
0
Size of the Causal Effect (3
>
Source: IOM (2008).
Figure A-3. Example posterior distribution for the determination of Sufficient.
Again, using the Bayesian model to illustrate the idea of Equipoise and
Figure A-4 shows a posterior probability distribution that is an example
of belief compatible with the category Equipoise and Above.
A-12
-------
Posterior Over (3
Posterior Mass
Over an Effect
P(P)
0
Size of the Causal Effect |3
>
Source: IOM (2007).
Figure A-4. Example posterior distribution for the determination of Equipoise and Above.
In this figure, unlike the one for evidence classified as Sufficient, there is
considerable mass over zero, which means that the scientific community
has considerable uncertainty as to whether exposure causes disease at all;
that is, whether |3 is greater than zero. At least half of the mass is to the
right of the zero, however, so the community judges causation to be at
least as likely as not, after they have seen and combined all the evidence
available.
Below Equipoise
To be categorized as Below Equipoise, the overall evidence for a causal
relationship should either be judged not to make causation at least as
likely as not, or not sufficient to make a scientifically informed
judgment.
This might occur:
¦ when the human evidence is consistent in showing an association, but the evidence is
limited by the inability to rule out chance, bias, or confounding with confidence, and
animal or mechanistic evidence is weak
¦ when animal evidence suggests a causal relationship, but human and mechanistic
¦ evidence is weak or inconsistent
¦ when mechanistic evidence is suggestive but animal and human evidence is weak or
inconsistent
¦ when the evidence base is very thin.
Against
To be categorized as Against, the overall evidence should favor belief
that there is no causal relationship from exposure to disease. For
example, if there is human evidence from multiple studies covering the
full range of exposures encountered by humans that are consistent in
showing no causal association, or there are is animal or mechanistic
evidence supporting the lack of a causal relationship, and combining all
of the evidence results in a posterior resembling Source: IOM (2008).
A-13
-------
Figure A-5 then the scientific community should categorize the evidence
as Against causation.
P( P)
Posterior Over p
Posterior Mass
Over an Effect
0 Size of the Causal Effect |3
Source: IOM (2007).
Figure A-5. Example posterior distribution for the determination of Against.
A.3.4. National Acid Precipitation Assessment Program Guidelines
The following guidelines in the form of questions were developed and published in 1991 by the
Oversight Review Board for the National Acid Precipitation Assessment Program (Washington, 1991) to
assist scientists in formulating presentations of research results to be used in policy decision processes.
Is the statement sound? Have the central issues been clearly identified?
Does each statement contain the distilled essence of present scientific
and technical understanding of the phenomenon or process to which it
applies? Is the statement consistent with all relevant evidence - evidence
developed either through NAPAP research or through analysis of
research conducted outside of NAPAP? Is the statement contradicted by
any important evidence developed through research inside or outside of
NAPAP? Have apparent contradictions or interpretations of available
evidence been considered in formulating the statement of principal
findings?
Is the statement directional and, where appropriate, quantitative? Does
the statement correctly quantify both the direction and magnitude of
trends and relationships in the phenomenon or process to which the
statement is relevant? When possible, is a range of uncertainty given for
each quantitative result? Have various sources of uncertainty been
identified and quantified, for example, does the statement include or
acknowledge errors in actual measurements, standard errors of estimate,
possible biases in the availability of data, extrapolation of results beyond
the mathematical, geographical, or temporal relevancy of available
information, etc. In short, are there numbers in the statement? Are the
numbers correct? Are the numbers relevant to the general meaning of the
statement?
A-14
-------
Is the degree of certainty or uncertainty of the statement indicated
clearly? Have appropriate statistical tests been applied to the data used in
drawing the conclusion set forth in the statement? If the statement is
based on a mathematical or novel conceptual model, has the model or
concept been validated? Does the statement describe the model or
concept on which it is based and the degree of validity of that model or
concept?
Is the statement correct without qualification? Are there limitations of
time, space, or other special circumstances in which the statement is
true? If the statement is true only in some circumstances, are these
limitations described adequately and briefly?
Is the statement clear and unambiguous? Are the words and phrases used
in the statement understandable by the decision makers of our society? Is
the statement free of specialized jargon? Will too many people
misunderstand its meaning?
Is the statement as concise as it can be made without risk of
misunderstanding? Are there any excess words, phrases, or ideas in the
statement which are not necessary to communicate the meaning of the
statement? Are there so many caveats in the statement that the statement
itself is trivial, confusing, or ambiguous?
Is the statement free of scientific or other biases or implications of
societal value judgments? Is the statement free of influence by specific
schools of scientific thought? Is the statement also free of words,
phrases, or concepts that have political, economic, ideological, religious,
moral, or other personal-, agency-, or organization-specific values,
overtones, or implications? Does the choice of how the statement is
expressed rather than its specific words suggest underlying biases or
value judgments? Is the tone impartial and free of special pleading? If
societal value judgments have been discussed, have these judgments
been identified as such and described both clearly and objectively?
Have societal implications been described objectively? Consideration of
alternative courses of action and their consequences inherently involves
judgments of their feasibility and the importance of effects. For this
reason, it is important to ask if a reasonable range of alternative policies
or courses of action have been evaluated? Have societal implications of
alternative courses of action been stated in the following general form?
"If this [particular option] were adopted then that [particular outcome]
would be expected."
Have the professional biases of authors and reviewers been described
openly? Acknowledgment of potential sources of bias is important so
that readers can judge for themselves the credibility of reports and
assessments.
A-15
-------
A.3.5. IARC Guidelines for Scientific Review and Evaluation
Categories
The following is excerpted from the International Agency for Research on Cancer Monographs on
the evaluation of carcinogenic risks to humans (IARC, 2006)
The available studies are summarized by the Working Group, with
particular regard to the qualitative aspects discussed below. In general,
numerical findings are indicated as they appear in the original report;
units are converted when necessary for easier comparison. The Working
Group may conduct additional analyses of the published data and use
them in their assessment of the evidence; the results of such
supplementary analyses are given in square brackets. When an important
aspect of a study that directly impinges on its interpretation should be
brought to the attention of the reader, a Working Group comment is
given in square brackets.
The scope of the IARC Monographs program has expanded beyond
chemicals to include complex mixtures, occupational exposures, physical
and biological agents, lifestyle factors and other potentially carcinogenic
exposures. Over time, the structure of a Monograph has evolved to
include the following sections:
1. Exposure data
2. Studies of cancer in humans
3. Studies of cancer in experimental animals
4. Mechanistic and other relevant data
5. Summary
6. Evaluation and rationale
In addition, a section of General Remarks at the front of the volume
discusses the reasons the agents were scheduled for evaluation and some
key issues the Working Group encountered during the meeting.
This part of the Preamble discusses the types of evidence considered and
summarized in each section of a Monograph, followed by the scientific
criteria that guide the evaluations.
Evaluation and rationale
Evaluations of the strength of the evidence for carcinogenicity arising
from human and experimental animal data are made, using standard
terms. The strength of the mechanistic evidence is also characterized.
A-16
-------
It is recognized that the criteria for these evaluations, described below,
cannot encompass all of the factors that may be relevant to an evaluation
of carcinogenicity. In considering all of the relevant scientific data, the
Working Group may assign the agent to a higher or lower category than
a strict interpretation of these criteria would indicate.
These categories refer only to the strength of the evidence that an
exposure is carcinogenic and not to the extent of its carcinogenic activity
(potency). A classification may change as new information becomes
available.
An evaluation of the degree of evidence is limited to the materials tested,
as defined physically, chemically or biologically. When the agents
evaluated are considered by the Working Group to be sufficiently closely
related, they may be grouped together for the purpose of a single
evaluation of the degree of evidence.
(a) Carcinogenicity in humans
The evidence relevant to carcinogenicity from studies in humans is
classified into one of the following categories:
Sufficient evidence of carcinogenicity. The Working Group considers
that a causal relationship has been established between exposure to the
agent and human cancer. That is, a positive relationship has been
observed between the exposure and cancer in studies in which chance,
bias and confounding could be ruled out with reasonable confidence. A
statement that there is sufficient evidence is followed by a separate
sentence that identifies the target organ(s) or tissue(s) where an increased
risk of cancer was observed in humans. Identification of a specific target
organ or tissue does not preclude the possibility that the agent may cause
cancer at other sites.
Limited evidence of carcinogenicity . A positive association has been
observed between exposure to the agent and cancer for which a causal
interpretation is considered by the Working Group to be credible, but
chance, bias or confounding could not be ruled out with reasonable
confidence.
Inadequate evidence of carcinogenicity . The available studies are of
insufficient quality, consistency or statistical power to permit a
conclusion regarding the presence or absence of a causal association
between exposure and cancer, or no data on cancer in humans are
available.
Evidence suggesting lack of carcinogenicity. There are several adequate
studies covering the full range of levels of exposure that humans are
known to encounter, which are mutually consistent in not showing a
positive association between exposure to the agent and any studied
cancer at any observed level of exposure. The results from these studies
alone or combined should have narrow confidence intervals with an
A-17
-------
upper limit close to the null value (e.g. a relative risk of 1.0). Bias and
confounding should be ruled out with reasonable confidence, and the
studies should have an adequate length of follow-up. A conclusion of
evidence suggesting lack of carcinogenicity is inevitably limited to the
cancer sites, conditions and levels of exposure, and length of observation
covered by the available studies. In addition, the possibility of a very
small risk at the levels of exposure studied can never be excluded.
In some instances, the above categories may be used to classify the
degree of evidence related to carcinogenicity in specific organs or
tissues.
When the available epidemiologic studies pertain to a mixture, process,
occupation or industry, the Working Group seeks to identify the specific
agent considered most likely to be responsible for any excess risk. The
evaluation is focused as narrowly as the available data on exposure and
other aspects permit.
(b) Carcinogenicity in experimental animals
Carcinogenicity in experimental animals can be evaluated using
conventional bioassays, bioassays that employ genetically modified
animals, and other in-vivo bioassays that focus on one or more of the
critical stages of carcinogenesis. In the absence of data from
conventional long-term bioassays or from assays with neoplasia as the
end-point, consistently positive results in several models that address
several stages in the multistage process of carcinogenesis should be
considered in evaluating the degree of evidence of carcinogenicity in
experimental animals.
The evidence relevant to carcinogenicity in experimental animals is
classified into one of the following categories:
Sufficient evidence of carcinogenicity. The Working Group considers
that a causal relationship has been established between the agent and an
increased incidence of malignant neoplasms or of an appropriate
combination of benign and malignant neoplasms in (a) two or more
species of animals or (b) two or more independent studies in one species
carried out at different times or in different laboratories or under
different protocols. An increased incidence of tumors in both sexes of a
single species in a well-conducted study, ideally conducted under Good
Laboratory Practices, can also provide sufficient evidence.
A single study in one species and sex might be considered to provide
sufficient evidence of carcinogenicity when malignant neoplasms occur
to an unusual degree with regard to incidence, site, type of tumor or age
at onset, or when there are strong findings of tumors at multiple sites.
Limited evidence of carcinogenicity . The data suggest a carcinogenic
effect but are limited for making a definitive evaluation because, e.g. (a)
the evidence of carcinogenicity is restricted to a single experiment; (b)
A-18
-------
there are unresolved questions regarding the adequacy of the design,
conduct or interpretation of the studies; (c) the agent increases the
incidence only of benign neoplasms or lesions of uncertain neoplastic
potential; or (d) the evidence of carcinogenicity is restricted to studies
that demonstrate only promoting activity in a narrow range of tissues or
organs.
Inadequate evidence of carcinogenicity . The studies cannot be
interpreted as showing either the presence or absence of a carcinogenic
effect because of major qualitative or quantitative limitations, or no data
on cancer in experimental animals are available.
Evidence suggesting lack of carcinogenicity: Adequate studies involving
at least two species are available which show that, within the limits of the
tests used, the agent is not carcinogenic. A conclusion of evidence
suggesting lack of carcinogenicity is inevitably limited to the species,
tumor sites, age at exposure, and conditions and levels of exposure
studied.
(c) Mechanistic and other relevant data
Mechanistic and other evidence judged to be relevant to an evaluation of
carcinogenicity and of sufficient importance to affect the overall
evaluation is highlighted. This may include data on preneoplastic lesions,
tumor pathology, genetic and related effects, structure-activity
relationships, metabolism and toxicokinetics, physicochemical
parameters and analogous biological agents.
The strength of the evidence that any carcinogenic effect observed is due
to a particular mechanism is evaluated, using terms such as 'weak,'
'moderate' or 'strong.' The Working Group then assesses whether that
particular mechanism is likely to be operative in humans. The strongest
indications that a particular mechanism operates in humans derive from
data on humans or biological specimens obtained from exposed humans.
The data may be considered to be especially relevant if they show that
the agent in question has caused changes in exposed humans that are on
the causal pathway to carcinogenesis. Such data may, however, never
become available, because it is at least conceivable that certain
compounds may be kept from human use solely on the basis of evidence
of their toxicity and/or carcinogenicity in experimental systems.
The conclusion that a mechanism operates in experimental animals is
strengthened by findings of consistent results in different experimental
systems, by the demonstration of biological plausibility and by coherence
of the overall database. Strong support can be obtained from studies that
challenge the hypothesized mechanism experimentally, by demonstrating
that the suppression of key mechanistic processes leads to the
suppression of tumor development. The Working Group considers
whether multiple mechanisms might contribute to tumor development,
whether different mechanisms might operate in different dose ranges,
whether separate mechanisms might operate in humans and experimental
A-19
-------
animals and whether a unique mechanism might operate in a susceptible
group. The possible contribution of alternative mechanisms must be
considered before concluding that tumors observed in experimental
animals are not relevant to humans. An uneven level of experimental
support for different mechanisms may reflect that disproportionate
resources have been focused on investigating a favored mechanism.
For complex exposures, including occupational and industrial exposures,
the chemical composition and the potential contribution of carcinogens
known to be present are considered by the Working Group in its overall
evaluation of human carcinogenicity. The Working Group also
determines the extent to which the materials tested in experimental
systems are related to those to which humans are exposed.
(d) Overall evaluation
Finally, the body of evidence is considered as a whole, in order to reach
an overall evaluation of the carcinogenicity of the agent to humans.
An evaluation may be made for a group of agents that have been
evaluated by the Working Group. In addition, when supporting data
indicate that other related agents, for which there is no direct evidence of
their capacity to induce cancer in humans or in animals, may also be
carcinogenic, a statement describing the rationale for this conclusion is
added to the evaluation narrative; an additional evaluation may be made
for this broader group of agents if the strength of the evidence warrants
it.
The agent is described according to the wording of one of the following
categories, and the designated group is given. The categorization of an
agent is a matter of scientific judgement that reflects the strength of the
evidence derived from studies in humans and in experimental animals
and from mechanistic and other relevant data.
Group 1: The agent is carcinogenic to humans.
This category is used when there is sufficient evidence of carcinogenicity
in humans. Exceptionally, an agent may be placed in this category when
evidence of carcinogenicity in humans is less than sufficient but there is
sufficient evidence of carcinogenicity in experimental animals and strong
evidence in exposed humans that the agent acts through a relevant
mechanism of carcinogenicity.
Group 2.
This category includes agents for which, at one extreme, the degree of
evidence of carcinogenicity in humans is almost sufficient, as well as
those for which, at the other extreme, there are no human data but for
which there is evidence of carcinogenicity in experimental animals.
Agents are assigned to either Group 2A {probably carcinogenic to
humans) or Group 2B (possibly carcinogenic to humans) on the basis of
A-20
-------
epidemiologic and experimental evidence of carcinogenicity and
mechanistic and other relevant data. The terms probably carcinogenic
and possibly carcinogenic have no quantitative significance and are used
simply as descriptors of different levels of evidence of human
carcinogenicity, with probably carcinogenic signifying a higher level of
evidence than possibly carcinogenic.
Group 2A: The agent is probably carcinogenic to humans.
This category is used when there is limited evidence of carcinogenicity in
humans and sufficient evidence of carcinogenicity in experimental
animals. In some cases, an agent may be classified in this category when
there is inadequate evidence of carcinogenicity in humans and sufficient
evidence of carcinogenicity in experimental animals and strong evidence
that the carcinogenesis is mediated by a mechanism that also operates in
humans. Exceptionally, an agent may be classified in this category solely
on the basis of limited evidence of carcinogenicity in humans. An agent
may be assigned to this category if it clearly belongs, based on
mechanistic considerations, to a class of agents for which one or more
members have been classified in Group 1 or Group 2A.
Group 2B: The agent is possibly carcinogenic to humans.
This category is used for agents for which there is limited evidence of
carcinogenicity in humans and less than sufficient evidence of
carcinogenicity in experimental animals. It may also be used when there
is inadequate evidence of carcinogenicity in humans but there is
sufficient evidence of carcinogenicity in experimental animals. In some
instances, an agent for which there is inadequate evidence of
carcinogenicity in humans and less than sufficient evidence of
carcinogenicity in experimental animals together with supporting
evidence from mechanistic and other relevant data may be placed in this
group. An agent may be classified in this category solely on the basis of
strong evidence from mechanistic and other relevant data.
Group 3: The agent is not classifiable as to its carcinogenicity to humans.
This category is used most commonly for agents for which the evidence
of carcinogenicity is inadequate in humans and inadequate or limited in
experimental animals.
Exceptionally, agents for which the evidence of carcinogenicity is
inadequate in humans but sufficient in experimental animals may be
placed in this category when there is strong evidence that the mechanism
of carcinogenicity in experimental animals does not operate in humans.
Agents that do not fall into any other group are also placed in this
category.
An evaluation in Group 3 is not a determination of non-carcinogenicity
or overall safety. It often means that further research is needed,
A-21
-------
especially when exposures are widespread or the cancer data are
consistent with differing interpretations.
Group 4: The agent is probably not carcinogenic to humans.
This category is used for agents for which there is evidence suggesting
lack of carcinogenicity in humans and in experimental animals. In some
instances, agents for which there is inadequate evidence of
carcinogenicity in humans but evidence suggesting lack of
carcinogenicity in experimental animals, consistently and strongly
supported by a broad range of mechanistic and other relevant data, may
be classified in this group.
(e) Rationale
The reasoning that the Working Group used to reach its evaluation is
presented and discussed. This section integrates the major findings from
studies of cancer in humans, studies of cancer in experimental animals,
and mechanistic and other relevant data. It includes concise statements of
the principal line(s) of argument that emerged, the conclusions of the
Working Group on the strength of the evidence for each group of studies,
citations to indicate which studies were pivotal to these conclusions, and
an explanation of the reasoning of the Working Group in weighing data
and making evaluations. When there are significant differences of
scientific interpretation among Working Group Members, a brief
summary of the alternative interpretations is provided, together with their
scientific rationale and an indication of the relative degree of support for
each alternative.
A.3.6. NTP: Report on Carcinogens
The criteria for listing an agent, substance, mixture, or exposure circumstance in the National
Toxicology Program's Report on Carcinogens (NTP, 2005) as follows:
Known to Be Human Carcinogen:
There is sufficient evidence of carcinogenicity from studies in humans*,
which indicates a causal relationship between exposure to the agent,
substance, or mixture, and human cancer.
Reasonably Anticipated to Be Human Carcinogen:
There is limited evidence of carcinogenicity from studies in humans*,
which indicates that causal interpretation is credible, but that alternative
explanations, such as chance, bias, or confounding factors, could not
adequately be excluded,
or
A-22
-------
there is sufficient evidence of carcinogenicity from studies in
experimental animals, which indicates there is an increased incidence of
malignant and/or a combination of malignant and benign tumors (1) in
multiple species or at multiple tissue sites, or (2) by multiple routes of
exposure, or (3) to an unusual degree with regard to incidence, site, or
type of tumor, or age at onset,
or
there is less than sufficient evidence of carcinogenicity in humans or
laboratory animals; however, the agent, substance, or mixture belongs to
a well-defined, structurally related class of substances whose members
are listed in a previous Report on Carcinogens as either known to be a
human carcinogen or reasonably anticipated to be a human carcinogen,
or there is convincing relevant information that the agent acts through
mechanisms indicating it would likely cause cancer in humans.
Conclusions regarding carcinogenicity in humans or experimental
animals are based on scientific judgment, with consideration given to all
relevant information. Relevant information includes, but is not limited to,
dose response, route of exposure, chemical structure, metabolism,
pharmacokinetics, sensitive sub-populations, genetic effects, or other
data relating to mechanism of action or factors that may be unique to a
given substance. For example, there may be substances for which there is
evidence of carcinogenicity in laboratory animals, but there are
compelling data indicating that the agent acts through mechanisms which
do not operate in humans and would therefore not reasonably be
anticipated to cause cancer in humans.
*This evidence can include traditional cancer epidemiology studies, data
from clinical studies, and/or data derived from the study of tissues or
cells from humans exposed to the substance in question that can be
useful for evaluating whether a relevant cancer mechanism is operating
in people.
A-23
-------
Annex B. Additional Information on the
Atmospheric Chemistry of SOx
B.1. Introduction
S02 is chiefly but not exclusively primary in origin. Primary S02 is emitted directly from sources,
whereas secondary S02 is formed as a product of atmospheric reactions. Secondary S02 is produced by
the photochemical oxidation of reduced sulfur compounds such as dimethyl sulfide (DMS) (CH3-S-CH3),
hydrogen sulfide (H2S), carbon disulfide (CS2), carbonyl sulfide (OCS), methyl mercaptan (CH3-S-H),
and dimethyl disulfide (CH3-S-S-CH3) which are all mainly biogenic in origin. Their sources are
discussed in Section B.3. Table B-l lists the atmospheric lifetimes of reduced sulfur species with respect
to reaction with various oxidants. Except for OCS, which is lost mainly by photolysis (x~6 months), these
species are lost mainly by reaction with OH and N03 radicals. Because OCS is relatively long-lived in the
troposphere, it can be transported upwards into the stratosphere.
Table B-1. Atmospheric lifetimes of SO2 and reduced sulfur species with respect to reaction with
OH, NO3, and CI radicals.
OH
NOs
CI
KX1012
T
KX1012
T
KX1012
T
so2
1.6
7.2d
NA
NA
CH3-S-CH3
5.0
2.3 d
1.0
1.1-h
400
29 d
h2s
4.7
2.2 d
NA
74
157 d
cs2
1.2
9.6 d
< 0.0004
> 116 d
< 0.004
NR
OCS
0.0019
17 y
< 0.0001
> 1.3 y
<0.0001
NR
CHs-S-H
33
8.4 h
0.89
1.2 h
200
58 d
CH3-S-S-CH3
230
1.2 h
0.53
2.1-h
NA
NA = Reaction rate coefficient not available.
NR = Rate coefficient too low to be relevant as an atmospheric loss mechanism. Rate coefficients were calculated at 298 K and 1 atmosphere,
yr = year
h = hour
OH = 1 x 106/cm3
N03 = 2.5 x 108/cm3
CI = 1 x 103/cm3.
1Rate coefficients were taken from JPL Chemical Kinetics Evaluation No. 14 (JPL2003)
Source: Seinfeld and Pandis (1998).
Crutzen (1976) proposed that OCS oxidation serves as the major source of sulfate (S042 ) in the
stratospheric aerosol layer sometimes referred to the ""Junge layer," (Junge et al., 1961) during periods
when volcanic plumes do not reach the stratosphere. However, the flux of OCS into the stratosphere is
probably not sufficient to maintain this stratospheric aerosol layer. Myhre et al. (2004) proposed instead
B-1
-------
that S02 transported upwards from the troposphere is the most likely source, as the upward flux of OCS is
too small to sustain observed S042" loadings in the Junge layer. In addition, in situ measurements of the
isotopic composition of sulfur (S) do not match those of OCS (Leung et al. 2002). Reaction with N03
radicals at night most likely represents the major loss process for DMS and methyl mercaptan. The
mechanisms for the oxidation of DMS are still not completely understood. Initial attack by N03 and OH
radicals involves H atom abstraction, with a smaller branch leading to OH addition to the S atom. The OH
addition branch increases in importance as temperatures decrease, becoming the major pathway below
temperatures of 285 K (Ravishankara, 1997). The adduct may either decompose to form methane sulfonic
acid (MSA), or undergo further reactions in the main pathway, to yield dimethyl sulfoxide (Barnes et al.,
1991). Following H atom abstraction from DMS, the main reaction products include MSA and S02. The
ratio of MS A to S02 is strongly temperature dependent, varying from about 0.1 in tropical waters to about
0.4 in Antarctic waters (Seinfeld and Pandis, 1998). Excess S042" (over that expected from the S042" in
seawater) in marine aerosol is related mainly to the production of S02 from the oxidation of DMS.
Transformations among atmospheric S compounds are summarized in Figure B-l
OCS
hv, O
Tropopause
\0H S(-2)
'/////\
OH, NO,
OTHER
OTHER
Source: Adapted from Berresheim et al. (1995).
Figure B-1. Transformations of sulfur compounds in the atmosphere.
B.1.1. Multiphase Chemical Processes Involving SOxand Halogens
Chemical transformations involving inorganic halogenated compounds effect changes in the
multiphase cycling of SOx in ways analogous to their effects on NOx. Oxidation of dimethylsulfide
(CH3)2S by BrO produces dimethylsulfoxide (CH3)2SO (Barnes et al., 1991; Toumi, 1994), and oxidation
by atomic chloride leads to formation of S02 (Keene et al., 1996). Dimethylsulfoxide and S02 are
precursors for methanesulfonic acid (CH3SO3H) and H2S04. In the MBL, virtually all H2S04 and
B-2
-------
CH3SO3H vapor condenses onto existing aerosols or cloud droplets, which subsequently evaporate,
thereby contributing to aerosol growth and acidification. Unlike CH3SO3H, H2S04 also has the potential
to produce new particles (Korhonen et al., 1999; Kulmala et al., 2000), which in marine regions is thought
to occur primarily in the free troposphere. Saiz-Lopez et al. (2004) estimated that observed levels of
bromine oxide (BrO) at Mace Head Atmospheric Research Station in Ireland, would oxidize (CH3)2S
about six times faster than OH and thereby substantially increase production rates of H2S04 and other
condensible S species in the MBL. Sulfur dioxide is also scavenged by deliquesced aerosols and oxidized
to H2S04 in the aqueous phase by several strongly pH-dependent pathways (Chameides and Stelson,
1992; Keene et al., 1998; Vogt et al., 1996). Model calculations indicate that oxidation of S(IV) by 03
dominates in fresh, alkaline sea salt aerosols, whereas oxidation by hypohalous acids (primarily HOC1)
dominates in moderately acidic solutions. Additional particulate non-sea salt (nss) S042" is generated by
S02 oxidation in cloud droplets (Clegg and Toumi, 1998). Ion-balance calculations indicate that most nss
S042" in short-lived (two to 48 h) sea salt size fractions accumulates in acidic aerosol solutions and/or in
acidic aerosols processed through clouds (Keene et al., 2004). The production, cycling, and associated
radiative effects of S-containing aerosols in marine and coastal air are regulated in part by chemical
transformations involving inorganic halogens (Von Glasow et al., 2002). These transformations include:
dry-deposition fluxes of nss S042" in marine air dominated, naturally, by the sea salt size fractions
(Huebert et al., 1996; Turekian et al., 2001); HC1 phase partitioning that regulates sea salt pH and
associated pH-dependent pathways for S(IV) oxidation (Keene et al., 2002; Pszenny et al., 2004); and
potentially important oxidative reactions with reactive halogens for (CH3)2S and S(IV). However, both
the absolute magnitudes and relative importance of these processes in MBL S cycling are poorly
understood.
Table B-2.
Relative contributions of various reactions to the total S(IV) oxidation rate within a
sunlit cloud, 10 min after cloud formation.
Reaction % of Totala
% of Totalb
GAS PHASE 1
OH + S02
3.5
3.1
AQUEOUS PHASE 1
03 + HSOs-
0.6
0.7
03 + SOs2"
7.0
8.2
H2O2 + SOs2"
78.4
82.1
CH3OOH + HSO3- 0.1
0.1
HN04+HS03-
9.0
4.4
HOONO + HSOs- <0.1
<0.1
HS05- + HS03-
1.2
<0.1
SO5- + SO32-
<0.1
<0.1
HSOs- + Fe2*
0.6
a In the absence of transition metals. Bin the presence of iron and copper ions.
Source: Adapted from Warneck (1999).
Iodine (I) chemistry has been linked to ultrafine particle bursts at Mace Head (O'Dowd et al., 1999;
O'Dowd et al., 2002). Observed bursts coincide with elevated concentrations of IO and are characterized
B-3
-------
by particle concentrations increasing from background levels to up to 300.000/cirf on a time scale of
seconds to minutes. This newly identified source of marine aerosol would provide additional aerosol
surface area for condensation of SOx and thereby presumably diminish the potential for nucleation
pathways involving H2S04. However, a subsequent investigation in polluted air along the New England,
USA coast found no correlation between periods of nanoparticle growth and corresponding concentrations
of I oxides (Fehsenfeld et al., 2006). The potential importance of I chemistry in aerosol nucleation and its
associated influence on sulfur cycling remain highly uncertain.
B.1.2. Mechanisms for the Aqueous Phase Formation of Sulfate
Warneck (1999) constructed a box model describing the chemistry of the oxidation of S02 and N02
including the interactions ofN and S species and minor processes in sunlit cumulus clouds. The relative
contributions of different reactions to the oxidation of S(IV) species to S(VI) and N02 to N03 10 min
after cloud formation are given in Table B-2. The two columns show the relative contributions with and
without transition metal ions. As can be seen from Table B-2, S02 within a cloud (gas + cloud drops) is
oxidized mainly by H202 in the aqueous phase, while the gas-phase oxidation by OH radicals is small by
comparison. A much smaller contribution in the aqueous phase is made by methyl hydroperoxide
(CH3OOH) because it is formed mainly in the gas phase and its Henry's Law constant is several orders of
magnitude smaller that of H202. After H202, HN04 is the major contributor to S(IV) oxidation. The pH
dependence of the oxidation rate of S(IV) in the presence of transition metal ions is illustrated in Figure
B-2.
1 1 1 1 ]
HA A
**
7/ _
//
/y.
//Mil"/'
>y
yy./
/
_
/
/
- /
./
/
1 1 1
1 1
PH
Source: Seinfeld and Pandis (1998).
Figure B-2. Comparison of aqueous-phase oxidation paths. The rate of conversion of S(IV) to
S(VI) is shown as a function of pH. Conditions assumed are: [S02(g)] = 5 ppb; [N02(g)]
= 1 ppb; [H202(g)] = 1 ppb; [03(g)] = 50 ppb; [Fe(lll)(aq)] = 0.3 jjM; [Mn(ll)(aq)] = 0.3 jjM.
B-4
-------
B.1.3. Multiphase Chemical Processes Involving SOxand Nhh
The phase partitioning of ammonia (NH3) with deliquesced aerosol solutions is controlled primarily
by the thermodynamic properties of the system expressed as follows:
Kb
NH3g^[NH3aq]^[NH4+] +KW/[H+]
where KH and Kb are the temperature-dependent Henry's Law and dissociation constants (62 M/atm and
1.8 x 10"5 M), respectively, for NH3, and Kw is the ion product of water (1.0 x 10"14 M) (Chameides,
1984). It is evident that for a given amount of NHX (NH3 + particulate NH/) in the system, increasing
aqueous concentrations of particulate H will shift the partitioning ofNH3 towards the condensed phase.
Consequently, under the more polluted conditions characterized by higher concentrations of acidic sulfate
aerosol, ratios of gaseous NH3 to particulate NH44" decrease (Smith et al., 2007). It also follows that in
marine air, where aerosol acidity varies substantially as a function of particle size, NH3 partitions
preferentially to the more acidic sub-micron size fractions where Kff and K/, are the temperature-
dependent Henry's Law and dissociation constants (62 M/atm and 1.8 x 10"5 M), respectively, forNH3,
and Kw is the ion product of water (1.0 x 10"14M) (Chameides 1984). It is evident that for a given amount
of NHX (NH3 + particulate NH/) in the system, increasing aqueous concentrations of particulate H will
shift the partitioning ofNH3 towards the condensed phase. Consequently, under the more polluted
conditions characterized by higher concentrations of acidic sulfate aerosol, ratios of gaseous NH3 to
particulate NH/ decrease (Smith et al., 2007). Under these conditions, the roughly equal partitioning of
total NH3 between the gas and particulate phases sustains substantial dry-deposition fluxes of total NH3 to
the coastal ocean (median of 10.7 |_imol//m2/day). In contrast, heavily polluted air transported from the
industrialized midwestern U.S. contains concentrations of nss S042" and total NH3 that are about a factor
of 3 greater, based on median values. Under these conditions, most total NH3 (> 85%) partitions to the
highly acidic sulfate aerosol size fractions and, consequently, the median dry-deposition flux of total NH3
is 30% lower than that under the cleaner northerly flow regime. The relatively longer atmospheric lifetime
of total NH3 against dry deposition under more polluted conditions implies that, on average, total NH3
would accumulate to higher atmospheric concentrations under these conditions and also be subject to
atmospheric transport over longer distances. Consequently, the importance of NHX removal via wet
deposition would also increase. Because of the inherently sporadic character of precipitation, we might
expect greater heterogeneity in NH3 deposition fields and any potential responses in sensitive ecosystems
downwind of major S-emission regions.
B.2. Transport of SOx in the Atmosphere
Crutzen and Gidel (1983), Gidel (1983), and Chatfield and Crutzen (1984) hypothesized that
convective clouds played an important role in rapid atmospheric vertical transport of trace species and
first tested simple parameterizations of convective transport in atmospheric chemical models. At nearly
the same time, evidence was shown of venting the boundary layer by shallow, fair weather cumulus
clouds (Greenhut et al., 1984; 1986). Field experiments were conducted in 1985 which resulted in
verification of the hypothesis that deep convective clouds are instrumental in atmospheric transport of
trace constituents (Dickerson et al., 1987). Once pollutants are lofted to the middle and upper troposphere,
they typically have a much longer chemical lifetime and with the generally stronger winds at these
altitudes, they can be transported large distances from their source regions.
B-5
-------
B.3. Emissions of SO2
As can be seen from Table B-3, emissions of S02 are due mainly to the combustion of fossil fuels
by electrical utilities and industry. Transportation related sources make only a minor contribution. As a
result, most S02 emissions originate from point sources. Since S is a volatile component of fuels, it is
almost quantitatively released during combustion, and emissions can be calculated on the basis of the
sulfur content of fuels to greater accuracy than for other pollutants such as NOx or primary PM.
Table B-3 Emissions of NOx, NH3, and SO2 in the U.S. by source and category, 2002.
2002 Emissions (Tg/yr)
NOx1
NH32
so2
2002 Emissions (Tg/yr)
NOx1
NH32
so2
Total all Sources
23.19
4.08
16.87
Other
<0.01
<0.01
0.02
Polymer & Resin Mfg
< 0.01
Fuel Combustion Total
9.11
0.02
14.47
Agricultural Chemical Mfg
Ammonium Nitrate/Urea Mfg.
0.05
0.02
<0.01
0.05
Fuel Combustion Electrical Utilities
5.16
<0.01
11.31
Other
0.02
Coal
4.50
<0.01
10.70
Paint, Varnish, Lacquer, Enamel Mfg
0.00
0.00
Bituminous
2.90
8.04
Pharmaceutical Mfg
0.00
0.00
Subbituminous
1.42
2.14
Other Chemical Mfg
0.03
<0.01
0.12
Anthracite & Lignite
0.18
0.51
Metals Processing
0.09
<0.01
0.30
Other
<0.01
Non-Ferrous Metals Processing
0.01
<0.01
0.17
Oil
0.14
<0.01
0.38
Copper
0.04
Residual
0.13
0.36
Lead
0.07
Distillate
0.01
0.01
Zinc
0.01
Gas
0.30
<0.01
0.01
Other
<0.01
Natural
0.29
Ferrous Metals Processing
0.07
<0.01
0.11
Process
0.01
Metals Processing
0.01
<0.01
0.02
Other
0.05
<0.01
0.21
Petroleum & Related Industries
0.16
<0.01
.38
Internal Combustion
0.17
<0.01
0.01
Oil & Gas Production
0.07
<0.01
0.11
Fuel Combustion Industrial
3.15
<0.01
2.53
Natural Gas
0.11
Coal
0.49
<0.01
1.26
Other
0.01
Bituminous
0.25
0.70
Petroleum Refineries & Related Industries
0.05
<0.01
0.26
Subbituminous
0.07
0.10
Fluid Catalytic Cracking Units
<0.01
0.16
Anthracite & Lignite
0.04
0.13
Other
<0.01
0.07
Other
0.13
0.33
Asphalt Manufacturing
0.04
0.01
Oil
0.19
<0.01
0.59
Other Industrial Processes
0.54
0.05
0.46
Residual
0.09
0.40
Agriculture, Food, & Kindred Products
0.01
<0.01
0.01
Distillate
0.09
0.16
Textiles, Leather, & Apparel Products
<0.01
<0.01
<0.01
Other
0.01
<0.01
0.02
Wood, Pulp & Paper, & Publishing Products
0.09
<0.01
0.10
Gas
1.16
0.52
Rubber & Miscellaneous Plastic Products
<0.01
<0.01
<0.01
Natural
0.92
Mineral Products
0.42
<0.01
0.33
Process
0.24
Cement Mfg
0.24
0.19
Other
<0.01
Glass Mfg
0.01
Other
0.16
<0.01
0.15
Other
0.10
0.09
Wood/Bark Waste
0.11
Machinery Products
<0.01
<0.01
<0.01
Liquid Waste
0.01
Electronic Equipment
<0.01
<0.01
<0.01
Other
0.04
Transportation Equipment
<0.01
<0.01
Internal Combustion
1.15
<0.01
0.01
Miscellaneous Industrial Processes
0.01
0.05
0.02
Fuel Combustion Other
0.80
<0.01
0.63
Solvent Utilization
0.01
<0.01
<0.01
Commercial/Institutional Coal
0.04
<0.01
0.16
Degreasing
<0.01
<0.01
<0.01
Commercial/Institutional Oil
0.08
<0.01
0.28
Graphic Arts
<0.01
<0.01
<0.01
Commercial/Institutional Gas
0.25
<0.01
0.02
Dry Cleaning
<0.01
<0.01
<0.01
Misc. Fuel Combustion (Exc. Residential)
0.03
<0.01
0.01
Surface Coating
<0.01
<0.01
<0.01
Residential Wood
0.03
<0.01
Other Industrial
<0.01
<0.01
<0.01
Residential Other
0.36
0.16
Nonindustrial
<0.01
Distillate Oil
0.06
0.15
Solvent Utilization Nec
<0.01
Bituminous/Subbituminous
0.26
<0.01
Storage & Transport
<0.01
<0.01
0.01
Other
0.04
<0.01
Bulk Terminals & Plants
<0.01
<0.01
<0.01
Industrial Process Total
1.10
0.21
1.54
Petroleum & Petroleum Product Storage
<0.01
<0.01
<0.01
Chemical & Allied Product Mfg
0.12
0.02
0.36
Petroleum & Petroleum Product Transport
<0.01
<0.01
<0.01
Organic Chemical Mfg
0.02
<0.01
0.01
Service Stations: Stage II
<0.01
<0.01
Inorganic Chemical Mfg
0.01
<0.01
0.18
Organic Chemical Storage
<0.01
<0.01
<0.01
Sulfur Compounds
0.17
B-6
-------
2002 Emissions (Tg/yr)
NOx1
NH32
SO2
2002 Emissions (Tg/yr)
NOx1
NH32
SO2
Organic Chemical Transport
0.01
<0.01
Light Commercial
0.04
Inorganic Chemical Storage
<0.01
<0.01
<0.01
Logging
<0.01
Inorganic Chemical Transport
<0.01
<0.01
Airport Service
<0.01
Bulk Materials Storage
0.01
<0.01
<0.01
Railway Maintenance
<0.01
Waste Disposal & Recycling
0.17
0.14
0.03
Recreational Marine Vessels
0.05
Incineration
0.06
<0.01
0.02
Non-Road Diesel
1.76
<0.01
0.22
Industrial
Recreational
0.00
Other
<0.01
Construction
0.84
Open Burning
0.10
<0.01
<0.01
Industrial
0.15
Industrial
<0.01
Lawn & Garden
0.05
Land Clearing Debris
Farm
0.57
Other
<0.01
Light Commercial
0.08
Public Operating Treatment Works
<0.01
0.14
<0.01
Logging
0.02
Industrial Waste Water
<0.01
<0.01
<0.01
Airport Service
0.01
Treatment, Storage, And Disposal Facility
<0.01
<0.01
<0.01
Railway Maintenance
<0.01
Landfills
<0.01
<0.01
<0.01
Recreational Marine Vessels
0.03
Industrial
<0.01
Aircraft
0.09
0.01
Other
<0.01
Marine Vessels
1.11
0.18
Other
<0.01
<0.01
<0.01
Diesel
1.11
Transportation Total
12.58
0.32
0.76
Residual Oil
Other
Highway Vehicles
8.09
0.32
0.30
Railroads
0.98
0.05
Light-Duty Gas Vehicles & Motorcycles
2.38
0.20
0.10
Other
0.32
<0.01
0.00
Light-Duty Gas Vehicles
2.36
0.10
Liquefied Petroleum Gas
0.29
Motorcycles
0.02
0.00
Compressed Natural Gas
0.04
Light-Duty Gas Trucks
1.54
0.10
0.07
Miscellaneous
0.39
3.53
0.10
Light-Duty Gas Trucks 1
1.07
0.05
Agriculture & Forestry
<0.01
3.45
<0.01
Light-Duty Gas Trucks 2
0.47
0.02
Agricultural Crops
<0.01
Heavy-Duty Gas Vehicles
0.44
<0.01
0.01
Agricultural Livestock
2.66
Diesels
3.73
<0.01
0.12
Other Combustion
0.08
0.10
Heavy-Duty Diesel Vehicles
3.71
Health Services
Light-Duty Diesel Trucks
0.01
Cooling Towers
Light-Duty Diesel Vehicles
0.01
Fugitive Dust
Off-Highway
4.49
<0.01
0.46
Other
Non-Road Gasoline
0.23
<0.01
0.01
Natural Sources
3.10
0.03
Recreational
0.01
1 Emissions are expressed in terms of NO2.
Construction
0.01
2 Natural emissions of non-methane volatile organic compounds, carbon monoxide,
Industrial
0.01
and oxides.
Lawn & Garden
0.10
Note: Subcategory values may not sum to category totals due to rounding.
Farm
0.01
Source: (EPA, 2006a)
The major natural sources of S02 are volcanoes, biomass burning, and DMS oxidation over the
oceans. S02 constitutes a relatively minor fraction (0.005% by volume) of volcanic emissions (Holland,
1978). The ratio of H2S to S02 is highly variable in volcanic gases. It is typically much less than one, as
in the Mt. St. Helens' eruption (Turco et al. 1983). However, in addition to being degassed from magma,
H2S can be produced if ground waters, especially those containing organic matter, come into contact with
volcanic gases. In this case, the ratio of H2S to S02 can be greater than one. H2S produced this way would
more likely be emitted through side vents than through eruption columns (Pinto et al., 1989). Primary
particulate S042" is a component of marine aerosol and is also produced by wind erosion of surface soils.
Volcanic sources of S02 in the U.S. are limited to the Pacific Northwest, Alaska, and Hawaii. Since
1980, the Mt. St. Helens volcano in the Washington Cascade Range (46.20°N, 122.18°W, summit 2549 m
asl) has been a variable source of S02. Its major effects came in the explosive eruptions of 1980, which
primarily affected the Northwest. The Augustine volcano near the mouth of the Cook Inlet in
southwestern Alaska (59.363°N, 153.43°W, summit 1252 m asl) has had variable S02 emission since its
last major eruptions in 1986. Volcanoes in the Kamchatka peninsula of eastern Siberia do not significantly
affect surface S02 concentrations in northwestern North America. The most serious effects in the U.S.
from volcanic S02 occur on the island of Hawaii. Nearly continuous venting of S02 from Mauna Loa and
Kilauea produces S02 in such large amounts that >100 km downwind of the island S02 concentrations
can exceed 30 ppb (Thornton and Bandy, 1993). Depending on wind direction, the west coast of Hawaii
B-7
-------
(Kona region) has had significant deleterious effects from S02 and acidic sulfate aerosols for the past
decade.
Emissions of S02 from burning vegetation are generally in the range of 1 to 2% of the biomass
burned (e.g., Levine and Pinto, 1998). Gaseous emissions are mainly in the form of S02 with much
smaller amounts of H2S and OCS. The ratio of gaseous N to S emissions is about 14, very close to their
ratio in plant tissue (Andrea et al., 1991). Gaseous emissions are mainly in the form of S02 with much
smaller amounts of H2S and OCS. The ratio of reduced N and S species such as NH3 and H2S to their
more oxidized forms, such as NO and S02, increases from flaming to smoldering phases of combustion,
as emissions of reduced species are favored by lower temperatures and reduced 02 availability.
Emissions of reduced S species are associated typically with marine organisms living either in
pelagic or coastal zones and with anaerobic bacteria in marshes and estuaries. Mechanisms for their
oxidation were discussed in Section B.l. Emissions of DMS from marine plankton represent the largest
single source of reduced S species to the atmosphere (e.g., Berresheim et al., 1995).
However, it should be noted that reduced S species are also produced by industry. For example,
DMS is used in petroleum refining and in petrochemical production processes to control the formation of
coke and carbon monoxide. In addition, it is used to control dusting in steel mills. It is also used in a
range of organic syntheses. It also has a use as a food flavoring component. It can also be oxidized by
natural or artificial means to dimethyl sulfoxide (DMSO), which has several important solvent properties.
B.4. Methods Used to Calculate S0X and Chemical
Interactions in the Atmosphere
Atmospheric chemistry and transport models are the major tools used to calculate the relations
among 03, other oxidants, and their precursors, the transport and transformation of air toxics, the
production of secondary organic aerosol, the evolution of the particle size distribution, and the production
and deposition of pollutants affecting ecosystems. Chemical transport models are driven by emissions
inventories for primary species such as the precursors for 03 and PM and by meteorological fields
produced by other numerical models. Emissions of precursor compounds can be divided into
anthropogenic and natural source categories. Natural sources can be further divided into biotic
(vegetation, microbes, animals) and abiotic (biomass burning, lightning) categories. However, the
distinction between natural sources and anthropogenic sources is often difficult to make as human
activities affect directly, or indirectly, emissions from what would have been considered natural sources
during the preindustrial era. Emissions from plants and animals used in agriculture have been referred to
as anthropogenic or natural in different applications. Wildfire emissions may be considered to be natural,
except that forest management practices may have led to the buildup of fuels on the forest floor, thereby
altering the frequency and severity of forest fires. Needed meteorological quantities such as winds and
temperatures are taken from operational analyses, reanalyses, or circulation models. In most cases, these
are off-line analyses, i.e., they are not modified by radiatively active species such as 03 and particles
generated by the model.
A brief overview of atmospheric chemistry-transport models is given in Section B.5. Uncertainties
in emissions estimates have also been discussed in AQCD for PM (EPA, 1996). Chemistry-transport
model evaluation and an evaluation of the reliability of emissions inventories are also presented in
Section B.5.
B-8
-------
B.5. Chemical-transport Models
Atmospheric chemical transport models (CTMs) have been developed for application over a wide
range of spatial scales ranging from neighborhood to global. Regional scale CTMs are used: (1) to obtain
better understanding of the processes controlling the formation, transport, and destruction of gas and
particle phase criteria and hazardous air pollutants; (2) to understand the relations between 03
concentrations and concentrations of its precursors such as NOx and VOCs, the factors leading to acid
deposition, and hence to possible damage to ecosystems; and (3) to understand relations among the
concentration patterns of various pollutants that may exert adverse health effects. CTMs are also used for
determining control strategies for 03 precursors. However, this application has met with varying degrees
of success because of the highly nonlinear relations between 03 and emissions of its precursors, and
uncertainties in emissions, parameterizations of transport, and chemical production and loss terms.
Uncertainties in meteorological variables and emissions can be large enough to lead to significant errors
in developing control strategies (e.g., Russell and Dennis, 2000; Sillman, 1995).
Global scale CTMs are used to address issues associated with climate change, stratospheric 03
depletion, and to provide boundary conditions for regional scale models. CTMs include mathematical
(and often simplified) descriptions of atmospheric transport, the transfer of solar radiation through the
atmosphere, chemical reactions, and removal to the surface by turbulent motions and precipitation for
pollutants emitted into the model domain. Their upper boundaries extend anywhere from the top of the
mixing layer to the mesopause (about 80 km in height), to obtain more realistic boundary conditions for
problems involving stratospheric dynamics. There is a trade-off between the size of the modeling domain
and the grid resolution used in the CTM that is imposed by computational resources.
There are two major formulations of CTMs in current use. In the first approach, grid-based, or
Eulerian, air quality models, the region to be modeled (the modeling domain) is subdivided into a three-
dimensional array of grid cells. Spatial derivatives in the species continuity equations are cast in terms of
finite-differences. A system of equations for the concentrations of all the chemical species in the model
are solved numerically at each grid point. Time dependent continuity (mass conservation) equations are
solved for each species including terms for transport, chemical production and destruction, and emissions
and deposition (if relevant), in each cell. Chemical processes are simulated with ordinary differential
equations, and transport processes are simulated with partial differential equations. Because of a number
of factors such as the different time scales inherent in different processes, the coupled, nonlinear nature of
the chemical process terms, and computer storage limitations, all of the terms in the equations are not
solved simultaneously in three dimensions. Instead, operator splitting, in which terms in the continuity
equation involving individual processes are solved sequentially, is used. In the second CTM formulation,
trajectory or Lagrangian models, a large number of hypothetical air parcels are specified as following
wind trajectories. In these models, the original system of partial differential equations is transformed into
a system of ordinary differential equations.
A less common approach is to use a hybrid Lagrangian/Eulerian model, in which certain aspects of
atmospheric chemistry and transport are treated with a Lagrangian approach and others are treaded in an
Eulerian manner (Stein et al., 2000). Each approach has its advantages and disadvantages. The Eulerian
approach is more general in that it includes processes that mix air parcels and allows integrations to be
carried out for long periods during which individual air parcels lose their identity. There are, however,
techniques for including the effects of mixing in Lagrangian models such as FLEXPART (e.g., Zanis et
al., 2003), ATTILA (Reithmeier and Sausen, 2002), and CLaMS (McKenna et al., 2002).
B.5.1. Regional Scale Chemical-Transport Models
Major modeling efforts within the EPA center on the Community Multiscale Air Quality modeling
system (CMAQ) (Byun and Ching, 1999; Byun and Schere, 2006). A number of other modeling platforms
B-9
-------
using Lagrangian and Eulerian frameworks have been reviewed in the 1996 AQCD for 03 (U.S. EPA,
1997) and in Russell and Dennis (2000). The capabilities of a number of CTMs designed to study local -
and regional-scale air pollution problems are summarized by Russell and Dennis (2000). Evaluations of
the performance of CMAQ are given in Arnold et al. (2003), Eder and Yu (2006), and Fuentes and Raftery
(2005). The domain of CMAQ can extend from several hundred km to the hemispherical scale. In
addition, both of these classes of models allow the resolution of the calculations over specified areas to
vary. CMAQ is most often driven by the MM5 mesoscale meteorological model (Seaman, 2000), though
it may be driven by other meteorological models (e.g., WRF, RAMS). Simulations of 03 episodes over
regional domains have been performed with a horizontal resolution as low as 1 km, and smaller
calculations over limited domains have been accomplished at even finer scales. However, simulations at
such high resolutions require better parameterizations of meteorological processes such as boundary layer
fluxes, deep convection and clouds (Seaman, 2000), and finer-scale emissions. Finer spatial resolution is
necessary to resolve features such as urban heat island circulations; sea, bay, and land breezes; mountain
and valley breezes, and the nocturnal low-level jet.
The most common approach to setting up the horizontal domain is to nest a finer grid within a
larger domain of coarser resolution. However, there are other strategies such as the stretched grid (e.g.,
Fox-Rabinovitz et al., 2002) and the adaptive grid. In a stretched grid, the grid's resolution continuously
varies throughout the domain, thereby eliminating any potential problems with the sudden change from
one resolution to another at the boundary. Caution should be exercised in using such a formulation,
because certain parameterizations that are valid on a relatively coarse grid scale (such as convection) may
not be valid on finer scales. Adaptive grids are not fixed at the start of the simulation, but instead adapt to
the needs of the simulation as it evolves. They have the advantage that they can resolve processes at
relevant spatial scales. However, they can be very slow if the situation to be modeled is complex.
Additionally, if adaptive grids are used for separate meteorological, emissions, and photochemical
models, there is no reason a priori why the resolution of each grid should match, and the gains realized
from increased resolution in one model will be wasted in the transition to another model. The use of finer
horizontal resolution in CTMs will necessitate finer-scale inventories of land use and better knowledge of
the exact paths of roads, locations of factories, and, in general, better methods for locating sources and
estimating their emissions.
The vertical resolution of these CTMs is variable, and usually configured to have higher resolution
near the surface and decreasing aloft. Because the height of the boundary layer is of critical importance in
simulations of air quality, improved resolution of the boundary layer height would likely improve air
quality simulations. Additionally, current CTMs do not adequately resolve fine scale features such as the
nocturnal low-level jet.
CTMs require time-dependent, three-dimensional wind fields for the period of simulation. The
winds may be either generated by a model using initial fields alone or with four-dimensional data
assimilation to improve the model's performance; fields (i.e., model equations can be updated periodically
or "nudged," to bring results into agreement with observations. Modeling efforts typically focus on
simulations of several days' duration, the typical time scale for individual 03 episodes, but there have
been several attempts at modeling longer periods. For example, Kasibhatla and Chameides (2000)
simulated a four-month period from May to September of 1995 using MAQSIR The current trend in
modeling applications is towards annual simulations. This trend is driven in part by the need to better
understand observations of periods of high wintertime PM (e.g., Blanchard et al., 2002) and the need to
simulate 03 episodes occurring outside of summer.
Chemical kinetics mechanisms (a set of chemical reactions) representing the important reactions
occurring in the atmosphere are used in CTMs to estimate the rates of chemical formation and destruction
of each pollutant simulated as a function of time. Unfortunately, chemical mechanisms that explicitly treat
the reactions of each individual reactive species are too computationally demanding to be incorporated
into CTMs. For example, a master chemical mechanism includes approximately 10,500 reactions
involving 3603 chemical species (Jenkin et al., 2003). Instead, "lumped" mechanisms, that group
compounds of similar chemistry together, are used. The chemical mechanisms used in existing
B-10
-------
photochemical 03 models contain significant uncertainties that may limit the accuracy of their
predictions; the accuracy of each of these mechanisms is also limited by missing chemistry. Because of
different approaches to the lumping of organic compounds into surrogate groups, chemical mechanisms
can produce somewhat different results under similar conditions. The CB-IV chemical mechanism (Gery
et al., 1989), the RADMII mechanism (Stockwell et al., 1990), the SAPRC (e.g., (Carter, 1990; Wang et
al., 2000a; b) and the RACM mechanism (Stockwell et al. 1997) can be used in CMAQ. Jimenez et al.
(2003b) provide brief descriptions of the features of the main mechanisms in use and they compared
concentrations of several key species predicted by seven chemical mechanisms in a box model simulation
over 24 h. The avg deviation from the avg of all mechanism predictions for 03 and NO over the daylight
period was less than 20%, and was 10% for N02 for all mechanisms. However, much larger deviations
were found for HN03, PAN, H02, H202, C2H4, and C5H8 (isoprene). An analysis for OH radicals was not
presented. The large deviations shown for most species imply differences between the calculated lifetimes
of atmospheric species and the assignment of model simulations to either NOx-limited or radical quantity
limited regimes between mechanisms. Gross and Stockwell (2003) found small differences between
mechanisms for clean conditions, with differences becoming more significant for polluted conditions,
especially for N02 and organic peroxy radicals. They caution modelers to consider carefully the
mechanisms they are using. Faraji et al. (2008) found differences of 40% in peak 1-h 03 in the Houston-
Galveston-Brazoria area between simulations using SAPRAC and CB4. They attributed differences in
predicted 03 concentrations to differences in the mechanisms of oxidation of aromatic hydrocarbons.
CMAQ and other CTMs (e.g., PM-CAMX) incorporate processes and interactions of aerosol-phase
chemistry (Mebust et al., 2003). There have also been several attempts to study the feedbacks of
chemistry on atmospheric dynamics using meteorological models, like MM5 (e.g., Grell et al., 2000; Liu
et al., 2001a; Liu et al., 2001b; Lu et al., 1997; Park et al., 2001b). This coupling is necessary to simulate
accurately feedbacks such as may be caused by the heavy aerosol loading found in forest fire plumes (Lu
et al., 1997; Park et al., 2001b), or in heavily polluted areas. Photolysis rates in CMAQ can now be
calculated interactively with model produced 03, N02, and aerosol fields (Binkowski et al., 2007).
Spatial and temporal characterizations of anthropogenic and biogenic precursor emissions must be
specified as inputs to a CTM. Emissions inventories have been compiled on grids of varying resolution
for many hydrocarbons, aldehydes, ketones, CO, NH3, and NOx. Emissions inventories for many species
require the application of some algorithm for calculating the dependence of emissions on physical
variables such as temperature and to convert the inventories into formatted emission files required by a
CTM. For example, preprocessing of emissions data for CMAQ is done by the Spare-Matrix Operator
Kernel Emissions (SMOKE) system. For many species, information concerning the temporal variability
of emissions is lacking, so long-term (e.g., annual or 03-season) averages are used in short-term, episodic
simulations. Annual emissions estimates are often modified by the emissions model to produce emissions
more characteristic of the time of day and season. Significant errors in emissions can occur if an
inappropriate time dependence or a default profile is used. Additional complexity arises in model
calculations because different chemical mechanisms are based on different species, and inventories
constructed for use with another mechanism must be adjusted to reflect these differences. This problem
also complicates comparisons of the outputs of these models because one chemical mechanism may
produce some species not present in another mechanism yet neither may agree with the measurements.
In addition to wet deposition, dry deposition (the removal of chemical species from the atmosphere
by interaction with ground-level surfaces) is an important removal process for pollutants on both urban
and regional scales and must be included in CTMs. The general approach used in most models is the
resistance in series method, in which where dry deposition is parameterized with a Vd , which is
represented as Vd = (ra + rb + rc)_1 where ra, rb, and rc represent the resistance due to atmospheric
turbulence, transport in the fluid sublayer very near the elements of surface such as leaves or soil, and the
resistance to uptake of the surface itself. This approach works for a range of substances, although it is
inappropriate for species with substantial emissions from the surface or for species whose deposition to
the surface depends on its concentration at the surface itself. The approach is also modified somewhat for
aerosols: the terms rb and rc are replaced with a surface vd to account for gravitational settling. In their
B-11
-------
review, Wesely and Hicks (2000) point out several shortcomings of current knowledge of dry deposition.
Among those shortcomings are difficulties in representing dry deposition over varying terrain where
horizontal advection plays a significant role in determining the magnitude of ra and difficulties in
adequately determining a Vd for extremely stable conditions such as those occurring at night (e.g., Mahrt,
1998). Under the best of conditions, when a model is exercised over a relatively small area where dry
deposition measurements have been made, models still commonly show uncertainties at least as large as
± 30% (e.g., Brook et al., 1996; Massman et al., 1994; Padro, 1996). Wesely and Hicks (2000) state that
an important result of these comparisons is that the current level of sophistication of most dry deposition
models is relatively low, and that deposition estimates therefore must rely heavily on empirical data. Still
larger uncertainties exist when the surface features in the built environment are not well known or when
the surface comprises a patchwork of different surface types, as is common in the eastern U.S..
The initial conditions, i.e., the concentration fields of all species computed by a model, and the
boundary conditions, i.e., the concentrations of species along the horizontal and upper boundaries of the
model domain throughout the simulation must be specified at the beginning of the simulation. It would be
best to specify initial and boundary conditions according to observations. However, data for vertical
profiles of most species of interest are sparse. The results of model simulations over larger, preferably
global, domains can also be used. As may be expected, the influence of boundary conditions depends on
the lifetime of the species under consideration and the time scales for transport from the boundaries to the
interior of the model domain (Liu et al., 2001a; Liu et al., 2001b).
Each of the model components described above has an associated uncertainty, and the relative
importance of these uncertainties varies with the modeling application. The largest errors in
photochemical modeling are still thought to arise from the meteorological and emissions inputs to the
model (Russell and Dennis, 2000). Within the model itself, horizontal advection algorithms are still
thought to be significant source of uncertainty (e.g., Chock and Winkler, 1994), though more recently,
those errors are thought to have been reduced (e.g., Odman and Ingram, 1996). There are also indications
that problems with mass conservation continue to be present in photochemical and meteorological models
(e.g.,(Odman and Russel, 1999); these can result in significant simulation errors. The effects of errors in
initial conditions can be minimized by including several days "spin-up" time in a simulation to allow the
model to be driven by emitted species before the simulation of the period of interest begins.
While the effects of poorly specified boundary conditions propagate through the model's domain,
the effects of these errors remain undetermined. Because many meteorological processes occur on spatial
scales which are smaller than the model grid spacing (either horizontally or vertically) and thus are not
calculated explicitly, parameterizations of these processes must be used and these introduce additional
uncertainty.
Uncertainty also arises in modeling the chemistry of 03 formation because it is highly nonlinear
with respect to NOx concentrations. Thus, the volume of the grid cell into which emissions are injected is
important because the nature of 03 chemistry (i.e., 03 production or titration) depends in a complicated
way on the concentrations of the precursors and the OH radical as noted earlier. The use of ever-finer grid
spacing allows regions of 03 titration to be more clearly separated from regions of 03 production. The use
of grid spacing fine enough to resolve the chemistry in individual power-plant plumes is too demanding
of computer resources for this to be attempted in most simulations. Instead, parameterizations of the
effects of sub-grid-scale processes such as these must be developed; otherwise serious errors can result if
emissions are allowed to mix through an excessively large grid volume before the chemistry step in a
model calculation is performed. In light of the significant differences between atmospheric chemistry
taking place inside and outside of a power plant plume (Ryerson et al., 1998; Sillman, 2000), inclusion of
a separate, meteorological module for treating large, tight plumes is necessary. Because the
photochemistry of 03 and many other atmospheric species is nonlinear, emissions correctly modeled in a
tight plume may be incorrectly modeled in a more dilute plume. Fortunately, it appears that the chemical
mechanism used to follow a plume's development need not be as detailed as that used to simulate the rest
of the domain, as the inorganic reactions are the most important in the plume see (e.g., Kumar and
Russell, 1996). The need to include explicitly plume-in-grid chemistry only down to the level of the
B-12
-------
smallest grid disappears if one uses the adaptive grid approach mentioned previously, though such grids
are more computationally intensive. The differences in simulations are significant because they can lead
to significant differences in the calculated sensitivity of 03 to its precursors (e.g., Sillman, 1995).
Because the chemical production and loss terms in the continuity equations for individual species
are coupled, the chemical calculations must be performed iteratively until calculated concentrations
converge to within some preset criterion. The number of iterations and the convergence criteria chosen
also can introduce error.
B.5.2. Intra-urban Scale Dispersion Modeling
The grid spacing for regional applications of air quality CTMs, typically between 1 and 12 km2, is
usually too coarse to resolve spatial variations on urban neighborhood scales. The link between regional
scale models and the population exposure models described in Annex C.2 is made with the classical
dispersion models. Several models could be used to simulate concentration fields near local sources, each
with its own set of strengths and weaknesses. For example, AERMOD is a steady-state plume model
formulated by a committee of the American Meteorological Society and EPA as a replacement to the ISC3
dispersion model. (Technical information on the current version of AERMOD, including model
evaluation databases, can be found at http://www.epa.gov/scram001/dispersion_prefrec.htm.) In the stable
boundary layer (SBL), steady-state dispersion models like AERMOD assume the distributions of
pollutant concentrations to be Gaussian in both the vertical and horizontal dimensions. In the convective
boundary layer, the horizontal distribution is also assumed to be Gaussian, but the vertical distribution is
described with a bi-Gaussian probability density function (pdf). Dispersion of emissions from line sources
like highways is treated as the sum of emissions from a series of contiguous rectangular area or volume
sources. AERMOD has provisions to be applied to flat and complex terrain, and with multiple source
types including point, area, and volume sources in both urban and rural settings with air dispersion
algorithms based on planetary boundary layer turbulence structure and scaling concepts. A large number
of applications have been carried out and evaluated for both surface and elevated sources, and simple and
complex terrain in rural and urban areas. However, AERMOD like similar models was not designed to
treat the formation and dispersal of secondary products.
B.5.3. Global-scale CTMs
The importance of global transport of 03 and 03 precursors and their contribution to regional 03
levels in the U.S. is slowly becoming apparent. There are presently on the order of 20 three-dimensional
global models that have been developed by various groups to address problems in tropospheric chemistry.
These models resolve synoptic meteorology, 03-NOx-CO-hydrocarbon photochemistry, have
parameterizations for wet and dry deposition, and parameterize sub-grid scale vertical mixing processes
such as convection. Global models have proven useful for testing and advancing scientific understanding
beyond what is possible with observations alone. For example, they can calculate quantities of interest
that cannot be measured directly, such as the export of pollution from one continent to the global
atmosphere or the response of the atmosphere to future perturbations to anthropogenic emissions.
Global simulations are typically conducted at a horizontal resolution of about 200 km2. Simulations
of the effects of transport from long-range transport link multiple horizontal resolutions from the global to
the local scale. Finer resolution will only improve scientific understanding to the extent that the governing
processes are more accurately described at that scale. Consequently, there is a critical need for
observations at the appropriate scales to evaluate the scientific understanding represented by the models.
B-13
-------
During the recent IPCC-AR4 tropospheric chemistry study coordinated by the European Union
project Atmospheric Composition Change: the European Network of excellence (ACCENT), 26
atmospheric CTMs were used to estimate the impacts of three emissions scenarios on global atmospheric
composition, climate, and air quality in 2030 (Dentener et al., 2006a). All models were required to use
anthropogenic emissions developed at IIASA (Dentener et al., 2005) and GFED version 1 biomass
burning emissions (Van der Werf et al., 2003) as described in Stevenson et al. (2006). The base
simulations from these models were evaluated against a suite of present-day observations. Most relevant
to this assessment report are the evaluations with O , and N02, and for N and sulfur deposition (Dentener
et al., 2006a; Stevenson et al., 2006; van Noije et al., 2006); see Figure B-3, sulfate deposition.
1000
800 -
-------
B.5.5. CTM Evaluation
The comparison of model predictions with ambient measurements represents a critical task for
establishing the accuracy of photochemical models and evaluating their ability to serve as the basis for
making effective control strategy decisions. The evaluation of a model's performance, or its adequacy to
perform the tasks for which it was designed can only be conducted within the context of measurement
errors and artifacts. Not only are there analytical problems, but there are also problems in assessing the
representativeness of monitors at ground level for comparison with model values which represent
typically an average over the volume of a grid box.
Evaluations of CMAQ are given in Arnold et al. (2003) and Fuentes and Raftery (2005).
Discrepancies between model predictions and observations can be used to point out gaps in current
understanding of atmospheric chemistry and to spur improvements in parameterizations of atmospheric
chemical and physical processes. Model evaluation does not merely involve a straightforward comparison
between model predictions and the concentration field of the pollutant of interest. Such comparisons may
not be meaningful because it is difficult to determine if agreement between model predictions and
observations truly represents an accurate treatment of physical and chemical processes in the CTM or the
effects of compensating errors in complex model routines. Ideally, each of the model components
(emissions inventories, chemical mechanism, meteorological driver) should be evaluated individually.
However, this is rarely done in practice.
B.6. Sampling and Analysis of SOx
B.6.1. Sampling and Analysis for SO2
S02 molecules absorb ultraviolet (UV) light at one wavelength and emit UV light at longer
wavelengths. This process is known as fluorescence, and involves the excitation of the S02 molecule to a
higher energy (singlet) electronic state. Once excited, the molecule decays non-radiatively to a lower
energy electronic state from which it then decays to the original, or ground, electronic state by emitting a
photon of light at a longer wavelength (i.e., lower energy) than the original, incident photon. The process
can be summarized by the following equations
S02 + hvj -^S02*
S02* -» S02 + hv2
where S02* represents the excited state of S02, hvu and hv2 represent the energy of the excitation and
fluorescence photons, respectively, and hv2 < hvi. The intensity of the emitted light is proportional to the
number of S02 molecules in the sample gas.
In commercial analyzers, light from a high intensity UV lamp passes through a bandwidth filter,
allowing only photons with wavelengths around the S02 absorption peak (near 214 nm) to enter the
optical chamber. The light passing through the source bandwidth filter is collimated (to improve image
quality) using a UV lens and passes through the optical chamber, where it is detected on the opposite side
of the chamber by the reference detector. A photomultiplier tube (PMT) is offset from and placed
perpendicular to the light path to detect the S02 fluorescence. Since the S02 fluorescence (330 nm) is at a
wavelength that is different from the excitation wavelength, an optical bandwidth filter is placed in front
of the PMT to filter out any stray light from the UV lamp. A lens is located between the filter and the
B-15
-------
PMT to focus the fluorescence onto the active area of the detector and optimize the fluorescence signal.
The LOD for a non-trace level S02 analyzer is 10 ppb (CFR, 2006). The S02 measurement method is
subject to both positive and negative interference.
B.6.1.1. Other Techniques for Measuring SO2
A more sensitive S02 measurement method than the UV-fluorescence method was reported by
Thornton et al. (2002), who reported use of an atmospheric pressure ionization mass spectrometer. The
high measurement precision and instrument sensitivity were achieved by adding isotopically labeled S02
(34S1602) continuously to the manifold as an internal standard. Field studies showed that the method
precision was better than 10% and the limit of detection was less than 1 ppt for a sampling interval of Is.
S02 can be measured by LIF at around 220 nm (Matsumi et al., 2005). Because the laser
wavelength is alternately tuned to an S02 absorption peak at 220.6 and trough at 220.2 nm, and the
difference signal at the two wavelengths is used to extract the S02 concentration, the technique eliminates
interference from either absorption or fluorescence by other species and has high sensitivity (5 ppt in 60
sec). S02 can also be measured by the same DOAS instrument that can measure N02.
Photoacoustic techniques have been employed for S02 detection, but they generally have detection
limits suitable only for source monitoring (Gondal, 1997; Gondal and Mastromarino, 2001).
Chemical Ionization Mass Spectroscopy (CIMS) utilizes ionization via chemical reactions in the
gas phase to determine an unknown sample's mass spectrum and identity. High sensitivity (10 ppt or
better) has been achieved with uncertainty of -15% when a charcoal scrubber is used for zeroing and the
sensitivity is measured with isotopically labeled 34S02 (Hanke et al., 2003; Hennigan et al., 2006; Huey et
al., 2004).
B.6.2. Sampling and Analysis for SO42-, NO3, and NH4+
Sulfate is commonly present in PM2 5. Most PM2 5 samplers have a size-separation device to
separate particles so that only those particles approximately 2.5 (.im or less are collected on the sample
filter. Air is drawn through the sample filter at a controlled flow rate by a pump located downstream of the
sample filter. The systems have two critical flow rate components for the capture of fine particulate: (1)
the flow of air through the sampler must be at a flow rate that ensures that the size cut at 2.5 |_im occurs;
and (2) the flow rate must be optimized to capture the desired amount of particulate loading with respect
to the analytical method detection limits.
When using the system described above to collect S042" sampling artifacts can occur because of:
(1) positive sampling artifact for S042, nitrate (N03), and particulate ammonium due to chemical
reaction; and (2) negative sampling artifact for N03 and ammonium (NH/) due to the decomposition and
evaporation.
There are two major PM speciation ambient air-monitoring networks in the U.S.: the Speciation
Trend Network (STN), and the Interagency Monitoring of Protected Visual Environments (IMPROVE)
network. The current STN samplers include three filters: (1) Teflon for equilibrated mass and elemental
analysis including elemental sulfur; (2) a HN03 denuded nylon filter for ion analysis including N03 and
S042, (3) a quartz-fiber filter for elemental and organic carbon. The IMPROVE sampler, which collects
two 24-h samples per week, simultaneously collects one sample of PMi0 on a Teflon filter, and three
samples of PM2 5 on Teflon, nylon, and quartz filters. PM2 5 mass concentrations are determined
gravimetrically from the PM2 5 Teflon filter sample. The PM2 5 Teflon filter sample is also used to
determine concentrations of selected elements. The PM2 5 nylon filter sample, which is preceded by a
denuder to remove acidic gases, is analyzed to determine N03 and sulfate aerosol concentrations. Finally,
the PM2 5 quartz filter sample is analyzed for OC and EC using the thermal-optical reflectance (TOR)
B-16
-------
method. The STN and the IMPROVE networks represent a major advance in the measurement of N03,
because the combination of a denuder (coated with either sodium carbonate [Na2C03] or MgO) to remove
HN03 vapor and a Nylon filter to adsorb HN03 vapor volatilizing from the collected ammonium nitrate
particles overcomes the loss of N03 from Teflon filters.
The extent to which sampling artifacts for particulate NH/ have been adequately addressed in the
current networks is not clear. Recently, new denuder-filter sampling systems have been developed to
measure S042, N03, and NH/ with an adequate correction of NH/ sampling artifacts. The denuder-filter
system, Chembcomb Model 3500 speciation sampling cartridge developed by Rupprecht & Patashnick
Co, Inc. could be used to collect N03, S042, and NH/ simultaneously. The sampling system contains a
single-nozzle size-selective inlet, two honeycomb denuders, the aerosol filter and two backup filters
(Keck and Wittmaack, 2005). The first denuder in the system is coated with 0.5% Na2C03 and 1%
glycerol and collects acid gases such as HC1, S02, HONO, and HN03. The second denuder is coated with
0.5% phosphoric acid in methanol for collecting NH3. Backup filters collect the gases behind denuded
filters. The backup filters are coated with the same solutions as the denuders. A similar system based on
the same principle was applied by Possanzini et al. (1999). The system contains two NaCl-coated annular
denuders followed by other two denuders coated with Na2C03/glycerol and citric acid, respectively. This
configuration was adopted to remove HN03 quantitatively on the first NaCl denuder. The third and fourth
denuders remove S02 and NH3, respectively. A polyethylene cyclone and a two-stage filter holder
containing three filters is placed downstream of the denuders. Aerosol fine particles are collected on a
Teflon membrane. A backup nylon filter and a subsequent citric acid impregnated filter paper collect
dissociation products (HN03 and NH3) of ammonium nitrate evaporated from the filtered particulate
matter.
Several traditional and new methods could be used to quantify elemental S collected on filters:
energy dispersive X-ray fluorescence, synchrotron induced X-ray fluorescence, proton induced X-ray
emission (PIXE), total reflection X-ray fluorescence, and scanning electron microscopy. Energy
dispersive X-ray fluorescence (EDXRF)(Method 10-3.3, EPA, 1997b; see 2004 PM CD for details) and
PIXE are the most commonly used methods. Since sample filters often contain very small amounts of
particle deposits, preference is given to methods that can accommodate small sample sizes and require
little or no sample preparation or operator time after the samples are placed into the analyzer. X-ray
fluorescence (XRF) meets these needs and leaves the sample intact after analysis so it can be submitted
for additional examinations by other methods as needed. To obtain the greatest efficiency and sensitivity,
XRF typically places the filters in a vacuum which may cause volatile compounds (nitrates and organics)
to evaporate. As a result, species that can volatilize such as ammonium nitrate and certain organic
compounds can be lost during the analysis. The effects of this volatilization are important if the PTFE
filter is to be subjected to subsequent analyses of volatile species.
Polyatomic ions such as S042, N03, and NH4+ are quantified by methods such as ion
chromatography (IC) (an alternative method commonly used forNH4+ analysis is automated colorimetry).
All ion analysis methods require a fraction of the filter to be extracted in deionized distilled water for
S042" and Na2C03/NaHC03 solution for N03 and then filtered to remove insoluble residues prior to
analysis. The extraction volume should be as small as possible to avoid over-diluting the solution and
inhibiting the detection of the desired constituents at levels typical of those found in ambient PM2 5
samples. During analysis, the sample extract passes through an ion-exchange column which separates the
ions in time for individual quantification, usually by an electroconductivity detector. The ions are
identified by their elution/retention times and are quantified by the conductivity peak area or peak height.
In a side-by-side comparison of two of the major aerosol monitoring techniques (Hains et al.,
2007), PM2 5 mass and major contributing species were well correlated among the different methods with
correlation coefficients in excess of 0.8. Agreement for mass, S042, OC, TC, and NH4+ was good while
that for N03 and BC was weaker. Based on reported uncertainties, however, even daily concentrations of
PM2 5mass and major contributing species were often significantly different at the 95% confidence level.
Greater values of PM2 5 mass and individual species were generally reported from Speciation Trends
Network methods than from the Desert Research Institute Sequential Filter Samplers. These differences
B-17
-------
can only be partially accounted for by known random errors. The authors concluded that the current
uncertainty estimates used in the STN network may underestimate the actual uncertainty.
The reaction of S02 (and other acid gases) with basic sites on glass fiber filters or with basic coarse
particles on the filter leads to the formation of S042" (or other nonvolatile salts, e.g., N03, chloride). These
positive artifacts lead to the overestimation of total mass, and S042, and probably also N03
concentrations. These problems were largely overcome by changing to quartz fiber or Teflon filters and
by the separate collection of PM25. However, the possible reaction of acidic gases with basic coarse
particles remains a possibility, especially with PMi0 and PMi0_2.5 measurements. These positive artifacts
could be effectively eliminated by removing acidic gases in the sampling line with denuders coated with
NaCl or Na2C03.
Positive sampling artifacts also occur during measurement of particulate NH/. The reaction of NH3
with acidic particles (e.g. 2NH3 + H2S04 —> (NFL^SC^), either during sampling or during transportation,
storage, and equilibration could lead to an overestimation of particulate NH4+ concentrations. Techniques
have been developed to overcome this problem: using a denuder to remove NH3 during sampling and to
protect the collected PM from NH3 (Brauer et al., 1991; Keck and Wittmaack, 2006; 1988b; Koutrakis et
al., 1988a; Possanzini et al., 1999; Suh et al., 1992; Winberry et al., 1999). Hydrogen fluoride, citric acid,
and phosphorous acids have been used as coating materials for the NH3 denuder. Positive artifacts for
particulate NH4 can also be observed during sample handling due to contamination. No chemical
analysis method, no matter how accurate or precise, can adequately represent atmospheric concentrations
if the filters to which these methods are applied are improperly handled. Ammonia is emitted directly
from human sweat, breath and smoking. It can then react with acidic aerosols on the filter to form
ammonium sulfate, ammonium bisulfate and ammonium nitrate if the filter was not properly handled
(Sutton et al., 2000). Therefore, it is important to keep filters away from NH3 sources, such as human
breath, to minimize neutralization of the acidic compounds. Also, when filters are handled, preferably in a
glove box, the analyst should wear gloves that are antistatic and powder-free to act as an effective
contamination barrier.
Continuous methods for the quantification of aerosol S compounds first remove gaseous S (e.g.,
S02, H2S) from the sample stream by a diffusion tube denuder followed by the analysis of particulate S
(Cobourn et al., 1978; Durham et al., 1978; Huntzicker et al., 1978; Tanner et al., 1980). Another
approach is to measure total S and gaseous S separately by alternately removing particles from the sample
stream. Particulate S is obtained as the difference between the total and gaseous S (Kittelson et al., 1978).
The total S content is measured by a flame photometric detector (FPD) by introducing the sampling
stream into a fuel-rich, hydrogen-air flame (e.g., Farwell and Rasmussen, 1976; Stevens et al., 1969) that
reduces sulfur compounds and measures the intensity of the chemiluminescence from electronically
excited sulfur molecules (S2*). Because the formation of S2* requires two S atoms, the intensity of the
chemiluminescence is theoretically proportional to the square of the concentration of molecules that
contain a single S atom. In practice, the exponent is between 1 and 2 and depends on the S compound
being analyzed (Dagnall et al., 1967; Stevens et al., 1971). Calibrations are performed using both particles
and gases as standards. The FPD can also be replaced by a chemiluminescent reaction with 03 that
minimizes the potential for interference and provides a faster response time (Benner and Stedman, 1989;
1990). Capabilities added to the basic system include in situ thermal analysis and sulfuric acid speciation
(Cobourn et al., 1978; Cobourn and Husar, 1982; Huntzicker et al., 1978; Tanner et al., 1980).
Sensitivities for particulate S as low as 0.1 (ig/m3, with time resolution ranging from 1 to 30 min, have
been reported. Continuous measurements of particulate S content have also been obtained by on-line XRF
analysis with resolution of 30 min or less (Jaklevic et al., 1981). During a field-intercomparison study of
five different S instruments, Camp et al. (1982) reported four out of five FPD systems agreed to within
± 5% during a 1-week sampling period.
B-18
-------
Annex C. Modeling Human Exposure
C.1. Introduction
Predictive (or prognostic) exposure modeling studies are assessments that start from emissions and
demographic information and explicitly consider the physical and chemical processes of environmental
and microenvironmental transport and fate, in conjunction with human activities, to estimate inhalation
intake and uptake. This type of comprehensive modeling effort, although conducted for other pollutants,
has not previously been undertaken for SO21. Previous efforts for S02 exposure assessment have relied on
statistical (diagnostic) analyses that have been reported using data obtained in various field exposure
studies. However, existing prognostic modeling systems for the assessment of inhalation exposures can in
principle be directly applied to, or adapted for, S02 studies; specifically, such systems include APEX,
SHEDS, and MENTOR-1A, to be discussed in the following sections. Nevertheless, it should be
mentioned that such applications will be constrained by data limitations, such as the degree of ambient
concentration characterization (e.g., concentrations at the local level) and quantitative information on
indoor sources and sinks.
Predictive models of human exposure to ambient air pollutants such as S02 can be classified and
differentiated based upon a variety of attributes. For example, exposure models can be classified as:
¦ models of potential (typically maximum) outdoor exposure versus models of actual
exposures (the latter including locally modified microenvironmental exposures, both
outdoor and indoor);
¦ Population Based Exposure Models (PBEM) versus Individual Based Exposure Models
(IBEM);
¦ deterministic versus probabilistic (or statistical) exposure models; and
¦ observation-driven versus mechanistic air quality models (see Section C.4 for discussions
about the construction, uses and limitations of this class of mathematical models.
Some points should be made regarding terminology and essential concepts in exposure modeling,
before proceeding to the overview of specific developments reported in the current research literature:
First, it must be understood that there is significant variation in the definitions of many of the terms
used in the exposure modeling literature; indeed, the science of exposure modeling is a rapidly evolving
field and the development of a standard and commonly accepted terminology is an ongoing process (e.g.,
WHO, 2004).
Second, it should also be mentioned that, very often, procedures that are called exposure modeling,
exposure estimation, etc. in the scientific literature, may in fact refer to only a sub-set of the complete set
of steps or components required for a comprehensive exposure assessment. For example, certain self-
identified exposure modeling studies focus solely on refining the sub-regional or local spatio-temporal
dynamics of pollutant concentrations (starting from raw data representing monitor observations or
regional grid-based model estimates). Though not exposure studies per se, such efforts have value and are
included in the discussion of the next sub-section, as they provide potentially useful tools that can be used
in a complete exposure assessment. On the other hand, formulations which are self-identified as exposure
models but actually focus only on ambient air quality predictions, such as chemical-transport models, are
not included in the discussion that follows.
Third, the process of modeling human exposures to ambient pollutants is very often identified
explicitly with population-based modeling, while models describing the specific mechanisms affecting the
1 OAQPS is proposing to apply the APEX exposure model to S02 as part of this NAAQS review.
C-1
-------
exposure of an actual individual (at specific locations) to an air contaminant (or to a group of co-
occurring gas and/or aerosol phase pollutants) are usually associated with studies focusing specifically on
indoor air chemistry modeling.
Finally the concept of microenvironments, introduced in earlier sections of this document, should
be clarified further, as it is critical in developing procedures for exposure modeling. In the past,
microenvironments have typically been defined as individual or aggregate locations (and sometimes even
as activities taking place within a location) where a homogeneous concentration of the pollutant is
encountered. Thus a microenvironment has often been identified with an ideal (i.e. perfectly mixed)
compartment of classical compartmental modeling. More recent and general definitions view the
microenvironment as a control volume, either indoors or outdoors, that can be fully characterized by a set
of either mechanistic or phenomenological governing equations, when appropriate parameters are
available, given necessary initial and boundary conditions. The boundary conditions typically would
reflect interactions with ambient air and with other microenvironments. The parameterizations of the
governing equations generally include the information on attributes of sources and sinks within each
microenvironment. This type of general definition allows for the concentration within a microenvironment
to be non-homogeneous (non-uniform), provided its spatial profile and mixing properties can be fully
predicted or characterized. By adopting this definition, the number of microenvironments used in a study
is kept manageable, but variability in concentrations in each of the microenvironments can still be taken
into account. Microenvironments typically used to determine exposure include indoor residential
microenvironments, other indoor locations (typically occupational microenvironments), outdoors near
roadways, other outdoor locations, and in-vehicles. Outdoor locations near roadways are segregated from
other outdoor locations (and can be further classified into street canyons, vicinities of intersections, etc.)
because emissions from automobiles alter local concentrations significantly compared to background
outdoor levels. Indoor residential microenvironments (kitchen, bedroom, living room, etc. or aggregate
home microenvironment) are typically separated from other indoor locations because of the time spent
there and potential differences between the residential environment and the work/public environment.
Once the actual individual and relevant activities and locations (for Individual Based Modeling), or
the sample population and associated spatial (geographical) domain (for Population Based Modeling)
have been defined along with the temporal framework of the analysis (time period and resolution), the
comprehensive modeling of individual/population exposure to S02 (and related pollutants) will in general
require seven steps (or components, as some of them do not have to be performed in sequence) that are
listed below. This list represents a composite based on approaches and frameworks described in the
literature over the last twenty-five years (WHO). 2005; EPA, 1992; 1997a; Georgopoulos and Lioy, 1994;
Georgopoulos et al., 2005; Georgopoulos and Lioy, 2006; Ott, 1982; Price et al., 2003) as well as on the
structure of various inhalation exposure models that have been used in the past or in current studies to
specifically assess inhalation exposures. Figure C-l, adapted from Georgopoulos et al. (2005),
schematically depicts the sequence of steps summarized here.
1) Estimation of the background or ambient levels of both S02 and related pollutants. This is
done through either (or a combination of):
a) multivariate spatio-temporal analysis of fixed monitor data, or
b) emissions-based, photochemical, air quality modeling (typically with a regional, grid-
based model such as Models-3/CMAQ or CAMX) applied in a coarse resolution mode.
2) Estimation of local outdoor pollutant levels of both S02 and related pollutants. These levels
could typically characterize the ambient air of either an administrative unit (such as a census
tract, a municipality, a county, etc.) or a conveniently defined grid cell of an urban scale air
quality model. Again, this may involve either (or a combination of):
C-2
-------
a) spatio-temporal statistical analysis of monitor data, or
b) application of either an intra-urban scale dispersion model (such as AERMOD or
CALPUFF) or an urban multi-scale, grid based model (such as CMAQ or CAMX) at its
highest resolution (typically around 2-4 km), or
c) correction of the estimates of the regional model using some scheme that adjusts for
observations and/or for subgrid chemistry and mixing processes.
3) Characterization of relevant attributes of the individuals or populations under study
(residence and work locations, occupation, housing data, income, education, age, gender,
race, weight, and other physiological characteristics). For Population Based Exposure
Modeling (PBEM) one can either:
a) select a fixed-size sample population of virtual individuals in a way that statistically
reproduces essential demographics (age, gender, race, occupation, income, education) of
the administrative population unit used in the assessment (e.g., a sample of 500 people is
typically used to represent the demographics of a given census tract, whereas a sample of
about 10,000 may be needed to represent the demographics of a county), or
b) divide the population-of interest into a set of cohorts representing selected subpopulations
where the cohort is defined by characteristics known to influence exposure.
4) Development of activity event (or exposure event) sequences for each member of the sample
population (actual or virtual) or for each cohort for the exposure period. This could utilize:
a) study-specific information, if available
b) existing databases based on composites of questionnaire information from past studies
c) time-activity databases, typically in a format compatible with EPA's Consolidated Human
Activity Database (McCurdy et al., 2000)
5) Estimation of levels and temporal profiles of both S02 and related pollutants in various
outdoor and indoor microenvironments such as street canyons, roadway intersections, parks,
residences, offices, restaurants, vehicles, etc. This is done through either:
a) linear regression of available observational data sets,
b) simple mass balance models (with linear transformation and sinks) over the volume (or a
portion of the volume) of the microenvironment,
c) lumped (nonlinear) gas or gas/aerosol chemistry models, or
d) detailed combined chemistry and Computational Fluid Dynamics modeling.
6) Calculation of appropriate inhalation rates for the members of the sample population,
combining the physiological attributes of the (actual or virtual) study subjects and the
activities pursued during the individual exposure events.
C-3
-------
7) Calculation of target tissue dose through biologically based modeling estimation (specifically,
respiratory dosimetry modeling in the case of S02 and related reactive pollutants) if sufficient
information is available.
i.a. Emissions: NEI (NET, NTI), State; Processing with SMOKE,
EMS-HAP, MOBILE/MOVES, NONROAD, FAAED, BEIS, etc.
i.b. Meteorology: NWS, NCDC; Modeling MM5, RAMS, CALMET
i.e. Land Use/Land Cover, Topography: NLDC (USGS), etc.
1 Estimate background
levels of air pollutants
through
a. multivariate spatio-
temporal analysis of
monitor data
b. emissions-based air
quality modeling
(with regional,
grid-based models:
Models-3/ CMAQ, CAMx
and REMSAD)
3 Develop database of
individual subjects
attributes (residence &
work location, housing
characteristics, age,
gender, race, income, etc.)
a. collect study-specific
information
b. supplement with
available relevant local,
regional, and national
demographic
information
2 Estimate local outdoor
pollutant levels that
characterize the ambient air of
an administrative unit (such as
census tract) or a conveniently
defined grid through
a. spatiotemporal statistical
analysis of monitor data
b. application of urban scale
model at high resolution
c. subgrid (e.g. plume-in-grid)
modeling
d. data/model assimilation
4 Develop activity event
(or exposure event)
sequences for each individual
of the study for the exposure
period
a. collect study-specific
information
b. supplement with other
available data
c. organize time-activity
database in format
compatible with CHAD
ii.a. Emissions: EMS-HAP
ii.b. Local Meteorology - Local
Effects: RAMS, FLUENT
5 Estimate levels and
temporal profiles of
pollutants in various
microenvironments (streets,
residences, offices, restaurants,
vehicles, etc.) through
a. regression of observational
data
b. simple linear mass balance
c. lumped (nonlinear) pE
gas/aerosol chemistry models
d. combined chemistry & CFD
(DNS, LES, RANS) models
6 Calculate appropriate
inhalation rates for the
members of the sample
population, combining the
physiological attributes of the
study subjects and the
activities pursued during the
individual exposure events
Calculate
exposures/
intakes
7 Biologically
based
target tissue
dose modeling
Study-specific survey
(also US Census,
US Housing Survey)
/ Study-specific survey
(or default from
CHAD, NHAPS)
ICRP and Other
Physiological & METS
Databases
Figure C-1. Schematic description of a general framework identifying the processes (steps or
components) involved in assessing inhalation exposures and doses for individuals
and populations. In general terms, existing comprehensive exposure modeling
systems such as SHEDS, APEX, and MENTOR-1A follow this framework.
Implementation of the above framework for comprehensive exposure modeling has benefited
significantly from recent advances and expanded availability of computational technologies such as
Relational Database Management Systems (RDBMS) and Geographic Information Systems (GIS)
(Georgopoulos et al.. 2005; Purushothaman and Georgopoulos, 1997; 1999b; a).
In fact, only relatively recently comprehensive, predictive, inhalation exposure modeling studies
for 03, PM, and various air toxics, have attempted to address/incorporate all the components of the
general framework described here. In practice, the majority of past exposure modeling studies have either
incorporated only subsets of these components or treated some of them in a simplified manner, often
focusing on the importance of specific factors affecting exposure. Of course, depending on the objective
of a particular modeling study, implementation of only a limited number of steps may be necessary. For
example, in a regulatory setting, when comparing the relative effectiveness of emission control strategies,
the focus can be on expected changes in ambient levels (corresponding to those observed at NAAQS
monitors) in relation to the density of nearby populations. The outdoor levels of pollutants, in conjunction
with basic demographic information, can thus be used to calculate upper bounds of population exposures
associated with ambient air (as opposed to total exposures that would include contributions from indoor
C-4
-------
sources) useful in comparing alternative control strategies. Though the metrics derived would not be
quantitative indicators of actual human exposures, they can serve as surrogates of population exposures
associated with outdoor air, and thus aid in regulatory decision making concerning pollutant standards and
in studying the efficacy of emission control strategies. This approach has been used in studies performing
comparative evaluations of regional and local emissions reduction strategies in the eastern U.S. (Foley et
al., 2003; Georgopoulos et al., 1997; Purushothaman and Georgopoulos, 1997).
C.2. Population Exposure Models: Their Evolution and
Current Status
Existing comprehensive inhalation exposure models consider the trajectories of individual human
subjects (actual or virtual), or of appropriately defined cohorts, in space and time as sequences of
exposure events. In these sequences, each event is defined by time, a geographic location, a
microenvironment, and the activity of the subject. EPA offices (OAQPS and NERL) have supported the
most comprehensive efforts in developing models implementing this general concept (see, e.g., Johnson,
2002). These families of models are the result: National Exposure Model and Probabilistic National
Exposure Model (NEM/pNEM, Whitfield et al., 1997) Hazardous Air Pollutant Exposure Model
(HAPEM, Rosenbaum, 2005); Simulation of Human Exposure and Dose System (SHEDS, Burke et al.,
2001); Air Pollutants Exposure Model (APEX, EPA, 2006b); and Modeling Environment for Total Risk
Studies (MENTOR, Georgopoulos et al., 2005; Georgopoulos and Lioy, 2006). European efforts have
produced some formulations with similar general attributes as the above U.S. models but, generally,
involving simplifications in some of their components. Examples of European models addressing
exposures to photochemical oxidants (specifically, 03) include the Air Pollution Exposure Model
(AirPEx, Freijer et al., 1998), which basically replicates the pNEM approach and has been applied to the
Netherlands, and the Air Quality Information System Model (AirQUIS, Clench-Aas et al., 1999).
The NEM/pNEM, SHEDS, APEX, and MENTOR for One-Atmosphere studies (MENTOR-1 A)
families of models provide exposure estimates defined by concentration and breathing rate for each
individual exposure event, and then average these estimates over periods typically ranging from one hour
to one year. These models allow simulation of certain aspects of the variability and uncertainty in the
principal factors affecting exposure. An alternative approach is taken by the HAPEM family of models
that typically provide annual avg exposure estimates based on the quantity of time spent per year in each
combination of geographic locations and microenvironments. The NEM, SHEDS, APEX, and MENTOR-
type models are therefore expected to be more appropriate for pollutants with complex chemistry such as
S02, and could provide useful information for enhancing related health assessments.
More specifically, regarding the consideration of population demographics and activity patterns:
¦ pNEM divides the population of interest into representative cohorts based on the
combinations of demographic characteristics (age, gender, and employment), home/work
district, residential cooking fuel and replicate number, and then assigns an activity diary
record from the CHAD to each cohort according to demographic characteristic, season,
day-type (weekday/weekend) and temperature.
¦ HAPEM6 divides the population of interest into demographic groups based on age, gender
and race, and then for each demographic group/day-type (weekday/weekend) combination,
selects multiple activity patterns randomly (with replacement) from CHAD and combines
them to find the averaged annual time allocations for group members in each census tract
for different day types.
¦ SHEDS, APEX, and MENTOR-1A generate population demographic files, which contain a
user-defined number of person records for each census tract of the population based on
proportions of characteristic variables (age, gender, employment, and housing) obtained
C-5
-------
from the population of interest, and then assign a matching activity diary record from
CHAD to each individual record of the population based on the characteristic variables.
It should be mentioned that, in the formulations of these models, workers may commute
from one census tract to another census tract for work. So, with the specification of
commuting patterns, the variation of exposure concentrations due to commuting between
different census tracts can be captured.
The conceptual approach originated by the SHEDS models was modified and expanded for use in
the development of MENTOR-1 A. Flexibility was incorporated into this modeling system, such as the
option of including detailed indoor chemistry and other relevant microenvironmental processes, and
providing interactive linking with CHAD for consistent definition of population characteristics and
activity events (Georgopoulos et al., 2005).
Table C-1. The Essential Attributes of the pNEM, HAPEM, APEX, SHEDS, and MENTOR-1A
pNEM
HAPEM
APEX
SHEDS
MENTOR-1A
Exposure Estimate
Hourly averaged
Annual averaged
Hourly averaged
Activity event based
Activity event based
Characterization of the
High-End Exposures
Yes
No
Yes
Yes
Yes
Typical Spatial
Scale/Resolution
Urban areas/Census
tract level
Ranging from urban to
national/ Census tract
level
Urban area/Census tract
level
Urban areas/Census
tract level
Multiscale/ Census tract
level
Temporal
Scale/Resolution
Annual /
one hour
Annual /
one hour
Annual /
one hour
Annual /
event based
Annual /
activity event based time
step
Population Activity
Patterns Assembly
Top-down approach
Top-down approach
Bottom-up "person-
oriented" approach
Bottom-up "person-
oriented" approach
Bottom-up "person-
oriented" approach
Microenvironment Non-steady-state and
Concentration Estimation steady-state mass
balance equations (hard-
coded)
Linear relationship
method (hard-coded)
Non-steady-state mass
balance and linear
regression (flexibility of
selecting algorithms)
Steady-state mass
balance equation
(residential) and linear
regression (non-
residential) (hard-coded)
Non-steady-state mass
balance equation with
indoor air chemistry
module or regression
methods (flexibility of
selecting algorithms)
Microenvironmental (ME)
Factors
Random samples from
probability distributions
Random samples from
probability distributions
Random samples from
probability distributions
Random samples from
probability distributions
Random samples from
probability distributions
Specification of Indoor
Source Emissions
Yes (gas-stove, tobacco
smoking)
Available; set to zero in
HAPEM6
Yes (multiple sources
defined by the user)
Yes (gas-stove, tobacco
smoking, other sources)
Yes (multiple sources
defined by the user)
Commuting Patterns
Yes
Yes
Yes
Yes
Yes
Exposure Routes
Inhalation
Inhalation
Inhalation
Inhalation
Multiple (optional)
Potential Dose
Calculation
Yes
No
Yes
Yes
Yes
Physiologically Based
Dose
No
No
No
Yes
Yes
Variability/Uncertainty
Yes
No
Yes
Yes
Yes (Various "Tools")
The essential attributes of the pNEM, HAPEM, APEX, SHEDS, and MENTOR-1 A models are
elaborated in Table C-1.
C-6
-------
NEM/pNEM implementations have been extensively applied to 03 studies in the 1980s and 1990s.
The historical evolution of the pNEM family of models of OAQPS started with the introduction of the
first NEM model in the 1980s (Biller et al., 1981). The first such implementations of pNEM/03 in the
1980s used a regression-based relationship to estimate indoor 03 concentrations from outdoor
concentrations. The second generation of pNEM/03 was developed in 1992 and included a simple mass
balance model to estimate indoor 03 concentrations. A report by Johnson (2002) describes this version of
pNEM/03 and summarizes the results of an initial application of the model to 10 cities. Subsequent
enhancements to pNEM/03 and its input databases included revisions to the methods used to estimate
equivalent ventilation rates, to determine commuting patterns, and to adjust ambient 03 levels to simulate
attainment of proposed NAAQS. During the mid-1990s, the EPA applied updated versions of pNEM/03 to
three different population groups in selected cities: (1) the general population of urban residents, (2)
outdoor workers, and (3) children who tend to spend more time outdoors than the average child. This
version of pNEM/03 used a revised probabilistic mass balance model to determine 03 concentrations over
one-h periods in indoor and in-vehicle microenvironments (Johnson, 2002).
In recent years, pNEM has been replaced by (or "evolved to") the APEX. APEX differs from
earlier pNEM models in that the probabilistic features of the model are incorporated into a Monte Carlo
framework (U.S. EPA, 2006; Langstaff, 2007). Like SHEDS and MENTOR-1A, instead of dividing the
population-of-interest into a set of cohorts, APEX generates individuals as if they were being randomly
sampled from the population. APEX provides each generated individual with a demographic profile that
specifies values for all parameters required by the model. The values are selected from distributions and
databases that are specific to the age, gender, and other specifications stated in the demographic profile.
The EPA has applied APEX to the study of exposures to 03 and other criteria pollutants; APEX can be
modified and used for the estimation of S02 exposures, if required.
Reconfiguration of APEX for use with S02 or other pollutants would require significant literature
review, data analysis, and modeling efforts. Necessary steps include determining spatial scope and
resolution of the model; generating input files for activity data, air quality and temperature data; and
developing definitions for microenvironments and pollutant-microenvironment modeling parameters
(penetration and proximity factors, indoor source emissions rates, decay rates, etc.) (ICF Consulting,
2005). To take full advantage of the probabilistic capabilities of APEX, distributions of model input
parameters should be used wherever possible.
C.3. Ambient Concentrations of SO2 and Related
Air Pollutants
As mentioned earlier, background and regional outdoor concentrations of pollutants over a study
domain may be estimated through emissions-based mechanistic modeling, through ambient data based
modeling, or through a combination of both. Emissions-based models calculate the spatio-temporal fields
of the pollutant concentrations using precursor emissions and meteorological conditions as inputs and
using numerical representations of transformation reactions to drive outputs. The ambient data based
models typically calculate spatial or spatio-temporal distributions of the pollutant through the use of
interpolation schemes, based on either deterministic or stochastic models for allocating monitor station
observations to the nodes of a virtual regular grid covering the region of interest. The geostatistical
technique of kriging provides various standard procedures for generating an interpolated spatial
distribution for a given time, from data at a set of discrete points. Kriging approaches were evaluated by
Georgopoulos et al. (1997) in relation to the calculation of local ambient 03 concentrations for exposure
assessment purposes, using either monitor observations or regional/urban photochemical model outputs. It
was found that kriging is severely limited by the nonstationary character of the concentration patterns of
reactive pollutants; so the advantages of this method in other fields of geophysics do not apply here. The
C-7
-------
above study showed that the appropriate semivariograms had to be hour-specific, complicating the
automated reapplication of any purely spatial interpolation over an extended time period.
Spatio-temporal distributions of pollutant concentrations such as 03, PM, and various air toxics
have alternatively been obtained using methods of the Spatio-Temporal Random Field (STRF) theory
(Christakos and Hristopulos, 1998). The STRF approach interpolates monitor data in both space and time
simultaneously. This method can thus analyze information on temporal trends which cannot be
incorporated directly in purely spatial interpolation methods such as standard kriging. Furthermore, the
STRF method can optimize the use of data which are not uniformly sampled in either space or time. STRF
was further extended within the Bayesian Maximum Entropy (BME) framework and applied to 03
interpolation studies (Christakos and Hristopulos, 1998; Christakos and Kolovos, 1999; Christakos,
2000). It should be noted that these studies formulate an over-arching scheme for linking air quality with
population dose and health effects; however, they are limited by the fact that they do not include any
microenvironmental effects. MENTOR has incorporated STRF/BME methods as one of the steps for
performing a comprehensive analysis of exposure to 03 and PM (Georgopoulos et al., 2005).
The issue of subgrid variability (SGV) from the perspective of interpreting and evaluating the
outcomes of grid-based, multiscale, photochemical air quality simulation models is discussed in Ching et
al. (2006), who suggest a framework that can provide for qualitative judgments on model performance
based on comparing observations to the grid predictions and its SGV distribution. From the perspective of
Population Exposure Modeling, the most feasible/practical approach for treating subgrid variability of
local concentrations is probably through l)the identification and proper characterization of an adequate
number of outdoor microenvironments (potentially related to different types of land use within the urban
area as well as to proximity to different types of roadways) and 2) then, concentrations in these
microenvironments will have to be adjusted from the corresponding local background ambient
concentrations through either regression of empirical data or various types of local atmospheric
dispersion/transformation models. This is discussed further in the next section.
C.4. Characterization of Microenvironmental
Concentrations
Once the background and local ambient spatio-temporal concentration patterns have been derived,
microenvironments that can represent either outdoor or indoor settings when individuals come in contact
with the contaminant of concern (e.g., S02) must be characterized. This process can involve modeling of
various local sources and sinks, and interrelationships between ambient and microenvironmental
concentration levels. Three general approaches have been used in the past to model microenvironmental
concentrations:
¦ Empirical (typically linear regression) fitting of data from studies relating ambient/local
and microenvironmental concentration levels to develop analytical relationships.
¦ Parameterized mass balance modeling over, or within, the volume of the
microenvironment. This type of modeling has ranged from very simple formulations, i.e.
from models assuming ideal (homogeneous) mixing within the microenvironment (or
specified portions of it) and only linear physicochemical transformations (including sources
and sinks), to models incorporating analytical solutions of idealized dispersion
formulations (such as Gaussian plumes), to models that take into account aspects of
complex multiphase chemical and physical interactions and nonidealities in mixing.
¦ Detailed Computational Fluid Dynamics (CFD) modeling of the outdoor or indoor
microenvironment, employing either a Direct Numerical Simulation (DNS) approach, a
Reynolds Averaged Numerical Simulation (RANS) approach, or a Large Eddy Simulation
C-8
-------
(LES) approach, the latter typically for outdoor situations (see, e.g., Chang and Meroney,
2003; Chang, 2006; Milner et al., 2005).
Parameterized mass balance modeling is the approach currently preferred for exposure modeling
for populations. As discussed earlier, the simplest microenvironmental setting corresponds to a
homogeneously mixed compartment, in contact with possibly both outdoor/local environments as well as
other microenvironments. The air quality of this idealized microenvironment is affected mainly by the
following processes:
¦ Transport processes: These can include advection/convection and dispersion that are
affected by local processes and obstacles such as vehicle induced turbulence, street
canyons, building structures, etc.
¦ Sources and sinks: These can include local outdoor emissions, indoor emissions, surface
deposition, etc.
¦ Transformation processes: These can include local outdoor as well as indoor gas and
aerosol phase chemistry, such as formation of secondary organic and inorganic aerosols.
Exposure modeling also requires information on activity patterns to determine time spent in various
microenvironments and estimates of inhalation rates to characterize dose. The next two subsections
describe recent work done in these areas.
C.4.1. Characterization of Activity Events
An important development in inhalation exposure modeling has been the consolidation of existing
information on activity event sequences in the Consolidated Human Activity Database (CHAD)
(McCurdy et al., 2000; McCurdy, 2000). Indeed, most recent exposure models are designed (or have been
re-designed) to obtain such information from CHAD which incorporates 24-h time/activity data
developed from numerous surveys. The surveys include probability-based recall studies conducted by
EPA and the California Air Resources Board, as well as real-time diary studies conducted in individual
U.S. metropolitan areas using both probability-based and volunteer subject panels. All ages of both
genders are represented in CHAD. The data for each subject consist of one or more days of sequential
activities, in which each activity is defined by start time, duration, activity type (140 categories), and
microenvironment classification (110 categories). Activities vary from one min to one h in duration, with
longer activities being subdivided into clock-hour durations to facilitate exposure modeling. A distribution
of values for the ratio of oxygen uptake rate to body mass (referred to as metabolic equivalents or METs)
is provided for each activity type listed in CHAD. The forms and parameters of these distributions were
determined through an extensive review of the exercise and nutrition literature. The primary source of
distributional data was Ainsworth et al. (1993), a compendium developed specifically to facilitate the
coding of physical activities and to promote comparability across studies.
C.4.2. Characterization of Inhalation Intake and Uptake
Use of the information in CHAD provides a rational way for incorporating realistic intakes into
exposure models by linking inhalation rates to activity information. As mentioned earlier, each cohort of
the pNEM-type models, or each (virtual or actual) individual of the SHEDS, MENTOR, APEX, and
HAPEM models, is assigned an exposure event sequence derived from activity diary data. Each exposure
event is typically defined by a start time, a duration, assignments to a geographic location and
microenvironment, and an indication of activity level. The most recent versions of the above models have
defined activity levels using the activity classification coding scheme incorporated into CHAD. A
probabilistic module within these models converts the activity classification code of each exposure event
to an energy expenditure rate, which in turn is converted into an estimate of oxygen uptake rate. The
C-9
-------
oxygen uptake rate is then converted into an estimate of total ventilation rate (VE) expressed in
Liters/min. Johnson (2001) reviewed briefly the physiological principles incorporated into the algorithms
used in pNEM to convert each activity classification code to an oxygen uptake rate and describes the
additional steps required to convert oxygen uptake to (VE).
McCurdy (1997; 2000) has recommended that the ventilation rate should be estimated as a function
of energy expenditure rate. The energy expended by an individual during a particular activity can be
expressed as EE = (MET)(RMR) in which EE is the average energy expenditure rate (kcal/min) during
the activity and RMR is the resting metabolic rate of the individual expressed in terms of number of
energy units expended per unit of time (kcal/min). MET (the metabolic equivalent of tasks) is a ratio
specific to the activity and is dimensionless. If RMR is specified for an individual, then the above
equation requires only an activity-specific estimate of MET to produce an estimate of the energy
expenditure rate for a given activity. McCurdy et al. (2000) developed distributions of MET for the
activity classifications appearing in the CHAD database.
An issue that should be mentioned in closing is that of evaluating comprehensive prognostic
exposure modeling studies, for either individuals or populations, with field data. Although databases that
would be adequate for performing a comprehensive evaluation are not expected to be available any time
soon, there have been a number of studies, reviewed in earlier sections of this chapter, which can be used
to start building the necessary information base. Some of these studies report field observations of
personal, indoor, and outdoor levels and have also developed simple semi-empirical personal exposure
models that were parameterized using the observational data and regression techniques.
In conclusion, though existing inhalation exposure modeling systems have evolved considerably in
recent years, limitations of available modeling methods and data in relation to potential S02 studies
should be taken into account. Existing prognostic modeling systems for inhalation exposure can in
principle be directly applied to, or adapted for, S02 studies; APEX, SHEDS, and MENTOR-1A are
candidates. However, such applications would be constrained by data limitations such as ambient
characterization at the local scale and by lack of quantitative information for indoor sources and sinks.
C-10
-------
Annex D. Controlled Human Exposure
This Annex summarizes the findings of human clinical studies that have been published since the
previous review. Descriptions of older studies were presented in the 1994 Supplement to the Second
Addendum to the 1982 AQCD for Sulfur Oxides (U.S. EPA, 1994a), and are not described in great detail
in this document.
Table D-1. Effects of medications on S02-induced changes in lung function among human
subjects.
Study Concentration Duration Subjects
Bigbyand 0.25- 8.0 ppm 4min
Boushey
(1993)
Lazarus 0.25-8.0 ppm 4min
et al.
(1997)
Gong 0.75 ppm 10 min
et al.
(2001)
Exposure Status
Subjects exposed via mouthpiece to filtered air
and increasing concentrations of SO2 during
eucapnic hyperventilation (20 L/min). Exposures
occurred following pretreatment with a
leukotriene receptor antagonist (zafirlukast 20
mg) or placebo.
Exposure to SO2 or clean air following 3 days of
treatment with montelukast (leukotriene receptor
antagonist) or placebo (each subject exposed 4
times). Exposures conducted in an exposure
chamber during moderate levels of exercise (35
L/min).
Effects
Treatment with nedocromil sodium significantly
increased the concentration of SO2 required to
produce an 8 unit increase in sRaw. Increasing the
dose of nedocromil sodium from 2 mg to 8 mg did
not significantly affect the response.
Compared with placebo, zafirlukast significantly
increased the SO2 concentration required to
produce an 8 unit increase in sRaw. This effect was
observed with challenges occurring both at 2 and
10 h following treatment.
Both ipratropium bromide and morphine reduced
the responsiveness to SO2, significantly increasing
the SO2 concentration required to reduce specific
airway conductance by 35%. Similarly
indomethacin was observed to attenuate airway
responsiveness to SO2, however, this effect was
smaller than what was observed with either
ipratropium bromide or morphine.
Observed a significant protective effect of
salmeterol xinafoate at 1 and 12-h post-dosing.
Following exercise/S02 exposure at 1,12,18, and
24 h, FEV1 decreased (versus preexposure) by 7,
12, 25, and 26%, respectively. Exercise with SO2
resulted in an approximate 26% decrease in FEV1
at all time points with placebo.
Reported a statistically significant S02-induced
increase in eosinophil count in induced sputum.
Measures of lung function (FEV1 and sRaw), as
well as respiratory symptoms and eosinophil count
all showed significant improvement after
pretreatment with montelukast.
10 asthmatics Increasing concentrations of SO2 during voluntary
eucapnic hyperpnea (20 L/min) preceded by
administration of the anti-inflammatory agent
nedocromil sodium (baseline, placebo, 2 mg, 4
mg, 8 mg).
12 asthmatics
Field etal. 0.25-8.0 ppm 3 min 31 asthmatics Increasing concentrations of SO2 (including clean
(1996) air exposure) in an exposure chamber during
voluntary eucapnic hyperpnea (35 L/min)
preceded by administration of placebo, the
anticholinergic bronchodilator ipratropium
bromide (15 subjects), morphine (opioid agonist)
(15 subjects), or the anti-inflammatory agent
indomethacin (16 subjects).
Gong 0.75ppm 10 min 10 asthmatics Subjects exposed to SO2 or clean air in a
etal. chamber while performing light exercise (29
(1996) L/min) at 1,12,18, and 24 h after pretreatment
with salmeterol xinafoate (^-adrenergic agonist)
or placebo (each subject exposed 4 times).
11 asthmatics
D-1
-------
Table D-2. Summary of new studies of controlled human exposure to SO2.
Study Concentration Duration Subjects
Exposure Status
Effects
Trenqa
et al.
(2001)
0.1, 0.25 ppm 10 min 17 asthmatics
S02-sensitive asthmatics exposed to
SO2 via mouthpiece while performing
mild to moderate levels of exercise.
Exposures preceded by 45 min expo-
sures to filtered air or O3 (0.12 ppm),
with or without pretreatment with
dietary antioxidants.
Exposure to O3 slightly increased bronchial
responsiveness to SO2 as measured by FEV1 and peak
expiratory flow. Pretreatment with dietary antioxidants
was shown to have a protective effect on respiratory re-
sponse, particularly among individuals with greater
sensitivity to SO2.
Devalia 0.2 ppm 6h 10 asthmatics Exposures to filtered air, as well as
etal. 0.2 ppm SO2 and 0.4 ppm NO2,
(1994) conducted separately and in
combination in an exposure chamber
(subjects at rest). All subjects
sensitive to inhaled house dust mite
antigen.
Neither SO2 nor NO2, alone or in combination, signi-
ficantly affected FEV1. The combination of SO2 and NO2
significantly reduced the amount of inhaled allergen
(60.5% change, p = 0.015) required to produce a 20%
decrease in FEV1 (PD20FEVi). Both SO2 and NO2alone
reduced PD20FEVi, but this reduction was not statis-
tically significant (32.2% (p = 0.506), and 41.2%
(p = 0.125), respectively).
Rusznak 0.2 ppm 6h 13 asthmatics Exposures to filtered air and a combi-
etal. nation of 0.2 ppm SO2 and 0.4 ppm
(1996) NO2 in an exposure chamber (sub-
jects at rest). All subjects sensitive to
Confirmed findings of Devalia et al. and further observed
that the combination of SO2 and NO2 enhanced airway
responsiveness to an inhaled allergen up to 48 h post-
exposure (maximal response at 24 h).
Tunnicliffe
et al.
(2001)
0.2 ppm
1 h
12 healthy adults,
12 asthmatics
Exposures (head dome) at rest to
filtered air and 0.2 ppm SO2.
Among healthy subjects, an S02-induced increase in
heart rate variability (total power) was observed, while a
reduction in heart rate variability with SO2 versus air was
observed in asthmatics.
Tunnicliffe
et al.
(2003)
0.2 ppm
1 h
12 healthy adults,
12 asthmatics
Exposures (head dome) at rest to
filtered air and 0.2 ppm SO2.
Exposures to SO2 at 0.2 ppm did not have a significant
effect on lung function, respiratory symptoms, markers of
inflammation, or antioxidant levels in healthy adults or
mild asthmatics.
Routledqe
etal.
(2006)
0.2 ppm
1 h
20 older adults with
coronary artery
disease (age 52-74),
20 healthy older
adults (age 56-75)
Exposures (head dome) at rest to
filtered air, as well as 0.2 ppm SO2
and ultra-fine carbon particles (50
(jg/m3), separately and in
combination.
In healthy subjects, exposure to SO2 alone significantly
decreased heart rate variability 4 h post-exposure
compared to clean air. No effect was observed in
subjects with coronary artery disease. The combination
of SO2 and carbon particles did not affect heart rate
variability in either group. SO2 was not observed to affect
markers of inflammation or coagulation.
Nowak 0.25-2.0 ppm 3 min
et al.
(1997)
786 adults Mouthpiece exposures to filtered air
and increasing concentrations of SO2
during eucapnic hyperventilation (40
L/min).
Among individuals who were not hyperresponsive to
methacholine, less than 1% were found to be hyperre-
sponsive to SO2. However, more than 22% of the indivi-
duals who were hyperresponsive to methacholine were
also hyperresponsive to SO2. Individuals were consi-
dered hyperresponsive to SO2 when exposure resulted in
a 20% or greater decrease in FEV1 versus baseline.
Trenqa
et al.
(1999)
0.5 ppm
10 min 47 asthmatics
Subjects exposed to SO2 via
mouthpeice while performing light to
moderate levels of exercise.
An S02-induced decrease in FEV1 of at least 8% was
observed in 53% of the subjects (range 8-44%).
Increases in respiratory symptoms were significantly
associated with decreases in FEV1. Among S02-sensitive
subjects, severity of asthma (as defined by medication
use) was not a significant predictor of the level of
response. It is not clear whether the response was
adjusted for the effects of exercise in clean air.
Winterton 0.5 ppm
et al.
(2001)
10 min 62 asthmatics
Subjects exposed to SO2 via
mouthpiece while performing light to
moderate levels of exercise.
Subjects who experienced at least at 12% decrease in
FEV1 following exposure were considered to be sensitive
to SO2. Out of 58 subjects who were genotyped for the
polymorphism at position -308 in the promoter region of
TNF-a, 21% (N: 12) were sensitive to SO2. Sensitivity to
SO2 was found to be associated with the homozygous
wild type allele (GG) (12 of 12 responders vs. 28 of 46
subjects who were not responsive to SO2).
Gong et al.
(1995)
0.5,1.0 ppm 10 min 14 asthmatics
Exposure to SC^and filtered air were
conducted in an exposure chamber
during low, moderate, and heavy
levels of exercise (target ventilation
ranges of 20-29, 30-39, and 40-49
L/min).
For the average individual, increasing SO2 concentration
resulted in a significant decrement in lung function (de-
crease in FEV1 and increase in sRaw) as well as a signi-
ficant increase in respiratory symptoms. Increasing SO2
conc. had a greater effect on lung function and
respiratory symptoms than increasing level of exercise.
D-2
-------
Annex E. Toxicological Studies
This Annex summarizes the findings of animal toxicology studies that have been published since
the previous review. Descriptions of older studies were presented in the 1982 AQCD for Sulfur Oxides
(U.S. EPA, 1982), and are not described in great detail in this document.
Table E-1. Respiratory System - Effects of SO2 on lung function.
Study
Concentration
Duration
Species
Effects
ACUTE AND SUBACUTE
Lewis and
Kirchner
(1984)
10 or 30 ppm
(26.2 or 78.6
mg/m3);
intratracheal
5 min
Mongrel dog; male and fe- Initial transient bronchoconstriction approximately 10 min in duration
male; age and weight NR; followed by a gradual change in pulmonary mechanics (43% increase
N = 5-15/group in airway resistance and 30% decrease in dynamic compliance) 4 hrs
following 30 ppm but not 10 ppm SO2.
Barthelemy
et al. (1988)
0.5 or 5 ppm
(1.3 or 13.1
mg/m3);
intratracheal
45 min
Rabbit; sex NR; adult;
mean 2.0 kg; N = 5-9/
group; rabbits were
mechanically ventilated
Lung resistance increased by 16% and 50% in response to 0.5 and
5 ppm SO2, respectively. Bivagotomy had no effect on 5 ppm SO2-
induced increases in lung resistance. Reflex bronchoconstrictive
response to phenyldiguanide (intravenously administered) was elimi-
nated by exposure to SO2 but SO2 had no effect on lung resistance
induced by intravenously- administered histamine. Authors concluded
that (1) vagal reflex is not responsible for S02-induced increase in lung
resistance at 45 min; (2) transient alteration in tracheobronchial wall
following SO2 exposure may have reduced accessibility of airway
nervous receptors to phenyldiguanide.
Amdur et al.
(1983)
~1 ppm
(2.62 mg/m3); head
only
1 h
Hartley guinea pig; male;
age NR; 200-300 g; N =
8-23/group
An 11% increase in pulmonary resistance and 12% decrease in
dynamic compliance were observed. Neither effect persisted into the 1
h period following exposure. No effects were observed for breathing
frequency, tidal volume, or min volume.
Conner et
al. (1985)
1 ppm (2.62
mg/m3); nose only
3 h/day for 6 days;
animals evaluated up to
48 h post-exposure
Hartley guinea pig; male;
age NR; 250-320 g; N =
s 18 group/time point
No effect was observed on residual volume, functional reserve
capacity, vital capacity, total lung capacity, respiratory frequency, tidal
volume, pulmonary resistance, pulmonary compliance, diffusing
capacity for CO or alveolar volume at 1 or 48 h after last exposure.
SUBCHRONICAND CHRONIC
Douglas et
al. (1994)
5 ppm
(13.1 mg/m3);
whole body
2 h/day for 13 wks from
birth
New Zealand White rabbit; No effects on lung resistance, dynamic compliance, transpulmonary
male and female; 1 day pressure, tidal volume, respiration rate, or min volume.
old; weight NR; N = 3-4/
group; immunized against
Alternaria tenuis
Scanlon
et al. (1987)
15 or 50 ppm
(39.3 or
131 mg/m3);
intratracheal expo-
sure
2 h/day 4 or 5 days/wk, for Mongrel dog; adult; sex
5 mos (low dose group) or NR; 10-20 kg;
10-11 mos (high dose N = 3-4/group (3 hyper-
group); authors stated that responsive, 3 hypore-
physiological changes sponsive, and 1 avg
were observed within 5 responsive)
mos; 7-9 mo recovery
period
At 15 ppm, there was no clinical evidence of bronchitis; pulmonary
resistance increased by 35-38% in 2 of 3 dogs, and dynamic lung
compliance decreased in 1 of 3 dogs, but physiological changes were
not significant for the group as a whole. At 50 ppm, cough and mucus
hypersecretion were observed; symptoms ceased during the recovery
period. Pulmonary resistance increased by 56% during the treatment
period and an additional 28% during the recovery period for a total
increase of 99%; dynamic lung compliance decreased in 2 of 4 dogs
and increased in 1 of 4 dogs during treatment but there were no
significant changes in the group as a whole. Authors considered
15 ppm to be the lower limit of exposure that failed to produce
physiological changes.
Smith et al.
(1989)
1 ppm
(2.62 mg/m3);
whole body
5 h/day 5 days/wk for 4
mos
Sprague-Dawley rat;
male; young adult; initial
weight NR;
N = 12-15/data point
Physiological tests were conducted in anesthetized animals both
during spontaneous breathing and during paralysis. SO2 exposure
resulted in an 11% decrease in residual volume and reduced
quasistatic compliance in paralyzed animals. Authors noted that
because residual volume was decreased only in paralyzed rats and
the magnitude of effect was very small, it may have been due to
chance. Quasistatic compliance values were observed to be very high
in controls, and may have accounted for effect in the treatment group.
E-3
-------
Table E-2. Respiratory System - Inflammatory responses following SO2 exposure.
Study Concentration Duration
Species
Effects
ACUTE/SUBACUTE/SUBCHRONIC
Clarke etal. 10ppm 4h
(2000) (26.2 mg/m3); nose
only
Outbred Swiss mouse;
female; age, weight
NR;
N = 10/experimental
value
No evidence was seen of inflammatory response in terms of total cell number,
lymphocyte/polymorphonuclear leukocytes differentials, or total protein level
taken from BAL fluid.
Mengetal. 5.35,10.7, or 4 h/day for 7 days
(2005a) 21.4ppm (14, 28, or
56 mg/m3); whole
body
Kunming albino In lung tissue, in vivo SO2 exposure (low, mid concentrations) significantly
mouse; male; age NR; elevated levels of the pro-inflammatory cytokines interleukin-6 and tumor
18-22 g; necrosis factor-a, but did not affect levels of the anti-inflammatory cytokine
N = 10/group transforming growth factor-(31. In serum, the only effect observed was a
low-dose elevation of tumor necrosis factor-a.
Langley-
Evans et al.
(1996)
5, 50, or 100 ppm
(13.1,131,or
262 mg/m3); whole
body
5 h/day for 7-28
days
Wistar rat; male; 7 wks
old; weight NR;
N = 4-5/treatment
group, 8 controls
No lung injury was observed and evidence of inflammatory response was only
observed in the 100 ppm group. A 4-fold increase in BAL fluid leukocyte
numbers was observed in the 100 ppm group at day 14; the increase lessened
at days 21 and 28 but remained higher than controls. The number of
macrophages in BAL fluid was increased at day 28 in the 100 ppm group.
Neutrophil numbers were 120 times higher than controls at day 14 in the
100 ppm group but returned to normal by day 21. Blood neutrophils were
depleted in rats exposed to 50 ppm on days 7-21 but were increased in rats
exposed to 5 ppm (significant) and 100 ppm (non-significant) at day 14. Lung
epithelial permeability was not affected.
Conner et
al. (1989)
1 ppm (2.62 mg/m3);
nose only
3 h/day for 1-5
days; bronchoal-
veolar lavage per-
formed each day
Hartley guinea pig;
male; age NR;
250-320 g; N = 4
No change in numbers of total cells and neutrophils, protein levels or enzyme
activity in lavage fluid following SO2 exposure.
Park et al.
(2001a)
Li et al.
(2007)
0.1 ppm
(0.26 mg/m3); whole
body; with and
without exposure to
ovalbumin
2 ppm (5.24 mg/m3)
with and without
exposure to
ovalbumin
5 h/day for 5 days Dunkin-Hartley guinea After bronchial challenge, the ovalbumin/S02-exposed group had significantly
pig; male; age NR; increased eosinophil counts in BAL fluid compared with all other groups,
250-350 g; N = including the SO2 group. The bronchial and lung tissue of this group showed
7-12/group infiltration of inflammatory cells, bronchiolar epithelial damage, and mucus and
cell plug in the lumen.
1 h/day for 7 days Wistar rat; male; age Increased number of inflammatory cells in BAL fluid, increased levels of
NR MUC5AC and ICAM-1 and an enhanced histopathological response compared
with those treated with ovalbumin or SO2 alone.
E-4
-------
Table E-3. Respiratory System - Effects of SO2 exposure on airway responsiveness and allergic
sensitization.
Study
Concentration
Duration
Species
Effects
ANTIGEN SENSITIZATION / ALLERGIC REACTIONS - ACUTE / SUBACUTE
Abraham 5ppm (13.1 mg/m3]
et al. (1981) head only
4 h
Sheep; sex and age
NR; mean weight 38
± 7 kg; N = 7/group
Acute exposure to 5 ppm SO2 did not produce significant airway changes
(pulmonary resistance, static compliance, dynamic compliance, tidal volume,
breathing frequency) in either normal or allergic (sensitized to Ascaris suum
antigen) sheep, nor increase airway reactivity (measured as pulmonary
resistance increase after aerosolized carbachol provocation) in normal sheep.
However, 5 ppm SO2 did significantly increased airway reactivity in allergic
sheep, which have antigen-induced airway responses similar to humans with
allergic airway disease; may model airway responses to SO2 in a sensitive
human subpopulation.
Parketal. 0.1 ppm (0.26 mg/m3); 5 h/day for 5 days Dunkin-Hartley After bronchial challenge, the ovalbumin/S02-exposed group had significantly
(2001a) whole body; with and guinea pig; male; age increased enhanced pause (an indicator of airway obstruction) compared with
without exposure to NR; 250-350 g; N = all other groups, including the SO2 group. Authors concluded that low level
ovalbumin 7-12/group SO2 may enhance the development of ovalbumin-induced asthmatic reactions
in guinea pigs.
Riedel et al.
0.1, 4.3, or 16.6 ppm 8 h/day for 5 days Perlbright-White Bronchial provocation with ovalbumin was conducted every other day for 2
(0,0.26,11.3, or Guinea pig; female; wks, starting at 1 wk after last exposure. Numbers of animals displaying
43.5 mg/m3); whole age NR; 300-350 g; N symptoms of bronchial obstruction after ovalbumin provocation increased in
body; animals were = 5 or 6/group all SO2 groups compared to air-exposed groups. Anti-ovalbumin antibodies
sensitized to ovalbu- (14 controls) (IgG total and lgG1) were increased in BAL fluid and serum of S02-exposed
min on the last 3 days animals compared to air-exposed controls; statistical significance attained for
of exposure. IgG total in BAL fluid at a 4.3 ppm SO2 and in serum at all SO2 concen-
trations. Results indicate that subacute exposure to even low concentrations
of SO2 can potentiate allergic sensitization of the airway.
ANTIGEN SENSITIZATION / ALLERGIC REACTIONS
-SUBCHRONIC
Kitabatake
et al. (1992;
1995)
5 ppm (13.1 mg/m3); 4 h/day, 5 days/wk, Hartley guinea pig;
whole body; sensitized 6 wks male; age NR;
with Candida albicans -200 g;
on day 1 and wk 4 N = 12/group
Respiratory challenge to Candida albicans was conducted 2 wks after last
exposure. At 15 h after challenge an increased number of S02-exposed
animals displayed prolonged expiration, inspiration, or both. Authors
concluded that SO2 exposure increased dyspneic symptoms.
GENERAL BRONCHIAL REACTIVITY STUDIES - ACUTE
Amdur et al.
(1988)
1 ppm 1 h
Guinea pig; N=8
Airway responsiveness to acetylcholine was measured 2 h following SO2
exposure. No changes were observed.
Douglas
et al. (1994)
5 ppm (13.1 mg/m3); 2h
whole body
New Zealand White
rabbit; sex NR; appar-
ently 3 mos old;
2.2-3.1 kg; N=6/group
No effect on airway responsiveness to inhaled histamine, as measured by
provocation concentrations of histamine required to increase pulmonary
resistance by 50% and decrease dynamic compliance by 35%.
Lewis and
Kirchner
(1984)
10 or 30 ppm (26.2 or
78.6 mg/m3); intratra-
cheal
5 min; a second Mongrel dog; male
exposure was and female; age and
conducted 20 days weight NR;
later, after N = 5-15/group
exposure to the
antiallergic drug
GENERAL BRONCHIAL REACTIVITY STUDIES - CHRONIC
Scanlon
et al. (1987)
15 or 50 ppm (39.3 or
131 mg/m3); intratra-
cheal
2 h/day, 4 or 5
days/wk for 5 mos
(low dose group)
or 10-11 mos (high
dose group); phy-
siological changes
observed within 5
mos; 7-9 mo
recovery period.
Mongrel dog; adult;
sex NR; 10-20 kg;
N = 3-4/ group
(3 hyperresponsive,
3 hyporesponsive,
and 1 avg respon-
sive)
No effect was observed at 10 ppm. At 30 ppm hyperresponsiveness and
hypersensitivity to aerosolized methacholine and 5-hydroxytrypfamine was
observed for up to 24 h following exposure. Twenty days later, pretreatment
with aerosolized 4% Wy-41,195 or disodium cromoglycate (antiallergic drugs)
at high doses lessened the methacholine-induced hypersensitivity observed
after exposure to 30 ppm SO2. Calculations used to determine hyper-
responsive and hyperreactivity were not clear.
Bronchial reactivity in response to inhaled histamine or methacholine was not
affected in either treatment group, as determined by the concentration of
histamine or methacholine required to double pulmonary resistance or the
concentrations required to decrease dynamic compliance by 65% (ED65).
E-5
-------
Table E-4. Respiratory System - Effects of SO2 layered on metallic or carbonaceous particles.
Study
S02
Metal or Carbon
Effects
Lam et al. (1982)
Hartley guinea pig;
male; age
NR;240-300 g;
N = 7-16/group
Exposure: 3 h
0 or ~1 ppm
(2.6 mg/m3); whole
body
Zinc oxide:
0.8, 2.7, 6.0, or 7.8 mg/m3
(0.05 micron projected area
diameter, GSD 2.0) (sulfate,
sulfite, and sulfur trioxide
detected)
Vital capacity: No effect with exposure to 7.8 mg/m3 zinc oxide alone and 2.7 mg/m3 zinc
oxide in combination with SO2, but decreased with exposure to 0.8 and 6.0 mg/m3 zinc
oxide in combination with SO2.
Total lung capacity: No effect with exposure to 7.8 mg/m3 zinc oxide alone, but
decreased with exposure to 6.0 mg/m3 zinc oxide in combination with SO2.
Diffusion capacity for CO and ratio of diffusion capacity for CO to total lung capacity or
alveolar volume: No effect with exposure to 7.8 mg/m3 zinc oxide alone, but decreased
with exposure to 2.7 and 6.0 mg/m3 zinc oxide in combination with SO2.
Alveolar volume: No effect with exposure to 7.8 mg/m3 zinc oxide alone, but decreased
with exposure to 6.0 mg/m3 zinc oxide in combination with SO2.
Amdur et al. (1983)
Hartley guinea pig;
male; age NR;
200-300 g;
N = 8-23/group
Exposure: 1 h
~1 ppm
(2.6 mg/mr
only
Zinc oxide: 0 mg/m3 Pulmonary function: SO2 exposure alone resulted in an 11% increase in resistance and
head 12% decrease in compliance.
Oppm Zinc oxide:-1-2 mg/m3
(0.05 micron projected area
diameter, GSD 2.0); mixed
at 24 °C and 30% RH
Pulmonary function: Zinc oxide exposure alone resulted in a 9% decrease in compliance
that persisted 1 h after exposure.
~1 ppm
(2.6 mg/m3)
only
head
Zinc oxide: -1-2 mg/m3;
mixed at 24 °C and 30%
RH
Zinc oxide: -1-2 mg/m3;
mixed at 480 °C and 30%
RH
Zinc oxide: -1-2 mg/m3;
mixed at 480 °C and 80%
RH with addition of water
vapor downstream
Pulmonary function: A12% decrease in compliance and decreased tidal volume that
persisted 1 h after exposure, and decreased min volume. There was no evidence of new
compound formation. Authors concluded that effects on tidal volume and min volume
most likely represented an additive effect.
Pulmonary function: A12% decrease in compliance and decreased tidal volume that
persisted 1 h after exposure and a 12% increase in resistance and decreased min
volume. There was no evidence of new compound formation.
Pulmonary function: A13% decrease in compliance that persisted 1 h after exposure and
a 29% increase in resistance. Sulfite formation was observed.
Zinc oxide: -1-2 mg/m3;
mixed at 480 °C and 30%
RH with addition of water
vapor during mixing.
Pulmonary function: A19% increase in resistance that persisted 1 h after exposure,
decreased tidal volume immediately after exposure, and a 26% decrease in compliance
1 h after exposure. Sulfate, sulfite, and sulfur trioxide formation was observed.
Chen et al. (1991) 1.10-1.25 ppm Copperoxide:
Hartley guinea pig;
male; age
NR;275-375 g;
N = 8/group
Exposure: 1h
(2.9-3.3 mg/m3);
head only
0 or 1.16-2.70 mg/m3
(< 0.1 micron)
Pulmonary resistance: Increased 32-47% during exposure and at 1 and 2 h
postexposure when SO2 and copper oxide were mixed at 37 °C, a condition that resulted
in formation of 0.36 pmol/m3 sulfite on the copperoxide particles. No effect was
observed when the compounds were mixed at 1411 °C, a condition that led to the
formation of sulfate on the copperoxide particles.
Dynamic lung compliance: No effect when mixed under conditions that led to the
formation of either sulfate or sulfite on particles.
Chen et al. (1992)
Hartley guinea pig;
male; age NR;
290-410 g;
N = 6-9/g roup
Exposure: 1h
1.02 ppm (2.7
mg/m3); head only
Zinc oxide:
0 mg/m3
0
2.76 mg/m3(0.05 micron
median diameter, GSD 2.0)
1.10 ppm
0.87 mg/m3
(2.9 mg/m3)
1.08 ppm
2.34 mg/m3
(2.8 mg/m3)
Baseline pulmonary resistance at 2 h following exposure: No effect in any group.
AHR to acetylcholine: No effect with exposure to SO2 or zinc oxide alone; compared to
furnace controls (3% argon). Hyperresponsiveness increased in both groups exposed to
S02-layered zinc oxide particles.
Jakab et al. (1996) 10 ppm (26.2 0 mg/m3
Cll-„ mg/m3); nose only
Swiss mouse, 3 " '
AM Fc receptor-mediated phagocytosis of sheep red blood cells at 3 days after
exposure: Dose-dependent reductions in AM phagocytosis were observed at each
E-6
-------
Study SO2 Metal or Carbon
Effects
female; 5 wks old;
20-23 g; N =
5/group
Exposure: 4 h
0 ppm Carbon black: 10 mg/m3
(0.3 micron, GSD 2.7)
5 ppm
10 mg/m3 (formed 6 |jg
(13.1 mg/m3)
sulfate at 85% humidity)
10 ppm
10 mg/m3 (formed 13.7 |jg
(26.2 mg/m3)
sulfate at 85% humidity)
20 ppm
10 mg/m3 (formed 48.7 |jg
(52.4 mg/m3)
sulfate at 85% humidity)
10 ppm
0 mg/m3
(26.2 mg/m3);
nose only
0 ppm
Carbon black: 10 mg/m3
(10% humidity)
0 ppm
Carbon black: 10 mg/m3 in
85% humidity to generate 8
(jg/m3 acid sulfate
10 ppm
Carbon black: 10 mg/m3 in
(26.2 mg/m3)
10% humidity to generate
41 (jg/m3 acid sulfate
10 ppm
Carbon black: 10 mg/m3 in
(26.2 mg/m3)
85% humidity to generate
137 (jg/m3 acid sulfate
1 ppm
Carbon black: 1 mg/m3 in
(2.62 mg/m3)
85% humidity to generate
20 (jg/m3 acid sulfate
1 ppm
Zinc oxide: 0or6 mg/m3
(2.6 mg/m3); nose
(0.05 micron projected area
only
diameter, GSD 2.0)
concentration of SO2 mixed with carbon black aerosol at 85% relative humidity, the only
conditions under which SO2 significantly chemisorbed to carbon black aerosol and
oxidized to sulfate. AM phagocytic activity was reduced somewhat immediately after
exposure (Day 0) to carbon black and 10 or 20 ppm SO2, was minimal on Days 1 and 3,
began increasing on Day 7, and was fully recovered by Day 14. No effects were
observed with exposure to SO2 or carbon black alone. The data indicate that
environmentally relevant respirable carbon particles can act as effective vectors for
delivering toxic amounts of acid SO42- to distal parts of the lung.
Clarke et al. (2000)
Outbred Swiss
mouse; female;
age and weight
NR;
N= 10 or 12/
experimental value.
Exposure: 4 h once
or for 4, 5, or 6
days
Inflammatory response after a single 4 h exposure: There was no effect on total cell
number, lymphocyte/PMN differentials, or total protein levels in BAL fluid in any group.
AM Fc receptor-mediated phagocytosis after a single 4 h exposure: Suppressed by acid
sulfate coated particles (at ~140 (jg/m3) at 1, 3, and 7 days postexposure; values
returned to normal by Day 14.
Intrapulmonary bactericidal activity toward Staphylococcus aureus: Decreased by a
single 4 h exposure to sulfate coated particles (at -140 (jg/m3) at 1 and 3 days
postexposure, with recovery by day 7. Suppression was also observed after 5 and 6
days of repeated exposure to -20 |jg/m3 sulfate-coated particles, a condition more
relevant to potential ambient human exposures.
Conner etal.
(1985)
Hartley guinea pig;
male; age NR;
250-320 g;
N = 5-18/group/
time point
Exposure: 3 h/day
for 6 days;
Animals evaluated
for up to 72 h
following exposure
Right lung to body weight ratio: No effect of SO2alone. Increased at 48 h in group
exposed to zinc oxide/S02.
Right lung wet to dry weight ratio: No effect of SO2alone. Increased at 1 h after exposure
in zinc oxide/S02 group.
Lung morphology: No lesions were observed in SO2 group. In group exposed to zinc
oxide/S02 there was an increased incidence of alveolar duct inflammation consisting of
interstitial cellular infiltrate, increased numbers of macrophages, and replacement of
squamous alveolar epithelium with cuboidal cells. Frequency and severity of lesions
were greatest immediately following exposure and by 72 h following exposure, lesions
were mild and infrequent.
Tracheal secretory cell concentration: No effects with either exposure.
Epithelial permeability: No effects with either exposure scenario.
DNA synthesis (3H-thymidine uptake) terminal bronchial cells: Unaffected by SO2.
Increased at 24 and 72 h after exposure to zinc oxide/S02.
Lung volumes: Unaffected by SO2 exposure. Functional reserve capacity, vital capacity,
and total lung capacity were decreased from 1 to 72 h following exposure to zinc
oxide/S02.
Diffusion capacity for CO: Unaffected by SO2 exposure. Decreased by -40-50% from
1 to 24 h following zinc oxide/S02exposure.
Alveolar volume: Unaffected by SO2 exposure. Decreased by-10% from 1 to 24 h
following exposure to zinc oxide/S02.
Pulmonary mechanics: Respiratory frequency, tidal volume, pulmonary resistance,
pulmonary compliance were unaffected by either exposure.
Author conclusion: Changes were identical to those reported in a previous study in which
guinea pigs were exposed to zinc oxide alone. Sulfur compounds deposited on surface
are less important than zinc oxide particle.
E-7
-------
Study SO2 Metal or Carbon
Effects
Amdur et al. (1988)
Guinea pig, sex,
age, and weight
NR;
N = 8-9/g roup
Exposure: 3 h/day
for 5 days
Exposure: 1 h
1 ppm
(2.6 mg/m3)
only
Zinc oxide: 1 or 2.5 mg/m3
head (0.05 micron CMD, GSD
2.0)
Sulfate was generated at
7 and 11 |jg/m3 at each
respective dose; sulfuric
acid level was reported at
21 and 33 |jg/m3 at each
respective dose.
Pulmonary diffusing capacity: No effect with exposure to 1 ppm SO2 or 2.5 mg/m3 zinc
oxide alone (data not shown by authors). Significant and dose related decreases on
exposure days 4 and 5 at 7 |jg/m3 sulfate (20% less than control) and days 2-5 at
11 (jg/m3 sulfate (up to 40% less than control).
Bronchial sensitivity to acetylcholine: No effect of 1 ppm SO2 or 2.5 mg/m3 zinc oxide
alone. Increased with SO2 administered in combination with either zinc oxide dose. The
authors noted that responses were similar to those produced by 200 |jg/m3 sulfuric acid
of similar particle size, thus indicating the importance of surface layer.
Shami et al. (1985)
Fischer-344 rat;
male and female;
18-19 wks old,
weight NR; N =
2/sex/group at
each evaluation
time period
Exposure: 2 h/day
for 4 days, followed
by 2 days without
exposure, followed
by 5 more days of
exposure; animals
were evaluated for
up to 28 days
following exposure
5 ppm (13 mg/m3);
nose only
22 mg/m3 gallium oxide (0.2
micron volume median
diameter, GSD not re-
ported), with and without
addition of 7 mg/m3 B[c]P
Tracheal and large airways morphology: No effects observed with coexposure to gallium
oxide and SO2.
Pulmonary morphology: Increased numbers of non-ciliated cells in terminal bronchial
epithelium were observed in the SCWgallium oxide/benzo(a)pyrene group. Mild
peribronchial and perivascular mononuclear inflammatory cell infiltrate, small
hyperplastic epithelial cells in alveoli, and alveolar septal hypertrophy were observed in
the SCVgallium oxide group, with and without B[c]P; effects were more prominent with
B[c]P exposure.
Cell proliferation (3H-thymidine uptake) in trachea and large airways: Increased on days
1 and 14 in SO2/ gallium oxide group; basal cells primarily labeled. Increased on day 8 In
the S02/gallium oxide/B[c]P group.
Cell proliferation (3H-thymidine uptake) in terminal bronchioles: Increased on day 14 in
SCWgallium oxide group; Clara cells primarily labeled. Increased on day 11 in the
SCWgallium oxide/B[c]P group.
Types of 3H-thymidine-labeled cells in the alveolar region: Type 2 cells were primarily
labeled in the alveolar region through 14 days of exposure in the SCVgallium oxide
group. Labeling was increased in Type II, Type I, and endothelial cells on day 8 in the
SCWgallium oxide/B[c]P group.
Wolff et al. (1989)
F344/Crl rat; male
and female;
10-11 wks old;
weight NR; N =
6/sex/group
Exposure: 2 h/d,
5 d/wk, 4 wks
5 ppm (13 mg/mr
nose only
Gallium oxide: 0 or
27 mg/m3 (-0.20 micron
MMD, GSD -1.5-2), with
and without 7.5 mg/m3 of
1-NP and B[c]P
Pulmonary particle clearance: No effect was observed with exposure to SO2 alone;
clearance was slowed only by gallium oxide, with or without coexposure to S02or the
other compounds; SO2 in combination with the PAH had no effect on clearance rate.
Authors concluded that toxicity was dominated by gallium oxide.
E-8
-------
Table E-5. Respiratory System - Effects of mixtures containing SO2 and O3.
Study
S02
03
Effects
ACUTE/SUBACUTE
Abraham et al. (1986)
Sheep; sex NR; adult; 23-50 kg;
N = 6
Exposure: 5 h/day for 3 days
3 ppm
(7.9 mg/m3); head
only
0.3 ppm
Tracheal mucus velocity: Decreased by 40% immediately after exposure and 25%
at 24 h postexposure to the mixture of the 2 compounds. The effects of either
compound alone were NR.
Ciliary beat frequency: No effect
CHRONIC/SUBCHRONIC
Aranyi et al. (1983)
CD1 mouse; female; 3-4 wks old; weight
NR;
N = 360/group total
(14-154/group in each assay)
Exposure: 5 h/day, 5 days/wk for up to
5.0 ppm
(13.2 mg/m3) in
addition to
1.04 mg/m3
ammonium
sulfate; whole
body
0.2 mg/m3
(0.10 ppm)
Mortality rate after Streptococcus aerosol challenge: Increased in groups exposed
to O3 alone and mixture of O3, SO2, and ammonium sulfate.
Alveolar macrophage bactericidal activity towards inhaled K. pneumoniae'.
Increased trend (non-significant) in O3 group but significantly increased in mixture
group.
Counts, viability, and ATP levels in cells obtained by pulmonary lavage: No effect
103 days
of either treatment.
Raub et al. (1983)
3olden hamster; m
= 14 or 15/group; mild emphysema was
induced in some animals by intratracheal
administration ofelastase
Exposure: 23 h/day, 7 days/wk, for 4wks
1 ppm
(2.62 mg/m3)
whole body
1 ppm in
addition to
3 ppm
trans-2-butene
Lung volumes: End expiratory volume, residual volume, total lung capacity and
vital capacity were unaffected in the mixture versus air exposure group in normal
or emphysematous hamsters.
Respiratory system compliance: Unaffected in the mixture versus air exposure
group in normal or emphysematous hamsters.
Distribution of ventilation (N2 washout slope): The N2 slope decreased in the
mixture versus air exposure group in both normal and emphysematous hamsters.
Diffusion capacity for CO: Significantly increased in the mixture versus
air-exposed normal animals. Although the text reported an increase in the mixture
versus air-exposed emphysematous animals, Figure 3 of the study indicated that
the effect was very small and did not obtain statistical significance. Significantly
lower in emphysematous versus normal hamsters exposed to the mixture. The
authors noted a significant interaction between exposure to the mixture and
emphysema.
Histopathology: Inflammatory lesions were found in the lungs of emphysematous
hamsters exposed to air or the mixture. Hyperplasia incidence was higher in
emphysema hamsters exposed to the mixture versus air. Inflammatory lesions
were similar in emphysematous hamsters exposed to air or the mixture. Data were
not shown for histopathology data.
Overall author conclusion: Animals with impaired lung function may have
decreased capacity to compensate for the pulmonary insult caused by exposure to
a complex pollutant mixture.
E-9
-------
Table E-6. Respiratory System - Effects of SO2 and sulfate mixtures.
Study
SO2
Sulfate
Effects
ACUTE
Mannix et al. (1982)
Sprague Dawley rat, male, age NR,
-200 g, N = 8/group
5 ppm
(13.1 mg/m3); nose
only
Sulfate aerosol 1.5
(0.5 micron MMAD,
GSD 1.6)
Lung clearance of radiolabeled tracer particles: No significant effect was
observed with the mixture of the two compounds at 80-85% humidity.
4 h exposure
CHRONIC/SUBCHRONIC
Smith et al. (1989)
Sprague-Dawley rat, male, young
adult, initial weight NR, N = 12-15/data
point
Exposure: 5 h/day, 5 days/wk for 4 or
8 mos; half the animals in the 8-mo
group were allowed to recover for
3 mos.
1 ppm
(2.62 mg/m3); whole
body
0
(NH4)2S04
0.5 mg/m3
(MMAD = 0.42-0.44 =
0.04 micron, GSD
2.2-2.6)
1 ppm (2.62 mg/m3) 0.5 mg/m3
Morphological observations at 4 mos exposure in "normal" rats:
Bronchiolar epithelial hyperplasia and increased numbers of non-ciliated
epithelial cells were observed in rats exposed to either compound alone
¦ but coexposure to both compounds did not magnify the effects. An
increase in alveolar chord length was observed in the (NH^SCU group
and no further changes were observed with coexposure to SO2.
Morphological observations at 4 mos exposure in rats treated with
elastase to induce an emphysema-like condition: Bronchiolar epithelial
_ hyperplasia was decreased in groups exposed to either compound alone
or the mixture of the two compounds. A decrease in alveolar chord length
was observed in the (NH^SCU group and no further changes were
observed with coexposure to SO2.
Morphological observations at 8 mos exposure in "normal" rats: An
increase in non-ciliated epithelial cells and alveolar birefringence (an
indication of alveolar interstitial fibrosis) was observed only in the group
exposed to (NbU^SCU.
Morphological observations at 8 mos exposure in rats treated with
elastase: An increase in lung volume per body weight and emphysema
incidence was observed in groups treated with either compound alone or
in combination; alveolar chord length was increased only in the group
exposed to the mixture of compounds.
Morphological observations at 12 mos exposure in normal rats: Increased
alveolar chord length was observed only in the (NH^SCU group.
Morphological observations at 12 mos exposure in rats treated with
elastase: An increase in absolute lung volume was observed only in the
group treated with the mixture of both compounds.
Lung function effects at 4 mos exposure in normal rats: A decrease in
residual volume was observed in the SO2 group and decreased
quasistatic compliance was observed in the SO2 group and in the
(NH4)2S04 group, but the effects were not observed with the mixture.
Lung function effects at 4 mos exposure in elastase-treated rats: Ratio of
residual volume/total lung capacity and N2 washout was decreased in the
SO2 group and in the (NLU^SCU group, but the effects were not observed
with the mixture.
Overall conclusions: In general, pollutant effects were minimal and
transient, and appeared obscured or repressed in elastase-treated
groups; (NH^SCUwas more bioactive than SO2, with little evidence of
mixture additivity (in several instances, effects seen with one or both
pollutants individually were not seen with the mixture).
E-10
-------
Table E-7. Respiratory System - Effects of actual or simulated air pollution mixtures.
Study
Exposed
Control
Effects
ACUTE/SUBACTUTE
Mautz et al. (1988)
Sprague-Dawley rat,
male, age NR, 240-280
g, N = 6-9/group
Exposure: 4 h
Air pollutant mixture at full con-
centration (tested in 2 studies):
0.35 ppm 63,1.3 ppm NO2,
2.5 ppm (6.6 mg/m3) SO2,10 |jg/m3
manganese sulfate, 500 |jg/m3
ferric sulfate, 500 |jg/m3 ammo-
nium sulfate, 500 |jg/m3 carbon
aerosol. Mixture also tested 14 and
'/concentrations. For aerosols:
MMAD = 0.3-0.48 micron with
GSD: 2.6-4.6. Nose-only exposure.
Compounds formed: sulfate, ni-
trate, hydrogen ion, nitric acid.
Clean air Breathing pattern: Effect of full concentration mixture in 2 studies - increased
breathing frequency, trend or significant decrease in tidal volume, decreased or
unaffected O2 consumption, increased or unaffected ventilation equivalent for O2.
Effect of half concentration mixture - increased min ventilation. Quarter
concentration - no significant effects.
Histopathology: Full concentration - area of type 1 parenchymal lung lesions
increased in 1 of 2 experiments; area of type 2 parenchymal lung lesions were
increased in both experiments. Effects were equivalent to those observed with O3
exposure alone. Half and quarter concentrations - no effects.
Mucociliary clearance: No effect on early or late clearance of 85Kr-labeled
polystyrene particles.
Nasal epithelial injury (measured by tritiated thymidine uptake): No effect at any
concentration.
Phalen and Kleinman
(1987)
Sprague-Dawley rat,
male, age NR,
200-225 g, N =
5-13/group/
time period
Exposure: 4 h/day for 7
or 21 days
SUBCHRONIC/ CHRONIC
2.55 ppm (6.7 mg/m3) SO2,
0.3 ppm O3,1.2 ppm NOx, 150
(jg/m3 ferric oxide, 130 (jg/m3 nitric
acid, 2.0 |jM/m3 hydrogen ion, and
500 (jg/m3 total Fe3-, Mn2*, and
NH42* combined; nose only
Purified air
Saldiva et al. (1992)
Wistar rat, male, 2 mos
old, weight NR, N =
14-30/group
Exposure: 6 mos
Urban air: Sao Paulo, mean levels
of air pollutants measured 200 m
from the police station where rats
were kept: 29.05 |jg/m3
(0.011 ppm) SO2; 1.25 ppm CO,
11.08 ppb O3, 35.18 |jg/m3 particu-
lates
Rural air: Atibaia,
an agricultural
town 50 km from
Sao Paulo was
considered the
control; air pollut-
ant levels were not
measured
Bronchoalveolar epithelial permeability to 99mTc-diethylenetriaminepentaacetate: No
effect at either time period.
Nasal mucosal permeability to 99mTc-diethylenetriaminepentaacetate: No effect at
either time period.
Macrophage rosette formation: Decreased (indicating damage to Fc receptors) for
up to 4 days after 7- or 21-day exposure; magnitude of effect greater following
21-day exposure. By day 4 post-exposure, numbers began increasing and by day 7
were equivalent to control values.
Macrophage phagocytic activity: In rats exposed for 7 days, decreased activity was
observed for 2 days post-exposure. No effects observed after 21-day exposure.
Death: 37 of 69 Sao Paulo rats died before study end; autopsy of 10 animals
identified pneumonia as the cause of death; 10/56 Atibaia animals died.
Respiratory mechanics: Nasal resistance was higher in Atibaia animals.
No differences were observed for pulmonary resistance or dynamic lung elastance.
Mucus properties: Sao Paulo animals' tracheal mucus output was lower, relative
speed of tracheal mucus was slower, ratio between viscosity and elasticity was
higher for nasal mucus, and rigidity of tracheal mucus was increased.
Bronchoalveolar lavage: In lavage fluid from Sao Paulo animals, there were
increased numbers of cells, lymphocytes, polymorphonuclear cells.
Histochemical evaluation: Hyperplasia was observed in respiratory epithelium of
rats housed in Sao Paulo.
Ultrastructural studies: Animals housed in Sao Paulo had a higher frequency of cilia
abnormalities including composite cilia, microtubular defect, vesiculation, and de-
creased microvelocity of luminal membrane.
Lemos et al. (1994)
Rats from the same
cohort as Saldiva et al.
(1992). N = 15/group
Exposure: 6 mos
Urban air: Sao Paulo, mean levels
of pollutants measured 200 m from
police station where rats were kept:
29.05 |jg/m3 (0.011 ppm)S02; 1.25
ppm CO, 11.08 pb O3,35.18 (jg/m3
particulates
Rural air: Atibaia,
agricultural town
50 km from Sao
Paulo, considered
control; air pollut-
ant levels not
measured
Nasal passage pathology: Rats housed in Sao Paulo had increased nasal
epithelium volume, larger amounts of mucosubstances stored in epithelium, and
more acidic mucus secretions in lamina propria glands.
Pereira et al. (1995)
4 groups of rats
housed: 3 mos in Sao
Paulo, 3 mos in Sao
Paulo followed by 3
mos in Atibaia, 3 mos in
Atibaia, or 6 mos at
Atibaia.
Wistar rats, male, 1.0-
1.5 mos old, weight
NR, N = 30/group
Urban air: Sao Paulo, levels of air
pollutants measured were: -8-50
(jg/m3 (0.003-0.019 ppm) SO2,
-0.1-0.45 ppm nitrogen dioxide,
-4.8-7 ppm carbon monoxide, and
-50-120 |jg/m3PM
Rural air: Atibaia,
an agricultural
town 50 km from
Sao Paulo was
considered the
control; air pollut-
ant levels were not
measured
Lung responsiveness to methacholine: Increased respiratory system elastance
resulting from increased sensitivity to methacholine in rats housed in Sao Paulo for
3 mos compared to all the other groups. No exposure-related effects were observed
for respiratory system resistance.
E-11
-------
Table E-8. Effects of meteorological conditions on SO2 effects.
Study
S02
Condition
Effects
Barthelemy et al. (1988)
Rabbit, sex NR, adult, mean 2.0 kg, N =
5-10/group; animals were mechanically
ventilated
Exposure: 45 min
Halinen et al. (2000a)
Duncan-Hartley guinea pig, male, age and
weight NR, N = 7-12/group, mechanically
ventilated; animals were hyperventilated during
cold air and SO2 exposure to simulate exercise.
In pre-exposure period: 15-min exposure to
warm humid air, 10-min exposure to cold dry
air, and 15-min exposure to warm humid air. In
the SO2 exposure period: 10-min exposures to
each SO2 concentration in cold dry air or with
cold dry air alone were preceded and followed
by 15-min exposures to warm humid air
0.5 or 5 ppm (1.31 Drop in air Lung resistance: Exposure to cool air for 20 min resulted in a -54%
or 13.1 mg/m3); temperature from mean increase in lung resistance. Exposure to SO2 for 20 min
intratracheal 38 °C to 15 °C increased lung resistance by 16% at 0.5 ppm and 50% at 5 ppm. The
difference in lung resistance from warm to cold air was halved (27%)
by exposure to 0.5 ppm and was not significant at 5 ppm. The authors
concluded that transient alteration in tracheobronchial wall following
SO2 exposure may have reduced accessibility of airway nervous
receptors to cold air.
1.0, 2.5, or 5 ppm Drop in Peak expiratory flow: Percent decreases were significantly greater with
(2.62, 6.55, or intratracheal exposures to SO2 in dry air at concentrations of 1.0 ppm (-32.7%) and
13.1 mg/m3); temperatures 2.5 ppm (-35.6%) than with exposure to cold dry air (-27%); decrease
apparently from -35.5 °C to at 5 ppm SO2 in cold dry air (-25.3%) was similar to that with cold dry
intratracheal -27 °C air. The effects did not persist following exposures.
Tidal volume: Percent decreases were significantly greater with
exposure to SO2 in cold dry air at concentrations of 1.0 ppm (-22.4%)
and 2.5 ppm (-28.3%) than with exposure to cold dry air (-18.1%);
decrease at 5 ppm SO2 in cold dry air (-17.8%) was similar to that of
cold dry air. The effects did not persist following exposures.
Bronchoalveolar lavage: The clean dry air group had significantly more
macrophages, lymphocytes, and increased protein concentration in
lavage than the warm humid air control. The cold dry air + SO2 group
had fewer macrophages than the clean dry air group and higher
protein concentration than controls.
Histopathology: Increased incidence of eosinophilic infiltration within
and below tracheal epithelium with exposure to cold dry air or SO2 in
cold dry air.
Halinen et al. (2000b)
Duncan-Hartley guinea pig, male, age and
weight NR, N = 8-9/group, mechanically
ventilated; animals were hyperventilated during
cold air and SO2 exposure to simulate exercise
Exposure: 60 min
1 ppm
(2.62 mg/m3);
apparently
intratracheal
Drop in Peak expiratory flow: Non-significant decreases compared to baseline
intratracheal (4.5-10.8%) at 10 and 20 min of exposure to cold dry air. With
temperatures exposure to SO2 in cold dry air: decreased significantly (11.4%, i.e.,
from -37 °C to bronchoconstriction) compared to baseline at 10 min of exposure but
-26 °C recovered from 20 to 60 min of exposure. The effect with SO2
exposure was not statistically significant compared to that of cold dry
air alone.
Tidal volume: Decreased from baseline throughout most of the
exposure period with cold dry air or SO2 in cold dry air; response with
S02was more shallow than that of cold dry air alone, but statistical
significance compared to cold dry air was obtained only at 60 min of
exposure.
Bronchoalveolar lavage: Decreased neutrophil numbers in the SO2
group compared to the warm humid air group but no significant
difference compared to the cold dry air group.
Histopathology: No effect in lung or tracheobronchial airway.
General conclusions: Functional effects on the lower respiratory tract
were weaker than in the previous study with 10-min exposures
(Halinen et al., 2000a).
E-12
-------
Table E-9. Cardiovascular effects of SO2 and metabolites.
Study Cone.
Duration
Species
Effects
IN VITRO EXPOSURE
Nieand Bisulfite/sulfite, 1:3
Meng (2005) molar/molar, 10
pM
NR
Ventricular myocytes isolated Effects of the 10 |jM bisulfite/sulfite mixture on sodium current included
from Wistar rat; adult; a shift of steady state inactivation curve to a more positive potential, a
200-300 g; N = 8 shift of the time-dependent recovery from inactivation curve to the left,
accelerated recovery, and shortened inactivation and activation time
constants. It was concluded the bisulfite/sulfite mixture stimulated
cardiac sodium channels.
Nieand Bisulfite/sulfite, 1:3
Meng (2006) molar/molar, 10
pM
ACUTE/SUBACUTE
NR
Ventricular myocytes isolated Effects of the 10 |jM bisulfite/sulfite mixture on voltage-dependent
from 200-300 g; N = 8 L-type calcium currents included a shift of steady-state activation and
inactivation to more positive potentials, accelerated recovery from
inactivation, and shortened fast and slow time inactivation constants.
Authors stated that their results suggested the possibility of cardiac
injury following SO2 inhalation.
Halinen
et al.
(2000a)
1.0, 2.5, or 5 ppm
(2.62, 6.55, or
13.1 mg/m3) in
cold dry air;
apparently intra-
tracheal
In pre-exposure per-
iod: 15 min exposure
to warm humid air, 10
min to cold dry air,
and 15 min to warm
humid air. In exposure
period: 10 min
exposures to each
SO2 concentration or
cold dry air were pre-
ceded and followed by
15 min exposures to
warm, humid air.
Duncan-Hartley guinea pig;
male; age and weight NR; N
= 7-12/ group, mechanically
ventilated; animals were
hyperventilated during cold
air and SO2 exposure to
simulate exercise
Arterial blood pressure increased transiently during exposure to 5 ppm
SO2 in cold dry air. No analyses were done to determine if the effects
on blood pressure were caused by exposure to cold air or SO2.
Halinen
et al.
(2000b)
1 ppm (2.62
mg/m3) in cold dry
air; apparently
intratracheal
60 min
Duncan-Hartley guinea pig;
male; age and weight NR;
N = 8-9/group, mechanically
ventilated; animals were
hyperventilated during cold
air and SO2 exposure to
simulate exercise
Blood pressure and heart rate increased similarly with exposure to cold
dry air or SO2 in cold dry air. Blood pressure generally increased during
the first 10-20 min of exposure and remained steady from that point
forward. The increase in heart rate was gradual. No analyses were
done to determine if the effects on blood pressure were caused by
exposure to cold air or SO2.
Nadziejko
et al. (2004)
1 ppm
(2.62 mg/m3);
nose only
4 h
F344 rat, male, 18 mos old,
weight NR, N = 20
(crossover design)
SO2 exposure had no effect on spontaneous arrythmia frequency in
aged rats. Authors urged caution in the interpretation of effects
because occurrence of arrhythmias in aged rats was sporadic and
variable from day to day.
Meng et al.
(2003a)
10, 20, or 40 ppm
(26.2, 52.4 or
105 mg/m3); whole
body
6 h
Wistar rat; male; 7-8 wks old; A dose-related decrease in blood pressure was observed at a 20 ppm.
180-200 g;
N = 10/group
Meng et al.
(2003a)
10, 20, or 40 ppm
(26.2, 52.4, or
104.8 mg/m3);
whole body
6 h/day for 7 days
Wistar rat, male; 7-8 wks old;
180-200 g;
N = 10/group
Dose-related decreases in blood pressure were observed on exposure
day 3 in the 10 ppm group, exposure days 2-6 in the 20 ppm group,
and all exposure days in the 40 ppm group. The authors noted possible
adaptive mechanism in the low but not the high dose group.
Langley-
Evans et al.
(1996)
5, 50, or 100 ppm
(13.1,131, or 262
mg/m3); whole
body
5 h/day for 7-28 days Wistar rat; male; 7 wks old;
weight NR; N = 4-5/treatment
group, 8 controls
GSH was depleted in the heart at 5 and 100 ppm. At 50 ppm, GSH
level decreased in heart at 7 days and returned to normal by 14 days.
No effects were observed for other GSH-related enzymes. Injury and
inflammation were not assessed in heart, but assessment in lung
revealed no effect.
Mengetal. 8.4, 21, or 43 ppm
(2003b) (22, 56, or 112
mg/m3); whole
body
6 h/day for 7 days
Kunming albino mouse; male
and female; 5 wks old;
19 ± 2 g;
N = 10/sex/group
Changes observed in heart (concentrations of effect) included: lower
SOD activity in males and females (a 8.4 ppm), higher TBARS level in
males and females (a 8.4 ppm), lower GPx activity in males (8.4 and
21 ppm; also 43 ppm according to text) and lower GSH level in males
(43 ppm). Authors concluded that SO2 induced oxidative damage in
hearts of mice.
Wu and 8.4, 24.4, or
Meng (2003) 56.5 ppm (22, 64,
or 148 mg/m3);
whole body
6 h/day for 7 days
Kunming-strain mouse;
male; age NR; 18-20 g;
N = 10/group
GSH, GST, and G6PD activities were decreased in the heart at 56.5
ppm.
E-13
-------
Table E-10. Hematological effects of SO2.
Study
Concentration
Duration
Species
Effects
ACUTE/SUBACUTE
Baskurt
(1988)
0.87 ppm
(2.36 mg/m3); whole
body
24 h
Swiss Albino rat; male;
age NR; 250-300 g;
N = 51, 50
Effects of SO2 exposure included increased hematocrit, sulfhemoglobin and
osmotic fragility and decreased whole blood and packed cell viscosities. RBC
number, hemoglobin, mean corpuscular volume, mean corpuscular
hemoglobin concentration, and plasma viscosity were not significantly
altered.
SUBCHRONIC
Gumu§lu
et al.
(1998)
10 ppm
(26.2 mg/m3); whole
body
1 h/day, 7 days/wk
for 8 wks
Swiss-Albino rat; male;
2.5-3.0 mos old; weight
NR; N = 30 (14 controls,
16 treated)
Decreased CuZn SOD activity, increased GPx and GST activity, and
increased TBARS formation were observed in RBC of treated rats. No
significant effect on G6PD or catalase levels was observed.
Yargigoglu 10 ppm (26.2 1 h/day, 7 days/wk Albino rat, male; 3,12,
etal. mg/m3); whole body for6wks and 24 mos old; mean
(2001) weight 213-448 g;
N = 10/group
Enzyme and GSH activity (GPx, catalase, GSH, and GST) were increased
and CuZn SOD activity was decreased in RBCs of all experimental groups
compared to controls. RBCs in older rats had lower levels of all antioxidant
enzymes and increased TBARS activity compared to younger rats.
Langley- 100 ppm (286 5 h/day for 28 days Wistar rat, male; 7 wks
Evans mg/m3); whole body. old; weight NR;
etal. Units were initially N = 4-16
(1997; reported as |jg/m3
2007) but were corrected
per correspondence
w/author.
Dams were fed diets containing casein at 180 [control], 120, 90, or 60 g/kg
during pregnancy and their offspring were exposed to air or SO2 as adults. In
blood of offspring, SO2 exposure significantly reduced the numbers of
circulating total leukocytes and lymphocytes in the 180 and 120 g/kg dietary
groups; neutrophils numbers were not affected in any group. GSH levels in
the 180 and 60 g/kg (but not the two intermediate) dietary groups were
reduced by SO2 exposure.
Etliketal. 10 ppm 1 h/day for 30 days Guinea pig; sex and age SO2 exposure resulted in RBC membrane lipoperoxidation (elevated levels of
(1995) (26.2 mg/m3); whole NR; 250-450 g; malonyldialdehyde) and other oxidative damage (elevated osmotic fragility
body N= 12/group ratios and levels of methemoglobin and sulfhemoglobin). All effects signifi-
cantly (p < 0.05) mitigated by injections ofVitamin E+C three times perwk.
Agar et al. 10 ppm 1 h/day, 7 days/ Swiss Albino rat; male; 3
(2000) (26.2 mg/m3); whole wk for 6 wks mos old; weight NR; N =
body 10/group in 4 groups
RBC parameters were monitored in non-diabetic rats, non-diabetic rats
exposed to SO2, alloxan-induced diabetic rats, and diabetic rats exposed to
SO2. In both non-diabetic and diabetic rats exposed to SO2, levels of GPx,
catalase, GSH, GST, and TBARS were elevated in RBC while those of SOD
were reduced.
Etliketal. 10ppm 1 h/day for 45 days Rat; sex and age NR;
(1997) (26.2 mg/m3); whole 214-222 g;
body N = 6-8/g roup
SO2 exposure significantly elevated levels of methemoglobin, sulfhemoglobin
and malonyldialdehyde, the latter of which was substantially reversed by
Vitamin E+C treatment. RBC osmotic fragility was increased by SO2, and
again partially mitigated by Vitamin E+C. SO2 elevated RBC, white blood cell,
hemoglobin and hematocrit values, but not mean corpuscular volume, mean
corpuscular hemoglobin or mean corpuscular hemoglobin concentration.
Vitamin E+C exposure did not affect these parameters.
E-14
-------
Table E-11. Carcinogenic effects of SO2.
Study Concentration Duration
Effects
PULMONARY EFFECTS
Gunnison 0,10, or 30 ppm (0, SO2: 21 wk, 5 Rat, Purpose was to investigate carcinogenic/cocarcinogenic effects of SO2 inhalation or
et al. (1988) 26.2, or 78.6 day/wk (minus Sprague-Dawley; dietary-induced high levels of systemic sulfite/bisulfite in conjunction with tracheal
mg/m3) SO2 (whole holidays), 6 h/day male; 9 wk old; installation of B[a]P. High drinking water levels of W in conjunction with low-Mo feed
body) ± 1 mg B[a]P High W, low Mo -315- 340 g; induce sulfite oxidase deficiency in rats, and thus high systemic levels of sulfite and
0,100 or 400 ppm diet: 21 wk, 7 N = 20-74/group bisulfite (at 0,100 or 400 ppm VV, mean plasma sulfite was 0, 0 or 44 |jM, while mean
W, or [400 ppm W day/wk tracheal sulfite + bisulfite was 33, 69 or 550 nmol/g wet wt). Mortality in B[a]P groups
+ 40 ppm Mo] in a B[a]P: 15 wk, once (-50% after -380-430 d) was due almost exclusively to SQCAof the respiratory tract;
low-Mo diet, perwk starting wk 4 survival rate was excellent for other groups (-50% mortality after -620-700 d). Results
± B[a]P (See indicate no SQCA was induced in any of the SO2 inhalation or systemic sulfite + bisulfite
Effects col- groups, nor were incidences in the B[a]P groups enhanced by such coexposures. This
umn) ± B[a]P lack of cocarcinogenicity does not support the hypothesis that SO2 exposure could
elevate systemic sulfite/bisulfite, generating GSSO3H, which would inhibit GST and
reduce intracellular GSH, thus interfering with a major detoxication pathway for B[a]P
and enhancing its carcinogenicity. Authors note that due to the high incidence of animals
with tumors in the two B[a]P only groups (65/72 and 63/72), cocarcinogenicity of SO2 or
sulfite + bisulfite could only have been demonstrated by shortening of tumor induction
time and/or increased rate of SQCA appearance—neither were observed.
Ohyama
et al. (1999)
0, 0.2 mL C, or
SO2 and/or NO2:
Rat, SPF
(0.2 mL DEP+C
10 mo, 16 h/day
F344/Jcl; male;
± [4 ppm (10.48
CBP or DEcCBP:
6 wk old; wt NR;
mg/m3) SO2 or
4 wk, once/wk by
N = 23-30/group
6 ppm (11.28
intratracheal
in 6 groups
mg/m3) NO2 or
infusion
4 ppm SO2 + 6
ppm NO2]; whole
body)
[Note: 0.2 mL
CBP= 1mg; 0.2 mL
DEcCBP =1 mg
CBP + 2.5 mg
DEP]
Purpose was to study effects of DEP on rat lung tumorigenesis and possible tumor
promoting effects of SO2 or NO2singly or together. Alveolar hyperplasia and adenoma
were significantly (p < 0.01-0.05) increased over controls in the CBP group, but not the
DEcCBP group. This was ascribed to induction of alveolitis and AM infiltration (a tumor
response specific to rat and of questionable relevance to humans) in the former group,
but apparently prevented by DEP in the latter. Alveolar bronchiolization near small
hyaline masses of deposited DEcCBP was observed in all DEcCBP groups, the masses
presumably allowing long-term exposure to DEP extracts by contacted alveolar
epithelium. DNAadducts were found only in the three gas-exposed groups. Discounting
the CBA group, elevated alveolar hyperplasia was seen only in the DEcCBP + NO2
group, and elevated incidences of alveolar adenoma in the DEcCBP + SO2 and
particularly the DEcCBP + NO2 groups; neither effect was observed with coexposure to
both gases—speculated by the authors to perhaps result from inhibition of the stronger
NO2 promotion by HSO3-. Thus, SO2 appears to have weaker capacity than NO2 for
promoting tumor induction (and perhaps genotoxicity) by DEP extract, and may
antagonize such effects by NO2 during coexposure of the gases.
Ito et al. 0, C, or (25 mg SO2 and/or NO2: Rat, SPF Fisher Purpose was to study effects of Tokyo air SPM, with or without coexposure to SO2 or
(1997) SPM+C±[4ppm 11 mo, 16 h/day 344; male; 5 wk N02or their combination, on the development of proliferative lesions of PEC. PEC
(10.48 mg/m3) SO2 C ± SPM: 4wk, old; wt NR; hyperplasia was significantly (p < 05) increased by exposure to SPM, but coexposure to
or 6 ppm (11.28 once/wk by intra- N = 5/group in 6 either gas or their mixture was without additional effect. No PEC papillomas were
mg/m3) NO2 or 4 tracheal injection groups observed in control groups, while a few were seen in the SPM groups, irrespective of gas
ppm SO2 + 6 ppm coexposures. Thus, SO2 demonstrated no tumor promotion or cocarcinogenic properties.
NO2I); whole body [Study did not describe the nature of the carbon (C) used.]
Heinrich
et al. (1989)
0 or [10 ppm (26.2
mg/m3) SO2 +
5 ppm (9.4 mq/m3)
NO2] ± [3 or
6 mg/kg bw of
DEN]; exposure to
gases whole body
S02 + N02: 6,10.5,
15, or 18 mo, 5
day/wk, 19 h/day
DEN: once by
s.c. injection, ~2wk
after the start of
inhalation exposure
Hamster, Syrian
golden; both
sexes; 10wkold;
bw NR; N =
40/sex per each
of 12 exposure
groups
NONPULMONARY EFFECTS
Klein et al.
(1989)
0 or 6 ppm (0 or
15.72 mg/m3) SO2,
±0.2 ppm (600
(jg/m3); whole
body; NDMA
20 mo, 5 day/wk,
4 h/day
Rat,
Sprague-Dawley;
female; age and
wt NR; N =
36/group in 4 rel-
evant groups
The principle focus of this large study was to examine whether two inhaled diesel-ex-
haust emission preparations (± particulates) could enhance the tumorigenesis of injected
DEN. Ancillary aim was to see whether inhalation of the irritant SO2+NO2 mixture could
cause similar enhancement of DEN tumorigenicity. Gas mixture exposure did not affect
bw gain, but slightly shortened survival times (although significantly only for females).
Apart from effects attributed to DEN, serial sacrifices showed progressive increases in
ciliated tracheal cell aberrations and in number of tracheal mucosal cells. In the lung, gas
mixture-related effects were limited to a progressing alveolar lesion involving lining with
bronchiolar epithelium and the presence of some pigment-containing AM, and to a mild,
diffuse thickening of the alveolar septa. SO2+NO2 exposure did not by itself elevate
tumor rate in the upper respiratory tract, nor did it enhance increases induced by DEN.
Thus the mixture appeared to have no tumor inducing or promoting effects.
This is a preliminary report for observations after 20 mo (800 h inhalation in 200
exposures, with calculated inhaled cumulative doses of 77 mg SO2 and 2-3 mg NDMA
per rat). The effects of NDMA ± SO2 inhalation were studied. Group mortality was as
follows: control (3/36), SO2 (5/36), NDMA (4/36), NDMA + SO2 (7/36). The only tumors
observed were nasal: control (0), SO2 (0), NDMA (1), NDMA + SO2 (3). No observable
parameters, including body wt gain, were affected by the additional SO2 exposure;
assessment of tumor incidence effects could not yet be performed.
E-15
-------
Table E-12. Nervous system effects of SO2 and metabolites.
Study
Concentration
Duration
Species
Effects
IN VITRO /
EX VIVO
Du and 1,10, 50, or 100 Not specified Wistar rat; sex NR; Exposure to SO2 derivatives (sulfite, bisulfite) reversibly increased the
Meng |jM SO2 derivatives 6-12 days old; weight amplitude of potassium channel TOCs in a dose-dependent and volt-
(2004a) (1:3, NaHSOs to and number NR; typical age-dependent manner. Compared to controls, 10 |jM SO2 shifted inactivation
Na2S03) observations made on of depolarization toward more positive potentials without significantly affecting
60 isolated the activation process. By increasing maximal TOC conductance and delaying
hippocampal neurons TOC inactivation, micromolar concentrations of SO2 derivatives may increase
per concentration the excitability of hippocampal neurons and thus contribute to the enhanced
neuronal activity associated with SO2 intoxication.
Du and 1 or 10 (jM SO2 2-4 min Wistar rat; both sexes;
Meng derivatives 10-15 days old; weight
(2004b) (1:3, NaHSOs to and number NR; N =
Na2S03) 6-13 isolated dorsal
root ganglion neurons
averaged per endpoint
Maximum sodium current amplitudes for both TTX-S and TTX-R channels were
increased by exposure to SO2 derivatives (10 or 1 (jM, respectively), with
amplitudes diminished at more negative evoking potentials and enhanced at
less negative or positive potentials. SO2 derivatives (a) slowed both current
activation and inactivation for both types of sodium channels; (b) shifted
activation currents to more positive potentials, increasing threshold voltages for
action potential generation and contributing to reduced neuron excitability; and
(c) caused even larger counteracting positive shifts in inactivation voltages
tending to increase dorsal root ganglion neuron excitability. On balance, the
data suggest micromolar concentrations of sulfite/bisulfite can increase the
excitability of dorsal root ganglion neurons, providing a basis forS02-
associated neurotoxicity.
Du and 0.01, 0.1, 0.5, or 1 Not specified, but Wistar rat; both sexes;
Meng |jM SO2 derivatives brief ("added to the 10-15 days old; weight
(2006) (1:3, NaHSOs to external solution and number NR; N =
Na2S03) just before each 6-15 isolated dorsal
experiment") root ganglion neurons
averaged per endpoint
In isolated dorsal root ganglion neurons, SO2derivatives increased HVA-Ica
amplitudes in a concentration- and depolarizing voltage-dependent manner
(ECsowas -0.4 (jM) by altering Ca channel properties. This effect was partially
reversible by SO2 derivative washout, and was PKI-inhibitable, indicating
involvement of PKA and secondary messengers. Additionally, exposure caused
a positive shift in reversal potential. SO2 derivatives also delayed activation and
inactivation of Ca channels, but the latter was more pronounced, thus overall
prolonging action potential duration and increasing HVA-Ica. Exposure also
slowed the fast component and accelerated the slow component of recovery
from Ca channel inactivation. Thus, s 1 pM sulfite/bisulfite caused prolonged
opening and altered properties of Ca channels, elevated HVA-Ica, and
abnormal Ca signaling with neuronal cell injury. Authors speculate these effects
may correlate with SO2 inhalation toxicity, perhaps leading to abnormal
regulation via peripheral neuron Ca channels of nociceptive impulse
transmission.
ACUTE / SUBACUTE / SUBCHRONIC
Wu and
Meng
(2003)
8.4, 24.4, or
56.5 ppm (22, 64,
or 148 mg/m3);
whole body
6 h/day for 7 days Kunming-strain mouse; Decreased glutathione, G6PD, and GST activities were observed in the brain
male; age NR; at 24.4 and 56.5 ppm.
18-20 g; N = 10/group
Haideretal. 10 ppm (26.2 1 h/day for 21 or
(1981) mg/m3); whole body 24 days
Guinea pig; sex NR; The effects of SO2 exposure on lipid profiles, lipid peroxidation and lipase
adult; 250-500 g; activity in three regions of the brain (cerebral hemisphere, CH; cerebellum, CB;
N = 12/group brain stem, BS) were examined. Significant (p < 0.001-0.05) findings include
(6/subgroup) reductions in total lipids (CH, BS; also CB, but nonsignificant) and free fatty
acids (CH, CB, BS). PL were elevated in CH, but reduced in CB; Choi was
elevated in CH, but reduced in CB and BS; and esterified fatty acids were
elevated in CB, but reduced in CH and BS. Levels of malonaldehyde and
lipase activity were elevated in all regions. Results indicate that subacute brief
exposures to SO2 can lead to degradation of brain lipids, with the exact nature
of the lipid alterations dependent upon brain region.
E-16
-------
Study Concentration Duration Species
Effects
Haideretal. 10ppm (26.2 1 h/dayfor
(1982) mg/m3); whole body 30 days
Charles Foster rat; The effects of SO2 exposure on lipid profiles, lipid peroxidation and lipase
male; adult; activity in three regions of the brain (cerebral hemisphere, CH; cerebellum, CB;
150-200 g; brain stem, BS) were examined. Significant (p < 0.001-0.05) findings include
N = 12/group reductions in total lipids (CH, BS, CB), while PL were elevated only in CB. Choi
(6/subgroup) was elevated in CH and CB, but not BS; and gangliosides were elevated in CB
and BS, but reduced in CH. Lipid peroxidation (malonaldehyde formation) was
elevated in whole brain and all regions (although nonsignificantly in BS), as
was lipase activity in CH, the only tissue examined. Despite regional
differences in PL and Choi changes, Chol/PL ratios were elevated in all three
brain regions (again nonsignificantly in BS). Results are somewhat different
than those seen in guinea pig (Haider et al., 1981), but again suggest that
subacute brief exposures to SO2 can lead to degradation of brain lipids, with
the exact nature of the lipid alterations dependent upon brain region.
Haiderand 10 ppm (26.2 1 h/day for Guinea pig; sex and The effects of alternating SO2+H2S exposure on lipid profiles, lipid
Hasan mg/m3)S02 30 days (alter- age NR; 250-400 g; peroxidation and lipase activity in four regions of the brain (cerebral
(1984) alternated with nating SO2 or H2S) N = 18/group in 2 hemisphere, CH; basal ganglia, BG; cerebellum, CB; brain stem, BS) and in
20 ppm groups (6/group in the spinal cord (SC) were examined. Significant (p < 0.001-0.05) findings
(14.7 mg/m3) H2S; some subgroups) include reductions in total lipids and Choi, and elevated lipid peroxidation
whole body (malonaldehyde formation) and lipase activity, in all brain regions and SC.
Chol/PL ratios were also reduced in all tissues (but nonsignificantly in BG and
CB). For other parameters (PL, free fatty acids, esterified fatty acids, and
gangliosides), changes were observed in most tissues but were region-specific.
Results indicate that subacute brief, alternating exposures to SO2 or H2S lead
to degradation of brain lipids, again with the exact nature of the lipid alterations
dependent upon brain/spinal cord region. Additionally, some of the effects
observed for this mixture vary from those seen with SO2 alone (Haider et al.,
1981; 1982).
Agaret al.
(2000)
10 ppm (26.2
mg/m3) (± iv alloxan
to induce
experimental type 1
diabetes); whole
body
1 h/day, 7 days/wk
for 6 wks
Swiss albino rat; male;
3 mos old; weight NR;
N = 10/group in
4 groups
SUBCHRONIC/ CHRONIC
Kugukatay 10 ppm (26.2 1 h/day, 7 days/wk
et al. (2003) mg/m3) (± iv alloxan for 6 wks
to induce
experimental type 1
diabetes); whole
body
Rat; male; 3 mos old;
weight not reported; N
= 10/group in 4 groups
In retina tissue, exposure elevated SOD activity and reduced GPx and catalase
activities. TBARS were elevated only in non-diabetic rats exposed to SO2. In
brain tissue, exposure elevated SOD and reduced GPx activities in both
non-diabetics and diabetics, while catalase activities were not affected; TBARS
were elevated in both non-diabetics and diabetics. With respect to VEPs,
exposure prolonged latencies in 4 of 5 VEP components in non-diabetics and 5
of 5 in diabetics, while reducing virtually all peak-to-peak amplitudes in
non-diabetics and diabetics. For many endpoints, SO2 effects were additive to
those resulting from the induced diabetic condition. In summary, brain and
retinal anti-oxidant and lipid peroxidation status, as well as neuro-visual
performance were affected by subchronic exposure to brief periods of 10 ppm
SO2, and these effects were exacerbated by a diabetic condition.
In brain tissue, SO2 exposure elevated SOD and reduced GPx activities in both
non-diabetics and diabetics, while catalase activities were not affected; TBARS
were elevated in both non-diabetics and diabetics. With respect to afferent
peripheral nerve pathways (SEPs), exposure prolonged latencies in 4 of 4 SEP
components in both non-diabetics and diabetics; also altered were some
inter-peak latencies (non-diabetics and diabetics) and some peak-to-peak
amplitudes (non-diabetics only). In some cases, SO2 effects were additive to
those resulting from the induced diabetic condition. In summary, brain
anti-oxidant and lipid peroxidation status, as well as afferent peripheral nerve
pathways, were affected by subchronic exposure to 10 ppm SO2, and these
effects were exacerbated by a diabetic condition. Authors suggest that SO2
exposure could potentiate the incidence and/or severity of diabetes.
E-17
-------
Study Concentration Duration Species
Effects
Yargigoglu 10 ppm (26.2 1 h/day, 7 days/wk Swiss albino rat; male; Effects of aging ± SO2 exposure on levels of lipid peroxidation (TBARS),
et al. (1999) mg/m3); whole body for6wks 3,12, or 24 mos old; antioxidant enzyme status (catalase, GPx, SOD), and afferent peripheral nerve
weight NR; N = pathways (SEPs) were monitored in the brain of young (Y, 3 mo), middle-aged
10/group in 6 groups (M, 12 mo) and old (0,24 mo) rats. In addition to age-related changes, SO2
exposure significantly (p < 0.0001-0.02) elevated TBARS and SOD, while
reducing GPx (Y, M, O); catalase levels were not affected. Of 4 monitored SEP
component peaks, SO2 significantly (p < 0.01-0.05) prolonged latencies in
groups Y (4/4) and M (1/4), but not in O (0/4). Peak-to-peak amplitudes were
decreased in Y, (2/3) and increased in M (1/3), but not affected in O (0/3).
Taken together, these data indicate that subchronic exposure to brief periods of
10 ppm SO2 can impact afferent peripheral nerve pathways and the lipid
peroxidation and antioxidant enzyme status of the brain.
Kilic (2003) 10ppm 1 h/day, 7 days/wk Swiss albino rat; male; Effects of aging ± SO2 exposure on levels of lipid peroxidation (TBARS),
(26.2 mg/m3); whole for 6 wks 3,12, or 24 mos old; antioxidant enzyme status (catalase, GPx, SOD), and visual system function
body weight NR; N = (VEPs) were monitored in the brain and eye (retina and lens) ofyoung (Y, 3
10/group in 6 groups mo), middle-aged (M, 12 mo) and old (O, 24 mo) rats. In addition to age-related
changes, SO2 exposure significantly (p < 0.0001-0.04) elevated TBARS in
brain and lens (Y, M, O), and in retina (Y); reduced GPx in brain (Y) and lens
(Y, M, O); reduced catalase in retina (Y, M, O); and elevated SOD in brain (Y,
M), retina (Y, M, O) and lens (M, O). Of 5 monitored VEP component peaks,
SO2 prolonged latencies in groups Y (4/5), M (3/5) and O (1/5). Taken together,
these data indicate that subchronic exposure to brief periods of 10 ppm SO2
can impact the visual system and the lipid peroxidation and antioxidant enzyme
status of the brain and eye.
NEUROBEHAVIOR
Petruzzi
et al. (1996)
5,12, or 30 ppm
(13.1,31.4, or
78.6 mg/m3); whole
body
Near continuous
(80% of time)
exposure from
9 days before
mating through the
12-14th day of
pregnancy
CD-1 mouse, adult
male and female
parental animals were
exposed (N =
10/group/sex)
(N = 8 litters/group,
fostered by unexposed
dams at birth) were
evaluated at 2-18 days
of age; adult male
offspring also
evaluated (N =
8/g roup)
Adults: Observation of behavior outside the exposure chamber on exposure
days 3, 6, and 9 revealed dose-related increases in digging and decreases in
grooming by females in the 30 ppm group on exposure day 9; non-dose related
increases were observed for crossing and wall rearing by females in the
30 ppm group on exposure day 9. Observance of behaviors in 2 breeding
pairs/group in the 12 and 30 ppm groups revealed increased rearing and social
interaction in the 30 ppm group shortly after the start of exposure, followed by
return to baseline levels; effects were generally of greater magnitude in males.
E-18
-------
Table E-13. Reproductive and developmental effects of SO2.
Study
Concentration
Duration
Species
Effects
Meng and
Bai (2004)
8.4, 21, or 43 ppm
(22, 56, or 112
mg/m3); whole body
6 h/day for 7 days
Kunming albino mouse;
male; 5 wks old; 19 ± 2
g; N = 10/group
Changes observed in mouse testes (concentrations of effects) included
decreased activities of SOD (43 ppm, possibly at 21 ppm according to text)
and GPx (a 21 ppm), increased catalase activity (8.4 and 21 ppm), decreased
GSH level (a 21 ppm), and increased TBARS levels (a 8.4 ppm). The authors
concluded that SO2 can induce oxidative damage in testes of mice.
Gunnison
et al.
(1987)
10 or 30 ppm (26.2
or 78.6 mg/m3);
whole body
6 h/day,
~5 days/wk for 21
wks (total of 99
days)
Sprague-Dawley CD
rat; male; 8 wks old;
weight NR;
N = 70/group in
3 groups (inhalation
series)
No significant (p < 0.05) effect on testes histopathology was found, although
there was a very slight and probably biologically insignificant increase in
relative testes weight. (0.61 ± 0.02 vs. 0.56 ± 0.02, % body weight).
Singh 32 or 65 ppm (83.8 Gestation day 7-18 CD-1 mouse dams
(1989) or 170 mg/m3); whole were exposed;
body numbers of dams
exposed and offspring
evaluated not indicated
No significant effects were observed for number of live pups born/litter. Pup
birth weight was lower at 65 ppm. Righting and negative geotaxis reflexes
were delayed at both concentrations.
Petruzzi
et al.
(1996)
5,12, or 30 ppm
(13.1,31.4, or
78.6 mg/m3); whole
body
Near continuous
(80% of time)
exposure from
9 days before
mating through the
12-14th day of
pregnancy
CD-1 mouse; adult
male and female
parental animals were
exposed
(N = 10/group/sex) and
male and female
offspring
(N = 8 litters/group,
fostered by unexposed
dams at birth) were
evaluated at 2-18 days
of age; adult male
offspring also evaluated
(N = 8/group)
Decreased food and water intake were observed in parental males and
females of the 12 and 30 ppm groups at the start of mating (exposure days
9-13). No effects observed for mating or successful pregnancies. There were
no effects on litter sizes, sex ratio, or neonatal mortality (data not shown by
authors). No effects observed for birth weight, postnatal body weight gain,
somatic and neurobehavioral development (e.g., eyelid and ear opening,
incisor eruption, and reflex development); no postnatal developmental data
were shown by authors. No effects observed in passive avoidance testing of
adult males.
Fiore et al.
(1998)
5,12, or 30 ppm
(13.1,31.4, or
78.6 mg/m3); whole
body
Near continuous
(90% of time)
exposure from
9 days before
mating through the
14th day of
pregnancy
CD-1 mouse; adult
male and female
parental animals were
exposed and adult
male offspring (fostered
by unexposed dams at
birth) were evaluated at
-120 days of age, N =
11-12 offspring/group
In 20-min encounters with unexposed males, prenatally-exposed males com-
pared to controls displayed (dose(s) of effect, time of testing effect observed)
increased duration of self grooming (5 ppm, 15-20 min), decreased frequency
and duration of tail rattling (a 5 ppm at 5-10 min and 12 ppm at 10-15 min),
and decreased duration of defensive postures (a 12 ppm, 0-5 min). Authors
also noted a non-significant decrease in freezing (apparently at all dose
levels) and non-significant increases in social exploration (apparently at all
doses) and rearing (apparently at a 12 ppm).
Douglas 5 ppm (13.1 mg/m3); 2 h/day for 13 wks New Zealand White
etal. whole body rabbit; male and
(1994) female;
N = 3-4/group;
1-day-old; immunized
against Alternaria
tenuis
Following subchronic exposure beginning in the neonatal period, there were
no effects on lung resistance, dynamic compliance, transpulmonary pressure,
tidal volume, respiration rate, or min volume.
E-19
-------
Table E-14. Endocrine system effects of SO2.
Study Concentration Duration
Species
Effects
Lovatiet 5 or 10 ppm (13.1 or 24 h/day for
al. (1996) 26.2 mg/m3); whole 15 days
body
Sprague-Dawley CD rat;
male; age NR; 250-275
g;
N = 9/subgroup in 9
subgroups
Subjects were rats fed standard diet (normal) or high cholesterol diet, and rats
with streptozotocin-induced diabetes fed standard diet. In diabetic rats, there
was no effect on glucose levels. Exposure to a 5 ppm lowered plasma insulin
level in normal and hypercholesterolemic diet groups, but elevated it
(non-significantly) in diabetic rats. In each rat model, inhalation of SO2 at
levels without overt effects affected insulin levels. Specific effects varied
according to diet or diabetes.
Agaret al.
(2000)
10 ppm (26.2
mg/m3); whole body
1 h/day, 7 days/wk
for 6 wks
Swiss Albino rat; male; 3
mos old; weight NR; N =
10/group
Effects were compared in non-diabetic rats and rats with alloxan-induced
diabetes. SO2 increased blood glucose in diabetic and non-diabetic rats.
Kugukatay
et al.
(2003)
10 ppm (26.2
mg/m3); whole body
1 h/day, 7 days/wk
for 6 wks
Rat, male; 3 mos old;
weight NR;
N = 10/group in 4 groups
Effects were compared in normal rats and rats with alloxan-induced diabetes.
SO2 elevated blood glucose levels in both non-diabetics and diabetics.
Table E-15. Liver and gastrointestinal effects of SO2.
Study
Concentration
Duration
Species
Effects
SUBACUTE /SUBCHRONIC
Mengetal. 7.86, 20, or 40 ppm (22,
(2003c) 56, or 112 mg/m3) per
author conversion;
whole body
6 h/day for 7 days
Kunming albino mouse,
male and female; 5 wks
old; 19 ± 2 g;
N = 6/sex/subgroup
Effects observed in stomach (concentration of effect) included:
increase in SOD activity (7.86 ppm, males only) and TBARS
level (a 7.86 ppm) and decreases in SOD (a 20 ppm, males
only) and GPx activities (a 20 ppm, males only) and GSH level
(40 ppm). Effects observed in intestine were increases in
catalase activity (a 20 ppm in males, 40 ppm in females) and
TBARS level (a 20 ppm) and decreases in SOD (a 7.86 ppm)
and GPx (a 20 ppm) activities and GSH level (a 7.86 ppm).
Wu and Meng
(2003)
8.4, 24.4, or 56.5 ppm
(22, 64, or 148 mg/m3);
whole body
6 h/day for 7 days
Kunming-strain mouse;
male; age NR; 18-20 g;
N = 10/group
No effects were observed in the liver at 8.4 or 24.4 ppm. GST
and G6PD activities and GSH level were decreased at 56.5 ppm.
Bai and Meng 5.35,10.70, or 6 h/day for 7 days
(2005b) 21.40 ppm (14, 28, or 56
mg/m3); whole body
Wistar rat; male; age NR;
180-200 g;
N = 6/group in 4 groups
Significant and concentration-dependent changes in mRNA (mid
and high concentrations) and protein expression (all
concentrations) included increases for bax and p53
apoptosis-promoting genes, and decrease for bcl-2
apoptosis-repressing gene. Authors speculated potential impact
on human apoptosis-deficient diseases.
Qin and Meng 5.35,10.70, 6 h/day for 7 days
(2005) or 21.40 ppm (14, 28, or
56 mg/m3); whole body
Wistar rat; male; age NR;
180-200 g;
N = 6/g roup in 4 groups
SO2 caused significant concentration-dependent reductions in
liver enzyme activities and gene expression for CYP1A1 and
CYP1A2. Effects were seen at the mid and high concentrations
(only high for CYP1A1 enzyme activity), but not the low. Authors
speculate that underlying mechanisms may involve oxidative
stress and/or cytokine release, and may represent an adaptive
response to minimize cell damage.
E-20
-------
Study Concentration Duration Species
Effects
Lovati et al.
(1996)
5 or 10 ppm (13.1 or
26.2 mg/m3); whole
body
24 h/day for
15 days
Sprague-Dawley CD rat;
male; age NR;
250-275 g;
N = 9/subgroup
Subjects were rats fed standard diet (normal) or high cholesterol
diet, and rats with streptozotocin-induced diabetes fed standard
diet. SO2 (a 5 ppm) elevated plasma triglycerides in normal and
hypercholesterolemic groups, while 10 ppm lowered plasma high
density lipoprotein cholesterol in hypercholesterolemic rats. In
diabetic rats, 10 ppm SO2 lowered triglycerides and free fatty
acids without affecting high density lipoprotein cholesterol or total
cholesterol. In the liver, SO2 elevated triglycerides in normal and
hypercholesterolemic groups (at 10 ppm), but lowered it in
diabetic rats (at a 5 ppm); esterified cholesterol was elevated in
normal rats (at 10 ppm), but lowered in diabetic rats (at a 5ppm),
and free cholesterol was unchanged in all groups. In normal rats,
triglycerides secretion rate was inhibited by 10 ppm SO2. SO2
caused several changes in plasma apolipoprotein composition in
normal and hypercholesterolemic groups, but not in diabetic rats.
Leukotriene parameters were not affected. Thus, in each rat
model, inhalation of S02at levels without overt effects affected
plasma and tissue lipid content. Specific effects varied according
to diet or diabetes.
Langley-Evans 5, 50, or 100 ppm (13.1, 5 h/day for
et al. (1996) 131, or 262 mg/m3); 7-28 days
whole body
Wistar rat; male; 7 wks old; GSH was depleted in the liver at 5 and 100 ppm but not at
weight NR; 50 ppm. With respect to GSH-related enzymes, exposure to
N = 4-5/treatment group, 5 ppm decreased GRed and GST activity in the liver. Exposure to
8 controls 50 ppm did not affect liver GST, but decreased liver GRed and
GPx.
Langley-Evans
et al. (1997);
Langley-Evans
(2007)
100 ppm (286 mg/m3); 5 h/day for
whole body 28 days
Units were incorrectly
reported as |jg/m3 in the
study but were corrected
according to information
provided by study author
Wistar rat; male; 7 wks old; Adult rats exposed to air or SO2 were born to dams fed diets with
weight NR; N = 4-16 varying casein contents (180 [control], 120, 90 or 60 g/kg) during
gestation. In the liver, SO2 exposure elevated GSH level in the
120 g/kg dietary group but lowered it in the 60 g/kg dietary group.
SO2 did not affect liver GST in any group. SO2 increased GCS
levels in the 180 and 90 g/kg groups, GPx in the 60 g/kg group,
and GRed in the 120 and 90 g/kg groups. This study provides
information for an extremely high concentration level but is being
acknowledged here with the unit corrected to verify that a
low-concentration level study was not missed.
Gunnison et al.
(1987)
10 or 30 ppm (26.2 or
78.6 mg/m3); whole
body
6 h/day,
~5 days/wk for
21 wks (total of
99 days)
Sprague-Dawley CD rat;
male; 8 wks old; weight
NR; N = 70/group in
3 groups (inhalation series)
No effects on relative liver weight or histopathology were found.
Agar etal. (2000) 10 ppm (26.2 mg/m3); 1 h/day, 7 days/wk Swiss Albino rat; male; Effects were compared in non-diabetic rats, non-diabetic rats
whole body for 6 wks 3 mos old; weight NR; exposed to SO2, alloxan-induced diabetic rats, and diabetic rats
N = 10/group exposed to SO2. SO2 increased blood glucose in all groups, but
did not affect total cholesterol, high density lipoprotein
cholesterol, low density lipoprotein cholesterol, very low density
lipoprotein cholesterol, or triglyceride levels in either normal or
diabetic rats.
Kugukatay et al.
(2003)
10 ppm (26.2 mg/m3);
whole body
1 h/day, 7 days/wk Rat; male; 3 mos old; Effects compared in normal rats and rats with alloxan-induced
for 6 wks weight NR; N = 10/group in diabetes. Among the significant effects observed, SO2 exposure
4 groups enhanced the body weight loss seen in the diabetic group, but
did not affect body weight gain in the control group. SO2 elevated
blood glucose levels in both controls and diabetics, but lowered
triglycerides only in diabetics. Cholesterol parameters were not
affected.
E-21
-------
Table E-16. Renal effects of SO2.
Study
Concentration
Duration
Species
Effects
Wu and Meng
(2003)
8.4, 24.4, or 56.5 ppm
(22, 64, or 148 mg/m3)
6 h/day for 7 days
Kunming-strain mouse;
male; age NR; 18-20 g;
N = 10/group
GST was decreased in the kidney at 24.4, or 56.5 ppm (64 and
148 mg/m3) and G6PD activity was decreased at 56.5 ppm
(148 mg/m3). Kidney GSH levels were reduced at all exposure
levels.
Langley-Evans
et al. (1996)
5, 50, or 100 ppm (13.1,
131, or 262 mg/m3)
5 h/day for 7-28
days
Wistar rat; male; 7 wks
old; weight NR;
N = 4-5/treatment group,
8 controls
GSH was depleted in the kidney in the 5 and 100 ppm groups but
not in the 50 ppm group. No effects were observed for other
GSH-related enzymes.
Table E-17. Respiratory System - Effect of SO2 on morphology.
Study
Concentration
Duration
Species
Effects
ACUTE/
SUBACUTE
Conner
et al.
(1985)
1 ppm (2.6
mg/m3); nose
only
3 h/day for 6 days;
animals evaluated
for up to 72 h
following exposure
Hartley guinea pig; male;
age NR; 250-320 g;
N = 14/group/time point
In combined group of S02-exposed animals and furnace gas controls, no
alveolar lesions were observed.
SUBCHRONIC/ CHRONIC
Wolff
et al.
(1989)
5 ppm (13
mg/m3); nose
only
2 h/day, 5 days/wk
for 4 wks
F344/Crl rat; male and
female; 10-11 wks old;
weight NR;
N = 3/sex/ group
No nasal or pulmonary lesions.
Smith
et al.
(1989)
1 ppm (2.62
mg/m3); whole
body
5 h/day, 5 days/wk
for 4 or 8 mos; half
the animals in the
8-mo group were
allowed to recover
for 3 mos.
Sprague-Dawley rat; male;
young adult; initial weight
NR; N = 12-15/data point
At 4 mos of SO2 exposure, increases were observed for incidence of bronchial
epithelial hyperplasia (80 vs. 40% in controls) and numbers of nonciliated
epithelial cells (31.1 vs. 27.7% in controls); neither effect persisted past 4 mos
of exposure.
Gunnison
et al.,
(1987)
10 and 30 ppm
6 h/day and 5
day/wk for 21 wks
Sprague-Dawley CD rat;
8 week of age
Mild epithelial hyperplasia in the trachea and larger bronchi, mucoid
degeneration and desquamation of epithelium of the larger bronchi.
E-22
-------
Table E-18. Respiratory System - Effects of SO2 exposure on host lung defenses.
Study
Concentration
Duration
Species
Effects
CLEARANCE
-SUBCHRONIC
Wolff et al.
(1989)
5 ppm (13.1
mg/m3); nose only
2 h/day, 5 days/wk
for 4 wks
F344/Crl rat; male and
female; 10-11 wks old;
weight NR; N =
6/sex/group
There was no effect on pulmonary clearance of radiolabeled aluminosilicate
particles (MMAD 1.0 micron).
IMMUNE RESPONSES - ACUTE,
(SUBACUTE
Jakab et al.
(1996)
10 ppm (26.2
mg/m3); nose only
4 h
Specific pathogen-free
white Swiss mouse; fe-
male; 5 wks old; 20-23
g;
N = 5/ group
No effect was observed on in situ Fc receptor-mediated phagocytosis of sheep
red blood cells by AM, which was assessed 3 days after exposure to SO2.
Clarke et al.
(2000)
10 ppm (26.2
mg/m3) SO2; nose
only
4 h
Outbred Swiss mouse;
female; age and
weight NR;
N = 10/experimental
value
No effect on in situ AM phagocytosis (data not shown) or on intrapulmonary
bactericidal activity toward Staphylococcus aureus.
Azoulay-
Dupuis et al.
(1982)
10 ppm (26.2
mg/m3); whole
body
24 h, 1 wk, 2 wks,
or 3 wks
OF1 mouse; female;
age NR; mean 20.6 g;
N = 768 (32/group)
Respiratory challenge with Klebsiella pneumoniae resulted in increased
mortality and decreased survival time in the 1, 2, and 3 wk SO2 exposure
groups compared to controls. Differences did not correlate with exposure
length.
EX VIVO
Blanquart
et al. (1995)
0,5,10, 20, 30, or
50 ppm (0,13.1,
26.2, 52.4, or
131 mg/m3) SO2
ODD
Fauve de Bourgogne
rabbit; 1 mo old;
tracheal epithelium
explants
Relative to control cultures, cell viability was not reduced at 5 and 10 ppm, but
was at 30 ppm (-70%) and 50 ppm (-60%). Ciliary beat frequency was
significantly reduced (p < 0.05) at 10-30 ppm, and was correlated with swollen
mitochondria and depletion of cellular ATP, as well as with blebbing of ciliated
or m icrovi 11 i-covered cells and with aggregation and flattening of cilia.
Riechelmann 2.9,5.7,8.6,11.5, 30min Guinea pig; sex, age,
et al. (1995) or14.3ppm and weight NR;
(7.5,15,22.5,30, N = 4-8/group
or 37.5 mg/m3); ex
vivo exposure of
trachea
Knorstetal. 2.5,5.0,7.5,10.0, 30min Guinea pig; sex, age,
(1994) or 12.5ppm (6.6, andweightNR;
13.1,19.7, 26.2, or N = 4-7/group
32.8 mg/m3); ex
vivo exposure of
trachea
No remarkable morphologic abnormalities in the tracheal mucociliary system
of the 2.9 ppm group, though slight vacuolization, rare membrane blebs, and
slightly widened intercellular spaces were observed. Abnormalities in the 5.7
and 8.6 ppm groups were similar and included loosened contact to the basal
membrane, extensive intracellular edema and vacuolization, swollen
mitochondria, polypoid extrusions and huge blebs in the cell membrane and
ciliary membrane, widened intercellular space, and disrupted tight junctions.
Additional abnormalities in the 11.5 and 14.3 ppm groups included marked
epithelial sloughing, occasionally disrupted cell membranes and microtubules,
and frequently disrupted ciliary membranes. Tracheal mucociliary activity was
significantly decreased in all exposure groups (from 8.7 ± 1.0 Hz [controls] to
4.0 ± 1.1, 3.4 ± 2.7,1.8 ± 2.2,1.5 ± 1.8, and 2.0 ± 1.2 Hz in the 7.5,15, 22.5,
30, and 37.5 mg/m3 groups, respectively).
63% decrease in tracheal mucociliary activity at 2.5 ppm with dose-dependent
decrease to 81% at 7.5 ppm; higher concentrations did not further decrease
mucociliary activity. Ciliary beat frequency decreased by 45% at 5.0 ppm with
dose-dependent decrease to 72% at 12.5 ppm. All reductions are relative to
baseline values; no effect on controls for either parameter.
E-23
-------
Table E-19. Genotoxic effects of SO2 and metabolites.
Study Concentration Duration Species/System
Effects
"POINT MUTATION"11N VITRO
Pool-Zobel 0 or 50 ppm (131 48 h
Rat, Sprague-Dawley,
etal. mg/m3)S02or
female, liver enzyme
(1990) the equivalent
preparations
agar concen-
tration ofS032-,
15 (jg/ml)
CYTOGENETIC AND DNA DAMAGE2 IN VITRO
Pool etal. 0,20, 50 or 1-24 h
Hamster, Syrian
(1988b) 200 ppm (0,
golden; fetal lung cells
52.4,131 or 524
(FHLC, gestational
mg/m3) SO2
day 15)
0,0.1,0.2 or 0.4
Rat, Sprague- Dawley;
mM SO32"
male; age NR; ~200g,
0 or 2.5 |_imol
hepatocytes
HSO3- per
Chinese hamster
microtiter plate
ovary cell line
well
transformed by SV40,
0,0.1,0.2 or 0.4
CO60 cells
mM SO42-
Precinorm U (human
0 or 10 |_imol
serum standard)
MgS04 per tube
In vitro induction of reverse mutation in cultures of S. typhimurium strain TA98 was not
affected by incubating the bacterial-B[a]P-liver S9 enzyme activation system in the
presence ofSCWsulfite. An ancillary finding from the 0 |jg B[a]P control exposures is that
SCWsulfite itself did not appear mutagenic.
Toxicity and genotoxicity of SO2, sulfite/bisulfite and sulfate (also NCWNOx) were variously
assessed in several in vitro test systems. It was noted that medium pH remained stable at
[SO2] s 200 ppm. Precinorm LDH activity was substantially inhibited by 50 ppm SO2 after
1-3 h, and by 0.1 mM sulfite ion almost immediately, but not by 0.1 mM sulfate ion; AST was
modestly inhibited after 5 h by 200 ppm SO2; other monitored enzymes were not affected.
While trypan blue exclusion was not affected, SO2 cytotoxicity to FHLC was demonstrated
at 20 ppm by reduced plating efficiency; at 50 ppm, enzyme activity leaked into culture
medium was reduced only forAP and especially LDH (not other enzymes). 200 ppm SO2
did not induce DNA damage (single-strand breaks) by itself in either FHLC or rat hepato-
cytes, but did somewhat reduce that induced by AMMN. In hepatocytes, incubation with
MgSCU also caused a small reduction in AMMN-induced DNA damage. A 1 h exposure to
200 ppm SO2 did not induce selective amplification of SV40 DNA in CO60 cells, nor affect
that induced by DMBA or B[a]P. However, while also not affecting induction by DMBA or
B[a]P, HSO3" added directly to the medium for 24 h did induce SV40 DNA amplification on
its own. Authors appear to suggest this might result from arrest of cells in mid-S phase,
which leads to DNA amplification. Thus, principal findings include inhibition of LDH by SO2
or sulfite that could impair the cellular energy system; such an impairment could be respon-
sible (possibly along with SO42- conjugation of reactive intermediates) for the observed
inhibition of AMMN-induced DNA damage by SO2. Further, SO2 does not appear by itself to
induce DNA damage.
Shi and
Mao
(1994)
3 mM SO32-
40 min
(test tube
reactions)
dG or DNA
Test tube reaction mixtures that caused sulfite to oxidize to sulfur trioxide radical resulted in
the hydroxylation of dG (8-OHdG) and the generation of DNA double strand breaks.
Shi (1994)
5 mM SO32- (as
Na2S03)
1.5 h
(test tube
reaction)
Dg
Test tube reaction of sulfite ion with H2O2 shown to generate hydroxyl radicals capable of
hydroxylating dG to the DNA damage marker, 8-OHdG. Furthermore, incubation of sulfite
with nitrite or various transition metal ions was shown to generate sulfur trioxide anion
radical.
CYTOGENETIC AND DNA DAMAGE2 ACUTE / SUBACUTE
Ruan et al.
(2003)
0 ppm SO2 (+0 ± SSO ip on Kunming mouse; male Subacute inhalation of 10.7 ppm (28 mg/m3) SO2 induced a significant (p < 0.001) 10-fold
or 8 mg/kg bw
SSO) or 10.7
ppm (28 mg/m3)
SO2 (+0,2, 4,6
or 8 mg/kg bw
SSO); whole
body
days 1-3; and female;
then SO2 for 20-25 g; N =
5 day (Days 6/sex/conc.
4-8),
6 h/day
-6 wk old; increase in mouse bone marrow MNPCE, which was partially mitigated in dose-dependent
fashion by pretreatment with SSO, a complex natural anti-oxidant substance. SO2 exposure
also resulted in organ:bw ratios that increased for liver and kidney, decreased for lung and
spleen, and remained unchanged for heart. Such ratio changes were largely mitigated by
SSO pretreatment.
Meng 0,5.35,10.7, 4 h/day for Kunming mouse; male In vivo exposure caused significantly (p < 0.01-0.001) increased frequencies of bone
etal. 21.4, or 32.1 7 days and female; ~6wk old; marrow MNPCE similarly in both sexes at all concentrations in a dose-dependent manner,
(2002) ppm (0,14, 28, 20-25 g; N = and with only minimal cytotoxicity at the 3 highest concentrations. The level of MNPCE (%)
56, or 84 mg/m3) 10/sex/conc. even at the low SO2 conc. was triple that ofthe control value. Thus, subacute inhalation of
SO2; whole body SO2 at noncytotoxic concentrations (though still notably higher than most human
exposures) was clastogenic in mice.
E-24
-------
Study Concentration Duration Species/System
Effects
Meng 0,5.35,10.7, 6 h/day for Kunming mouse; male Following in vivo exposure to SO2, it was shown by the single cell gel electrophoresis
etal. 21.4,or 32.1 7 days and female; ~5wk old; (comet) assay that such exposure induced significant (p < .001-.05) dose-dependent DNA
(2005b) ppm (0,14, 28, 18-20 g; N = damage (presumed mostly to be single-strand breaks and alkali-labile sites) in cells isolated
56, or 84 mg/m3) 6/sex/conc. from brain, lung, liver, intestine, kidney, spleen, and testicle, as well as in lymphocytes, and
SO2; whole body beginning at the lowest concentration (except male intestine—lowest response at 10.7 ppm
[28 mg/m3]). Results demonstrate that SO2 can cause systemic DNA damage in many
organs, not just the lung. Authors note that potential occupational exposures and the fact
that the obligate nose-breathing mouse removes -95% of inhaled SO2 in its nasal passages
make this experimental concentration range relevant to possible human exposures.
Pool etal. 0 or 50 ppm 24 h/day,
(1988a) (131 mg/m3) SO2 7 day/wk,
for 2 wks
Rat, Sprague-Dawley;
female; 4 mo old; wt
NR; N = 5/group
CYTOGENETIC AND DNA DAMAGE2SUBCHRONIC / CHRONIC
Ohyama
0, 0.2 mL C, or
et al.
(0.2 mL DEP+C
(1999)
±[4 ppm (10.48
mg/m3) SO2 or
6 ppm (11.28
mg/m3) NO2 or
4 ppm SO2+ 6
ppm NO2]);
whole body
[Note: 0.2 mL
C=1 mg; 0.2 mL
DEcCBP=1 mg
C + 2.5 mg DEP]
SO2 and/or
N02:
16 h/day for
10 mos
C or
DEP+C:
4wk,
once/wk by
intratracheal
infusion
Assessments were conducted on isolated primary lung and liver cells, or on blood serum. In
vivo SO2 exposure did not affect viability (trypan blue exclusion) of cells either immediately
after isolation or after 1 h incubation with 1% DMSO (used for enzyme leakage assays). In
contrast to controls, hepatocytes from S02-exposed rats released no LDH activity into
DMSO-medium after 1 h, and AST activity was reduced. Other enzyme (AP, ALT, GT)
activity releases were not affected in lung cells, and none were in hepatocytes. In blood
serum, the only effect was a marked increase in LDH activity. The only significant
(p < 0.001- 0.01) exposure effects on lung or liver activities (in x 9000 g supernatants of cell
homogenates) of xenobiotic metabolizing enzymes (AHH, NDMA-D, GST) were elevated
NDMA-D in the liver and reduced GST in the lung. Single-strand DNA breakage induced by
three nitroso compounds (AMMN, NDMA, NMBzA) was reduced in hepatocytes from
S02-exposed rats. Authors discuss possible mechanisms for the observed effects, and note
they are similar to in vitro effects reported elsewhere (Pool et al.,
Rat, SPF F344/Jcl;
male; 6 wk old; wt NR;
N = 23-30/group in 6
groups
Purpose was to study effects of DEP on rat lung tumorigenesis and possible tumor
promoting effects of SO2 or NO2 singly or together. (See Table E-11 for tumor-related
effects.) DEP extract-DNA adducts were found only in the three gas-exposed groups.
Chromatograms revealed two different adducts, one of which appears somewhat more
abundant with SO2 coexposure, the other substantially more so with NO2; combined
coexposure of both gases with DEP+C produced an adduct chromatogram appearing to be
a composite of those for the individual gases. Thus, SO2 and NO2 appear capable of
promoting the genotoxicity of DEP extract, though perhaps not in identical fashion.
1 Encompasses classical mutant selection assays based upon growth conditions under which mutants (or prototrophic revertants), but not the wi Id type (or auxotrophic) popu lation
treated with the test agent, can successfully grow (e.g., "Ames test," CHO/HGRPT or mouse lymphoma L5178Y/TK mammalian cell systems, various yeast and Drosophila systems,
etc.); while most viable mutation events detected in these assays are typically "point" mutations (DNA base substitutions, small deletions orframeshifts, etc.), some may involve larger
losses/rearrangements of genetic material.
2Encompasses CA, induction of MN or SCE, aneuploidy/polyploidy, DNA adduct and crosslink formation, DNA strand breakage, etc.
E-25
-------
Table E-20. Respiratory System - Effects of SO2 and metabolites on biochemistry.
Study
Concentration
Duration
Species
Effects
IN VITRO—PRIMARY / NONPRIMARY
Li and Meng.
(2007)
0.1 NaHSOs and Na2S03
1:3
4 h, followed by harvest at 0- BEP2D cell line of
24 h human bronchial
epithelial cells
Increased mRNA and protein levels of MUC5AC and IL-
13
Menzel etal. 0, 0.1, 2, 20, or 40 mM (0, 4, 80,
(1986) 800, or 1600 (jg/mL) S032-
-1 min - 96 h
Rat,
Sprague-Dawley;
200-250g; sex,
age, and n NR;
lung cells and
liver cells.
Human
lung-derived cell
line, A549
OXIDATION AND ANTIOXIDANT DEFENSES - (SUBACUTE / SUBCHRONIC)
Mengetal. 8.4, 21, or 43 ppm (22, 56, or 112 6 h per day for 7 days
(2003b) mg/m3); whole body
Kunming albino
mouse; male and
female; 5 wks old;
19 ± 2 g; N =
10/sex/group
This study focused on intracellular covalent reactions of
sulfite with primarily proteinaceous sulfhydryl compounds
in cells isolated from rat lung and rat liver (for some
comparative purposes), as well as in the human
lung-derived cell line, A549. Sulfitolysis of protein disulfide
bonds results in formation of cysteine S-sulfonate, and
sulfitolysis of GSSG in formation of GSSO3H. The latter
was formed in dose-dependent fashion upon the addition
of sulfite to A549 cells. In addition to fibronectin and
albumin, this study identified a third sulfite-binding protein
in rat lung cytosol. GSSO3H was shown to be a potent
competitive inhibitor of GST in rat lung, liver and A549
cells. Results suggest that SO2 could affect the detoxica-
tion of PAHs and other xenobiotics via formation of
GSSO3H and subsequent inhibition of GST and enzymatic
conjugation of GSH with reactive electrophiles.
Changes observed in lung tissue (concentrations of effect)
included higher SOD activity in males (8.4 ppm) and
females (8.4 and 21 ppm), lower SOD activity in males (21
and 43 ppm) and females (43 ppm), increased GPx
activity in males and females (8.4 ppm), decreased GPx
activity in males and females (a 21 ppm), decreased
catalase activity in males (43 ppm), decreased reduced
GSH level in males and females (a 8.4 ppm), increased
TBARS level in males (a 8.4 ppm) and females
a 21 ppm). Authors concluded that SO2 induced oxidative
damage in lungs of mice.
Wu and 8.4, 24.4, or 56.5 ppm (22, 64, or 6 h/day for 7 days
Meng (2003) 148 mg/m3); whole body
Kunming-strain Glucose-6-phosphate dehydrogenase and GST activity
mouse; male; age were decreased in lung at 24.4 and 56.5 ppm. Lung GSH
NR; 18-20 g; levels were reduced in the 8.44 and 56.5 ppm exposure
N = 10/group groups. Administration ofbuckthorn seed oil increased
GST and decreased TBARS activity compared to mice
exposed to SO2 alone.
Langley- 5, 50, or 100 ppm
Evans et al. (13.1,131, or 262 mg/m3);
(1996) whole body
5 h/day for 7-28 days Wistar rat; male; 7 In the 5 and 100 ppm groups, GSH in BAL fluid decreased
wks old; weight at 7 days and increased at 21 days; at 28 days GSH
NR; returned to normal in the 5 ppm group and further
N = 4-5/treatment increased in the 100 ppm group. GSH was depleted in the
group, 8 controls lung, at 5 and 100 ppm but not at 50 ppm. With respect to
GSH-related enzymes, exposure to 5 ppm lowered GCS,
GPx, GST, and GRed activity in the lung. Effects in the
100 ppm group were similar to the 5 ppm group, except
that lung GPx was not reduced. Exposure to 50 ppm did
not affect lung GST, but reduced the number of
inflammatory cells in circulation and decreased GCS, GPx,
GRed, and GT in the lung. Authors concluded that
sulfitolysis of glutathione disulphide occurs in vivo during
SO2 exposure and that SO2 is a potent glutathione
depleting agent, even in the absence of pulmonary injury.
E-26
-------
Study Concentration
Duration
Species
Effects
Gumu§lu
et al. (2001)
10 ppm (26.2 mg/m3); whole body 1 h/day, 7 days/wk for 6 wks
Swiss albino rat; Effects of age on S02-induced oxidative effects in lung
male; tissue were observed in young (3-mo-old), middle aged
3,12, or 24 mos (12-mo-old), and old (24-mo old) rats. SO2 exposure
old; 210-450 g; N significantly elevated TBARS, SOD, GPx, and GST in all
= 9-11/group in 6 age groups; reduced catalase in young and middle-aged
groups rats, but did not affect catalase in old rats. In rats not
exposed to SO2, SOD, GPx and GST increased with age
and catalase decreased with age. General observations in
S02-exposed animals were increases in SOD, GPx, and
TBARS with age. The authors noted that while lipid
peroxidation increased with age, relative TBARS increases
in response to SO2 were inversely correlated with age (i.e.,
largest percent increase seen in young rats).
Langley- -101 ppm (by study author
Evans et al. calculations 286 mg/m3); whole
(1997; 2007) body
Note: The study mistakenly listed
units of (jg/m3 and it was verified
with the authors that the units
should have been listed as mg/m3.
5 h/day for 28 days
Wistar rat, male,
7 wks old, weight
NR, N = 4-16
DIFFERENTIAL GENE EXPRESSION - SUBACUTE
Qin and 5.35,10.70, or 21.40 ppm (14, 28, 6 h/day for 7 days
Meng (2005) or 56 mg/m3); whole body
Wistar rat; male;
age NR;
180-200 g; N =
6/group in
4 groups
This study explored the effects of maternal diet protein
restriction during gestation on offspring lung enzyme
responses after SO2 exposure in adulthood. Adult offspring
representing different maternal dietary concentrations of
casein (180 [control], 120, 90 or 60 g/kg) were exposed
either to air or SO2. GSH levels in BAL fluid and the lung
were not affected either by maternal diet or SO2 exposure.
In the lung GRed and GT were not affected by SO2 in any
maternal diet group; GPx was reduced only in the 120 g/kg
maternal diet group; GCS was elevated in the 180 and 60
g/kg groups; and GST was reduced in the 180,120 and 90
g/kg groups (to the level seen in both the air- and
S02-exposed 60 g/kg maternal diet groups). This study
does not provide information relevant to ambient
exposures, but is being mentioned in this table to record
that a low-concentration level study was not overlooked.
Repeated acute exposure caused significant (p <
0.001-0.05) concentration-dependent reductions in
enzyme activities and gene expression in the lung for both
CYP1A1 and CYP1A2. Effects were seen at the mid and
high concentrations, but not the low. Authors speculate
that underlying mechanisms may involve oxidative stress
and/or cytokine release, and may represent an adaptive
response to minimize cell damage.
BaiandMeng 5.35,10.70, or 21.40 ppm (14, 28, 6 h/day for 7 days
(2005a) or 56 mg/m3); whole body
Wistar rat; male; SO2 exposure caused significant concentration-dependent
age NR; 180-200 changes in the mRNA (mid and high concentrations) and
g; N = 6/g roup in protein expression (all concentrations in lung, but
4 groups statistical significance not indicated) of apoptosis-related
genes: increases for bax and p53 apoptosis-promoting
genes, and decreases for the apoptosis-repressing gene
bcl-2. Caspase-3 activity (occurring early in apoptosis
process) was also increased at the mid and high
concentration.
E-27
-------
Table E-21. Lymphatic system effects of S02 and SO2 mixtures.
Study
Concentration
Duration
Species
Effects
SUBCHRONIC
/ CHRONIC
Smith et al.
(1989)
1 ppm
(2.62 mg/m3);
whole body
5 h/day, 5
days/wk for
4 mos
Sprague- Dawley rat; male;
young adult; initial weight
NR; N=12-15/data point
No significant effects were reported for spleen weight or mitogen-induced
activation of peripheral blood lymphocytes or spleen cells (data not shown by
authors).
Aranyi et al.
(1983)
5.0 ppm (13.2
mg/m3)
SO2+1.04
mg/m3
ammonium sul-
fate + 0.10 ppm
(0.2 mg/m3) O3;
whole body
5 h/day 5
days/wk for
up to 103
days
CD1 mouse; female;
3-4 wks old; weight NR; N =
360/group total
(14-154/group in each
assay)
Cytostasis of MBL-2 leukemia target cells by peritoneal macrophage was
increased in groups exposed to O3 alone or a mixture of the three compounds
but was significantly higher with the mixture than with O3 alone at a
macrophage:target cell ratio of 10:1; no significant effects were observed with
macrophage:target cell ratio of 20:1. Reduction in splenic lymphocyte
blastogenesis in response to phytohemagglutinin and concanavalin A occurred
after exposure to O3 alone, but increased response occurred after exposure to
the mixture; no response to alloantigen occurred after exposure to O3 alone but
increased response occurred after exposure to mixture; there were no effects on
S. typhosa lipopolysaccharidewith either exposure scenario.
E-28
-------
Annex F. Epidemiologic Studies
This Annex summarizes the findings of epidemiologic studies that have been published since the
previous review. Descriptions of older studies were presented in the 1982 AQCD for Sulfur Oxides (U.S.
EPA, 1982), and are not described in great detail in this document.
Table F-1. Short-term exposure to SO2 and respiratory morbidity in field/panel studies.
Study
Method
Pollutant Data
Findings
Adamkiewicz et al. Panel study of 29 non-smoking adults 24-h avg SO2:12.5 ppb
(2004)
Steubenville, OH
Period of Study:
Sep 2000-Dec 2000
(median age 70.7 yrs) in Steubenville,
OH. Participants provided breath
samples weekly to determine the
association between air pollution levels
and eNO. ANOVA and GLM used in
analysis.
Max: 50.9; 25th: 5.4
75th: 16.9; IQR: 11.5
1-h max SO2:14.8 ppb
Max: 233.9; 25th: 3.7
75th: 15.2; IQR: 11.5
Copollutants:
N02 NO 03 PM2.5
No associations were observed with SO2.
Single pollutant models
Effect (95% CI) per IQR increase in SO2:
Current hour: 0.09 (-0.22, 0.41)
24-h moving avg: 0.24 (-0.61,1.10)
Delfino et al.
(2003a)
Los Angeles, CA
Period of Study:
Nov 1999-Jan 2000
Panel study of 22 Hispanic children with
asthma aged 10 to 16 yrs. Participants
performed twice-daily PEF
measurements and filled out symptom
diaries. Analyses of symptoms conducted
using GEE with exchangeable
correlation. Linear mixed model used for
PEF analyses. GEE models controlled
for respiratory infections (data available
for 20 subjects) and temperature.
1-h max SO2: 7.0 ppb
(SD 4.0); IQR: 4.0
8-h max SO2:4.6 ppb
(SD 3.0); IQR: 2.5
Copollutants:
03 (r = -0.19)
N02 (r = 0.89)
CO (r = 0.69)
PM10 (r = 0.73)
EC (r = 0.87)
OC (r = 0.83)
VOCs
None of the VOCs or gaseous pollutants associated with PEF.
Current-day, but not previous-day, SO2 concentrations associated
with symptom score >1 and >2.
OR for symptom score >1 per IQR increase in SO2:1-h max SO2:
Lag 0:1.31 (1.10,1.55); Lag 1:1.11 (0.91,1.36)
8-h max S02: Lag 0: 1.23 (1.06,1.41); Lag 1:1.11 (0.97,1.28)
OR for symptom score >2 per IQR increase in SO2:1-h max SO2:
Lag 0:1.37 (0.87, 2.18); Lag 1: 0.76 (0.35,1.64)
8-h max S02: Lag 0: 1.36 (1.08,1.71); Lag 1: 0.91 (0.51,1.60)
Mortimer et al.
(2002)
Eight urban areas:
Baltimore, MD;
Bronx, NY; Chicago,
IL; Cleveland, OH;
Detroit, Ml; East
Harlem, NY; St.
Louis, MO;
Washington, DC
Period of Study:
Jun-Aug 1993
Panel study of 846 asthmatic children 4-9 3-h avg SO2
yrs from the National Cooperative Inner- (8 a.m.-11 a.m
City Asthma Study (NCICAS). Study
children either had physician-diagnosed
asthma and symptoms in the past
12 mos or respiratory symptoms consis-
tent with asthma that lasted more than 6
wks during the previous yr. Respiratory
symptoms recorded in daily diary and
included cough, chest tightness, and
wheeze. Mixed effects models and GEE
models used to evaluate the effect of air
pollutants on PEF and respiratory symp-
toms. Models adjusted for day of study,
previous 12-h avg temperature, urban
area, diary number, rain in the past 24 h.
for all
8 areas (shown in figure):
22 ppb
Avg intradiary range:
53 ppb
Copollutants:
03 (r = 0.29)
NO2 PM10
None of pollutants associated with evening PEF or evening
symptoms. Using single-pollutant model, SO2 had little effect on
morning PEF (data not shown). Significant associations between
moving avg of 1 - to 2-day lag of SO2 and incidence of morning
asthma symptoms.
OR for morning symptoms associated with 20 ppb increase in 3-h
avg SO2 concentration (Lag 1-2 day):
8 urban areas: Single-pollutant model: 1.19 (1.06,1.35)
S02 with 03 model: 1.18(1.05,1.33)
7 urban areas: Single-pollutant model: 1.22 (1.07,1.40)
S02 with 03 and NO2 model:1.19 (1.04,1.37)
3 urban areas: Single-pollutant model: 1.32 (1.03,1.70)
SO2 with O3, NO2, and PM10 model:1.23 (0.94,1.62)
F-1
-------
Study
Method
Pollutant Data
Findings
Neas et al. (1995)
Uniontown, PA
Period of Study:
Summer 1990
Panel study of 83 fourth-fifth graders in
Uniontown, Pennsylvania. Participants
reported twice-daily PEF and presence of
cold, cough, or wheeze. During summer
of 1990, there were 3,582 child-days.
PEF analyzed with autoregressive linear
regression model that included a separ-
ate intercept for evening measurements,
trend, temperature and 12-h avg air
pollutant concentration, weighted by the
number of hours child spent outdoors
during the previous 12 h.
12-h avg SO2:10.2 ppb
Max: 44.9; IQR: 11.1
Daytime 12-h avg SO2
(8 am-8 pm): 14.5 ppb
Overnight 12-h avg SO2
(8 pm-8 am): 5.9 ppb
Copollutants:
PM10 PM2.5 O3
total sulfate particles
particle-strong acidity
(r= 0.44)
Incidence of new evening cough episodes significantly
associated with the preceding daytime 12-h avg SO2. Mean
deviation in PEF not associated with SO2.
Effects associated with 10 ppb increase in 12-h avg SO2: Change
in mean deviation in PEF: —0.63 L/min (-1.33, 0.07)
OR for evening cough: 1.19 (1.00,1.42)
Concentration weighted by proportion of hours spent outdoors
during prior 12-h:
Change in mean deviation in PEF: -1.25 L/min (-2.75, 0.25)
OR for evening cough: 1.53 (1.07, 2.20)
Newhouse et al.
(2004)
Tulsa, OK
Period of Study:
Sep-Oct 2000
Panel study of 24 patients 9-64 yrs with
physician-diagnosed asthma. Subjects
performed twice-daily PEF (morning and
evening) measurements, and recorded
medications, symptoms. Simple linear
regression, forward stepwise multiple
regression, correlation analysis per-
formed. Multiple regression analyses
used to develop predictive models for
other environmental factors. Analyses
produced complex models with different
predictor variables for each symptom.
24-h avg SO2:0.01 ppm
Range: 0.00, 0.02
Copollutants:
PM2.5 CO O3 pollen
fungal spores
Of the atmospheric pollutants, avg and max O3 were most
significant factors that influenced symptoms. Quantitative results
not provided for SO2.
Avg or max SO2 found to be negative predictors of asthma in
subgroup analyses of women and nonsmokers and rhinitis in all
patients. Avg SO2 also negative predictor of evening PEF.
Quantitatively useful effect estimates not provided.
Ross et al. (2002)
East Moline, IL
Period of Study:
April-Oct 1994
Panel study of 59 asthmatics 5-49 yrs.
Analysis based on 40 subjects, due to
withdrawal or failure to provide requested
health data. Study assessed the effect of
single and combined exposures to air
pollutants and airborne allergens on PEF,
symptom scores and medication use
frequency. Multivariate linear-regression
models with 1 st order autoregression
used for analysis of daily means of mean
-standardized PEF, symptom scores and
asthma medication use; logistic regres-
sion used for dichotomized data for
symptom score and medication use, log-
linear models for log-transformed symp-
tom scores and medication use
frequency.
24-h avg SO2:
3.4 ppb (SD 3.1)
Median: 2.8
IQR: 2.4
Range: 0, 27.3
Copollutants:
PM10O3NO2 pollen fungi
No associations observed with SO2.
No effect estimates provided.
F-2
-------
Study
Method
Pollutant Data
Findings
Schildcrout et al.
(2006)
Albuquerque, NM;
Baltimore MD;
Boston MA; Denver,
CO; San Diego, CA;
Seattle, WA; St.
Louis, MO; Toronto,
Ontario, Canada
Period of Study:
Nov 1993-Sept 1995
Meta-analysis of 8 panel studies with 990
children of the Childhood Asthma
Management Program (CAMP), during
the 22-mos prerandomization phase to
investigate effects of criteria pollutants on
asthma exacerbations (daily symptoms
and use of rescue inhalers). Poisson
regression and logistic regression
models used in analyses. Within city
models controlled for day of wk, ethnicity,
annual family income, flexible functions
of age and log-transformed sensitivity to
the methacholine challenge using natural
splines with knots fixed at 25th, 50th, and
75th percentiles. Also controlled for
confounding due to seasonal factors. All
city-specific estimates included in
calculations of study-wide effects except
Albuquerque where SO2 data were not
collected.
24-h avg SO2: Median
(10th, 25th, 75th, 90th
percentile):
Albuquerque: NA
Baltimore: 6.7 ppb
(3.2,4.7,9.8,14.2)
Boston: 5.8 ppb
(2.7,3.7,9.1,14.1)
Denver: 4.4 ppb
(1.2,2.5,6.7,9.5)
San Diego: 2.2 ppb
(1.2,1.7,3.1,4.4)
Seattle: 6.0 ppb
(3.7, 4.7, 7.5, 9.5)
St. Louis: 7.4 ppb
(3.9,5.3,10.7,13.6)
Toronto: 2.5 ppb
(0.2,1.0,4.8,8.8)
Copollutants:
03 (-0.03 < r < 0.44)
N02 (0.23 s rs 0.68)
PM10 (0.31
-------
Study
Method
Pollutant Data
Findings
Boezen et al. (1998) Panel study of 189 adults (48-73 yrs) w/
Amsterdam and and w/out chronic respiratory symptoms
.. I ,, in urban and rural areas to investigate
Netherlands whether BHRand PEF-variabilitycan be
used to identity subjects who are
Winter of 1993-1994 susceptible to air pollution. Spirometry
and methacholine challenge were
performed and subjects with a fall in
FEVi of 20% or greater were considered
BHR. Subjects performed twice-daily
peak flow for 3 mos. A subject's basal
PEF variability calculated over an 8-day
period with low air pollution. PEF
variability expressed as (highest PEF-
lowest PEF/mean) or amplitude % mean
PEF. After calculation of daily PEF
variability, number of days where the
amplitude % mean was greater than 5%
was determined. This resulted in 2
groups of subjects; those with amplitude
% mean PEF of 5% or less every day in
the 8-day period, and those with an
amplitude % mean PEF greater than 5%
on at least 1 day. Effects of air pollutants
on prevalence of symptoms assessed
with logistic regression models that
adjusted for autocorrelation of the
residuals, daily min temp, time trend and
weekends/holidays.
24-h avg SO2
Urban
Mean: 11.8 |jg/m3
Range: 2.7, 33.5
Rural
Mean: 8.2
Range: 0.8, 41.5
Copollutants:
PM10BS N02
No association between SO2 and respiratory symptoms in
subjects with no BHR, BHR at a cumulative dose of methacholine
s 2.0 mg or s 1.0 mg. In subjects with amplitude % mean PEF >
5% any day and those with amplitude % mean PEF > 5% for >
33% of days, SO2 was associated with the prevalence of phlegm.
Odds ratio (per 40 |jg/m3 SO2)
Subjects with no BHR
URS: 0.86 (0.73, 1.03). LRS: 1.15 (0.90,1.46)
Cough: 1.01 (0.84,1.21). Phlegm: 1.01 (0.86,1.20)
BHR at cumulative dose of methacholine s 2.0 mg:
URS: 1.11 (0.78, 1.56). LRS: 1.03 (0.72,1.47)
Cough: 0.89 (0.66,1.19). Phlegm: 1.03 (0.78,1.37)
BHR at cumulative dose of methacholine s 1.0 mg:
URS: 1.02 (0.65, 1.61). LRS: 0.96 (0.63,1.47)
Cough: 0.96 (0.64,1.44). Phlegm: 1.00 (0.68,1.46)
Amplitude % mean PEF s 5%:
URS: 0.82 (0.62, 1.08). LRS: 1.38 (0.93, 2.03)
Cough: 0.72 (0.52, 0.98). Phlegm: 0.79 (0.59,1.05)
Amplitude % mean PEF > 5%, any day:
URS: 1.04 (0.88, 1.23). LRS: 1.14 (0.96,1.36)
Cough: 1.07 (0.90,1.26). Phlegm: 1.23 (1.05,1.43)
Amplitude % mean PEF > 5%, > 33% of days:
URS: 1.10(0.85, 1.41). LRS: 1.14(0.91,1.42)
Cough: 1.14(0.89,1.47). Phlegm: 1.36 (1.14,1.63)
F-4
-------
Study
Method
Pollutant Data
Findings
Boezen et al. (1999)
Amsterdam and
Meppel (urban and
rural), the
Netherlands
Period of Study:
3 winters of 1992-
1995
Panel study of 632 children (7 to 11 yrs)
living in rural and urban areas of the
Netherlands, to investigate whether
children with bronchial hyperrespon-
siveness (BHR) and relatively high serum
concentrations of total IgE were suscep-
tible to air pollution. 459 children had
complete data. Methacholine challenge
performed to determine BHR. Serum
total IgE higher than the median (60kU/L)
were defined as relatively high. Peak flow
was measured twice daily and lower and
URS were recorded daily for 3 mos.
Association between symptoms and air
pollutants assessed using logistic
regression that adjusted for daily min
temp, linear, quadratic and cubic time
trend, weekends and holidays, and
incidence of influenza. Examined 0,1,2
Lags and 5 day mean of air pollutants.
1992-9:
Urban areas-
Mean: 22.5 (jg/m3,
Range: 1.4, 61.3
Rural areas-
Mean: 9.8
Range: 1.3, 34.2
1993-4:
Urban areas-
Mean: 11.8,
Range: 2.7, 33.5
Rural areas-
Mean: 8.2,
Range: 0.8, 41.5
1994-5:
Urban areas-
Mean: 8.3,
Range: 0.6, 24.4
Rural areas-
Mean: 4.3,
Range: 0.5,17.0
Copollutants:
PM10 BS N02
For children with BHR and relatively high serum total IgE, the
prevalence of LRS was associated with increases in PM10, BS,
SO2, and NO2. In the group with no BHR and relatively low IgE,
and the group with BHR and low IgE, there was no consistent
association between air pollutants with symptoms or decreased
PEF. In children with no BHR but relatively high serum total IgE,
there was a 28% to 149% increase in the prevalence of LRS per
40 |jg/m3S02.
Odds ratio (per 40 |jg/m3 SO2)
Children with BHR and relatively high IgE (N = 121):
LRS: Lag 0:1.45 (1.13,1.85)
Lag 1:1.41 (1.09,1.82). 5-day mean: 2.25 (1.42, 3.55)
URS: Lag 0:1.17 (0.99,1.38). Lag 1:1.06 (0.90,1.25)
>10% morning PEF decrease
Lag 0:1.09 (0.89,1.34). Lag 1:1.00 (0.81,1.23)
>10% evening PEF decrease
Lag 0:1.06 0.86,1.30). Lag 1: 0.83 (0.68,1.02)
No BHR and low IgE (N = 167):
LRS: Lag 0:1.12 (0.76,1.66). Lag 1: 0.61 (0.39, 0.94)
URS :Lag 0:1.01 (0.89,1.13). Lag 1:1.08 (0.96,1.22)
>10 morning PEF decrease
Lag 0:1.02 (0.89,1.16). Lag 1:1.00 (0.87,1.15)
>10% evening PEF decrease
Lag 0:1.10 (0.97,1.25). Lag 1:1.06 (0.93,1.21)
With BHR and low IgE (N = 67):
LRS: Lag 0: 0.72 (0.41,1.28). Lag 1:1.03 (0.56,1.91)
URS: Lag 0: 0.82 (0.62,1.09). Lag 1: 0.84 (0.64,1.12)
>10% morning PEF decrease
Lag 0: 0.74 (0.51,1.07). Lag 1: 0.96 (0.67,1.37)
>10% evening PEF decrease
Lag 0:1.23 (0.88,1.73). Lag 1:1.32 (0.96,1.82)
No BHR and high IgE (N = 104):
LRS: Lag 0:1.44 (1.17,1.77)
Lag 1:1.28 (1.00,1.64). 5-day mean: 2.49 (1.54, 4.04)
URS: Lag 0: 0.98 (0.84,1.14). Lag 1:1.01 0.87,1.18)
>10% morning PEF decrease
Lag 0: 0.92 (0.79,1.08). Lag 1:1.03 (0.89,1.21)
>10% evening PEF decrease
Lag 0:1.00 (0.85,1.17). Lag 1:1.05 (0.90,1.23)
F-5
-------
Study
Method
Pollutant Data
Findings
Boezen et al. (2005)
Amsterdam,
Meppel, Nunspeet,
The Netherlands
Period of Study:
Two winters 1993-
1995
Panel study of 327 elderly patients (50 to
70 yrs) to determine susceptibility to air
pollution by AHR, high total
immunoglobulin (IgE), and sex.
Methacholine challenges were performed
and subjects with greater than or equal to
20% fall in FEVi after inhalation of up to
2.0 mg methacholine were considered
AHR+. Subjects with total serum IgE >
20 kll/L were defined as high total IgE
(lgE+). Twice daily PEF measurements
and daily symptoms recorded for 3 mos.
Data analysis performed using logistic
regression with modeling of first-order
autocorrelation in the residuals that ad-
justed for daily min temperature, time
trend, weekend/ holidays and influenza
incident for the rural and urban areas and
the two winters separately. Subjects were
classified as lgE+AHR+, lgE+AHR-,
IgE- AHR+ or IgE- AHR+. Examined
effects of pollutants on the same day,
Lag 1, Lag 2 and the 5-day mean
concentration of Lag 0 to Lag 4
preceding that day. Groups that had
effect estimates for PM10, BS, SO2, and
NO2 that were outside the 95% CI of the
effect estimates for the AHR-/lgE-
(control group) were considered to have
increased susceptibility to air pollution.
24-h avg SO2 (|jg/m3)
winter
Winter 1993/1994
Urban:
Mean: 11.8 |jg/m3
Median: 10.2
Range: 2.7, 33.5
Rural:
Mean: 8.2
Median: 4.4
Range: 0.8, 41.5
Winter 1994/1995
Urban:
Mean: 8.3
Median: 7.4
Range: 0.6, 24.4
Rural:
Mean: 4.3
Median: .7
Range: 0.5,17.0
Copollutants:
PM10 BS N02
No consistent associations between the prevalence of LRS or
>10% fall in evening PEF and air pollution in any of the four
groups. In the AHR+/lgE+ group, the prevalence of URS was
associated with SO2 at 1 day lag, and the prevalence of >10% fall
in morning PEF with SO2 at Lag 1, Lag 2 and 5-day mean (avg of
Lag 0 to Lag 4). For females who were AHR+/lgE+, the preva-
lence of >10% fall in PEF was associated with SO2 Lag 1, Lag 2
and 5-day mean. In subjects with AHR-/lgE+ the prevalence of
URS was associated with SO2 the previous day and the mean of
Lag 0 to Lag 4. The effect estimate was outside the 95% CI of
the estimate for the control group AHR-/lgE-. No consistent
positive associations found between prevalences of URS, cough
or >10% fall in morning PEF and air pollutants in subjects with
AHR+/lgE- or AHR-/lgE-. Based on results of the study, authors
conclude that subjects with AHR+/lgE+ were the most responsive
to air pollution.
No AHR and low IgE (N = 125):
URS: Lag 0:0.99 (0.93,1.05)
Lag 1:1.02 (0.97,1.08). 5-day mean: 0.99 (0.88,1.12)
Cough: Lag 0:1.03 (0.98,1.08). Lag 1: 0.97 (0.93,1.02)
>10% fall in morning PEF. Lag 1:1.00 (0.92,1.08)
No AHR and high IgE (N = 112):
URS: Lag 0:0.98 (0.92,1.03)
Lag 1:1.07 (1.01,1.12). 5-day mean: 1.15 (1.02,1.29)
Cough: Lag 0:1.01 (0.95,1.07). Lag 1:1.02 (0.96,1.08)
>10% fall in morning PEF . Lag 1:1.00 (0.92,1.08)
With AHR and low IgE (N = 42):
URS: Lag 0:1.05 (0.94,1.17). Lag 1:1.07 (0.96,1.19)
5-day mean: 1.04 (0.83,1.30)
Cough: Lag 0:1.03 (0.95,1.12). Lag 1:1.01 (0.93,1.09)
>10% fall in morning PEF. Lag 1: 0.99 (0.87,1.12)
With AHR and high IgE (N = 48):
URS: Lag 0:1.06 (0.97,1.15)
Lag 1:1.13 (1.05,1.23). 5-day mean: 1.18 (0.99,1.40)
Cough: Lag 0:1.02 (0.94,1.11). Lag 1:1.02 (0.94,1.10)
>10% fall in morning PEF
Lag 1:1.15 (1.04,1.27). Lag 2 :1.18 (1.07,1.30)
5-day mean: 1.26 (1.07,1.49)
With AHR and high IgE, by gender:
URS, males: Lag 0:1.09 (0.95,1.25)
Lag 1:1.12 (0.98,1.28). 5-day mean: 1.62 (1.25, 2.11)
URS, females: Lag 0:1.10 (0.97,1.24)
Lag 1:1.12 (0.99,1.25). 5-day mean: 1.02 (0.82,1.28)
>10% fall in morning PEF, males. Lag 1:1.04 (0.87,1.25)
Lag 2: 0.92 (0.77,1.10. 5-day mean: 0.88 (0.64,1.21)
>10% fall in morning PEF, females. Lag 1:1.18 (1.03,1.36)
Lag 2:1.24 (1.08,1.42). 5-day mean: 1.31 (1.03,1.66)
F-6
-------
Study
Method
Pollutant Data
Findings
Cuijpers et al.
(1994)
Maastricht, the
Netherlands
Period of Study:
Nov-Dec 1990
(baseline)
Aug 8-16
(smog episode)
The effects of exposure to summer smog
on respiratory health were studied in 535
children (age unspecified). During a
smog episode, 212 children were
randomly chosen to be reexamined for
lung function and symptoms. Only 112 of
the children had adequately completed
summer questionnaires and were used
for the symptom analysis. Lung function
measurements made with forced
oscillation technique were available for
212 children and valid spirometry was
available for 208 children. Corrected
baseline lung function compared using
paired t test and difference in the preva-
lence in symptoms during baseline and
episode compared.
24-h avg SO2
Baseline 55 |jg/m3
Summer episode 23 |jg/m3
Copollutants:
N02 BS 03 PM10
Acid aerosol H*
Small decrements in FEV1 and FEF25-75 found in the 212 children
during the episode compared to baseline. However, there was
also a significant decrease in resistance parameters. No
increases observed in the prevalence of acute respiratory
symptoms.
Change in lung function and impedance between baseline and
smog episode:
FEV1:-0.032 L(SD 0.226), p< = 0.05
FEF25-75: -0.086 L/s (SD 0.415), p < = 0.01
Resistance at 8 Hz: -0.47 cm H2O (L/s)
(SD 1.17), p< = 0.05
Desqueyroux et al.
(2002a)
Paris, France
Period of Study:
Oct 1995-Nov 1996
Panel study of 39 Parisian adults with
severe COPD (avg age 67 yrs) to
determine if air pollution affects health
outcomes. Episodes of exacerbation
were based on regular physician
appointments and patient-initiated
consultations. Exacerbation was
confirmed as a decrease in "vesicular"
breath sound, bronchial obstruction,
tachycardia/arrhythmia, or cyanosis.
Examined with logistic-regression
analysis on the basis of GEE. Examined
lag effects of 0 to 5 days.
24-h avg SO2 (|jg/m3)
Summer:
Mean: 7 (SD 5)
Range: 2, 27
Winter:
Mean: 19 (SD 12)
Range 3, 81
Copollutants:
NO2 PM10 O3
No association between episodes of symptom exacerbation and
SO2, regardless of the lag.
Mean 24-h avg SO2 (per 10 (jg/m3)
OR on incident episodes of exacerbation of COPD:
Lag 1: 0.98 (0.64,1.33). Lag 2: 0.96 (0.66,1.40)
Lag 3: 0.91 (0.63,1.33). Lag 4: 0.89 (0.61,1.29)
Lag 5: 0.87 (0.63,1.20). Lag 1-5: 0.83 (0.47,1.50)
Multipollutant model with O3 and SO2
24-h avg SO2:
Lag 1-5:0.64(0.19,2.19)
Desqueyroux et al.
(2002b)
Paris, France
Period of Study:
Nov 1995-Nov 1996
Panel study of 60 patients with moderate
to severe physician-diagnosed asthma
(mean age 55 yrs). Asthma attacks were
noted by physician at each consultation
(regular or emergency). Asthmatic
attacks defined as need to increase
twofold the dose of beta2 agonist.
24-h avg SO2 (|jg/m3)
Summer: 7 (SD 5)
Range: 2, 27
Winter 19 (SD 12)
Range: 3, 81
Copollutants:
PM10 NO2 O3
No association between asthma attacks and SO2 for any lag or
season.
Mean 24-h avg SO2 (per 10 |jg/m3).OR on incident of asthma
attacks:
Lag 1: 0.98 (0.76,1.27). Lag 2: 0.92 (0.72,1.19)
Lag 3:1.01 (0.82,1.23). Lag 4:1.01 (0.86,1.19)
Lag 5:1.05 (0.85,1.29). Lag 1-5: 0.99 (0.76,1.30)
Forsberg et al.
(1993)
Pitea, Northern
Sweden
Period of Study:
Mar to Apr
Panel study of 31 asthmatic patients (9 to 24-h avg SO2 (|jg/m3)
71 yrs) to assess relationship between
daily occurrence of asthma symptoms
and fluctuations in air pollution and
meteorological conditions. Subjects
recorded symptoms (shortness of breath,
wheezing, cough, phlegm) for
14 consecutive days.
Mean: 5.7
Range: 1.3,12.9
Correlations:
N02 (r = 0.24)
BS (r= 0.70)
No significant association observed with SO2. Positive asso-
ciation between severe shortness of breath and BS.
Regression coefficient and 90% CI
Subjects with shortness of breath (N = 28):
0.0345 (-0.49, 0.118)
Subjects with 5 or more incident episodes of severe shortness of
breath (N = 10): -0.0266 (-0.140, 0.087)
F-7
-------
Study
Method
Pollutant Data
Findings
Higgins et al. (1995)
United Kingdom
Study Period: NR
Panel study of 75 patients with physician
diagnosed asthma or chronic bronchitis
(avg age 50, range 18 to 82 yrs) to
determine if air pollution affects respira-
tory function and symptoms. Subjects
asked to keep symptom records and
perform PEF for 28 days. PEF values
recorded every 2 h beginning at 02.00 h
each day. Methacholine challenge
performed on each subject. Those with
PM20 FEV1 of < 12.25 |_imol were consid-
ered as methacholine reactors. PEF
variability was calculated as the ampli-
tude % Mean: (highest-lowest PEF
value/mean) x 100. 75 patients had PEF
records, 62 completed satisfactory
symptom questionnaires.
Max 24-h avg SO2
117 (jg/m3
Copollutants:
O3 NO2
The amplitude % mean was significantly associated with
increasing levels of SO2, on the same day for all subjects and
among reactors. Mean daily PEF and min PEF associated with
SO2 among reactors only. Significant associations also observed
with wheeze and SO2 on the same day, at 24-h lag, and 48-h lag
for all subjects and meta-choline reactors; and with
bronchodilator use for all subjects at 24-h lag.
Regression coefficient per 10 |jg/m3 SO2
All subjects:
Mean PEF (L/min): Same day 0.021 (0.031);
24-h Lag 0.003 (0.033); 48-h Lag 0.021 (0.032)
Minimum PEF(L/min): Same day 0.062 (0.039);
24-h Lag -0.048 (0.041); 48-h Lag -0.001 (0.040)
Amplitude (% mean): Same day: 0.167 (0.072);
24-h Lag 0.191 (0.76); 48-h Lag 0.022 (0.075)
Wheeze: Same day 1.14 (1.03,1.26);
24-h Lag 1.22 (1.09,1.37); 48-h Lag 1.14 (1.02,1.27)
Dyspnoea: Same day 1.03 (0.94,1.14);
24-h Lag 1.07 (0.96,1.18); 48-h Lag 0.94 (0.85,1.05)
Cough: Same day 1.03 (0.95,1.12);
24-h Lag 1.04 (0.95,1.13); 48-h Lag 1.02 (0.94,1.12)
Bronchodilator use: Same day 1.11 (0.97,1.26);
24-h Lag 1.16 (1.01,1.34); 48-h Lag 1.12 (0.98,1.27)
Reactors:
Mean PEF (l/min): Same day 0.087 (0.054);
24-h Lag -0.44 (0.058); 48-h Lag 0.012 (0.057)
Minimum PEF(L/min): Same day 0.168 (0.071)
24-h Lag -0.078 (0.076); 48-h Lag -0.026 (0.075)
Amplitude (% mean): Same day 0.157 (0.120);
24-h Lag 0.083 (0.127); 48-h Lag 0.005 (0.126)
Wheeze: Same day 1.26 (1.08,1.47);
24-h Lag 1.57 (1.30,1.89); 48-h Lag 1.24 (1.06,1.45)
Dyspnoea: Same day 1.04 (0.90,1.20);
24-h Lag 1.17 (1.00,1.37); 48-h Lag 1.03 (0.89,1.20)
Cough: Same day 1.09 (0.96,1.24);
24-h Lag 1.05 (0.91,1.20); 48-h Lag 1.00 (0.87,1.15)
Bronchodilator use: Same day 1.18 (0.99,1.42);
24-h Lag 1.23 (1.02,1.50); 48-h Lag 1.31 (1.09,1.58)
Hiltermann et al.
(1998)
Bilthoven, The
Netherlands
Period of Study:
Jul-Oct 1995
Panel study of 60 adult (18 to 55 yrs)
nonsmoking patients with intermittent to
severe persistent asthma to examine the
association of summertime air pollution
(O3 and PM10) with respiratory symptoms,
medication use and PEF. Subjects were
followed over 96 days. Twice daily PEF,
respiratory symptoms, and medication
use and whether they were exposed to
environmental tobacco smoke were re-
corded daily. Analysis controlled for time
trends, aeroallergens, environmental
tobacco smoke exposures, day of wk and
temperature. Examined Lag effects of 0
to 2 days.
24-h avg SO2 (|jg/m3)
Mean: 6.2
Range: 0.1,16.2
Correlation with BS
r= 0.53
Correlation with
copollutants:
03 (r = 0.30)
PM10 (r = 0.37)
N02 (r = 0.49)
BS (r= 0.53)
SO2 not included in the analysis since levels were negligible
during the study period (< 17 (jg/m3)
Effect estimates not provided.
F-8
-------
Study
Method
Pollutant Data
Findings
Hoek and
Brunekreef (1992)
The Netherlands
Period of Study:
Winter only, 1987-
1990
Panel study of 1,078 children (7 to 11
yrs) to examine the effects of low-level
winter air pollution with respiratory
symptoms. Pulmonary function test were
performed 6 to 10 times on predeter-
mined days. Parents filled out symptom
diary that was turned in every 2 wks.
Symptoms include hoarseness, cough,
cough with phlegm, wheeze, runny/
stuffed nose, aching throat, shortness of
breath, chest tightness, eye irritation, and
sneezing. Association of symptom
prevalence and symptom incidence was
analyzed using individual regression
slopes.
24-h avg SO2 (|jg/m3)
Mean (SD): 14.9(14.5)
Range: 0.4, 94.3
Correlation with
copollutants:
N02 (r = 0.46)
PMiofr =0.50)
S042" (r =0.41)
NOs- (r =0.39)
HONO (r =0.40)
No association between pulmonary function and SO2
concentration.
SE and medians of individual regression slopes
FVC Same day 0.26 (0.21). Lag 1: 0.54 (0.18), p < 0.05
FEV1 Same day 0.15 (0.20). Lag 1: 0.21 (0.17)
PEF Same day -0.83 (1.07). Lag 1: -0.54 (0.87)
MMEF Same day -0.58 (0.53). Lag 1: -0.44 (0.44)
Odds ratio and 95% confidence interval:
Cough. Same day 1.10 (0.78,1.55). Lag 1: 0.80 (0.55,1.18)
LRS. Same day 1.25 (0.81,1.93). Lag 1:1.73 (1.11, 2.72)
URS. Same day 1.28 (0.96,1.70). Lag 1:1.11 (0.82, 1.51)
Any respiratory symptoms.
Same day 0.76 (0.56,1.03). Lag 1:1.09 (0.83,1.44)
Hoek and
Brunekreef (1993)
Wageningen,
The Netherlands
Period of Study:
Winter 1990-1991
Panel study of 112 children (7 to 12 yrs,
non-urban) to assess effects of winter air
pollution pulmonary function and
respiratory symptoms. Parents filled out
symptom diary that was turned in every 2
wks. Pulmonary function test performed
by technician every 3 wks. Additional
pulmonary function tests performed when
S02was predicted to be higher than
125 (jg/m3 or NO2 > 90 |jg/m3.
Daily concentrations
presented in graph;
Highest 24-h avg cone
SO2:105 (jg/m3 (air
pollution episode)
Copollutants:
PM10 BS N02
During the winter episode, pulmonary function of schoolchildren
was significantly lower than baseline. Significant negative asso-
ciations between SC^and FVC, FEV1 and MMEF. No significant
associations found with prevalence of respiratory symptoms.
Authors noted that it is not clear which components of episode
mix responsible for association and that the concentrations of
acid aerosol and S02were too low for direct effects to be likely.
SO2 moderately correlated with PM10 (r = 0.69) and BS (r = 0.63)
but not NO2 (r = 0.28).
Mean of individual regression slopes and SE
FVC Same day -0.55 (0.10), p < 0.05
Lag 1: -0.74 (0.15) p < 0.05.1 wk -0.94 (0.20) p < 0.05
FEV1 Same day -0.51 (0.09) p < 0.05
Lag 1: -0.21 (-0.63) p < 0.05.1 wk -0.78 (0.18) p < 0.05
PEF Same day-0.64 (-0.44)
Lag 1: -0.21 (0.63). 1 wk -0.34 (0.81) p < 0.05
MMEF. Same day -0.54 (0.20)
Lag 1:-0.40 (0.29). 1 wk-0.61 (0.37)
Prevalence of acute respiratory symptoms regression coefficient
from time-series model and SE
Cough. Same day 0.02 (0.18);
Lag 1: -0.14 (0.19); 1 wk 0.13 (0.76)
URS. Same day 0.12 (0.16);
Lag 1: -0.02 (0.17); 1 wk -0.24 (0.76)
LRS. Same day 0.06 (0.26);
Lag 1: -0.11 (0.29); 1 wk-0.54 (0.92)
Any respiratory symptoms. Same day 0.01 (0.13);
Lag 1: -0.03 (0.13); 1 wk-0.11 (0.60)
Hoek and
Brunekreef (1995)
Deurne and
Enkhuizen, The
Netherlands
Period of Study:
Mar-Jul 1989
Panel study of 300 children (7-11 yrs) to
examine the effects of photochemical air
pollution on acute respiratory symptoms.
Occurrence of respiratory symptoms
recorded by parents in daily diary.
Symptoms included cough, shortness of
breath, upper and LRS, throat and eye
irritation, headache and nausea.
Association of symptom prevalence and
incidence assessed using first order
autoregressive, logistic regression
model.
Daily concentration of SO2
< 43 (jg/m3
Copollutants:
O3 PM10 SO42 NO3"
Same day concentrations of SO2 and NO2 not associated with
symptom prevalence.
No effect estimates for SO2 provided.
Just et al (2002)
Paris, France
Period of Study:
1996
Panel study consisting of 82 medically 24-h avg (|jg/m3):
diagnosed asthmatic children, 7-15 yrs
old, followed for 3 mos (Jan-Mar).
Examined the association between air
pollution and asthma symptoms using
regression analyses based on
generalized estimating equations (GEE).
11.6(5.7)
Copollutants:
PM BS NO2O3
SO2 was not analyzed because it was only present at low
concentrations.
F-9
-------
Study
Method
Pollutant Data
Findings
Lagorio et al. (2006)
Rome, Italy
Period of Study:
May 24 to June 24,
1999 and Nov 18 to
Dec 22,1999
Panel study of 29 patients with either
COPD (N: 11, mean age 67 yrs), asthma
(N: 11, mean age 33 yrs) or ischemic
heart disease (N: 7, mean age 63 yrs) to
evaluate whether daily levels of air
pollutants have a measurable impact on
lung function in adults with preexisting
lung or heart disease.
24-h avg SO2 (|jg/m3)
Spring mean 4.7
SD 1.8
Winter mean 7.9
SD 2.2
Overall mean 6.4
SD: 2.6
Copollutants:
PM2.5, PM10-2.5, PM10, Cd,
Cr, Fe, Ni, Pb, Pt, V, Zn,
N02, CO, 03
Correlation with
copollutants:
PM2.5 (r = 0.34)
PM10-2.5 (r = -0.16)
PM10 (r = 0.21)
N02 (r = 0.01)
03 (r = -0.61)
CO (r = 0.65)
Because avg 24-h concentrations of SO2 were low and showed
little variability, SO2 was not considered in the analysis
Neukirch et al.
(1998)
Paris, France
Period of Study:
Nov 15,1992 to
May 9,1993
Panel study of 40 nonsmoking, mild to
moderate asthmatics (16 to 70 yrs, mean
46) to examine the short-term effects of
winter air pollution in asthma symptoms
and three daily peak flow measurements.
Patients were followed for 23 wks. Used
GEE models that controlled for autocor-
relation of responses, weather, and time
trends. Analysis conducted on entire
study population and for subgroup of
subjects who took inhaled B2 agonists as
needed. Assessed air pollution effect on
both incident and prevalence of symp-
toms, Z-transformed morning PEF and
daily PEF variability.
24-h avg SO2
Mean: 21.7 (13.5) |jg/m3
Range: 4.4, 83.8
Copollutants
N02, PM13, BS
Correlation with
copollutants:
N02 (r = 0.54)
PM13 (r = 0.83)
BS (r= 0.89)
Significant effects on incidence and prevalence of symptoms.
Effects at Lag days 3-6 and weekly avg exposures. Based on
group avg PEF of 407 l/min, a 50 |jg/m3 increase SO2 caused a
maximum decrease in morning PEF of 5.5%.
Odds ratio per 50 |jg/m3 SO2.
All subjects - Incident episodes:
Wheeze: Lag 5:1.66 (1.01, 2.70)
Nocturnal cough: Lag 3:1.60 (0.98, 2.62);
Lag 4:1.71 (0.86, 3.40); Lag 6:1.72 (1.16, 2.55)
Respiratory infections: Lag 3: 3.14 (1.30, 7.59);
Lag 4: 2.70 (1.36, 5.37); Lag 5: 2.79 (0.95, 8.21);
Wk: 8.52 (1.20, 60.5)
All subjects - Prevalent episodes:
Wheeze: Lag 5:1.35 (1.01,1.81);
Lag 6:1.39 (1.04,1.87); Wk: 1.64 (0.91, 2.94)
Nocturnal cough: Lag 6:1.34 (1.00,1.79)
Shortness of breath: Wk: 1.56 (1.06, 2.32)
Respiratory infections: Lag 4: 2.40 (1.33, 4.33);
Lag 5: 2.72 (1.67, 4.44); Lag 6: 2.94 (1.80, 4.79);
Wk: 6.30 (1.31, 30.2)
Subjects taking B2 agonists - Incident episodes:
Asthma attacks: Lag 6: 2.19 (0.91, 5.29)
Wheeze: Lag 5:1.84 (1.13, 3.00)
Nocturnal cough: Lag 3: 2.41 (1.47, 3.93);
Lag 4: 2.35 (0.88, 6.26); Lag 6:1.86 (1.14, 3.04)
Subjects taking B2 agonists - Prevalent episodes:
Asthma attacks: Lag 5:1.88 (0.95, 3.73);
Lag 6: 2.82 (1.57, 5.07)
Wheeze: Lag 5:1.51 (1.02, 2.23);
Lag 6:1.57 (1.06,2.32)
Nocturnal cough: Lag 3:1.73 (1.06, 2.82);
Lag 4: 2.28 (1.27, 4.11); Lag 5:1.91 (1.17, 3.12);
Lag 6:1.91 (1.17,3.12)
Shortness of breath: Lag 4:1.81 (1.22, 2.67);
Lag 5:1.65 (1.11, 2.44); Lag 6:1.61 (1.20,2.16);
Wk: 3.03 (1.26, 7.33)
Regression coefficients of the effects and SE (per 1 (jg/m3)
Z-transformed morning PEF: Lag 5: -0.450 (0.138) p = 0.001
Lag 6:-0.337 (0.164) p = 0.03
PEF daily variability, Lag 2: 0.025 (0.013) p = 0.05
F-10
-------
Study
Method
Pollutant Data
Findings
Peacock et al.
(2003)
Southern England
Period of Study:
Nov 1,1996 to Feb
14,1997
Panel study of 177 children (mean age
10.7 yrs, range 7 to 13) from three
schools (two urban and 1 rural location)
to investigate effects of winter air
pollution on respiratory function. Children
were followed for 13 wks. Used two
sources of air pollution in the rural area,
one that was "locally validated" and the
other "nationally validated."
24-h avg SO2 (ppb)
Rural
(nationally validated)
Mean: 5.1 (4.7)
Range: 0.0, 35.6
Rural
(locally validated)
Mean: 5.4 (5.1)
Range: 0.0, 39.1
Urban 1
Mean: 6.0 (6.0)
Range: 0.5, 32.5
Copollutants:
O3 NO2 PM10 SO4
No statistically significant association between winter SO2 and
PEFR, 0.70% decline in PEFR for a 10 ppb increase in the five-
day mean concentration of SO2 (community monitor)
Regression Coefficient (95% CI) per 1 ppb SO2
Community monitor
Lag 0: 0.05 (-0.05, 0.16). Lag 1: -0.04 (-0.13, 0.06)
Lag 2: -0.08 (-0.19, 0.04). Lag 0-4: -0.23 (-0.65, 0.18)
Local monitor
Lag 0: -0.01 (-0.10, 0.07). Lag 1: 0.02 (-0.05, 0.10)
Lag 2: -0.09 (-0.18, 0.01). Lag 0-4: -0.09 (-0.25, 0.07)
Odds of 20% decrement in PEF below the median - all children
Lag 0: 0.987 (0.958,1.017). Lag 1:1.007 (0.986,1.030)
Lag 2: 0.992 (0.963,1.023). Lag 0-4: 0.972 (0.887,1.066)
Odds of 20% decrement in PEF below the median - wheezy
children
Lag 0: 0.981 (0.925,1.041). Lag 1: 0.999 (0.957,1.042)
Lag 2: 0.995 (0.939 1.054). Lag 0-4:1.019 (0.890,1.167)
Peters et al. (1996)
Erfurt and Weimar,
former German
Democratic
Republic; Sokolov,
Czech Republic
Period of Study:
Sep 1990 to June
1992
Panel study of 102 adult (32 to 80 yrs)
and 155 children (7 to 15 yrs) with
asthma from the former German
Democratic Republic and Czech
Republic to investigate the acute effects
of winter type air pollution on symptoms,
medication intake and PEF. Used
regression analyses and distributed Lag
models.
Winter 1990/1991
Erfurt:
Mean: 125 |jg/m3
Max: 564 |jg/m3
IQR: 113 (jg/m3
Weimar
Mean: 236 |jg/m3
Max: 1018 |jg/m3
IQR: 207 |jg/m3
Sokolov
Mean: 90 |jg/m3
Max: 492 |jg/m3
IQR: 94 (jg/m3
Winter 1991/1992
Erfurt
Mean: 96 |jg/m3
Max : 462 pg/m3,
IQR: 80 (jg/m3
Weimar
Mean: 153 |jg/m3
Max: 794 |jg/m3
IQR: 130 (jg/m3
Sokolov
Mean: 71 |jg/m3
Max: 383 |jg/m3
IQR: 66 (jg/m3
Copollutants: TSP, PM10,
SO4, PSA (particle strong
acidity)
5-day mean concentration of SO2 associated with PEF and
symptoms in children (combined analysis from former German
Democratic Republic and Czech Republic).
Correlation coefficient between SO2 and TSP in Erfurt was r =
0.8, 0.9 during both winters and in Weimar during the first winter.
Correlation with TSP in Sokolov and in Weimar during the second
winter was r= 0.4, 0.5.
Combined analysis for children
% change in PEF
Concurrent day 0.18 (-0.44, 0.09) per 133 |jg/m3
5-day mean —0.90 (-1.35, -0.46) per 128 |jg/m3
% change in symptom score
Concurrent day -0.1 (-5.9, 5.7) per 133 |jg/m3
5-day mean 14.7 (0.8, 28.6) per 128 |jg/m3
Combined analysis for adults
% change in PEF
Concurrent day -0.20 (-0.53, 0.12) per 133 |jg/m3
5-day mean —0.28 (-0.72, 0.16) per 128 |jg/m3
Pikhart et al. (2000)
Czech Republic
Period of Study:
1993-1994
SAVIAH study of 3045 children by Median: 73.9 |jg/m3
questionnaire to determine association of 25th percentile: 63.5
SO2 to wheezing. Used ecological and 75th percentile: 95.5
multilevel analysis
Copollutant: NO2
Positive association of SO2 with wheezing
Odds Ratio (95% CI) per 10 |jg/m3 increase SO2
Logistic Regression:
Individual outcome and area exposure: 1.08 (0.98,1.20).
Individual outcome and individual exposure: 1.08 (0.98,1.19).
Ecological analysis: 1.05 (0.96,1.16)
F-11
-------
Study
Method
Pollutant Data
Findings
Pinter et al. (1996)
Tata Area, Hungary
Period of Study:
winter mos between
Dec 1993-Mar 1994
Longitudinal (children < 14 yrs) and
cross-sectional study (9 to 11 yrs) to
examine air pollution and respiratory
morbidity in children. In the longitudinal
prospective study, respiratory morbidity
was evaluated daily and on a weekly
basis. In cross-sectional study,
anthropometric parameters, physical
status, pulse and blood pressure, lung
function parameters, eosinophils in the
nasal smear, hematological
characteristics and urinary excretion of
some metabolites were examined and
measured. Anova and linear regression
used in analysis.
Avg SO2 exceeded the
limit of yearly avg 150
(jg/m3
Daily peaks reached as
high as 450 |jg/m3
No specific values given
Copollutant: NO2
Significant correlation between SO2 levels and acute daily
respiratory morbidity, but no correlation with weekly incidence.
Authors stated that in the cross-sectional study, almost all health
parameters were impaired but no results were shown.
Results only provided in graph. No p-values provided.
Ponka, 1990
Helsinki, Finland
1991
Survey study to compare weekly
changes in ambient SO2, NO2, and
temperature and the incidence of
respiratory diseases, and absenteeism
for children in day-care centers and
schools and for adults in the work place
during a 1-yr period (1987).
Avg weekly concentration
of SO2 (|jg/m3)
Mean: 21.1
SD: 11.7
Median: 17.0
Range: 9, 61.5
Mean of daily max
Mean: 53
SD: 20.8
Median: 48
Range: 25.9,130.3
Copollutant: NO2
Mean SO2 concentration correlated with the incidences of URI
and tonsillitis reported from health centers. SO2 also correlated
with absenteeism due to febrile illness among children in day
care centers and adults. When comparing incidences during the
low and high levels of SO2, the number of cases of URI and
tonsillitis reported from health centers increased as well as
absenteeism. After standardization for temperature, the only
difference that was statistically significant was the occurrence of
URI diagnosed at health centers. Frequency of URI was 15%
higher during high levels of SO2 compared to low.
Statistical significance of product moment correlation coefficients
(correlation coefficient) between SO2 and respiratory disease and
absenteeism
Respiratory tract infections diagnosed at health centers:
URI: SO2 arithmetic mean: p < 0.001 (0.553)
Mean of daily maximums: p = 0.0012 (0.437)
Tonsillitis: SO2 arithmetic mean: p = 0.0098 (0.355)
Mean of daily maximums: NS
Absenteeism due to febrile illness:
Day care centers:
SO2 arithmetic mean: p = 0.012 (0.404)
Mean of daily maximums: p = 0.048 (0.323)
School children:
SO2 arithmetic mean: NS. Mean of daily maximums: NS
Adults:
SO2 arithmetic mean: p < 0.0001 (0.644)
Mean of daily maximums: p < 0.0001 (0.604)
No significant correlation between SO2 and URI, tonsillitis, otitis,
or LRI in day care center children
Statistical significance of weekly frequency of respiratory tract
disease and absenteeism during low and high levels of SO2:
Respiratory infections diagnosed at health centers:
URI: SO2 arithmetic Mean: p < 0.001
Mean of daily max: p = 0.0005
Tonsillitis:
SO2 arithmetic mean: 0.0351. SO mean of daily max: NS
Absenteeism due to febrile illness:
Day care center children: p = 0.0256
School children: p = 0.0014. Adults: p = 0.0005
F-12
-------
Study
Method
Pollutant Data
Findings
Roemer et al. (1993) Panel of 73 children (mean age 9.3 yrs,
Bennekom and
Wageningen,
Netherlands
Period of Study:
Winter 1990-1991
range 6 to 12 yrs) with chronic
respiratory symptoms to investigate
effects of winter air pollution on lung
function, symptoms and medication use.
Subjects performed twice-daily PEF
measurements, largest of three PEF
readings used in regression analysis.
Both incidence and prevalence of
symptoms analyzed, using logistic
regression.
Daily concentrations of
SO2 shown in graph
Highest 24-h avg
concentration SO2:105
(jg/m3
Copollutants:
NO2PM10 BS
Correlation with
copollutants:
N02 (r = 0.26)
PM10 (r = 0.65)
BS (r= 0.63)
Positive association between incidence of phlegm and runny
nose with SO2 on the same day. Significant association also
found between evening PEF and SO2 on, the same day, previous
day and 1 wk (avg of same day and 6 days before). The use of
bronchodilators also associated with SO2.
Mean of individual regression coefficient
Morning PEF: Same day -0.021 (0.024);
Lag 1 -0.024 (0.031); Wk -0.50 (0.069)
Evening PEF: Same day -0.048 (0.018) p < 0.05;
Lag 1 -0.039 (0.021) p < 0.10; Wk -0.110 (0.055) p < 0.05
Prevalence of symptoms (per 50 |jg/m3 SO2)
Asthma attack: Same day 0.008 (0.012);
Lag 1 0.016 (0.011); 1 Wk 0.058 (0.027) p < 0.05
Wheeze: Same day 0.033 (0.17) p < 0.10;
Lag 1 0.042 (0.016) p < 0.05; Wk 0.069 (0.032) p < 0.05
Waken with symptoms: Same day 0.033 (0.019) p < 0.10;
Lag 1 0.032 (0.018) p < 0.10; Wk 0.058 (0.045)
Shortness of breath: Same day 0.029 (0.016) p < 0.10;
Lag 10.016 (0.015); Wk 0.044 (0.035)
Cough: Same day 0.018 (0.025);
Lag 1 0.012 (0.023); Wk 0.072 (0.066)
Runny nose: Same day 0.070 (0.026) p < 0.05;
Lag 1 -0.11 (0.025); Wk 0.153 (0.074) p < 0.05
Phlegm: Same day 0.011 (0.022);
Lag 1 0.014 (0.020); Wk -0.005 (0.056)
Roemer et al. (1998)
14 European
Centers:
Amsterdam, The
Netherlands;
Athens, Greece;
Berlin and Hettstadt,
Germany; Budapest,
Hungary; Krakow
and Katowice,
Poland; Kuopi,
Finland; Malmo and
Umea, Sweden;
Oslo, Norway; Pisa,
Italy; Prague and
Teplice, Czech
Republic
Period of Study:
Winter 1993-1994
Multicenter panel study of the acute
effects of air pollution on respiratory
health of 2,010 children (aged 6 to 12
yrs) with chronic respiratory symptoms.
Results from individual centers were
reported by Kotesovec et al. (1998),
Kalandidi et al. (1998), Haluszka et al.
(1998), Forsberg et al. (1998), Clench-
Aas et al. (1998), and Beyer et al. (1998).
Calculated effect estimates of air
pollution on PEF or the daily prevalence
of respiratory symptoms and
bronchodilator use from the panel-
specific effect estimates
Range: 2.7 |jg/m3 (Umea,
urban), 113.9 |jg/m3
(Prague, urban)
Copollutants:
PM10, BS N02
No clear associations between PM10, BS, SO2, or NO2 and
morning PEF, evening PEF, prevalence of respiratory symptoms,
or bronchodilator use could be detected. Previous day PM10 was
negatively associated with evening PEF, but only in locations
where BS was high compared to PM10 concentrations.
No consistent differences in effect estimates between subgroups
based on urban versus suburban, geographical location or mean
levels of PM10, BS, SO2, and NO2. Combined effect estimates
with 95% CI of air pollution on PEF.
Morning: Lag 0: 0.2 (-0.2, 0.6)
Lag 1:0.2 (-0.2, 0.6). Lag 2: 0.6 (0.2,1.0)
7-day mean: 0.6 (-1.3, 2.5)
Afternoon: Lag 0: 0.1 (-0.3, 0.5)
Lag 1: 0.0 (-0.4, 0.4). Lag 2:0.1 (-0.4, 0.6)
7-day mean: 0.2 (-0.5, 0.9)
F-13
-------
Study
Method
Pollutant Data
Findings
Segala et al. (1998)
Paris, France
Period of Study:
Nov 15,1992 to
May 9,1993
Panel study of 84 children (7 to 15 yrs)
with physician diagnosed asthma to exa-
mine the effects of winter air pollution on
childhood asthma. For 25 wks, parents
recorded the presence or absence of
asthma attacks, upper or lower respira-
tory infections with fever, the use of
supplementary inhaled B2 agonist, the
severity of symptoms (wheeze, nocturnal
cough and shortness of breath). Children
also recorded PEF three times a day.
GEE models adjusted for age, sex,
weather and time trend. Investigated
effects of SO2 at 0 to 6 day Lags.
SO2 avg (SD):
21.7 (13.5) (jg/m3
Range:
(4.4, 83.8) (jg/m3
Copollutants:
N02 PM13 BS
Correlation with
copollutants:
N02 (r = 0.54)
PM13 (r = 0.43)
BS (r= 0.89)
SO2 associated with both incident and prevalent episodes of
asthma, use of supplementary beta 2 agonist, incident episodes
of nocturnal cough, prevalent episodes of shortness of breath
and respiratory infection.
OR per 50 |jg/m3 SO2 (Only effects at 0 and 1-days lag shown
below unless statistically significant)
Incident episodes: Mild asthmatics (N: 43)
Asthma: Lag 0: OR 2.86 (1.31, 6.27). Lag 1: 2.45 (1.01, 5.92)
Wheeze: Lag 0:1.47 (0.90, 2.41). Lag1:1.27 (0.48, 3.38)
Nocturnal cough: Lag 3:1.93 (1.18, 3.15).
Lag 4:2.12(1.43,3.13)
Respiratory infections: Lag 1:1.52 (0.38, 5.98)
Prevalent episodes: Mild asthmatics (N: 43)
Asthma: Lag 0:1.71 (1.15, 2.53). Lag 1:1.55 (0.86, 2.78)
Wheeze: Lag 4:1.48 (0.90, 2.41)
Shortness of breath : Lag 1:1.36 (0.92, 2.01)
Lag 2:1.45 (0.98, 2.14). Lag 3:1.52 (1.03, 2.25)
Lag 4:1.51 (1.02,2.24)
Respiratory infections: Lag 0:1.58 (0.72, 3.46)
Lag 1:1.91 (0.79, 4.62). Lag 2: 2.13 (0.97, 4.67)
Lag 3: 2.09 (1.05, 4.15). Lag 4: 2.05 (1.14, 3.68)
Beta2 agonist: Lag 4:1.63 (1.00, 2.66)
Beta2 agonist: Lag 4: 2.02 (1.02, 4.01). Lag 5:1.96 (0.99, 3.88)
Moderate asthmatics (N: 41)
Statistically significant (only) prevalent episodes:
Beta2 agonist: Lag 0: 3.67 (1.25,10.8). Lag 1: 4.60 (2.10,10.1)
Lag 2: 7.01 (3.53,13.9). Lag 3: 4.74 (1.96,11.5)
Soyseth et al.
(1995a)
Ardal and Laerdal
regions, Norway
Period of Study:
Winters, 1989-1992
Cross sectional study of 620 children
(ages 7 to 13 yrs) to determine whether
short term exposure to SO2 and fluoride
on the number of capillary blood
eosinophils is related to prevalence of
BHR. Ardal is located in a SO2 emitting
aluminum smelter and Laerdal is
nonindustrialized. Parents filled out
respiratory questionnaires. Clinical
examinations used skin prick tests and
spirometry to test for atopy and BHR,
respectively. Multiple regression and
logistic regression models used in
analyses.
24-h avg SO2 (ug/m3):
Median: 22.2
10th: 1.9
90th: 85.3
A significant positive association between BHR and SO2 was
observed in atopic children. Eosinophils and SO2 exposure also
had a positive correlation.
Odds ratio for BHR (per 10 |jg/m3 SO2)
Last 24-hours: 1.12(1.01,1.24)
Last 1-30 days: 0.94 (0.73,1.21)
Taggart et al. (1996)
Runcorn and
Widnes in NW
England
Period of Study:
Jul-Sep 1993
Panel study of 38 nonsmoking asthma
subjects (18 to 70 yrs) to investigate the
relationship between asthmatic BHR and
pulmonary function (PEF, FEV1, FVC)
and summertime ambient air pollution.
Used univariate nested (hierarchical)
analysis of variance to test hypothesis
that BHR or spirometry measurements
varied with air pollution levels. Analysis
was limited to within-subject variation of
(BHR, FEV1, or FVC).
24-h avg SO2
Max: 103.7 |jg/m3
Copollutants:
NO2, O3, smoke
Correlation with
copollutants:
03 (r = 0.13)
N02 (r = 0.65)
Smoke (r = 0.48)
No association between SO2 and FEV1 or FVC.
Changes in BHR correlated significantly with changes in 24-h
mean SO2, NO2, and smoke.
Percentage change in BHR per 10 |jg/m3 SO2
24-h mean SO2 -6.3 % (-13.6, 0.6)
48-h mean -2.9 % (-12.8, 8.2)
24-h Lag 7.4% (-4.5,20.8)
F-14
-------
Study
Method
Pollutant Data
Findings
Timonen and
Pekkanen (1997)
Kuopio (urban and
suburban), Finland
Period of Study:
Winter, 1994
Panel study of 169 children (7 to 12 yrs)
with asthma or cough symptoms living in
urban and suburban areas of Kuopio,
Finland to determine association
between air pollution and respiratory
health. In the urban areas there were 39
asthmatics and 46 with cough only; in the
suburban areas there were 35
asthmatics and 49 with cough who were
included in the final analysis. Twice daily
PEF and daily symptoms were recorded
for 3 mos. First order autoregressive
models used to assess associations
between air pollutants and PEF and
logistic regression models used for
symptom prevalences and incidences.
Analysis conducted on daily mean PEF
deviations. Mean morning or evening
PEF calculated for each child was
subtracted from the daily value of
morning or evening PEF. The daily
deviations were then averaged to obtain
daily mean PEF deviation for morning or
evening PEF.
24-h avg SO2 (|jg/m3^
Urban area:
Mean: 6.0
25th percentile: 2.6
50th percentile: 3.6
75th percentile: 7.1
Max: 32.0
Copollutants:
PM10 BS N02
Correlation coefficient with
S02
PM10 (r = 0.21)
BS (r= 0.20)
N02 (r = 0.22)
Among children with cough only, morning and evening deviations
in PEF in the urban panel was negatively associated with SO2.
SO2 was also associated with an increase in the incidence of
URS in children with cough only in the urban area. When
excluding the three highest SO2 days, these effects were no
longer statistically significant. No associations found between
SO2 and morning or evening PEF or respiratory symptoms in
children with cough only in the suburban panel.
Asthmatic: Lag 0: 0.198 (0.804). Lag 1: 0.382 (0.789)
Lag 2: 0.648 (0.715). 4-day Mean: 1.39 (1.14)
Odds ratio (per 10 |jg/m3)
URS: Lag 1:1.46 (1.07, 2.00). Lag 2:1.46 (1.14,1.87)
4-day Mean: 1.55 (1.08, 2.24)
Odds ratio when excluded 3 highest SO2 days (no 95% CI
provided, but effects were not significant)
Lag 1:1.13. Lag 2:1.46. 4-day Mean: 1.12
Regression coefficient (SE) (per 10 |jg/m3 SO2):
Morning PEF deviations
Children with cough alone: Lag 0: -0.229 (0.608)
Lag 1: -1.38 (0.564). Lag 2: -0.683 (0.523)
4-day Mean: -1.28 (0.633)
Evening PEF deviations
Children with cough alone. Lag 0: -1.84 (0.673)
Lag 1: -0.144 (0.711). Lag 2: -0.291 (0.613)
4-day Mean: -0.878 (0.868)
Asthmatics: Lag 0:1.28 (0.711). Lag 1: 0.575 (0.727)
Lag 2: 0.819 (0.642). 4-day Mean: 1.34 (1.05)
F-15
-------
Study Method Pollutant Data Findings
The correlation between SO2 and PM varied from 0.5 to 0.8
during first two winters. Correlation with NO2 about 0.50. In the
urban areas, SO2 was associated with > 10% decrements in
evening PEF, LRS and use of bronchodilator in children with
symptoms (n = 142). Most consistent associations found with
PM10, BS, and sulfate. No association found between SO2 and
prevalence of URS, cough, phlegm, and > 10% decrements in
morning PEF. In the nonurban areas, no associations found with
SO2. In children without symptoms, no consistent associations
with SO2. Authors concluded that children with symptoms are
more susceptible to particulate air pollution effects and that use
of medication for asthma did not prevent the adverse effects of
PM in children with symptoms.
Odds ratio (per 40 |jg/m3 SO2)
Children with symptoms:
Urban areas:
Evening PEF: Lag 0:1.32 (0.96,1.80).
Lag 1:0.83 (0.60,1.14). Lag 2:1.67 (1.28,2.19)
Symptoms of lower respiratory tract
Lag 0:1.35 (1.01,1.79). Lag 1:1.23 (0.93,1.64)
Symptoms of upper respiratory tract
Lag 0: 0.97 (0.82,1.14).Lag 1:1.10 (0.94,1.28)
Cough: Lag 0: 0.90 (0.77,1.05). Lag 1:1.12 (0.96,1.30)
Use of bronchodilator: Lag 0: 0.92 (0.72, 1.18)
Lag 1:1.45 (1.13,1.86)
Nonurban areas:
Evening PEF: Lag 0:1.20 (0.91,1.58). Lag 1: 0.89 (0.68,1.17)
Symptoms of lower respiratory tract. Lag 0: 0.91 (0.69,1.19)
Lag 1:0.91 (0.69,1.22)
Symptoms of upper respiratory. Lag 0: 0.94 (0.81,1.09)
Lag 1: 0.97 (0.83,1.13). 5-day Mean: 0.67 (0.47, 0.94)
Cough: Lag 0:1.08 (0.94,1.23). Lag 1: 0.98 (0.85,1.12)
Use of bronchodilator. Lag 0: 0.86 (0.59,1.25)
Lag 1:1.18 (0.80,1.74)
Odds ratio (per 40 |jg/m3 SO2)
Children without symptoms:
Urban areas:
Evening PEF: Lag 0:1.13 (0.88,1.47). Lag 1:1.16 (0.90,1.50)
URS: Lag 0: 0.92 (0.76,1.11). Lag 1:1.10 (0.91,1.34)
Lag 2: 0.83 (0.70, 0.99)
Cough: Lag 0: 0.93 (0.78,1.11). Lag 1:1.02 (0.84,1.23)
Nonurban areas:
Evening PEF: Lag 0:1.10 (0.87,1.39). Lag 1:1.07 (0.85,1.35)
URS: Lag 0:1.07 (0.92,1.25). Lag 1: 0.85 (0.72,1.00)
Cough: Lag 0: 0.86 (0.76, 0.97). Lag 1: 0.95 (0.83,1.08)
Two-Pollutant Models
Odds Ratio (95% CI) (per 40 |jg/m3 ppm SO2)
Evening PEF: SO2 + PM10. Lag 0:1.14 (0.80,1.61)
Lag 1:0.75 (0.51,1.09). Lag 2:1.56 (1.13, 2.13)
5 Day mean: 1.03 (0.50, 2.10)
van der Zee et al.
(1999)
The Netherlands:
Amsterdam and
Meppel (1993-1994)
Amsterdam and
Nunspeet
(1994-1995)
Bodegrven/Reeuwijk
and Rotterdam
(1992-1993)
Period of Study:
3 winters from 1992
to 1995
Panel study of 633 children (aged 7 to
11 yrs) with and without chronic
respiratory symptoms, living in urban and
nonurban areas in the Netherlands.
Volunteers measured daily PEF and
reported the occurrence of respiratory
symptoms and bronchodilator use in a
diary. Association between air pollution
and decrements in PEF, symptoms and
bronchodilator use evaluated with logistic
regression models that adjusted for first
order autocorrelation, min daily
temperature, day of wk, time trend,
incidence of influenza and influenza-like
illness.
Median and max 24-h avg
concentration (|jg/m3)
1992-1993
Urban 23 (152); Nonurban
8.9 (43)
1993-1994
Urban 11 (34); Nonurban
5.0 (42)
1994-1995
Urban 6.0 (24); Nonurban
3.6(17)
Copollutants:
PM10 BS
Sulfate NO2
F-16
-------
Study
Method
Pollutant Data
Findings
van der Zee (2000)
Netherlands,
3 winters from
1992 to 1995
Rotterdam
1992-1993
Panel study of 489 adults (aged 50 to
70 yrs) with and without chronic
respiratory symptoms, living in urban and
nonurban areas in the Netherlands.
Volunteers measured daily PEF and
reported the occurrence of respiratory
symptoms and bronchodilator use in a
diary. Association between air pollution
and decrements in PEF, symptoms and
bronchodilator use evaluated with logistic
regression models that adjusted for first
order autocorrelation, min daily
temperature, day of wk, time trend,
incidence of influenza and influenza-like
illness.
Avg (max) cone
1992/1993:
Urban 25 (61) |jg/m3
1993/1994
Urban 11 (34) |jg/m3
Nonurban 5.0 (42) |jg/m3
1994/1995
Urban 6.0 (24)
Nonurban 3.6 (17) |jg/m3
Copollutants
PM10 BS Sulfate NO2
Among symptomatic adults (n = 138) living in urban areas, the
prevalence of >20% decrement in morning PEF was associated
with SO2. Moreover, there were no associations found with
prevalence of bronchodilator use, LRS, >10% decrement in
morning PEF and >10% and >20% decrement in evening PEF.
In the nonurban areas, there was no consistent association
between air pollution and respiratory health. In the
nonsymptomatic adults, no consistent associations observed
between health effects and air pollutants, but a significant and
positive association was observed with URS in the nonurban
area at 1 day Lag.
Range of Spearman correlation coefficients between 24-h avg
cone SO2 and copollutants:
PM10: 0.31, 0.78. BS: 0.21, 0.75
Sulfate: 0.29, 0.69. N02: 0.47, 0.51
Odds ratio (per 40 |jg/m3 SO2) symptomatic adults
In urban areas
>10% decline in PEF; Morning
Lag 0: 0.86 (0.60,1.23); Lag 1: 0.97 (0.68,1.39)
>20% decline in PEF; Morning
Lag 0:1.33 (0.66, 2.71); Lag 1:1.98 (1.03-3.79)
LRS: Lag 0:1.01 (0.84,1.20)1 Lag 1: .97 (0.82,1.16)
5-day mean: 0.71 (95% CI: 0.53 to 0.95)
URS: Lag 0:1.15 (0.97,1.37)
Lag 1:1.06 (0.90,1.26)
Bronchodilator use: Lag 0:1.09 (0.93, 1.28)
Lag 1:1.05 (0.89,1.24). Lag 2: 0.85 (0.72, 0.99)
In nonurban areas
>10 % decline in PEF; Morning
Lag 0: 79 (0.48,1.29); Lag 1:1.08 (0.68,1.72)
>20% decline in PEF; Morning
Lag 0: 0.79 (0.22, 2.88); Lag 1: 71 (0.13, 4.02)
LRS: Lag 0:1.11 (0.94,1.30);
Lag 1:1.04 (0.88,1.22)
URS: Lag 0:0.97 (0.79,1.20);
Lag 1:1.20 (0.98,1.47)
Bronchodilator use: Lag 0:1.04 (0.91, 1.18);
Lag 1:1.08 (0.95,1.22)
Wardet al. (2002b)
Birmingham and
Sandwell, England
Period of Study:
Jan-Mar 1997,
May-Jul 1997
Children ages 9-yrs old in 5 different
schools were given a questionnaire and
administered PEF measurement in
summer and/or winter. Study used
bivariate correlation, linear and logistic
regressions for analysis
Median, Range (ppb):
Winter: 5.4 (2-18)
Summer: 4.7 (2-10)
Copollutants: NO2; O3;
PM10; PM2.5; H+; Ch; HCI;
HNOs; NH3; NH4+; NOs-;
S042"
Study does not provide evidence for day-to-day respiratory health
effects of pollutants.
Effect size, CI
Winter:
APEF morning (L/min): -0.60 (-2.51,1.32)
APEF afternoon: -0.32 (-2.71, 2.04). Cough: 0.92 (0.82,1.05)
III: 1.09(1.01,1.18). SOB: 1.02 (0.93,1.13)
Wake: 1.00 (0.91,1.10). Wheeze: 0.96 (0.85,1.07)
Summer:
APEF morning (L/min): 0.91 (-0.95, 2.78).
APEF afternoon: -0.89 (-2.61, 0.83). Cough: 1.018 (1.02,1.15)
III: 1.05 (0.96,1.14). SOB: 0.98 (0.87,1.10)
Wake: 1.00 (0.87,1.14). Wheeze: 1.05 (0.92,1.19)
F-17
-------
Study
Method
Pollutant Data
Findings
Ward et al. (2002a)
Birmingham and
Sandwell, England
Period of Study:
Jan-Mar 1997
May-July 1997
Panel study of 162 children (9 yrs at time
of enrollment) from two inner city
locations to investigate the association
between ambient acid species with PEF
and symptoms. Daily symptoms and
twice-daily peak flow measurements
were recorded over 8 wk periods in the
summer and winter. 39 of the children
reported wheezing in the past 12 mos.
Linear regression used for PEF and
logistic regression for symptoms.
24-h avg SO2
Winter:
Jan 13-Mar 10,1997
Median: 5.4 ppb Range: 2,
18 ppb
Summer:
May 19-July 14,1997
Median: 4.7 ppb
Range: 2,10 ppb
Copollutants:
N02, 03, PM10, H*, CI",
HCI, HNOs, NH3, NH4*,
NOs", S042-
SO2 concentrations were
not related to changes in
PEF or respiratory
symptoms
In the summer, changes in morning PEF were associated with
SO2 at 3-days lag and the 7-day mean SO2. Prevalence of cough
associated with SO2 on the same day. In the winter SO2 was only
associated with symptom of feeling ill on the same day. 24-h avg
SO2 (per 4.0 ppb in winter; per 2.2 ppb in summer). Data also
available for 3-,4-, and 7-day Lag
APEF (L/min)
Morning: Lag 0-day
Winter: -0.60 (-2.51, 1.32); Summer: 0.91 (-0.95, 2.78)
Afternoon: Lag 0-day
Winter: -0.32 (-2.71, 2.04); Summer: -0.89 (-2.61, 0.83)
Morning: Lag 1-day
Winter: 0.08 (-1.67,1.86); Summer: 0.29 (-1.56, 2.14)
Afternoon: Lag 1-day
Winter: -0.88 (-2.87, 1.10); Summer: -0.02 (-1.68,1.65)
Odds ratio for symptoms
Coughiag 0-day
Winter: 0.92 (0.81,1.05); Summer: 1.08 (1.02,1.15)
llliag 0-day
Winter 1.09 (1.01, 1.18); Summer 1.05 (0.96,1.14)
SOB: Lag 0-day
Winter: 1.02 (0.93,1.13); Summer: 0.98 (0.87,1.10)
Coughiag 1-day
Winter: 1.00 (0.87,1.15); Summer: 1.04 (0.97,1.11)
III: Lag 1-day
Winter: 1.03 (0.95,1.11); Summer: 1.02 (0.94,1.12)
SOB: Lag 1-day
Winter: 1.00 (0.90,1.09); Summer: 1.00 (0.89,1.13)
Wake at night with cough: Lag 0 day
Winter: 1.00 (0.91,1.10); Summer: 1.00 (0.87,1.14)
Wake at night with cough: Lag 1 day
Winter: 1.05 (0.96,1.15); Summer: 1.02 (0.89,1.16)
Wheeze: Lag 0 day
Winter: 0.96, (0.85,1.07); Summer: 1.05 (0.92,1.19)
Wheeze: Lag 1 day
Winter: 0.96 (0.86,1.07); Summer: 1.00 (0.88,1.13
Summer APEF 2.7 (1.03, 4.38) per 2.2 ppb SO2
Lag 3 days (p < 0.05)
Summer APEF 6.83 (0.98,12.69) per 2.2 ppb SO2
Lag 0-6 days (p < 0.05)
LATIN AMERICA
Pino et al. (2004)
Santiago, Chile
Period of Study:
1995-1997
Cohort study of 504 infants recruited at
4 mos of age and followed through the
first yr of life to determine the association
between air pollution on wheezing
bronchitis.
Mean concentration of SO2 No consistent association was found between the 24-h avg SO2
(ppb) and risk of wheezing bronchitis. However, after a 7-day lag, a 10-
IYIean. ^ g ppb increase in the 24-h avg SO2 was associated with a 21%
5Q. g'l increase in risk of wheezing bronchitis.
Median: 10.0 Increase in wheezing bronchitis (95% CI) per 10 ppb SO2
Copollutants:
PM2.5 NO2
21% (8, 39%)
Romieu et al. (1996) Panel study of 71 mildly asthmatic
Mexico City, Mexico
Period of Study:
Apr-Jul 1991
Nov 1991-Feb 1992
children (5 to 13 yrs) to assess the
relationship between air pollution and
childhood asthma exacerbation. Children
measured PEF three times daily and
recorded daily symptoms and medication
use. Examined both incidence and
prevalence of symptoms. LRS, cough,
phlegm, wheeze, and/or difficulty
breathing.
24-h avg SO2 (ppm)
Mean: 0.09
SD: 0.05
Range: 0.02, 0.20
Copollutants:
O3 PM10 PM2.5 NO2
SO2 concentrations were not related to changes in PEF or
respiratory symptoms.
APEF per 10-ppb increase in SO2
0.26 (-0.35,0.88,1.01) L/min
Odds ratio per 10-ppb SO2
Coughing: 0.96 (0.92,1.01)
LRS: 0.97 (0.94,1.01)
F-18
-------
Study
Method
Pollutant Data
Findings
Chen et al. (1999)
Three towns in
Taiwan:
Sanchun, Taihsi,
Linyuan
Period of Study:
May 1995-Jan 1996
Cross-sectional panel study of 895
children (8 to 13 yrs) to evaluate the
short-term effect of ambient air pollution
on pulmonary function. Single and
multipollutant models adjusted for sex,
height, BMI, community, temperature,
and rainfall. Examined 1, 2, and 7-day
lag effects.
Peak concentrations of
S02
Range: 0, 72.4 ppb
Day-time avg and 1-day
lag
Copollutants:
CO NOs PM10 (r= 0.68)
N02 (r = 0.71)
Daytime peak SO2 at 2 days lag significantly associated with FVC
using the single-pollutant model. Association also observed with
NO2 and CO with FVC. No PM10 effects. Only O3 effects
significant in multipollutant models.
AFVC (mL) daytime avg SO2
Lag 1: -3.18 (1.80); Lag 2: -2.70 (1.49); Lag 7: 0.61 (2.59)
Daytime peak SO2
Lag 1: -0.91 (0.73); Lag 2: -1.27 (0.59), p < 0.05;
Lag 7:-1.05 (1.29)
AFEV1 (mL) daytime avg SO2
Lag 1: -1.95 (1.69); Lag 2: -1.12 (1.41); Lag 7: -1.05 (1.29)
Daytime peak SO2
Lag 1: -0.57 (0.68); Lag 2: -0.64 (0.56); Lag 7: -1.96 (1.22)
Jadsri et al. (2006)
Thailand
Period of Study:
1993-1996
Spatial regression analysis of outpatient —
disease occurrence (respiratory system c ,, . .
diseases; ICD chapter 10) in 25 TSP NOx
communities in Rayong Province.
During summer, SO2 played a role in adverse health effects after
taking into account distance between community and health
providers. During winter, no relationship was found.
Min et al. (2008)
Korea
Panel study consisting of 867 smokers,
former smokers, and never smokers 20-
86 yrs old. Used linear regression
analysis, adjusting forage, height,
gender, and a diagnosis of asthma to
examine the combined effects of
cigarette smoking and SO2 on lung
function. Lung function measurements
used in this analysis included forced vital
capacity (FVC), forced expiratory volume
in 1 sec (FEV1), percent predicted value
of FVC (FVC % pred), and percent
predicted value of FEV1 (FEV1 % pred).
24-h avg (ppm): 0.006 Found a short, marked decrease in FVC and FEV1 in smokers
after exposure to SO2 that lasted for up to 30 h.
Study did not provide effect estimates.
Park et al. (2002)
Seoul, Korea
Period of Study:
Mar 2,1996 to
Dec 22,1999
Time-series analysis of school
absenteeism due to illness and air
pollution in one elementary school in
Seoul. School located in area with heavy
traffic. Avg enrollment in 1996 was 1,264.
24-h avg SO2
Mean: 9.19 ppb
SD: 4.61
Range: 2.68, 28.11
Copollutants:
PM10
NO2CO (r = 0.67) 03
SO2, PM10, and O3 associated with illness related school
absenteeism. SP2 and O3 are protective for non-illness related
absences.
Relative risk per IQR SO2 (5.68 ppb)
Total absences: 1.03 (1.02,1.05)
Non-illness related absences: 0.95 (0.92, 0.99)
Illness related absences: 1.09 (1.07,1.12)
2-pollutant model with O3:1.10 (1.08,1.13)
Park et al. (2005a)
Korea
Period of Study:
Mar to June 2002
Panel study of 69 patients (16 to 75 yrs)
diagnosed with asthma by bronchial
challenge or by bronchodilator response.
Patients recorded twice-daily PE,
symptoms at the end of each day (cough,
wheeze, chest tightness, shortness of
breath, sputum changes and the next
morning, night awakenings). During the
study period, 14 Asian dust days were
identified. GEE and generalized additive
Poisson regression model used in
analysis.
Daily avg SO2
Control Days:
0.0069
(0.0019) ppm
Dust days:
0.0052 (0.0010) ppm
Copollutants:
PM10, N02, CO, 03
During the dust days, SO2 levels were significantly lower
compared to control days. SO2 had no significant effect on PEF
variability or night symptoms.
Relative risk based on Poisson log-linear regression analysis
PEF variability (>20%) 0.76 (0.37,1.56)
Night symptoms: 0.98 (0.59,1.51)
F-19
-------
Study
Method
Pollutant Data
Findings
Xu et al. (1991)
Beijing, China
Three areas:
industrial,
residential, and
suburban (control)
Period of Study:
Aug 1986
Cross sectional survey of 1140 adults (40
to 69 yrs) who had never smoked living
in three areas of Beijing, to determine
respiratory health effects of indoor and
outdoor air pollution. A trained interviewer
obtained pulmonary function meas-
urements and determined history of
chest illnesses, respiratory symptoms,
cigarette smoking, occupational
exposure, residential history, education
level, and type of fuel used for cooking
and heating.
Annual mean
concentration of SO2
(pg/m3)
Residential: 128
Industrial: 57
Suburban: 18
Copollutants:
TSP
An inverse linear association found between Ln outdoor SO2 and
FEV1 and FVC after adjusting for age, height and sex.
Regression estimate and standard error
per Ln SO2 (|jg/m3)
Height-adjusted FEV1 (mL): -35.6 (17.3)
Height-adjusted FVC (mL): -131.4 (18.8)
Table F-2. Short-term exposure to SO2 and emergency department visits and hospital
admissions for respiratory diseases.
Methods
Pollutant Data
Findings
UN TED STATES
Gwynn* et al. (2000)
Buffalo and Rochester, NY
U.S.
Period of Study:
May 1988-October 1990
Days: 1,090
Hospital Admissions
Outcome(s) (ICD9):
Respiratory admissions:
Acute bronchitis/ bronchiolitis
(466); Pneumonia (480-4860);
COPD and Asthma (490-493,
496)
Age groups analyzed: 6
Study design: Time-series
N: 24
Statistical analyses: Poisson
regression with GLM and
GAM
Covariates: season, day of
wk, holiday, temperature,
relative humidity
Lag: 0-3 days
24-h avg SO2
(PPb):
Min: 1.63
25th: 8.4
Mean: 12.2
75th: 15.4
Max: 37.7
Copollutants:
H+ (r= 0.06)
S042- (r = 0.19)
PM10 (r = 0.19)
03 (r = 0.02)
N02 (r = 0.36)
CO (r= 0.11)
COH (r = 0.29)
Significant associations observed between several pollutants and various
health-effect outcomes make it difficult to discriminate the influence of a
single-pollutant. This is likely a result of the relatively high intercorrelations
among the various pollutants, as well as the possible interactive role of
several pollutants in the reported associations.
Increment: 25.5, 7.0 ppb
(Max-Mean; IQR)
SO2 alone:
Max-Mean RR 1.096 (t = 3.05) lag 0
IQR RR 1.025 (t = 3.05) lag 0
F-20
-------
Study
Methods Pollutant Data
Findings
Ito et al. (2007)
New York, NY
Period of Study:
Jan 1999-Dec 2002
ED Visits
Outcome(s): Asthma
Study design: Time-series
Statistical Analysis: Poisson
GLM
Age groups analyzed: All
ages
Covariates: Adjustment for
temperature (same day and
avg lag 1-3), dew point (same
day and avg lag 1-3)
# Hospitals: 11
Lag(s): Avg 0 and 1 day
All yr
24-h avg (ppb): 7.8
(4.6)
5th: 3
25th
50th
75th
95th
Warm mos
(Apr-Sep)
24-h avg (ppb): 5.4
(2.2)
5th: 3
In single-pollutant models, NO2 was found to have the most significant
association with asthma ED visits for all-yr and warm mos. SO2 was
significantly associated with asthma ED visits for all single-pollutant models
for all-yr and both the warm and cold mos. In copollutant models for the warm
mos, NO2 eliminated the association between SO2 and asthma ED visits.
This result is consistent with the monitor-to-monitor correlations, which
suggested that NO2 had less exposure error compared to SO2.
Warm Mos (Apr-Sep)
(Weather model including smoothing terms for same day temperature and
avg lag 1-3 day temperature.)
Relative Risk (95% CI) (per 6 ppb S02) 1.20 (1.13,1.28)
25th
50th
75th
95th
Cold Mos
(Oct-Mar)
24-h avg (ppb):
10.2 (5.1)
5th: 4
25th
50th
75th
95th
Copollutants:
PM2.5; N02; 03; CO
Jaffe et al. (2003)
3 cities, Ohio, U.S.
(Cincinnati, Cleveland,
Columbus)
Period of Study:
Jul 1991-Jun 1996
(June-Aug months only)
ED Visits
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: 5-34
Study design: Time-series; N:
4,416
Statistical analyses: Poisson
regression using a standard
GAM approach
Covariates: City, day ofwk,
wk, yr, min temperature,
overall trend, dispersion
parameter
Season: June to Aug only
Dose-response investigated:
Yes
Statistical package: NR
Lag: 0-3 days
24-h avg:
Cincinnati:
35.9 (25.1) (jg/m3
Range: 1.7,132
Cleveland:
39.4 (25.3) (jg/m3
Range: 2.6,167
Columbus:
11.1(8.5) (jg/m3
Range: 0, 56.8
Copollutants:
Cincinnati:
PM10 (r = 0.31)
N02 (r = 0.07)
03 (r = 0.14)
Cleveland:
PM10 (r = 0.29)
N02 (r = 0.28)
03 (r = 0.26)
Columbus:
PM10 (r = 0.42)
N02 (r = NR)
03 (r = 0.22)
Wide confidence intervals for data from Cleveland and Columbus make these
data not significant and unstable. Only data for Cincinnati was considered
statistically significant and demonstrated a concentration response function
that was positive.
No multipollutant models were utilized.
Increment: 50 |jg/m3
Cincinnati: 35% (9, 21) lag 2
Cleveland: 6% (-7, 21) lag 2
Columbus: 26% (-25, 213) lag 3
All cities: 12% (1,23)
Attributable risk from SO2
increment:
Cincinnati: 4.2%
Cleveland: 0.66%
Columbus: 2.94%
F-21
-------
Study
Methods Pollutant Data
Findings
Lin et al. (2004d)
New York (Bronx County),
U.S.
Period of Study: 6/1991 -
12/1993
Hospital Admissions
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: 0-14
Study design:
Case-control
N: 2,629 cases; 2,236
controls
Statistical analyses: logistic
regression
Covariates: race and ethnicity,
age, gender, season
Statistical package:
Lag: 0,1,2,3, 0-3
Cases:
24-h avg: 16.78
PPb
50th: 13.72
Range: 2.88,
66.35
Controls:
24-h avg: 15.57
PPb
50th: 13.08
Range: 2.88,
66.35
Quartile
Concentrations
(PPb):
Odds ratios for risk of hospitalization for asthma increased with each quartile
of SO2 concentration.
Lag 1, 2, or 3 all showed a concentration response that was positive for odds
ratio as each quartile was compared to the total exposure group
(trend p > 0.001).
Quartile (24-h avg)
Q2 OR 1.26 lag 3
Q3 OR 1.45 lag 3
Q4 OR 2.16 (1.77, 2.65) lag 3
Quartile (1-h max)
Q4 OR 1.86 (1.52, 2.27) lag 3
For a 4 ppb increase in SO2
(24-h avg)
RR 1.07 (1.04,1.11)
2.88, 8.37
9.37,13.38
13.5,20.91
20.21,66.35
Michaud et al. (2004)
Hilo, Hawaii
Period of Study: 2/21/1997-
5/31/2001
ED Visits 1-h max:
Outcome(s) (ICD9): COPD 1 92 (12.2) ppb
(490-496); Asthma (493, 495); Range: 0.0, 447
bronchitis (490, 491), other
COPD (492, 494, 496)
Age groups analyzed: All
Study design: Time-series
Statistical analyses:
Exponential regression
models
Covariates: temporal
variables, day ofwk,
meteorology
Statistical package: Stata,
SAS
Lag: 0,1,2,3 days
24-h avg:
1.97 (7.12) ppb
Range: 0.0,108.5
Copollutants: PM1
The lack of organic carbon shows the pure SO2 effect uncontaminated by
vehicle emissions.
Asthma is associated with vog, but vog is not a major cause of asthma in
Hawaii. The strongest association was with the mo of the yr.
Admission for asthma and respiratory conditions was higher in the winter
compared to the summer, based on admission per day (observational- not
statistical analysis).
Increment: 10 ppb
COPD: RR 1.04 (0.99,1.09) lag 1; RR 1.04 (1.00,1.09) lag 2;
RR 1.07 (1.03,1.11) lag 3
Asthma: RR 1.01 (1.00,1.10) lag 1; RR 1.02 (1.03,1.12) lag 2;
RR 1.02 (1.03,1.12) lag 3
Bronchitis: RR 1.01 (0.93,1.13) lag 1; RR 0.99 (0.88,1.05) lag 2;
RR 1.01 (1.00,1.14) lag 3
Other COPD: RR 1.00 (0.78,1.23) lag 1; RR 0.96 (0.62,1.11) lag 2;
RR 0.98 (0.75,1.16) lag 3
F-22
-------
Study
Methods Pollutant Data
Findings
Moolgavkar* et al. (1997)
U.S.: Minneapolis-St. Paul;
Birmingham
Period of Study:
Jan 1986-Dec 1991
Hospital Admissions
Outcome(s) (ICD9): COPD
including asthma
(490-496), Pneumonia
(480-487)
Age groups analyzed: 65+
Study design: Time-series
Statistical analyses: Semi-
parametric Poisson
regression, GAM
Covariates: day ofwk,
season, temporal trends,
temperature
Statistical package:
S Plus
Lag: 0-3 days
SO2 24-h avg
(PPb):
Minneapolis:
Mean: 4.82
10th
25th
50th
75th
90th
1.9
2.66
4.02
6.0
: 8.5
Birmingham:
Mean: 6.58
10th
25th
50th
75th
90th
2.2
3.7
6.0
8.6
11.6
Copollutants:
Minneapolis:
PM10 (r = 0.08)
N02 (r = 0.09)
CO (r = 0.07)
03 (r = -0.12)
Birmingham:
PM10 (r = 0.17)
CO (r = 0.16)
03 (r = 0.02)
SO2 with NO2 and PM10 were associated with hospital admissions. Evidence
of mixture effects was found. No single-pollutant was more important than the
other for respiratory admissions. Each pollutant was associated with
admissions except CO.
Consideration of four pollutants together showed the strongest association
with O3. No pollutant other than O3 was stable in its association with hospital
admissions.
No effects were reported for Birmingham. Positive results were only observed
in Minneapolis.
Increment: 3.5 ppb
Sum of Pneumonia and COPD: 1.6% (-0.1, 3.3) lag 2
Pneumonia Only
Minneapolis:
65+0.9% (-1.1,2.9) lag 2 20 df
0.5% (-1.5, 2.5)
lag 2 130 dfS
Moolgavkar (2000a)
Multicity, U.S.: Cook, Los
Angeles, Maricopa County,
(Phoenix)
Period of Study:
1987-1995
Hospital Admissions
Outcome(s) (ICD9): COPD
including asthma (490-496)
Age groups analyzed:
0-19,20-64, 65+
(LA only)
Study design: Time-series
Statistical analyses: Poisson
regression, GAM
Covariates: Day ofwk,
temporal trends, temperature,
relative humidity
Lag: 0-5 days
Cook:
Median: 6 ppb
25th: 4
75th: 8
Range: 0.5, 36
Los Angeles:
Median: 2 ppb
25th: 1
75th: 4
Range: 0,16
Maricopa:
Median: 2 ppb
25th: 0.5
75th: 4
Range: 0,14
Copollutants:
Cook:
PM10 (r = 0.42)
CO (r = 0.35)
N02 (r = 0.44)
03 (r = 0.01)
Los Angeles:
PM2.5 (r = 0.42)
PM10 (r = 0.41)
CO (r = 0.78)
N02 (r = 0.74)
03 (r = -0.21)
Maricopa:
PM10 (r = 0.11)
CO (r = 0.53)
N02 (r = 0.02)
03 (r = -0.37)
In Los Angeles there was a significant association with hospital admissions
for COPD.
SO2 may be acting as a surrogate for other pollutants since heterogeneous
responses found in different cities are inconsistent with a cause-effect model.
Increment: 10 ppb
COPD, >65 yrs
Cook lagO: 4.87 (t = 3.18) GAM-100
LA lag 0:2.84 (t = 13.32) GAM-30
LA lag 0:1.80 (t = 9.60)
GAM-100
LA lag 0:1.78 (t = 7.72)
NS-100
F-23
-------
Study
Methods Pollutant Data
Findings
NY DOH (2006)
Bronx and Manhattan, NY
Period of Study:
Jan 1999-Dec 2000
ED Visits
Outcome(s) (ICD9): Asthma
(493), for infants (466.1 and
786.09)
Study design: Time-series
N:« 36,000
Statistical Analysis: Poisson
regression with GLM
Statistical package: S-Plus
Age groups analyzed: All
ages
Covariates: Season, day-of-
wk, temperature
# Hospitals: 22
Lag(s): Avg 0- to 4-day lags
24-h avg (ppm):
0.011 (0.0072)
Copollutants:
PM10 PM2.5 OC
EC Cr Fe Pb
Mn Ni Zn H+
Sulfate O3
N02 S02
In single-pollutant models, PM2.5, SO2, O3, and NO2 were all found to be
significantly associated with asthma ED visits in a community, Bronx, with a
high prevalence of asthma. This association was maintained in both two- and
three-pollutant models for O3 and SO2.
Single-Pollutant Models
Relative Risk: (95% CI) (per 0.011 ppm SO2)
5-day moving avg
Manhattan: 0.99 (0.88,1.12); Bronx: 1.11 (1.06,1.17)
Relative Risk: (95% CI) (per 0.0072 ppm SO2)
Bronx: 1.07 (1.04,1.11)
Relative Risk based on Daily Max Hourly SO2: (95% CI) (per 0.0227 ppm
S02)
Manhattan: 0.96 (0.86,1.07); Bronx: 1.07 (1.03,1.12)
Relative Risk:(95% CI) (per 0.0072 ppm S02)-model excludes temperature
Manhattan: 0.99 (0.88,1.11); Bronx: 1.11 (1.06,1.17)
Relative Risk: (95% CI) (per 0.0072 ppm SO2)- by Gender
Manhattan: Male: 0.90 (0.75,1.07); Female: 1.08 (0.91,1.29)
Bronx: Male: 1.08 (1.00,1.17); Female: 1.14(1.06,1.23)
Relative Risk: (95% CI) (per 0.0072 ppm S02)-By Age
Manhattan: 0-4:0.82 (0.59,1.15); 5-18:1.03 (0.77,1.37);
19-34:1.01 (0.76,1.35); 35-64:1.04 (0.86,1.25); 65+: 0.88 (0.57,1.37)
Bronx: 0-4:1.13 (1.01,1.26); 5-18:1.03 (0.92,1.16); 19-34:1.06 (0.93,1.21);
35-64:1.18 (1.07,1.30); 65+: 1.12 (0.88, 1.42)
Two-Pollutant Models
Relative Risk: (95% CI) (per 0.0072 ppm SO2)
5-day moving avg
Manhattan: SO2+ Max8-h O3: 0.99 (0.88,1.12);
S02 + FRM PM2.5: 0.97 (0.85,1.11); SO2 + Max PM2.5:0.98 (0.85,1.12);
SO2 + NO2:1.01 (0.87,1.16)
Bronx: S02 + Max 8-h 03:1.11 (1.05,1.17);
SO2 +FRM PM2.5:1.11 (1.04,1.18); SO2 +Max PM2.5:1.09 (1.03,1.16);
SO2 + NO2:1.11 (1.04,1.17)
Norris et al. (1999)
Seattle, Washington
Period of Study:
1995-1996
ED Visits
Outcome(s) (ICD9): Asthma
(493)
Study design: Time-series
Statistical Analysis:
Semiparametric Poisson
regression model
Statistical package: S-Plus
Age groups analyzed: < 18
Covariates: adjustments for
day-of-wk indicator variables,
time trends, temperature, dew
point temperature
# Hospitals: 6
Lag(s): 0, 2
24-h avg (ppb): 6.0
(3.0)
Range: 1.0, 21.0
1-h max (ppb):
16.0(14.0)
Range: 2.0, 84.0
Copollutants:
PM10, Dry light
scattering,
N02,CO, 03
A significant association was found between asthma emergency department
visits in children and PM2.5 and CO. Estimates were not found to be different
between high and low hospital utilization areas. S02was negatively
associated with asthma emergency department visits in high utilization areas,
and positively associated in low utilization areas.
Relative Rates (95% CI) (per 3 ppb 24-h avg SO2; per 12 ppb 1-h max SO2)
High Utilization Areas:
24-h avg: 0.92 (0.83,1.03)
1-h max: 0.99 (0.89,1.10)
Low Utilization Areas
24-h avg: 1.09 (1.00,1.19)
1-h max: 1.09(1.00,1.19)
All Areas
24-h avg: 0.97 (0.91,1.04) lag 0
1-h max: 1.02 (0.95,1.09) lag 2
F-24
-------
Study
Methods Pollutant Data
Findings
Peel et al. (2005)
Atlanta, GA, U.S.
Period of Study:
1/93-8/2000
ED Visits
Outcome(s) (ICD 9): All
respiratory (460-6, 477,
480-6, 480-6, 490-3, 496);
Asthma (493); COPD (491-2,
496); Pneumonia (480-486);
Upper Respiratory Infection
(460-6, 477)
Age groups analyzed: All
Study design: Time-series
N:« 480,000
# of Hospitals: 31
Statistical analyses: Poisson
Regression, GEE, GLM, and
GAM (data not shown for
GAM)
Covariates: Day ofwk,
hospital entry/exit, holidays,
time trend; season,
temperature, dew point
temperature
Statistical package: SAS, S-
Plus
Lag: 0 to 7 days. 3 day
moving avgs.
1-h max: 16.5
(17.1) ppb
10th%: 2.0
90th%: 39.0
Copollutants: O3
N02
CO
PM2.5
Evaluated
multipollutant
models (data not
shown)
Estimates from distributed lag models (0-13 days) tend to be higher than for
3-day moving avg. Positive associations for URI and COPD with S02were
noted for unconstrained lags (0-13 days) that covered the previous two
weeks of exposure.
Increment: 20 ppb
All respiratory:
RR 1.008 (0.997,1.019) lag 0-2, 3-day moving avg
Upper Respiratory Infection (URI):
RR 1.010 (0.998,1.024) lag 0-2, 3-day moving avg
Asthma:
All: 1.001 (0.984,1.017) lag 0-2, 3-day moving avg
Pneumonia:
RR 1.003 (0.984,1.023) lag 0-2, 3-day moving avg
COPD:
RR 1.016 (0.985,1.049) lag 0-2, 3-day moving avg
24-h
avg
New Haven
Mean 78 |jg/m3
(29.E
PPb)
10th
23
25th
35
50th
78
75th
100
90th
159
Tacoma
Mean: 44 |jg/m3
(16.E
PPb)
10th
15
25th
26
50th
40
75th
56
90th
74
Copollutants:
PM10O3
24-h avg: 35 ppb
10th
13
25th
20
50th
31
75th
45
90th
61
Copollutants:
PM2.5O3
Schwartz (1995)
New Haven, CT
Tacoma, WA
U.S.
Period of Study:
1988-1990
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory admissions (460-
519)
Age groups analyzed: a65
Study design: Time-series
N: 13,470
Statistical analyses: Poisson
regression, log linear
regression using GLM and
GAM
Covariates: dewpoint, temp,
long-term trends, days ofwk
Statistical package: S- Plus
Lag: 0-1
In New Haven, risk associated with SO2 was not affected by inclusion of
PM2.5 in the model and the effect of PM2.5 was not strongly affected by
inclusion of SO2. This suggests that in New Haven, SO2 and PM2.5 acted
independently.
In Tacoma, 2-pollutant model analysis showed risk associated with SO2 was
attenuated by PM2.5. This suggested risks associated with SO2 and PM2.5
were not independent. Possibly, SO2 acts as a surrogate for PM2.5 in this city.
Increment: 50 |jg/m3 or 18.8 ppb
New Haven, CT
RR = 1.03 (C11.02, 1.05), lag 0-1. p < 0.001
2-pollutant model with PM10:
RR = 1.04 (C11.02, 1.06) p< 0.001
Tacoma, WA
RR = 1.06 (C11.01, 1.12), lag 0-1. p> 0.02
2-pollutant model with PM10:
RR = 0.99 (CI 0.93, 1.06) p> 0.5
Schwartz et al. (1996)
Cleveland, OH
Period of Study:
1988-1990
Hospital Admissions
Outcome(s) (ICD9): All
respiratory disease
Age groups analyzed: a 65
Study design: Time-series
Statistical analyses: Poisson
regression
Covariates: Season,
temperature, day ofwk
Statistical package:
Lag: 0-1
Significant associations were seen for PM2.5 and O3, with somewhat weaker
evidence for SO2.
Increment: 100 |jg/m3
RR 1.03 (0.99,1.06) lag 0-1
F-25
-------
Study
Methods Pollutant Data
Findings
Sheppard et al. (1999)
Reanalysis (2003)
Seattle, WA, U.S.
Period of Study:
Jan1987-Dec1994
Hospital Admissions
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: < 65
Study design: Time-series
N: 7,837
# of Hospitals: 23
Statistical analyses: Poisson
regression with adjustment for
auto-correlation.
Covariates:
Statistical package: S-Plus
Lag: 0,1,2,3
24-h avg: 8 ppb
IQR: 5 ppb
10th
25th
50th
75th
90th
3.0
5.0
8.0
10.0
13.0
Copollutants:
PM10 (r = 0.31)
PM2.5 (r = 0.22)
03 (r = 0.07)
CO (r = 0.24)
Sources of SO2 adjacent or near to monitoring site. Low concentrations. No
association with SO2 for asthma but positive association for appendicitis.
Increment: 5 ppb (IQR)
GAM with stricter criteria:
1.0% (-2.0, 3.0) lag 0
GLM with natural spline
smoothing:
0.0% (-3.0, 4.0) lag 0
Sinclair and Tolsma (2004)
Atlanta, GA
Period of Study:
8/1/1998-8/31/2000
ED Visits
Outcome(s): asthma, upper,
and lower respiratory
infections.
Study design: Time-series
investigation
Statistical Analysis: Single
pollutant Poisson general
linear modeling
Statistical package: SAS v.
8.02
Age Groups Analyzed: All
# Hospitals: 10
Lag(s): 0-8 days
1-hour Max Mean:
19.28 ppb
SD:16.28
Copollutants:
PM2.5PM10NO2
CO 03
No significant findings for child or adult asthma.
Significant negative associations with upper respiratory infections for 6-8 day
lag (RR = 0.98).
Significant positive association with lower respiratory infections for 0-2 day
lag (RR = 1.067).
Not provided.
Tolbert et al. (2007)
Atlanta, GA
Period of Study:
Jan 1993-Dec 2004
ED Visits
Outcome(s) (ICD9):
Cardiovascular (410-414,
427, 428, 433-437, 440, 443-
445, 451-453); Respiratory
(493,786.07,786.09,491,
492, 496, 460-465, 477, 480-
486,466.1,466.11,466.19)
Study design: Time-series
Statistical Analysis: Poisson
Generalized Linear Model
(GLM).
Statistical package: SAS
Age groups analyzed: All
ages
Covariates: Adjustment for
day-of-wk, hospital entry,
holidays, time, temperature,
dew point temperature
# Hospitals: 41
N: 238,360 (Cardiovascular)
1,072,429 (Respiratory)
Lag(s): 3-day moving avg
1-h max (ppb):
14.9
Range: 1.0,149.0
10th
25th
75th
90th
2.0
4.0
20.0
35.0
In single pollutant models, O3, PM10, CO, and NO2 significantly associated
with ED visits for respiratory outcomes.
Relative Risk (95% CI)
(per 16.0 ppb SO2)
1.003 (0.997,1.009)
Copollutants: PM10
PM2.5O3NO2CO
Sulfate
Total Carbon
Organic Carbon
EC
Water-Soluble
Metals
Oxygenated
Hydrocarbons
F-26
-------
Study
Methods Pollutant Data
Findings
Wilson et al. (2005)
Multicity, U.S.
(Portland, ME and
Manchester, NH)
Period of Study:
Jan 1996-Dec 2000
(Manchester)
Jan 1998-Dec 2000
(Portland)
ED Visits
Outcome(s) (ICD 9 codes): All
respiratory
(460-519); Asthma (493)
Age groups analyzed:
0-14 yrs; 15-64 yrs; a65 yrs
Study design: Time-series
Statistical analyses: Multiple
regression analysis standard
GAM with more stringent
criteria parameters
Covariates:
Time-trend, season, influenza,
temperature, humidity,
precipitation
Statistical package: S-Plus
Lag: 0-2
SO21-h max:
Mean, (SD) (ppb)
Portland
All yr: 11.1 (9.1)
Winter: 17.1 (12.0)
Spring: 10.0 (7.1)
Summer: 9.1 (8.0)
Fall: 9.7 (7.1)
Manchester
All yr: 16.5 (14.7)
Winter: 25.7 (15.8)
Spring: 14.8 (12.0)
Summer: 10.6
(15.1)
Fall: 14.6(11.1)
Copollutants:
O3 PM2.5
Elevated levels of SO2 were positively associated with elevated respiratory
and asthmatic ER visits. The significance of these relationships is not
sensitive to analytic or smoothing techniques.
Increment: 6.3 ppb (IQR) for Portland; IQR for Manchester
Portland:
All respiratory: All ages RR 1.05 (1.02,1.07) lag 0;
0-14 yrs RR 0.98 (0.93,1.02) lag 0; 15-64yrs RR 1.06 (1.03, 1.09) lag 0;
>65 yrs RR 1.10 (1.05,1.15) lag 0
Asthma: All ages RR 1.06 (1.01,1.12) lag 2;
0-14 yrs RR 1.03 (0.93,1.15) lag 2; 15-64yrs 1.07 (1.01,1.15) lag 2;
>65 yrs RR 1.07 (0.90,1.26) lag 2
Manchester:
All respiratory: All ages RR 1.01 (0.99,1.02) lag 0;
0-14 yrs RR 1.00 (0.96,1.04) lag 0; 15-64 yrs RR 1.00 (0.98, 1.03) lag 0;
>65 yrs RR 1.04(0.97,1.11) lag 0
Asthma: All ages RR 1.03 (0.98,1.09) lag 2;
0-14 yrs RR 1.11 (0.98,1.25) lag 2; 15-64 yrs 1.02 (0.96,1.08) lag 2;
>65 yrs RR 1.06 (0.83,1.36) lag 2
Bates et al. (1990)
Vancouver Region, BC,
Canada
Period of Study:
7/1/1984-10/31/1986
ED Visits
Outcome(s) (ICD 9): Asthma
(493);
Pneumonia (480-486);
Chronic bronchitis
(491,492,496);
Other respiratory (466)
Age groups analyzed:
All; 15-60
Study design:
# of Hospitals: 9
Statistical analyses: Pearson
correlation coefficients were
calculated between asthma
visits and environmental
variables
Season:
Warm (May-Oct);
Cool (Nov-Apr)
Covariates: NR
Lag: 0,1,2
May-Oct
SO21-h max:
Range: 0.0137,
0.0151 ppm
Nov-Apr
Range: 0.012,
0.0164 ppm
Number of
stations: 11
Copollutants:
May-Oct:
03 (r = 0.23)
N02 (r = 0.67)
COH (r = 0.34)
S04 (r = 0.46)
Nov-Apr:
03 (r = 0.47)
N02 (r = 0.61)
COH (r = 0.64)
S04 (r = 0.54)
SO2 effects depend on the season. In the summer a rise in ambient SO2
levels was seen to coincide with a rise in respiratory related hospital
admissions.
Correlation Coefficients:
Warm Season (May-Oct)
Asthma (15-60 yrs): r = 0.118 lag 0 p < 0.01; r = 0.139 lag 1
Respiratory (15-60 yrs): r = 0.134 lag 0 p < 0.001; r = 0.164 lag 1 p < 0.001
Cool Season (Nov-Apr)
Respiratory (1-14yrs): r = 0.205 lag 0 p < 0.001; r = 0.234 lag 1 p < 0.001;
r= 0.234 lag 2 p < 0.001
(15-60 yrs): r = 0.180 lag 0 p < 0.001; r = 0.214 lag 1 p < 0.001;
r= 0.215 lag 2 p < 0.001
(a 61 yrs 0: r = 0.257 lag 0 p < 0.001; r = 0.308 lag 1 p < 0.001;
r= 0.307 lag 2 p < 0.001
Asthma (a 61 yrs): r = 0.125 lag 0 p < 0.001; r = 0.149 lag 1 p < 0.001;
r= 0.148 lag 2 p < 0.001
Total ER admissions (a 61 yrs): r = 0.13 lag 1 p < 0.01;
r = 0.13 lag 2 p < 0.01
Burnett* et al. (1997a)
16 cities, Canada
Period of Study:
4/1981-12/1991
Days: 3,927
Hospital Admissions
Outcome(s) (ICD9): All
respiratory admissions (466,
480-6, 490-4, 496)
Study design: Time-series.
N: 720,519
# of Hospitals: 134.
Statistical analyses: random
effects relative risk regression
model
Covariates: Long-term trend,
season, day of wk, hospital
Statistical package: NR
Lag:0,1,2 day
1-h max SO2 (ppb)
Mean: 14.4. SD:
22.2
25th
50th
75th
95th
99th
03r = 0.04
Copollutants:
CO, N02,COH
Control of SO2 reduced but did not eliminate the O3 association with
respiratory hospital admissions.
Increment: 10 ppb
Single-pollutant
SO2 and respiratory admissions, p = 0.134
Multipollutant model (adjusted for CO, O3, NO2, COH, dew point):
RR 1.0055 (0.9982,1.0128) lagO
F-27
-------
Study
Methods Pollutant Data
Findings
Burnett* et al. (1997b)
Toronto, Canada
Period of Study:
Summer only, 1992-1994
Hospital Admissions
Outcomes (ICD 9 codes):
Respiratory tracheobronchitis
(480-6), COPD (491-4,496)
Study design: Time-series
Statistical analyses: Poisson
regression, GEE, GAM
Covariates: Temperature, dew
point temperature, Long-term
trend, season, influenza, day
ofwk
Season: Summers only
Lag: 0,1,2,3,4 days
Avg SO2: 7.9 ppb
CV: 64
Range: 0, 26
5th: 1
25th
50th
75th
95th
Number of
stations: 6-11
Copollutants:
CO (r = 0.37)
H+ (r= 0.45)
S04 (r = 0.42)
TP (r = 0.55)
FP (r = 0.49)
CP (r = 0.44)
COH (r = 0.50)
03 (r = 0.18)
N02 (r = 0.46)
Risks of hospitalization for respiratory disease were summed for O3, NO2,
and S02at 11% increase in admissions. The proportion associated with the
single-pollutant SO2 was 3.6%. CoH was the strongest predictor of
hospitalization indicating particle associated pollutants are responsible for
effects and outcomes measured.
Increment: 4.00 ppb (IQR)
Respiratory-percent increase 4.0% (t = 4.14) lag 0
Copollutant and multipollutant models RR
(t-statistic):
S02, COH: 1.012(1.10)
S02, H+: 1.022 (1.96)
S02, S04:1.021 (1.93)
S02, TP
S02, FP
S02, CP
1.021 (1.72)
1.022 (1.92)
1.023 (2.03)
S02,03, N02:1.019 (1.64)
Burnett* et al. (1999)
Toronto, Canada
Period of Study:
Jan1980-Dec1994
Hospital Admissions
Outcome(s) (ICD9): Asthma
(493); obstructive lung
disease (490-2, 496);
Respiratory infection (464,
466, 480-7, 494)
Study design: Time-series
N: m 60,000 (asthma)
Statistical analyses: Poisson
regression model with
stepwise analysis
Covariates: Long-term trends,
season, day ofwk, daily max
temperature, daily min
temperature, daily avg dew
point temperature, daily avg
relative humidity
Statistical package: S-Plus,
SAS
Lag: 0,1,2 days, cumulative
24-h Avg: 5.35 ppb The percent hospital admissions associated with SO2 increased for: asthma,
CV = 110; COPD, and respiratory infection. However, in multipollutant models
g^. g significant increases were only seen in asthma and respiratory infection. SO2
effects could be largely explained by other variables in the pollution mix as
demonstrated by the Multipollutant model. The greatest contribution of SO2 is
to respiratory infection. However, overall SO2 is a small factor in total
hospitalization response.
Increment: 5.35 ppb (Mean)
Single-pollutant model percent increase (t statistic)
Asthma: 1.01% (1.76) lag 0-2; OLD: 0.03% (0.05) lag 0-1
Respiratory infection: 2.40% (5.04) lag 0-2
Multipollutant model percent increase (SE)
Asthma: SO2 + CO + 03: 0.89% (SE < 2);
S02 + CO + 03 + PM2.5: 0.69% (SE < 2);
S02 + CO + 03 + PM10-2.5: 0.16% (SE < 2);
S02 + CO + 03 + PM2.5: 0.76% (SE < 2)
Respiratory infection: SO2 + NO2 + O3:1.85%;
S02 + N02 + 03 + PM2.5: 0.67 (SE < 2);
S02 + N02 + 03 + PM10-2.5:1.71 (SE > 3);
S02 + N02 + 03 + PM2.5:1.00 (SE > 2)
25th
50th
75th
95th
100th: 57
Number of
stations: 4
Copollutants:
PM2.5 (r = 0.46)
PM10-2.5 (r = 0.28)
PM10 (r = 0.44)
CO (r = 0.37)
N02 (r = 0.54)
03 (r = 0.02)
Burnett* et al. (2001)
Toronto, Canada
Period of Study:
1980-1994
Hospital Admissions
Outcome(s) (ICD9): Croup
(464.4), pneumonia (480-
486), asthma (493), acute
bronchitis/bronchiolitis (466)
Age groups analyzed:
< 2 yrs
Study design: Time-series
Statistical analyses: Poisson
regression with GAM
Covariates: Temporal trend,
day ofwk, temperature,
relative humidity
Statistical package:
S-Plus
Lag: 0-5 days
1-h max SO2 (ppb) SO2 had the smallest effect on respiratory admissions of all pollutants
considered.
Mean: 11.8
CV: 93
5th: 0
25th
50th
75th
95th
99th
100th: 110
Number of
stations: 4
Copollutants:
03 (r = 0.39) S02
CO PM2.5PM10-2.5
Increment: NR
All respiratory admissions:
Single-pollutant:
Percent increase: 3.1% (t= 1.900)
lag 3
Multipollutant (adjusted for O3):
Percent increase: 1.21% (t = 0.67)
lag 3
F-28
-------
Study
Methods Pollutant Data
Findings
Cakmak et al. (2007)
Canada
(Calgary, Edmonton, Halifax,
London, Ottawa, Saint John,
Toronto, Vancouver, Windsor,
Winnipeg)
Period of Study:
4/1/1993-3/31/2000
Hospital Admissions
Outcome(s) (ICD9):
Respiratory (466, 480-486,
490-494, 496)
Study design: Time-series
Statistical Analysis: Poisson
Statistical package: S-Plus
Age groups analyzed: All
ages
Covariates: Day-of-wk, mean
daily temperature, max daily
temperature, min daily
temperature, change in
barometric pressure, mean
relative humidity
N: 215,544
Lag(s): 2.6 days
24-h avg: 4.6 ppb SO2 associated with increased hospital admissions.
Range: 2.8 ppb to % increase
10.2 ppb (per 4.6 ppb SO2)
Copollutants: Overall
O3NO2CO Single-pollutant model: 1.1% (0.5, 1.8)
Multi-pollutant model: 0.5% (0.1, 0.9)
By Gender: Male: 0.4% (-0.2,1.1); Female: 0.9% (-0.4,2.1)
By Education: 35,905: 0.7% (-0.4,1.8)
¦0.1, 1.9):
-0.4, 1.4);
Fung et al. (2006)
Vancouver, BC, Canada
Period of Study:
6/1/95-3/31/99
Hospital Admissions
Outcome(s) (ICD9): All
respiratory hospitalizations
(460-519)
Age groups analyzed: 65+
Study design: (1) Time-series
(2) Case-crossover, (3) DM-
models (Dewanji and
Moolgavkar, 2000, 2002)
N: 40,974
Statistical analyses: (1)
Poisson, (2) conditional
logistic regression, (3) DM
method-analyze recurrent
data in which the occurrence
of events at the individual
level over time is available
Covariates: Day ofwk
Statistical package: S-Plus
and R
Lag: Current day, 3 and
5 day lag
SO2 24-h avg:
Mean: 3.46 ppb
SD: 1.82
IQR: 2.50 ppb
Range: 0.00,
12.50
Copollutants:
CO (r = 0.61)
COH (r = 0.65)
N02 (r = 0.57)
PM10 (r = 0.61)
PM2.5 (r = 0.42)
PM10-2.5 (r = 0.57)
03 (r = -0.35)
No significant association was found between hospital admissions and
current day SO2 levels (lag 0). Significant associations were found with SO2
using a 3, 5, and 7 day moving avg, with the strongest association observed
with a 7 day lag. The DM method produced slightly higher relative risks
compared to the Time-series and case crossover results.
Increment: 2.5 ppb (IQR)
SO2 Time-series: RR 1.013 (0.997,1.028) lag 0;
RR 1.030 (1.010,1.051) lag 0-3; RR 1.032 (1.008,1.056) lag 0-5;
RR 1.031 (1.003,1.060) lag 0-7
SO2 Case-crossover: RR 1.010 (0.992,1.027) lag 0;
RR 1.028 (1.005,1.050) lag 0-3; RR 1.030 (1.004,1.057) lag 0-5;
RR 1.028 (0.998,1.058) lag 0-7
S02 DM model: RR 1.013 (0.998,1.027) lag 0;
RR 1.034 (1.015,1.053) lag 0-3; RR 1.039 (1.016,1.061) lag 0-5;
RR 1.044 (1.018,1.070) lag 0-7
DM method produced slightly higher RR estimates on O3, SO2 and PM2.5
compared to time-series and case-crossover, and slightly lower RR estimates
on COH, NO2, and PM10, though the results were not significantly different
from one another.
Kesten et al. (1995)
Toronto, ON, Canada
Period of Study:
7/1/1991-6/30/1992
ED Visits
Outcome(s) (ICD 9): Asthma
(493)
Age groups analyzed: all ages
Study design: Time-series
N: 854
# of Hospitals: 1
Statistical analyses: Auto
regression
Statistical package: SAS
Lag: 1 or 7
SO2 24-h avg
No data was
provided for
concentration or
for correlation with
other pollutants.
Copollutants:
NO2 O3
API (TRS, CO,
TSP)
Fit of an auto-regression model with covariates linked to same day gave no
evidence of association between asthma and SO2.
Despite multiple attempts to correlate individual or combinations of pollutants
with air quality indices, no association was found between ER visits for
asthma and ambient daily, weekly, or monthly levels of SO2, NO2, or O3.
No relative risks were provided.
F-29
-------
Study
Methods Pollutant Data
Findings
Lin et al. (2003)
Toronto, ON, Canada
Period of Study:
1981-1993
Hospital Admissions
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed:
6-12
Study design:
Bi-directional case-crossover
N: 7,319
Statistical analyses:
Conditional logistic regression
Covariates: Daily max and
min temperatures and avg
relative humidity
Lag: Cumulative lag of 1-7
days.
SO2 24-h avg:
5.36 ppb SD: 5.90
Range: 0, 57.00
25th
50th
75th
1.00
4.00
8.00
SO2 is positively associated with asthma hospitalizations, although the
relationship varies in boys and girls.
Increment: 7 ppb (IQR)
Boys 6-12 yrs; Girls 6-12 yrs:
Lag 0: OR 1.00 (0.95,1.05); 1.04 (0.97,1.11); Lag 0-1: OR 0.99 (0.93,1.06);
1.04(0.95,1.13
Number of
stations: 4
Copollutants:
CO (r = 0.37)
N02 (r = 0.54)
PM10 (r = 0.44)
03 (r = -0.01)
PM2.5 (r = 0.46)
PM10-2.5 (r = 0.28)
Lag 0-2
Lag 0-3
Lag 0-4
Lag 0-5
Lag 0-6
OR 0.98 (0.90,1.06); 1.05 (0.95,1.16)
OR 0.96 (0.87,1.05); 1.09 (0.98,1.22)
OR 0.95 (0.86,1.05); 1.13 (1.00,1.28)
OR 0.93 (0.83,1.03); 1.17 (1.02,1.34)
OR 0.93 (0.83,1.04); 1.20 (1.04,1.39)
Multipollutant model with PM10-2.5 and PM2.5
Boys 6-12 yrs; Girls 6-12 yrs:
Lag 0: OR 0.98 (0.93,1.04); 1.06 (0.98,1.14)
Lag 0-1
Lag 0-2
Lag 0-3
Lag 0-4
Lag 0-5
Lag 0-6
OR 0.99 (0.91
OR 0.96 (0.88
OR 0.95 (0.85
OR 0.94 (0.84
OR 0.91 (0.80
OR 0.91 (0.80
1.06); 1.03 (0.93,1.14)
1.05); 1.04 (0.92,1.17)
1.05); 1.08 (0.95,1.23)
1.06); 1.12 (0.97,1.29)
1.04); 1.18 (1.00,1.38)
1.04); 1.28 (1.08,1.51)
Lin* et al. (2004c)
Vancouver, BC, Canada
Period of Study:
1987-1998
Hospital Admissions
Outcome(s) (ICD9):
Asthma (493)
Age groups analyzed: 6-12
Study design: Time-series
N: 3,754 (2,331 male, 1,423
female)
Statistical analyses: Semi-
parametric Poisson
regression with GAM (with
default and more stringent
criteria)
Covariates: Trend, day of wk,
Statistical package:
S-Plus
Lag: Cumulative
1-7 day
24-h avg SO2
(PPb)
Mean: 4.77
SD: 2.75
Min: 0
25th: 2.75
50th: 4.25
75th: 6.00
Max: 24.00
Number of
stations: 30
Copollutants:
CO (r = 0.67)
N02 (r = 0.67)
03 (r = -0.10)
Results presented are default GAM, but authors state that use of natural
cubic splines with a more stringent convergence rate produced similar results
Increment: 3.3 ppb (IQR)
Boys 6-12 yrs by SES status:
Lag 0RR 1.02(0.94,1.10); 1.
Lag 0-1 RR 1.03 (0.94,1.13)
Lag 0-2 RR 1.03 (0.93,1.15)
Lag 0-3 RR1.01 (0.90,1.13)
Lag 0-4 RR0.98 (0.88,1.10)
Lag 0-5 RR0.97 (0.86,1.10)
Lag 0-6 RR0.98 (0.86,1.12)
Girls 6-12 yrs by SES status:
Lag 0RR 1.05 (0.95,1.16); 1
Lag 0-1 RR 1.11 (0.99,1.25)
Lag 0-2 RR 1.11 (0.97,1
Lag 0-3 RR 1.18 (1.02,1
Lag 0-4 RR 1.18 (1.02,1
Lag 0-5 RR 1.19 (1.01,1
Lag 0-6 RR 1.15 (0.97,1
Low; High
03 (0.95,1.12)
; 1.06 (0.96,1.17)
; 1.06 (0.95,1.18)
; 1.04 (0.92,1.17)
; 1.02 (0.90,1.14)
; 1.02 (0.89,1.16)
; 1.05 (0.91,1.21)
Low; High
.07 (0.96,1.19)
1.07 (0.94,1.21)
1.07 (0.93,1.23)
1.02 (0.87,1.19)
0.99 (0.85,1.15)
0.95 (0.80,1.13)
0.98 (0.81,1.17)
Multipollutant model (adjusted for NO2)
Girls, Low SES: 1.17 (1.00,1.37), 4-day avg; 1.19 (1.00,1.42), 6-day avg
Lin et al. (2005)
Toronto, ON
Period of Study:
1998-2001
Hospital Admissions
Outcome(s) (ICD9):
Respiratory infections
(464,466,
480-487)
Age groups analyzed:
0-14
Study design: Case-crossover
N: 6,782
Statistical analyses:
Conditional logistic regression
Covariates:
Statistical package:
SAS 8.2
Lag: 0-6 days
24-h avg:
Mean: 4.73 ppb
SD: 2.58 ppb
Range: 1.00,
19.67
25th
50th
75th
3.00
4.00
6.00
Number of
monitors: 5
Copollutants:
PM2.5 (r = 0.47)
PM10-2.5 (r = 0.29)
PM10 (r = 0.48)
CO (r = 0.12)
N02 (r = 0.61 )
Asthma hospitalization for boys was associated with SO2 before the
adjustment for fine and coarse PM. Asthma hospitalization for girls was not
associated with SO2 for any lag.
Increment: 3 ppb (IQR)
Unadjusted Model:
Boys only: OR 1.06 (0.97,1.16) lag 0-3; OR 1.02 (0.92,1.13) lag 0-5
Girls only: OR 1.05 (0.94,1.16) lag 0-3; OR 1.07 (0.95,1.21) lag 0-5
Boys and Girls: OR 1.06 (0.99,1.13) lag 0-3; OR 1.04 (0.96,1.13) lag 0-5
Adjusted
Boys only: OR 1.11 (1.01,1.21) lag 0-3; OR 1.08 (0.97,1.21) lag 0-5
Girls only: OR 1.07 (0.96,1.19) lag 0-3; OR 1.12 (0.98,1.28) lag 0-5
Boys and Girls: OR 1.10 (1.02,1.18) lag 0-3; OR 1.10 (1.01,1.20) lag 0-5
Multipollutant model with PM2.5 and PM2.5
Boys only: OR 1.02 (0.90,1.15) lag 0-3; OR 0.99 (0.85,1.16) lag 0-6
Girls only: OR 1.09 (0.0.94,1.26) lag 4; OR 1.07 (0.90,1.28) lag 6
Boys and Girls: OR 1.05 (0.95,1.15) lag 4; OR 1.03 (0.91,1.16) lag 6
F-30
-------
Study
Methods Pollutant Data
Findings
Luginaah et al. (2005)
Windsor, ON, Canada
Period of Study:
4/1/95-12/31/00
Hospital Admissions
Outcome(s) (ICD9):
Respiratory admissions
(460-519)
Age groups analyzed: 0-14,
15-64, 65+, all ages
Study design: (1) Time-series
and (2) case-crossover
N: 4,214. # of Hospitals: 4
Statistical analyses:
(1 JPoisson regression, GAM
with natural splines (stricter
criteria), (2) conditional
logistic regression with Cox
proportional hazards model
Covariates: Temperature,
humidity, change in
barometric pressure, day of
wk
Statistical package: S-Plus
Lag: 1,2,3 days
SO2 avg 24-h
Max: 27.5 ppb,
SD: 16.5;
Range: 0,129
Number of
stations: 4
Copollutants:
N02 (r = 0.22)
CO (r = 0.16)
PM10 (r = 0.22)
COH (r= 0.14)
03 (r = -0.02)
TRS (r= 0.13)
The effect of SO2 on respiratory hospitalization varies considerably,
especially at low levels of exposure.
Increment: 19.25 ppb (IQR)
Time-series, females; males:
All ages: 1.041 (0.987,1.098); 0.953 (0.900,1.009) lag 1
0-14 yrs: 1.111 (1.011,1.221); 0.952 (0.874,1.037) lag 1
15-65 yrs: 1.031 (0.930,1.144); 0.971 (0.845,1.15) lag 1
65+ yrs: 1.030 (0.951,1.115); 0.9409 (0.860,1.029) lag 1
Case-crossover, females; males
All ages: 1.047 (0.978,1.122); 0.939 (0.874,1.009) lag 1
0-14 yrs: 1.119 (0.995,1.259); 0.923 (0.831,1.025) lag 1
15-65 yrs: 1.002 (0.879,1.141); 0.944 (0.798,1.116) lag 1
65+ yrs: 1.020 (0.924,1.126); 0.968 (0.867,1.082) lag 1
Stieb et al. (1996)
St. John, New Brunswick,
Canada
Period of Study:
1984-1992
(May-Sep only)
ED Visits. Outcome(s):
Asthma
ICD9 codes: NR
Age groups analyzed: 0-15,
>15
Study design: Time-series
N: 1,987
# of Hospitals: 2
Statistical analyses: SAS
NUN (Equivalent to Poisson
GEE)
Covariates: Day of wk, long-
term trends
Season: Summers only (May-
Sep)
Dose-response investigated?:
Yes
Statistical package: SAS
Lag: 0-3 days
1 -h max SO2 (ppb) SO2 did not affect the rate of asthma ED visits when O3 was included in the
model.
Mean: 38.1
Range: 0, 390
95th 110
Copollutants:
03 (r = 0.04)
N02 (r = -0.03)
S042" (r = 0.23)
TSP (r = 0.16)
Increment: NR
S02 + 03: p = -0.0030 (0.0027) lag 0
F-31
-------
Study
Methods Pollutant Data
Findings
Stieb* et al. (2000)
Saint John, New Brunswick,
Canada
Period of Study:
Retrospective:
7/92-6/94
Prospective:
7/94-3/96
ED Visits
Outcome(s): Asthma; COPD;
Respiratory infection
(bronchitis, bronchiolitis,
croup, pneumonia);
All respiratory
ICD9 codes: NR
Age groups analyzed: All
Study design: Time-series
N: 19,821
Statistical analyses: Poisson
regression, GAM
Covariates: Day ofwk,
selected weather variables in
each model
Season: All yr, summer only
Dose-response investigated:
Yes
Statistical package: S-Plus
Lag: all yr = 0; summer
only = 0-3
24-h avg:
Annual Mean: 6.7
(5.6) ppb
Non-linear effect of SO2 on summertime respiratory visits observed and log
transformation strengthened the association.
Increment: 23.8 ppb (mean)
1-h max:
Respiratory visits:
All yr: 3.9% lag 5
May to Sept: 3.9% lag 0-3
Multipollutant model (SO2,03, NO2)
95th: 18.0
Max: 60.0
Warm season:
Mean: 7.6 (5.2)
ppb
95th: 18.0
Max: 29.0
Annual Mean: 23.8 A" yr: 37% <15' 6 0' la9 5
(21.0) ppb Multipollutant model (In (NO2), O3, SO2 COH)
Max: 1610 May to Sept: 3 9% (1 1' 67) lag 0-3
Warm season:
Mean: 25.4 (17.8)
ppb
95th: 62.0
Max: 137.0
Copollutants:
CO (r= 0.31)
03 (r = 0.10)
N02 (r = 0.41)
TRS (r= 0.08)
PM10 (r = 0.36)
PM2.5 (r = 0.31)
H+ (r= 0.24)
S042" (r = 0.26)
COH (r= 0.31)
H2S (r = -0.01)
Assessed
multipollutant
models
Villeneuve et al., (2006b)
Toronto, ON, Canada
Period of Study:
1995-2000
Days: 2,190
GP Visits
Outcome(s) (ICD9): Allergic
Rhinitis (177)
Age groups analyzed: a65
Study design: Time-series
N: 52,691
Statistical analyses: GLM,
using natural splines (more
stringent criteria than default)
Covariates: Day ofwk,
holiday, temperature, relative
humidity,
aero-allergens
Season: All Yr; Warm, May-
Oct; Cool, Nov-Apr
Statistical package: S-Plus
Lag: 0-6
24-h avg: 4.7 ppb
SD: 2.8
IQR: 3.2 ppb
Range: 0, 24.8
Number of
stations: 9
Copollutants: NO2
03
CO
PM10
PM10-2.5
PM2.5
There were positive associations between allergic rhinitis and SO2 for
exposures occurring on the same day as physician visits, but only during the
winter time.
Increment: 10.3 ppb (IQR)
All results estimated from Stick Graph:
All Yr:
Mean increase: 1.7% (-0.4, 2.8) lag 0
Warm:
Mean increase: 0.3% (-1.9, 2.5) lag 0
Cool:
Mean increase: 1.9% (-0.2, 4.1) lag 0
F-32
-------
Study
Methods Pollutant Data
Findings
Yang et al. (2003b)
Vancouver, Canada
Period of Study:
1986-1998
Days: 4748
Hospital Admissions
Outcome(s) (ICD9): All
respiratory admissions (460-
519)
Study design:
Case-crossover
Age groups analyzed: < 3, a
65
Statistical analyses:
conditional logistic regression
Lag: 0-5 days
24-h avg SO2
(PPb):
Mean: 4.84
SD: 2.84
5th: 1.50
25th
50th
75th
2.75
4.25
6.25
100th: 24.00
IQR: 3.50
Number of
stations: 30
Copollutants:
CO N02
03 (r = -0.37)
COH
SO2 showed the weakest effect among children and the second weakest
effect among older adults when compared to all other pollutants considered
in the study.
Increment: 3.50 ppb (IQR)
All respiratory admissions < 3 yrs:
S02 alone: OR 1.01 (0.98,1.05) lag 2
S02 + 03: OR 1.01 (0.97,1.04) lag 2
S02 + 03 + CO + COH + N02: OR 0.98 (0.94,1.03) lag 2
All respiratory admissions a 65 yrs:
S02 alone: OR 1.02 (1.00,1.04) lag 0
S02 + 03: OR 1.02 (1.00,1.04) lag 0
S02 + 03 + CO + COH + N02: OR 1.01 (0.98,1.03) lag 0
Yang et al. (2005)
Vancouver, BC, Canada
Period of Study:
1994-1998
Days: 1826
Hospital Admissions
Outcome(s) (ICD9): COPD
excluding asthma (490-2,
494, 496)
Age groups analyzed: 65+
Study design: Time-series
N: 6,027
Statistical analyses: Poisson
regression with GAM (with
more stringent criteria)
Covariates: Temperature,
relative humidity, day of wk,
temporal trends, season
Statistical package:
S-Plus
Lag: 0-6 days, moving
averages
24-h avg: 3.79 ppb
SD: 2.12;
IQR: 2.75 ppb;
Range: 0.75,
22.67
Winter: 4.10 (2.87)
Spring: 3.40 (1.58)
Summer: 4.10
(1.79)
Fall: 3.56 (1.92)
Number of
stations: 5
Copollutants:
PM10 (r = 0.62)
N02 (r = 0.61)
CO (r = 0.67)
03 (r = -0.34)
This study produced a marginally significant association between COPD
hospitalization and 6-day SO2 exposure. Most previous studies have not
detected a significant effect of SO2 on respiratory ED visits or
hospitalizations.
Increment: 2.75 ppb (IQR)
COPD
>65 yrs, yr round: RR 1.00 (0.97,1.04) lag 0; RR 1.02 (0.98,1.06) lag 0-1;
RR 1.04 (0.99,1.08) lag 0-2; RR 1.04 (0.99,1.09) lag 0-3;
RR 1.05 (0.99,1.11) lag 0-4; RR 1.06 (1.00,1.13) lag 0-5;
RR 1.06 (0.99,1.13) lag 0-6
2-pollutant model: NO2: RR 0.99 (0.91,1.08) lag 0;
CO: RR 0.97 (0.87,1.07) lag 0-6; 03: RR 1.07 (1.00,1.14) lag 0-6;
PM10: 0.97 (0.88,1.06) lag 0-6
Multipollutant models: SO2, CO, NO2, O3, PM10: RR 0.94 (0.85,1.05);
S02, CO, N02,03: RR 0.96 (0.86,1.06)
Anderson et al. (1997)
Multicity, Europe
(Amsterdam, Barcelona,
London, Milan, Paris,
Rotterdam)
Period of Study:
1977-1989 for Amsterdam
and Rotterdam
1986-1992 for Barcelona
1987-1991 for London
1980-1989 for Milan
1987-1992 for Paris
Hospital Admissions
Outcome(s) (ICD 9): COPD-
unspecified bronchitis (490),
chronic bronchitis (491),
emphysema (492), chronic
airway obstruction (496)
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson regression,
meta-analysis
Covariates: Trend, season,
day ofwk, holiday, influenza,
temperature, humidity
Season:
Cool, Oct-Mar;
Warm, Apr-Sep
Lag: 0,1,2 days and
0-3 cumulative
24-h all yr avg
(|jg/m3):
Amsterdam: 21
Barcelona: 40
London:31
Milan: 53
Paris: 23
Rotterdam: 32
1-h max
Amsterdam: 50
Barcelona: 60
London: NR
Milan: NR
Paris: 47
Rotterdam: 82
Copollutants:
NO2BS TSPO3
The effect of SO2 varied considerably across the cities; however, the summer
estimate was significantly associated with COPD for the 1-h measure and
borderline significant for the daily mean. Both 24-h and 1-h SO2
concentrations were significantly associated with COPD ER admissions in
the warm season. Only cumulative lags of SO2 showed borderline
significance.
Increment: 50 |jg/m3
COPD-Warm season:
24 h avg 1.05 (1.01,1.10) 1-h 1.02 (1.00,1.04)
COPD-C00I season:
24 h avg 1.02 (0.98,1.05) 1-h 1.01 (0.99,1.03)
COPD-AII yr:
24-h avg 1.022 (0.981,1.055) lag 1
24-h avg 1.021 (0.998,1.045) lag 0-3, cumulative
1-h max 1.011 (0.994,1.029) lag 1
1-h max 1.015 (1.003,1.027) lag 0-3, cumulative
F-33
-------
Study
Methods Pollutant Data
Findings
Anderson et al. (1998)
Hospital Admissions
London, England
Outcome(s) (ICD 9):
Period of Study:
Asthma (493)
Apr 1987-Feb 1992
Age groups analyzed:
Days: 1,782
< 15,15-64, 65+
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson regression
Covariates: Time trends,
seasonal cycles, day ofwk,
public holidays, influenza
epidemics, temperature,
humidity
Season:
Cool (Oct- Mar);
Warm (Apr-Sep)
Statistical package: NR
Lag: 0,1,2 days
24-h avg SO2
(pg/m3)
Mean: 32.0
SD: 11.7
Range: 9,100
5th: 16
10th
25th
50th
75th
90th
95th
18
24
31
38
46
52
# of monitors: 2
Copollutants:
O3NO2BS
The strongest association between SO2 and asthma admissions was for
those a65 yrs in the cool season. A weaker association was observed for
children in the warm season and all yr. The adult population showed no
association.
In 2-pollutant models O3 was overall the strongest pollutant associated with
hospital admission with weaker associations with NO2 and BS. The most
consistent yr-round association for All ages was found with BS. When looking
at all ages combined, SO2 association remained significant in all 2-pollutant
models except with NO2, both for all yr and the summer (warm) season.
Increment: 10 ppb in 24-h avg SO2
0-14 yrs
Whole yr: 1.64% (0.29, 3.01) lag 1; 2.04% (0.29, 3.83) lag 0-3;
+ 031.77% (0.22, 3.36) lag 1; + NO21.23% (-0.22, 2.69) lag 1;
+ BS 1.66% (0.23, 3.12) lag 1
Warm season: 3.33% (1.09, 5.63) lag 1; 3.40% (0.41, 6.48) lag 0-3;
+ 033.35% (0.89, 5.87) lag 1; + N022.92% (0.58, 5.32) lag 1;
+ BS 3.66% (1.35, 6.02) lag 1
Cool season: 0.56% (-1.16, 2.32) lag 1; 1.24% (-0.95, 3.49) lag 0-2
15-64 yrs
Whole yr: -0.69% (-2.28, 0.94) lag 2; -0.71% (-2.69,1.30) lag 0-2
Warm season: -1.39% (-3.97,1.27) lag 0; -2.2% (-5.46,11.8) lag 0-2
Cool season: -0.24% (-2.28,1.84) lag 0; 0.20% (-2.28, 2.74) lag 0-2
a 65 yrs
Whole yr: 2.82% (-0.82, 5.96) lag 2; 3.06% (-0.72, 6.98) lag 0-3
Warm season: -2.62% (-7.31, 2.31) lag 2; -4.27% (-9.89,1.71) lag 0-3
Cool season: 5.85% (1.81,10.05) lag 2; 7.28% (2.19,12.62) lag 0-3;
+ 037.84% (2.48,13.48) lag 1; +I\I02 4.19% (-0.53, 9.13) lag 1;
+ BS 5.29% (0.42,10.40) lag 1
All Ages
Whole yr: 1.64% (0.54, 2.75) lag 1; 2.75% (1.22, 4.30) lag 0-3;
+ 031.48% (0.24, 2.73) lag 1; + SO21.14% (-0.04, 2.33) lag 1;
+ BS 1,54%(0.36, 2.73) lag 1
Warm season: 2.02% (0.22, 3.85) lag 1; 2.60% (0.02, 5.25) lag 0-3;
+ 031.91% (0.05, 3.81) lag 1; + N02 1.64% (-0.23, 3.56) lag 1
+ BS 2.18% (0.32, 4.07) lag 1
Cool season: 1.41% (0.0, 2.83) lag 1; 2.83% (0.89, 4.81) lag 0-3;
+ 03 -0.09% (-1.61,1.82) lag 1; + NO2 0.83% (-0.67, 2.34) lag 1;
+ BS 1.11% (-0.41, 2.66) lag 1
Anderson* et al. (2001)
West Midlands conurbation,
United Kingdom
Period of Study:
10/1994-12/1996
Hospital Admissions
Outcome(s) (ICD9): All
respiratory (460-519), Asthma
(493), COPD (490-496,
excluding 493)
Age groups analyzed: 0-14,
15-64, 65+
Study design: Time-series
Statistical analyses: followed
APHEA 2 protocol, GAM
Covariates: Season,
temperature, humidity,
epidemics, day ofwk,
holidays
Statistical package:
S-Plus 4.5 Pro
Lag: 0,1,2,3, 0-1, 0-2, 0-3
24-h avg: 7.2 ppb,
4.7 (SD)
Min: 1.9 ppb
: 59.8 ppb
: 3.3 ppb
: 12.3 ppb
# of monitors: 5
Max:
10th:
90th:
Copollutants:
PM10 (r = 0.55)
PM10-2.5 (r = 0.31)
PM2.5 (r = 0.52)
BS (r = 0.50)
S042" (r = 0.19)
N02 (r = 0.52)
03 (r = -0.22)
CO (r = -0.29)
When admissions were analyzed by subgroups, respiratory and asthma
admissions were positively correlated with SO2. SO2 significantly associated
with asthma and respiratory admissions for the 0 to 14-yr-age group;
however, little evidence of a seasonal interaction was observed.
Increment: 9 ppb (90th-10th)
All respiratory:
All ages: 1.3% (-0.7, 3.4) lag 0-1
0-14 yrs: 4.6% (1.40, 7.8) lag 0-1
15-64yrs: -0.9% (-4.8, 3.3) lag 0-1
a 65 yrs: -2.0% (-4.9,1.1) lag 0-1
COPD with asthma:
0-14 yrs: 10.9% (4.50,17.8) lag 0-1
15-64yrs: 2.4% (-5.5,10.9) lag 0-1
a 65 yrs: -4.2% (-8.9, 0.8) lag 0-1
F-34
-------
Study
Methods Pollutant Data
Findings
Atkinson et al. (1999b)
London, England
Period of Study:
1992-1994
Days: 1,096
Hospital Admissions
Outcome(s) (ICD9): All
respiratory (460-519); Asthma
(493); Asthma and COPD
(490-496); LRD (466,480-
486)
Age groups analyzed: all
ages, 0-14 yr, 15-64 yr and
a65 yr
Study design: Time-series
N: 165,032 (respiratory
admissions), 189,032
(cardiovascular admissions)
Statistical analyses: Poisson
regression following APHEA
protocol
Covariates: Long-term
seasonal patterns, day of wk,
temperature, humidity,
influenza.
Statistical package: SAS
Investigated Dose/Response:
Yes
Lag: 0,1,2,3 days
SO2- 24-h (|jg/m3)
Mean: 21.2 (7.8)
(jg/m3
Min: 7.4
10th
50th
90th
Max: 82.2
# of monitors: 5
Copollutants: O3,
CO, PM10, BS,
N02
Correlation
coefficients ranged
between
r = 0.5 and 0.6
Asthma was closely linked with PM, CO, NO2, and traffic pollution. When SO2
and PMiowere included in the same model, the magnitude of the individual
associations was reduced, as were their statistical significance. This
reduction occurred in children, adults and the elderly. The other pollutants all
had the effect of reducing the magnitude of the individual SO2 and PM2.5
associations, although their statistical significance was unaffected. This
indicates that both SO2 and PM2.5 were indicators of the same pollutant
mixture.
Increment: 18 |jg/m3
All respiratory: All ages 2.01% (0.29, 3.76) lag 1;
0-14 yrs 5.14% (2.59, 7.76) lag 1; 15-64yrs 1.90% (-0.79, 4.66) lag 3;
a 65 yrs 2.25 (-0.09, 4.65) lag 3
Asthma: All ages: 3.38 (0.42, 6.43) lag 1; 0-14 yrs: 6.74% (2.92,10.69) lag 1;
15-64yrs: 4.58% (-0.18, 9.57) lag 3; a 65 yrs: 6.31% (-1.59,14.83) lag 2
COPD and Asthma: a 65 yrs: 1.53% (-1.83, 5.00) lag 3
Lower Respiratory: a 65 yrs: 5.16% (1.19, 9.28) lag 3
Atkinson et al. (1999a)
London, United Kingdom
Period of Study:
1/92-12/94
ED Visits
Outcome(s) (ICD 9):
Respiratory ailments
(490-496), including asthma,
wheezing, inhaler request,
chest infection, COPD,
difficulty in breathing, cough,
croup, pleurisy, noisy
breathing
Age groups analyzed:
0-14; 15-64; a 65;
All ages
Study design: Time-series
N: 98,685
# of Hospitals: 12
Statistical analyses: Poisson
regression, APHEA protocol
Covariates: Long-term trend,
season, day of wk, influenza,
temperature, humidity
Statistical package: SAS
Lag: 0,1,0-2 and
0-3 days
24-h avg: 21.2
(jg/m3, SD: 7.8
10th: 13.0
50th: 19.8
90th: 31.0
Range: 7.4, 82.2
# of Stations: 5
Copollutants:
SO2O3 (8 h)
CO (24 h avg),
PM10 (24 h avg)
BS
SO2 was closely related to PM10, but 2-pollutant models showed that the
effect of SO2 was decreased by NO2 and PM10 inclusion. Inclusion of other
pollutants did not significantly decrease the influence of SO2 on ER
admissions in
2-pollutant models.
Increment: 18 |jg/m3 in 24-h
Single-pollutant model
Asthma only:
0-14 yrs 9.92% (4.75,15.34) lag 1
15-64yrs 4.19% (-0.53, 9.13) lag 1
All ages 4.95% (1.53, 8.48) lag 1
All respiratory:
0-14yrs 6.01% (2.98,9.12) lag 2
15-64yrs 2.72% (-0.18, 5.70) lag 3
65+ yrs -1.82% (-5.72, 2.25) lag 3
All Ages 2.81% (0.72, 4.93) lag 1
Copollutant models for asthma among children:
SO2 + NO2: 5.42% (0.18,10.93)
S02 + 03: 8.39% (3.82,13.17)
S02 + CO: 8.05% (3.45,12.86)
S02 + PM10: 5.63 (0.53,10.98)
S02 + BS: 8.03 (3.32,12.96)
F-35
-------
Study
Methods
Pollutant Data
Findings
Atkinson et al. (2001)
Hospital Admissions
1-h max of SO2
The inclusion of SO2 in the models only modified PM10 associations in the 0-
Multicity, Europe (Barcelona,
Outcome(s) (ICD 9): Asthma
(pg/m3)
to 14-yr age group.
Birmingham, London, Milan,
(493), COPD (490-496), All
Barcelona: NR
Increment: 10 |jg/m3 for PM10; change in SO2 not described.
Netherlands, Paris, Rome,
respiratory (460-519)
Birmingham: 24.3
Asthma, 0 to 14 yrs:
For PM10:1.2(0.2,2.3)
Stockholm)
Study design: Time-series
London: 23.6
Milan: 29.1
Netherlands: 8.5
Period of Study:
1997-1998
Statistical analyses: APHEA
For PM10 + S02: 0.8 (-3.7,5.6)
protocol, Poisson regression,
Paris: 17.7
Asthma, 15 to 64 yrs:
meta-analysis
Rome: 9.8
For PM10:1.1 (0.3,1.8)
Covariates: Season,
Stockholm: 3.8
For PM10 + S02:1.6 (0.6,2.6)
temperature, humidity,
Copollutants:
COPD + Asthma, a 65 yrs:
holiday, influenza
N02, 03, CO, BS,
For PM10:1.0(0.4,1.5)
Statistical package: NR
PM10
For PM10 + S02:1.3 (0.7,1.8)
Lag: NR
Correlation
All respiratory, a 65 yrs of age:
coefficients with
PM10:
Barcelona: 0.32
Birmingham: 0.77
London: 0.72
Milan: 0.64
Netherlands: 0.67
Paris: 0.63
Rome: 0.15
Stockholm: 0.36
For PM10: 0.9 (0.6,1.3)
For PM10 + S02:1.1 (0.7,1.4)
Boutin-Forzano et al. (2004)
Marseille, France
Period of Study:
4/97-3/98
ED Visits Mean: SO2:
Outcome(s): Asthma 22.5 |jg/m
ICD 9 Code(s): NR Range: 0.0, 94.0
Age groups analyzed: 3-49 ^('S)
Study design: Case-crossover O3 (r = -0.25)
N: 549
Statistical analyses: Logistic
regression
Covariates: Minimal daily
temperature, max daily
temperature, min daily relative
humidity, max daily relative
humidity, day ofwk
Statistical package: NR
Lag: 0-4 days
No association was observed between ER visits for asthma and SO2 levels.
Only single-pollutant models were utilized.
Increment: 10 |jg/m3
Increased ER visits:
OR 1.0023 (0.9946,1.0101) lag 0
OR 0.9995 (0.9923,1.0067) lag 1
OR 0.9996 (0.9923,1.0069) lag 2
OR 0.9970 (0.9896,1.0045) lag 3
OR 0.9964 (0.9889,1.0040) lag 4
Buchdahl et al. (1996)
London, United Kingdom
Period of Study:
3/1/92-2/28/93
ED visits.
Outcomes: Daily acute
wheezy episodes
Age groups analyzed: s 16
Study design: Case-control
N: 1,025 cases, 4,285
controls
# of Hospitals: 1
Statistical analyses: Poisson
regression
Covariates: Season,
temperature, wind speed
Season: Spring (Apr-Jun),
Summer (Jul-Sep), Autumn
(Oct-Dec), Winter (Jan-Mar)
Statistical package: Stata.
Lag: 0-7 days
SO21-h yr round
Mean: 22 ug/m3,
SD: 14
Spring: 20 (14)
Summer: 18 (22)
Fall: 24 (14)
Winter: 25 (14)
Copollutants:
N02 (r = 0.62)
03 (r = -0.28)
Variations in SO2 could not explain the U-shaped relationship between O3
and incidence of asthma.
Increment: 14 |jg/m3 (Std. Dev.)
No adjustments to model:
RR 1.16(1.10,1.23) lag not specified
Adjusted for temperature and season:
RR 1.12 (1.06,1.19) lag not specified
Adjusted for temperature, season and wind speed:
RR 1.08 (1.00,1.16) lag not specified
F-36
-------
Study
Methods Pollutant Data
Findings
Castellsague et al. (1995)
Barcelona, Spain
Period of Study:
1986-1989
ED visits
Outcome(s): Asthma
Age groups analyzed: 15-64
Study design: Time-series
# of Hospitals: 4
Statistical analyses: Poisson
regression
Covariates: Long time trend,
day ofwk, temperature,
relative humidity, dew point
temperature
Seasons : Winter: Jan-Mar;
Summer: Jul-Sep
Dose-response investigated:
Yes
Lag: 0,1-5 days and
cumulative; Summer: lag 2
days, Winter: lag 1 day
Mean SO2 (pg/m
Summer: 40.8
25th
25
50th
36
75th
54
95th
82
Winter: 52.0
25th
36
50th
49
75th
67
95th
94
# of Stations: 15
manual, 3
automatic
Copollutants:
NO2 O3
Interaction between pollutants and asthma emergency room visits was
influenced by soy-bean dust in the air. The daily mean of asthma visits and
level of SO2 were higher in the winter than in the summer. A positive but not
statistically significant increase in relative risk was found for SO2 in the
summer. SO2 levels were higher in the winter, but the RR was lower
compared to the RR in the summer. SO2 was not significantly associated with
asthma related ER visits.
Increment: 25 |jg/m3
Seasonal differences:
Summer: RR 1.052 (0.980,1.129) lag 2
Winter: RR 1.020 (0.960,1.084) lag 1
Dab et al. (1996)
Paris, France
Period of Study:
1/1/87-9/30/92
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory
(460-519), Asthma (493),
COPD
(490-496)
Age groups analyzed: All
ages
Study design: Time-series
Number of hospitals: 27
Statistical analyses: Poisson
regression, followed APHEA
protocol
Covariates: Temperature,
relative humidity, influenza,
long-term trend, season,
holiday, medical worker strike
Lag: 0,1,2 days, 0-3
cumulative
All Yr:
24-h avg: 29.7
(jg/m3
Median:23.0
5th: 7.0
99th: 125.0
1-h max: 59.9
Median: 46.7
5th: 14.0
99th: 232.7
Warm season
24-h avg: 20.1
Median:18.3
5th: 6.0
99th: 49.3
1-h max:42.7
Median:37.0
5th: 13.0
99th: 133.7
Cold season
24-h avg: 40.1
(jg/m3
Median:31.3
5th: 8.7
99th: 149.0
1-h max:78.3
Median:60.7
5th: 17.0
99th: 268.3
Copollutants:
NO2O3PM13BS
1-h max SO2 levels yielded lower relative risk when compared to 24-h avg
levels. COPD effects were only significantly associated with SO2 with no lag.
The strongest association was observed with PM13; 4.5% increase in
respiratory admission per 100 |jg/m3 increment. SO2 was a close second.
Neither analysis by age or by season showed a significant sensitivity for
hospital admissions. The strongest association for asthma admission for all
pollutants was with SO2 24-h avg of 7% (0.14,14.10), but 1-hr max level was
not significant. The strongest association for admission with COPD diagnosis
was also for 24-h avg of SO2 (9.9% [2.3,18]). Increment: 100 |jg/m3
All respiratory (1987-1990):
24-h avg RR 1.042 (1.005,1.080) lag 0-2
1-h max RR 1.018 (0.988,1.048) lag 0-2
Asthma (1987-1992):
24-h avg RR 1.070 (1.004,1.141) lag 2
1-h max RR 1.047 (0.998,1.098) lag 2
COPD:
24-h avg RR 1.099 (1.023,1.180) lag 0
1 -h max RR 1.051 (1.025,1.077) lag 0
F-37
-------
Study
Methods Pollutant Data
Findings
de Diego Damia et al. (1999) ED visits
Valencia, Spain Outcome(s) (ICD 9): Asthma
Period of Study:
3/1994-3/1995
(493)
Age groups analyzed: > 12
N: 515
# of Hospitals: 1
Statistical analyses: Stepwise
regression and ANOVA;
Linear regression
Covariates: Season and
temperature
Statistical package: SPS
24-h avg SO2
(|jg/m3)
Winter
Mean:
56.
Range: 30,
86
Spring
Mean:
47.
Summer
Mean:
40.
Autumn
Mean:
50.
The SO2 concentration was averaged for each season and quartiles of
concentration determined. Asthma visits that occurred in each season were
examined. There were no significant associations with asthma ER visits with
any season or with any quartile of SO2 exposure.
Mean number of asthma-related ED visits based on quartile of SO2
All yr: < 41 |jg/m3:8.6
41-50 (jg/m3: 9.1
51-56 (jg/m3:11.6
>56 (jg/m3:11.9
# of monitors: 1
Copollutants:
BS (r = 0.54)
Fusco* et al. (2001)
Rome, Italy
Period of Study:
1/1995-10/1997
Hospital Admissions
Outcomes
(ICD 9 codes): All Respiratory
(460-519, excluding
470-478); Acute respiratory
infections including
pneumonia (460-466, 480-
486), COPD
(490-492, 494-496), asthma
(493)
Age groups analyzed: All
ages, 0-14
Study design: Time-series
Statistical analyses: Poisson
regression with GAM
Covariates: Influenza
epidemics, day of study,
temperature, humidity, day of
wk, holidays
Statistical package: S-Plus 4
Lag: 0,1,2,3,4
24-h avg: 9.1
(5.8) (jg/m3
25th
50th
75th
5.1
7.9
12.0
# of monitors: 5
Copollutants:
03 (r = -0.35)
CO (r = 0.56)
N02 (r = 0.33)
Particles;
r = 0.25
SO2 did not have an effect on respiratory hospitalizations.
Increment: 6.9 |jg/m3 (IQR)
Respiratory conditions:
All ages: 0.4% (-1.3, 2.2) lag 0. 0.8% (-0.9, 2.4) lag 1. 0.3% (-1.3,1.8) lag 2
0-14 yrs: -0.7% (-4.0, 2.7) lag 0; -2.0 (-5.2,1.3) lag 1; -0.8 (-3.8, 2.3) lag 2
Acute respiratory infections: All ages: 0.4% (-2.1, 3.0) lag 0;
1.4% (-1.0, 3.9) lag 1; 1.2% (-1.0, 3.5) lag 2;
0-14 yrs:—0.1% (-3.9, 3.8) lag 0; -2.7% (-6.3,1.0) lag 1;
-1.2% (-4.5, 2.2) lag 2
Asthma: All ages: -1.5% (-6.6, 3.9) lag 0; -1.5% (-6.5, 3.7) lag 1;
2.5% (-2.2, 7.4) lag 2; 0-14yrs: -2.6 (-10.4, 6.0) lag 0;
4.3% (-3.5, 12.7) lag 1; 5.5% (-1.8,13.2) lag 2
COPD: All ages: 1.0% (-1.9, 4.0) lag 0;-1.1% (-3.9,1.8) lag 1;
-0.5% (-3.1, 2.1) lag 2
Galan et al. (2003)
Madrid, Spain
Period of Study:
1995-1998
ED Visits
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: All
Study design:Time-series
N: 4,827
Statistical analyses: Poisson
regression, (1) classic APHEA
protocol and (2) GAM with
stringent criteria
Covariates: Trend, yr, season,
day ofwk, holidays,
temperature, humidity,
influenza, acute respiratory
infections, pollen
Statistical package: NR
Lag: 0-4 days
24-h Mean:
23.6 (jg/m3
SD: 15.4
10th
25th
50th
75th
90th
9.2
12.3
18.7
31.3
43.9
Range: 5,121.2
# of Stations: 15
Copollutants:
PM10 (r = 0.581)
N02 (r = 0.610)
O3 (r = -0.547)
SO2 registered a predominately winter based pattern, and was positively
correlated with PM2.5, NO2. The lag that described the strongest association
was 3 days.
Multipollutant models were fitted for cold season pollutants. SO2 was the
most affected when PM2.5 was included in the model.
Parametric estimates using APHEA protocol produced similar results as
GAM.
The SO2 association may be due to the concealing effects of other pollutants.
PM2.5 accounted for most of the observed effects.
Increment: 10 |jg/m3
Asthma: RR lag 0 1.018 (0.984,1.054); RR lag 1 1.005 (0.972,1.039);
RR lag 2 1.002 (0.970,1.036); RR lag 3 1.029 (0.997,1.062);
RR lag 4 1.025 (0.994,1.058)
Multipollutant model: SO2/PM10 0.966 (0.925,1.009)
F-38
-------
Study
Methods Pollutant Data
Findings
Garty et al. (1998)
Tel Aviv, Israel
Period of Study:
1/1/1993-12/31/1993
ED Visits
Outcome(s): Asthma
ICD 9 Code(s): NR
Age groups analyzed: 1-18
Study design: Descriptive
study with correlations
N: 1,076
Statistical analyses: Pearson
correlation and partial
correlation coefficients
Covariates: Max and min
ambient temperatures,
relative humidity and
barometric pressure
Statistical package: Statistix
24-h mean of SO2
(estimated from
histogram): 27
(jg/m3
Range: 11,64
Copollutants:
N0xS02 O3
Asthma morbidity was higher in the autumn and winter than the rest of the yr.
The number of ER visits is Sep was exceptionally high.
The percent of total variance showed positive correlation between asthma
ER visits in children and high levels of NOx, SO2, and increased barometric
pressure. NOx enhances the effects of SO2, whereas O3 had a reverse
relation to SO2.
Air borne pollen was not a significant contributor to ER visits.
Correlation between SO2 and ER visits for asthma:
All yr: Daily data r = 0.24; Running mean for 7 days r = 0.53
Excluding Sep: Daily data r = 0.31; Running mean for 7 days r = 0.64
Hagen et al. (2000)
Drammen, Norway
Period of Study:
1994-1997
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory admissions (460-
519)
Age groups analyzed: All
ages
Study design: Time-series
Number of hospitals: 1
Statistical analyses: Poisson
regression with GAM
(adhered to HEI phase 1 .B
report)
Covariates: Time trends, day
ofwk, holiday, influenza,
temperature, humidity
Lag: 0,1,2,3 days
SO2 24-h avg
(|jg/m3): 3.64,
SD: 2.41
25th
50th
75th
2.16
2.92
4.38
# of Stations: 2
Copollutants:
PM10 (r = 0.42)
N02 (r = 0.58)
benzene (r= 0.29)
NO (r = 0.47)
03 (r = -0.24)
Formaldehyde
(r = 0.54)
Toluene (r= 0.48)
SO2 was significantly associated with respiratory hospital admissions. This
relationship was robust to the inclusion of PM2.5, but attenuated when both
PM2.5 and benzene were included in the model.
Increment: SO2: 2.22 |jg/m3 (IQR)
Single-pollutant model
Respiratory disease only:
1.056 (1.013,1.101)
All disease: 0.990 (0.974,1.007)
2-pollutant model with PM10
1.051 (1.005,1.099)
3-pollutant model with PM10 + Benzene
1.040 (0.993,1.089)
Hajat et al. (1999)
London, United Kingdom
Period of Study:
1992-1994
GP visits
Outcome(s) (ICD9): Asthma
(493); Lower respiratory
disease (464, 466, 476, 480-
3, 490-2, 485-7, 4994-6, 500,
503-5,510-5)
Age groups analyzed: 0-14;
15-64; 65+; all ages
Study design: Time-series
analysis
Statistical Analysis: Poisson
regression, APHEA protocol
Covariates: Long-term trends,
seasonality, day ofwk,
temperature, humidity
Season:
Warm, Apr-Sep; Cool, Oct-
Mar; All-yr
Dose-response investigated?
Yes
Statistical package: SAS
Lag: 0-3 days, cumulative
All yr
24-h avg: 21.2
(jg/m3, SD: 7.8
10th: 13.0
90th: 31.0
Warm:
24-h avg: 20.5
(jg/m3, SD: 6.5
10th: 13.4
90th: 28.4
Cool:
24-h avg: 22.0
(jg/m3, SD: 9.0
10th: 12.8
90th: 33.3
Copollutants:
N02 (r = 0.61)
BS (r = 0.57)
CO (r = 0.51)
PM10 (r = 0.63)
03 (r = -0.11)
This study showed weak, but consistent associations between SO2 and
consultations for asthma and other LRD, especially in children. Bubble plot
suggests a concentration-response relationship.
Increment: 18 |jg/m3
(90th-10th percentile)
Asthma: All ages: 3.6% (0.3, 6.9) lag 2; 4.4% (0.9, 7.9) lag 0-2;
0-14 yrs 4.9% (0.1, 9.8) lag 1; 4.4% (-0.7,9.7) lag 0-2
Warm: 9.0% (2.2,16.2) lag 1; Cool: 2.0% (4.5, 8.9) lag 1
15-64yrs 3.6% (-0.6, 8.0) lag 2; 3.5% (-1.0, 8.2) lag 0-3
Warm: 2.5% (-3.3, 8.7) lag 2; Cool: 4.5% (-1.4,10.7) lag 2
65+yrs 4.5% (-3.5, 13.1) lag 1; 4.8% (-2.9,13.2) lag 0-1
Warm: 7.5% (-4.0, 20.3) lag 1; Cool: 2.0% (-8.6, 13.9) lag 1
Lower respiratory disease:
All ages 1.8% (0.2, 3.4) lag 2; 2.2% (0.4, 4.1) lag 0-2;
0-14 yrs 4.5% (1.4, 7.8) lag 2; 5.7% (1.7, 9.7) lag 0-3;
Warm: 2.4% (-2.6, 7.7) lag 2; Cool: 5.8% (1.6,10.2) lag 2
15-64 yrs 1.5% (-0.7, 3.7) lag 1; 1.6% (-0.9, 4.1) lag 0-3
Warm: -0.5% (-3.8, 2.9) lag 1; Cool: 2.5% (-0.5, 5.5) lag 1
65 + -2.2% (-4.9, 0.6) lag 0; -1.4% (-4.4,1.7) lag 0-1
Warm: -3.1% (-6.9, 0.9) lag 0; Cool: -1.6 % (-5.3, 2.3) lag 0
2-pollutant model - Asthma: SO2 alone 4.9% (0.1, 9.8);
S02/035.9% (1.1, 10.9); S02/N022.7% (-2.7, 8.4);
S02/PM2.5 3.4% (-3.0,10.2)
2-pollutant model-Lower respiratory disease:
S02 alone 4.5% (1.4, 7.8); SO2/O3 4.8% (1.6, 8.1);
SO2/NO23.1% (-0.6, 6.9); S02/PM2.53.8% (0.4, 7.2)
F-39
-------
Study
Methods Pollutant Data
Findings
Hajat* et al. (2001)
London, United Kingdom
Period of Study:
1992-1994
GP visits
Outcome(s) (ICD9): Allergic
Rhinitis (477)
Age groups analyzed:
0-14; 15-64; 65+; all ages
Study design:Time-series
analysis
N: 4,214
Statistical Analysis: Poisson
regression, GAM
Covariates: Long-term trends,
seasonality, day ofwk,
temperature, humidity,
variation in practice
population, counts for lagged
allergic pollen measures, daily
number of consultations for
influenza
Dose-response investigated?
Yes
Statistical package:
S-Plus
Lag: 0-6 days, cumulative
24-h avg: 21.2 The number of allergic rhinitis admissions peaked in Apr and June. After 2-
(jg/m3, pollutant model analysis, SO2 still remained highly significant in the presence
SD: 7.8 of other pollutants. For both children and adults exposure-response
10th" 13 0 associations showed that risk levels off at higher SO2 levels.
90th: 31.0 Increment: 18 |jg/m3
Copollutants: (90th-10th percentile)
NO2 (r = 0.61) Single-pollutant model
BS (r = 0.57) < 1 to 14 yrs:
CO (r = 0.51) 24.5% (14.6, 35.2) lag 4
PM10 (r = 0.63) 24.9% (11.9, 39.4) lag 0-4
O3 (r = —0.11) 15 to 64 yrs:
14.3% (6.2,23.0) lag 3
15.5% (9.1,22.3) lag 0-5
>64 yrs-too small for analysis
2-pollutant models
< 1 to 14 yrs:
S02 & 03: 22.1% (12.0,33.1)
SO2&NO2: 28.5% (15.5, 42.9)
SO2&PM10: 27.2% (15.3, 40.2
15 to 64 yrs:
S02 & 03: 8.5% (3.4,13.9)
S02 & N02: 8.3% (1.7,15.3)
S02 & PM10: 6.7% (0.7,13.0)
Hajat* et al. (2002)
London, United Kingdom
Period of Study:
1992-1994
GP visits
Outcome(s) (ICD9): Upper
respiratory disease, excluding
Rhinitis
(460-3, 465, 470-5, 478)
Age groups analyzed:
0-14; 15-64; 65+; all ages
Study design:
Time-series analysis
Statistical Analysis: Poisson
regression, GAM
Covariates: Long-term trends,
seasonality, day ofwk,
holidays, temperature,
humidity, variation in practice
population, counts for lagged
allergic pollen measures, daily
number of consultations for
influenza
Season:
Warm, Apr-Sep;
Cool, Oct-Mar
Dose-response investigated?
Yes
Statistical package:
S-Plus
Lag: 0,1,2,3 days
All yr:
24-h avg: 21.2
ug/m3,
SD: 7.8
10th: 13.0
90th: 31.0
Warm:
24-h avg: 20.5
ug/m3,
SD: 6.5
10th: 13.4
90th: 28.4
Cool:
24-h avg: 22.0
ug/m3,
SD: 9.0
10th: 12.8
90th: 33.3
# of Stations: 3
Copollutants:
N02 (r = 0.61)
BS (r = 0.57)
CO (r = 0.51)
PM10 (r = 0.63)
03 (r = -0.11)
Increased consultations for URD were most strongly associated with SO2 in
children. For adults and the elderly the strongest associations were for PM10
and NO2. The most consistent lag in adults and the elderly for development
of URD was 2 days (one day after a pollution event).
Increment: 18 |jg/m3
(90th-10th percentile)
Single-pollutant model:
All yr: 0-14 yr: 3.5% (1.4, 5.8) lag 0; 15-64 yrs: 3.5% (0.5, 6.5) lag 1;
>65 yrs: 4.6% (0.4, 9.0) lag 2
Warm: 0-14yrs: 3.2% (-0.5, 7.0) lag 0; 15-64yrs: 4.6% (1.5, 7.7) lag 1;
a 65 yrs: 1.6% (-4.8, 8.5) lag 2
Cool: 0-14yrs: 5.5% (2.4, 8.7) lag 0; 15-64 yrs: 2.7 (0.0, 5.4) lag 1;
>65 yrs: 5.7% (0.4,11.4) lag 2
2-pollutant models
0-14 yrs: SO2&O3:1.0% (-2.2, 4.2); SO2& N02: 4.7% (2.2, 7.4);
SO2&PM10: 4.6% (2.1,7.2)
For 15-64 yrs: SO2 & 03: 3.7% (0.6, 7.0); SO2 & N02: 2.6% (-0.0, 5.2);
S02 & PM10: 2.4% (-0.1,5.0)
For >65yrs: SO2& 03: 9.0% (1.7, 16.9); SO2&NO2: 4.3% (-1.2,10.2);
S02 & PM10: 3.2% (-1.9,8.7)
F-40
-------
Study
Methods Pollutant Data
Findings
Llorca et al. (2005)
Torrelavega, Spain
Period of Study:
1992-1995
Days: 1,461
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory admissions (460-
519)
Age groups analyzed: All
ages
Study design:Time-series
Number of hospitals: 1
Statistical analyses: Poisson
regression
Covariates: Short and Long-
term trends
Statistical package: Stata
Lag: NR
24-h avg SO2:
13.3 ug/m3,
SD: 16.7
# of Stations: 3
Copollutants:
N02 (r = 0.588)
NO (r = 0.544)
TSP (r = -0.40)
SH2 (r = 0.957)
Associations between SO2 and admissions observed in the single-pollutant
model disappear in a 5-pollutant model. Only NO2 was significantly
associated with admissions.
No relation was described for sulfur compounds including H2S or SO2. The
concentration of SO2 changes with temperature changes, which may be
responsible for cardiac stress.
SO2 was not significantly associated with cardiac respiratory or cardio-
respiratory admissions
Increment: 100 |jg/m3
Single-pollutant model
All cardio-respiratory admissions: RR0.98 (0.89,1.07)
Respiratory admissions: 1.04 (0.90,1.19)
5-pollutant model
All cardio-respiratory admissions: RR0.98 (0.80,1.21)
Respiratory admissions: 0.89 (0.64,1.24)
Oftedal et al. (2003)
Drammen, Norway
Period of Study:
11/1994-12/2000
Hospital Admissions
Outcomes (ICD 10): All
respiratory admissions (J00-
J99)
Age groups analyzed: All
ages
Study design: Time-series
Statistical analyses: Semi-
parametric Poisson
regression, GAM with more
stringent criteria
Covariates: Temperature,
humidity, influenza
Lag: 2,3 days
Mean: 2.9 ug/m3,
SD: 2.1
IQR: 2.03 |jg/m3
Copollutants: PM10
N02
03
Benzene
Formaldehyde
Toluene
The study found positive associations between daily number of hospital
admissions for acute respiratory diseases and concentrations of SO2;
associations did not change substantially from the first to the second
3-yr period.
Increment: 2.03 |jg/m3 (IQR)
All respiratory disease
1.042 (1.011,1.073)
Ponce de Leon et al. (1996)
London, England
Period of Study:
04/1987-1988;
1991-02/1992
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory (460-519)
Age groups analyzed: 0-14,
15-64, 65+, all ages
Study design: Time-series
N: 19,901
Statistical analyses: APHEA
protocol, Poisson regression
GAM
Covariates: Long-term trend,
season, influenza, day of wk,
holiday, temperature, humidity
Season:
Cool, Oct-Mar; Warm: Apr-
Sep
Dose-response Investigated?:
Yes
Statistical package: SAS
Lag: 0,1, 2 days, 0-3
cumulative avg.
SO2 24-h avg:
32.2 (jg/m3,
SD: 12.6
5th: 15
10th: 18
25th: 24
50th: 31
75th: 39
90th: 47
95th: 54
# of stations: 2
Copollutants:
N02 (r = 0.44)
BS (r = 0.44)
03(r= -0.067)
Though significant effects were observed with SO2 in some age groups, they
were not consistent or similar in magnitude to those of O3.
Increment: 90th-10th percentile
(24-h avg: 29 (jg/m3).
All yr:
All ages 1.0092 (0.9926,1.0261) lag 1
0-14 yrs 1.0093 (0.9837,1.0356) Iag1
15-64 yr 1.0223 (0.9942,1.0511) lag 1
> 65 yr 1.0221 (0.9970,1.0478) lag 2
Warm season:
All ages 1.0111 (0.9864,1.0364) lag 1
0-14 yrs 1.0468 (1.0066,1.0885) lag 1
15-64 yr 0.9996 (0.9596,1.0411) lag 1
>65 yr 1.0124 (0.9772,1.0489) lag 2
Cool season:
All ages 1.0079 (0.9857,1.0306) lag 1
0-14yrs 0.9848 (0.9515,1.0192) lag 1
15-64yr 1.0389 (1.0010,1.0783) lag 1
>65 yr 1.0280 (0.9945,1.0625) lag 2
F-41
-------
Study
Methods Pollutant Data
Findings
Ponka (1991)
Helsinki, Finland
Period of Study:
1987-1989
Hospital Admissions
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: 0-14;
15-64; a 65 yrs
Study design: Time-series
N: 4,209
Statistical analyses:
Correlations and partial
correlations
Covariates: Min temperature
Statistical package:
Lag: 0-1
24-h avg: 19.2
(12.6) (jg/m3
Range: 0.2, 94.6
Number of
monitors: 4
Copollutants:
N02 (r = 0.4516)
NO (r = 0.4773)
03(r= -0.1778)
TSP (r= 0.1919)
CO
The frequency of all admissions for asthma was significantly correlated to
S02.
Child asthma admissions were not significantly correlated with SO2, but were
correlated to O3 and NO. S02was also significantly correlated with elderly
admissions. Increased hospitalization correlated with S02was also observed
for adults.
Hospital admissions were more strongly correlated with SO2 than other
pollutants. ER visits were more strongly correlated with a mixture of
pollutants (TSP, SO2, O3, and temperature).
Multipollutant model
co-linear results of SO2, CO, NO2, and NO suggest a mixture of pollutants is
responsible for asthma admissions.
Correlations between hospital admissions (HA) for asthma and pollutants and
temperature by ages.
0-14 yrs: HA: -0.01391; Emergency HA: 0.0332
15-64yrs: HA: 0.1039 p = 0.0006; Emergency HA: 0.1199 p < 0.0001
a 65 yrs: HA: 0.0796 p = 0.0085; Emergency HA: 0.1169 p < 0.0001
Partial correlations between admissions for asthma and SO2 were
standardized for temperature.
HA: 0.0770 p = 0.0172; Emergency HA: 0.1050; p = 0.0011
Ponka and Virtanen (1994)
Helsinki, Finland
Period of Study:
1987-1989
Days: 1096
Hospital Admissions
Outcome(s) (ICD 9): Chronic
bronchitis and emphysema
(493)
Age groups analyzed: < 65, a
65
Study design: Time-series
Statistical analyses: Poisson
regression
Covariates: Season, day of
wk, yr, influenza, humidity,
temperature
Season: Summer (Jun-Aug),
Autumn (Sep-Nov),
Winter (Dec-Feb),
Spring (Mar-May)
Lag: 0-7 days
24-h avg:
19 (jg/m3
SD: 12.6;
Range: 0.2, 95
# of stations: 2
Copollutants:
NO2O3TSP
SO2 was significantly associated with increased admissions for chronic
bronchitis and emphysema for patients < 65 yrs of age with a lag of 0 and 3
days.
In the steps leading to regression analysis no association was observed
between SO2 levels and the a65 population. Multipollutant models were only
used to examine NO2 and SO2.
SO2 had no significant association with morbidity caused by chronic
bronchitis and emphysema in the a 65 yr old population.
Increment: NR
Chronic bronchitis and emphysema:
< 65 yrs: RR 1.31 (1.01,1.70) lag 0; RR 0.96 (0.73,1.27) lag 1;
RR 0.78 (0.59,1.03) lag 2; RR 1.39 (1.05,1.86) lag 3;
RR 0.89 (0.68,1.16) lag 4; RR 1.28 (0.97,1.70) lag 5;
RR 0.91 (0.69,1.20) lag 6; RR 1.09 (0.84,1.40) lag 7
65+ yrs: NR
Ponka and Virtanen (1996)
Helsinki, Finland
Period of Study:
1987-1989
Hospital Admissions
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: 0-14,
15-64, 65+
Study design: Time-series
Statistical analyses:
Covariates: Long-term trend,
season, epidemics, day of wk,
holidays, temperature, relative
humidity
Statistical package:
Lag: 0-2
24-h avg (|jg/m3):
Winter: 26
Spring: 22
Summer: 13
Fall: 15
Copollutants:
NO2O3TSP
Significant associations were observed between daily SO2 concentrations
and daily counts of hospitalizations among 15- to 64-yr-old patients and
among those over 64 yrs old, but not among children. These effects were
observed when mean daily SO2 values were lower than the max value
recommended by WHO (125 (jg/m3).
Parameter estimates (PE) and standard error (SE) for a 1-unit increase:
Asthma:
15-64yrs: PE 0.2176 (0.1081) p = 0.44 lag 2;
PE 0.3086 (0.1545) p = 0.046 lag 0-3
Asthma:
65+ yrs: PE 0.2412 (0.0956) p = 0.012 lag 2
F-42
-------
Study
Methods
Pollutant Data
Findings
Prescott et al. (1998)
Edinburgh, United Kingdom
Period of Study:
10/92-6/95
Hospital Admissions
Outcome(s) (ICD 9):
Pneumonia (480-7), COPD +
Asthma (490-496)
Age groups analyzed:
< 65, 65+
Study design: Time-series
Statistical analyses: Poisson
log linear regression
Covariates: Trend, seasonal
and weekly variation,
temperature, wind speed, day
ofwk
Lag: 0,1, or 3 day rolling avg
S02:14.5 (9.0)
PPb
Min: 0 ppb
Max: 153 ppb
# of Stations: 1
Copollutants:
CO PM10NO2
OsBS
No effect of SO2 on hospitalizations observed in either age category.
Increment: 10 ppb
Respiratory admissions:
>65 yrs -2.5 (-11.0, 6.9) lag 0-2; < 65 yrs 0.0 (-8.3, 9.1) lag 0-2
Rossi et al. (1993)
Oulu, Finland
Period of Study: 10/1/1985-
9/30/1986
ED Visits
Outcome(s) (ICD 9): Asthma
(493)
Age groups analyzed: 15-85
Study design: Time-series
N: 232
Statistical analyses:
Pearson's and partial
correlation coefficients and
multiple regression with
stepwise discriminate analysis
Covariates: Temperature,
humidity
Statistical package: BMDP
software
Lag: 0,1,2,3
24-h avg:
10.0 (jg/m3
Range: 0, 56
1-h max:
31.0 (jg/m3
Range: 1, 24
# of monitoring
stations: 4
Copollutants:
N02 (r = 0.48)
TSP (r = 0.31)
H2S
Same day ER visits were correlated to daily SO2 levels, but the significance
was lost with longer lag periods.
When asthma visits were analyzed, SO2 was positively and significantly
correlated with asthma visits in the same wk and the wk after.
After regression analyses, SO2 became insignificant.
Pearson correlation coefficients
ED asthma visits and same day SO2:
r = 0.13 p < 0.01 lag 0
Weekly ED asthma visits and same wk SO2: r = 0.28 p < 0.05
Weekly ED asthma visits and next wk SO2: 0.30 p < 0.05
Multipollutant (NO2; TSP; H2S)
Regression coefficient:
All yr: p = 0.037, p = 0.535
Winter: p = -0.024, p = 0.710
Summer: p = -0.003, p = 0.991
F-43
-------
Study
Methods Pollutant Data
Findings
Schouten et al. (1996)
Multicity, The Netherlands
(Amsterdam, Rotterdam)
Period of Study:
04/01/77-09/30/89
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory (460-519), COPD
(490-2, 494, 496), Asthma
(493)
Age groups analyzed:
15-64, 65+, all ages
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson regression
Covariates: Long-term trend,
season, influenza, day of wk,
holiday, temperature, humidity
Season:
Cool, Nov-Apr;
Warm: May-Oct
Lag: 0,1,2 days; and
cumulative 0-1 and 0-3 day
lags
24-h avg SO2
Amsterdam
Mean/Med:
28/21 (jg/m3
Rotterdam
Mean: 40/32
(jg/m3
Daily 1-h max
Amsterdam
Mean/Med:
65/50 (jg/m3
Rotterdam
Mean/Med:
99/82 (jg/m3
# of stations: 1 per
city
Copollutants:
NO2BS 03
The relationship between short-term air pollution and hospital admissions
was not always consistent at low levels of exposure. One statistically
significant association between hospital admissions and asthma (all ages)
occurred in Amsterdam after a cumulative lag of
1-3 days in the summer. Higher SO2 levels were reported for the winter;
therefore, this association was not a concentration response.
In Rotterdam neither 1 day nor cumulative lags in the summer or winter
increased asthma admissions to statistical significance. Rotterdam had much
higher mean SO2 concentrations. There were no significant associations to
hospital admissions when higher pollution levels were prevalent.
The analysis of all respiratory hospital admissions for all ages in the entire
country (Netherlands) produced a statistically significant association for both
1-h and 24-h periods (100 (jg/m3).
Increment: 100 |jg/m3 increment.
All respiratory, Amsterdam
24-h avg: 15-64 yrs: RR 0.944 (0.864,1.032) lag 2;
RR 0.915 (0.809,1.035) lag 0-3; >65 yrs: RR 1.046 (0.965,1.134) lag 2;
RR 1.008 (0.899,1.131) lag 0-3
1-h max: 15-64 yrs: RR 0.989 (0.952,1.028) lag 2;
RR 0.977 (0.927,1.030) lag 0-3; >65 yrs: RR 1.022 (0.985,1.060) lag 2;
RR 1.010 (0.955,1.068) lag 0-3; RR 0.941 (0.863,1.026) lag 0-3
COPD, Amsterdam
24-h avg-all ages: RR 0.907 (0.814,1.011) lag 0;
RR 0.948 (0.838,1.072) lag 0-1
1-h max-all ages: RR 0.978 (0.933,1.026) lag 0;
RR 0.995 (0.940,1.053) lag 0-1
Asthma, Amsterdam
24-h avg-all ages: RR 0.802 (0.696, 0.924) lag 1;
RR 0.792 (0.654, 0.958) lag 0-3
1-h max-all ages: RR 0.995 (0.942,1.051) lag 0
All respiratory, Rotterdam
24-h avg: 15-64 yrs: RR 0.941 (0.855,1.036) lag 1;
RR 0.895 (0.787,1.019) lag 0-2;
>65 yrs (1977-1981): RR 1.027 (0.904,1.165) lag 2;
RR 1.011 (0.834,1.227) lag 0-3;
>65 yrs (1982-1984): RR 1.087 (0.890,1.328) lag 0;
RR 1.258 (0.926,1.710) lag 0-3;
>65 yrs (1985-1989): RR 1.045 (0.908,1.204) lag 0;
RR 0.968 (0.787,1.190) lag 0-3
1-h max: 15-64 yrs: RR 0.989(0.953,1025) lag 1;
RR 0.965 (0.915,1.018) lag 0-2;
>65 yrs (1977-1981): RR 0.892 (0.842, 0.945) lag 0;
RR 0.987 (0.907,1.074) lag 0-3;
>65 yrs (1982-1984): RR 1.005 (0.933,1.081) lag 0;
RR 1.062 (0.938,1.202) lag 0-3;
>65 yrs (1985-1989): RR 1.010 (0.955,1.068) lag 0;
RR 1.064 (0.992,1.141) lag 0-1
COPD, Rotterdam
24-h avg-all ages: RR 0.963 (0.874,1.059) lag 2;
RR 1.019 (0.887,1.172) lag 0-3
1-h max-all ages: RR 0.991 (0.955,1.029) lag 2;
RR 1.013 (0.953,1.076) lag 0-3
F-44
-------
Study
Methods Pollutant Data
Findings
Spix et al. (1998)
Multicity (Amsterdam,
London, Milan, Paris,
Rotterdam), Europe
Period of Study:
1977 and 1991
Hospital Admissions
Outcome(s) (ICD9): All
respiratory (460-519);
Asthma (493)
Age groups analyzed:
15-64, 65+
Study design: Time-series
Statistical analyses: Poisson
regression following APHEA
protocol. Pooled meta-
analysis adjusted for
heterogeneity
Covariates: trend, seasonality,
day ofwk, holiday,
temperature, humidity,
unusual events (strikes, etc.)
Lag: 1 to 3 days
SO2 daily avg
(|jg/m3)
London:29
Amsterdam: 21
Rotterdam: 25
Paris: 23
Milan: 66
Copollutants:
N02 03 BS TSP
Daily counts of adult respiratory admissions were not consistently associated
with daily mean levels of SO2. Heterogeneity between cities was likely due to
the number of stations or temperature. Only hospital admissions for a 65 yr
olds were significantly associated with SO2 in the warm season.
Increment: 50 |jg/m3
All cities, yr round
15-64yrs:RR 1.009 (0.992,1.025)
Warm RR 1.01 (0.98,1.04)
Cold RR 1.01 (0.97,1.07)
> 65 yrs RR 1.02 (1.005,1.046)
Warm RR 1.06 (1.01,1.11)
Cold RR 1.02 (0.99,1.04)
APHEA protocol pooled result from a65 yrs old from Europe
All respiratory
RR 1.02 (1.00,1.05)
Sunyeretal. (1997)
Multicity, Europe (Barcelona,
Helsinki, Paris, London)
Period of Study:
1986-1992
Hospital admissions/ED Visits
Outcome(s) (ICD 9): Asthma
(493)
Age groups analyzed: < 15,
15-64
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson regression,
GEE; meta-analysis
Covariates: Humidity,
temperature, influenza,
soybean, Long-term trend,
season, day ofwk
Season:
Cool, Oct-Mar;
Warm: Apr-Sep
Lag: 0,1,2,3 and cumulative
1-3
24-h median
(range) (|jg/m3;
Barcelona: 41
(2,160)
Helsinki: 16
(3, 95)
London:31
(9,100)
Paris: 23
(1,219)
# of stations:
Barcelona: 3
London: 4
Paris: 4
Helsinki: 8
Copollutants:
N02
BS
03
SO2 alone or as part of a mixture was a factor that exacerbated asthma
admissions.
In 2-pollutant models with SO2 and BS, the association of BS with SO2 was
attenuated for < 15 yr olds, compared to single-pollutant model associations.
In addition, the association of NO2 was also attenuated by the inclusion of
S02.
Increment: 50 |jg/m3 of 24-h avg for all cities combined.
Asthma:
15-64 yrs:
0.997 (0.961,1.034) lag 2
1.003 (0.959,1.050) lag 0-3, cum
< 15 yrs:
1.075 (1.026,1.126) lag 1
1.061 (0.996,1.131) lag 2-3, cum
2-pollutant models
SO2/BS:
<15 yrs 1.092 (1.031,1.156) lag 0-1
SO2/NO2
< 15 yrs 1.075 (1.019,1.135)
Sunyer* et al. (2003)
Multicity study (Birmingham
(B), London (L), Milan (M),
Netherlands (N), Paris (P),
Rome (R) and Stockholm (S),
Europe
Period of Study:
1992 and 1997
Hospital admissions/ED Visits
Outcome(s) (ICD 9): Asthma
(493); COPD and Asthma
(490-496); all respiratory
(460-519)
Age groups analyzed: All, 0-
14 yrs; 16-64 yrs; a 65 yrs
Study design:Time-series
Poisson regression with GAM
following APHEA 2 protocol
Covariates: temperature,
humidity, Long-term trend,
season
Lag: 0, 1
SO2 24-h avg and
SD (|jg/m3)
B 24.3 (12.7)
L 23.6 (23.7)
M 32.5 (37.5)
N 8.5 (7.7)
P 17.7 (12.5)
R 9.8 (9.9)
S 6.8 (6.2)
Copollutants:
PM10 (r = 0.64)
CO (r = 0.53)
The magnitude of association with asthma across the seven cities was
comparable to earlier studies of London, Helsinki and Paris.
Exposure factors may be important. Children may spend greater time
outdoors compared with adults. Pneumonia requires chronic exposure to
produce inflammatory response and infection, whereas asthma is an acute
response.
Increment: 10 |jg/m3
Asthma : 0-14yrs: 1.3% (0.4, 2.2); 15-64 yrs : 0.0% (-0.9,1.00)
COPD and Asthma: a 65 yrs : 0.6% (0.0,1.2)
All Respiratory: a 65 yrs 0.5% (0.1, 0.9)
Asthma: 0-14 yrs: SO2 + PM10: -3.7% (p > 0.1); SO2 + CO: -0.7% (p > 0.1)
F-45
-------
Study
Methods Pollutant Data
Findings
Sunyeretal. (1991)
ED Visits
24-h avg (SD):
An incremental change of 25 |jg/m3 in SO2 was correlated with an adjusted
Barcelona, Spain
Outcome(s) COPD
56.5 (22.5) (jg/m3
increase of 0.5 daily visits due to COPD.
Period of Study:
1985-1986
(ICD 9): 490-496
Age groups analyzed: > 14
98th: 114.3
Range: 17,160
SO2 and ER visits were more strongly correlated in warm weather.
Even at 24-h avg levels less than 100 pg/m3, effects of SO2 were statistically
Study design:
Time-series
1-h max (SD):
141.9 (98.8) (jg/m3
significant for COPD admissions.
Change in 24-h avg SO2 daily ER |jg/m3 admissions P-value
# of Hospitals: 4
Statistical analyses:
multivariate linear regression
98th: 461.3
Range: 14,720
Number of
monitors: 17
150 0.55 <0.01
100 0.7 <0.01
72 0.7 0.04
52 0.41 >0.05
Covariates: Meteorology,
season, day ofwk
Copollutants:
39-1.27 >0.05
BS, CO, N02,03
0.5 excess daily admissions per 25 |jg/m3 increment of SO2.
Statistical package:
Lag: 0 to 2 days
Sunyeretal. (1993)
ED Visits
S02, 24-h
SO2 concentrations were associated with the number of COPD ER
Barcelona, Spain
Period of Study:
Outcome(s) (ICD 9): COPD
(490-492;
494-496)
Winter Tertiles
(|jg/m3)
admissions in the winter and summer. An increase of 25 |jg/m3 in SO2
produced an adjusted change of-6% and 9%, respectively, in the number of
COPD emergencies in the winter and summer. Controlling for particulate
1985-1989
<40.4
matter resulted in a loss of significance. Co linearity of BS with SO2 was
Study design:
40.4, 61
observed.
Time-series
>61
Effects were expressed as adjusted changes in daily COPD ER admissions
Statistical analyses:
Winter Tertiles
based on an increment of 25 |jg/m3.
Autoregressive linear
(|jg/m3)
Winter: 6%
regression
<28.1
Summer: 9%
Statistical package:
28.1,46.1
>46.1
Copollutants: BS
Mean ER admissions for COPD (winter) were 15.8 (range 3, 34) and 8.3
Lag: 1,2
(range 1, 24) in the summer.
Tenias et al. (1998)
Valencia, Spain
Period of Study:
1993-1995
Seasons:
Cold: Nov-Apr
Warm: May-Oct
ED Visits
Outcome(s): Asthma
ICD 9 Code(s): NR
Age groups analyzed: > 14
Study design: Time-series
N: 734
Statistical analyses: Poisson
regression, APHEA protocol
Covariates: seasonality,
temperature, humidity, long-
term trend, day ofwk,
holidays, influenza
Season: Cold: Nov-Apr;
Warm: May-Oct
Dose-response investigated:
Yes
Statistical package: NR
Lag: 0-3 days
24 h avg: 26.6
(jg/m3
25th
50th
75th
95th
17.9
26.2
34.3
42.6
Cold: 31.7
Warm: 21.7
1-h max:
56.3 (jg/m3
25th
50th
75th
95th
36.3
52.2
72.2
95.2
Cold: 64.6
Warm: 48.2
# of Stations: 2
Copollutants:
24 h avg:
03(r= -0.431)
N02 (24 h av)
(r = 0.265)
N02 (1-h)
(r = 0.199)
1-h:
03(r= -0.304)
NO2 (24 h avg)
(r = 0.261)
N02 (1-h)
(r = 0.201)
SO2 showed the strongest correlation to asthma admissions during the warm
mos.
Multipollutant models showed that O3 and BS had a small effect on the
association between SO2 and asthma ER visits while NO2 greatly depressed
these effects. It is likely that NO2 was the dominant pollutant for respiratory
outcomes. SO2 was the "most vulnerable pollutant" to the presence of other
pollutants.
Increment: 10 |jg/m3
SO2 24-h avg:
All yr 1.050 (0.973,1.133) lag 0
Cold 1.032 (0.937,1.138) lagO
Warm 1.070 (0.936,1.224) lag 0
SO21-h max:
All yr 1.027 (0.998,1.057) lag 0
Cold 1.018(0.980,1.057) lagO
Warm 1.038 (0.990,1.090) lag 0
F-46
-------
Study
Methods Pollutant Data
Findings
Tenias et al. (2002)
Valencia, Spain
Period of Study:
1994-1995
ED Visits
Outcome(s): COPD
ICD 9 Code(s): NR
Age groups analyzed: > 14
Study design: Time-series
N: 1,298
# of Hospitals: 1
Statistical analyses: Poisson
regression, APHEA protocol;
basal models and GAM
Covariates: Seasonality,
annual cycles, temperature,
humidity, day ofwk, feast
days
Season:
Cold, Nov-Apr;
Warm, May-Oct
Dose-response investigated:
Yes
Statistical package: NR
Lag: 0-3 days
24 h avg: 26.6
(jg/m3
5th: 8.2
50th
75th
95th
26.2
34.3
42.6
Cold: 31.7
Warm: 21.7
1-h max: 56.3
(jg/m3
5th: 16.8
50th
75th
95th
52.2
72.2
95.2
Cold: 64.6
Warm: 48.2
Copollutants:
BS (r = 0.687)
N02 (r = 0.194)
CO (r = 0.734)
03(r= -0.431)
SO2 did not show any significant association with COPD ER visits for all
seasons analyzed.
SO2 did not affect O3 or CO association to ER admission for COPD when
assessed together in the Multipollutant model.
Possibility of a linear relationship between pollution and risk of emergency
cases could not be ruled out.
Increment: 10 |jg/m3.
24-h avg SO2:
All yr RR 0.971 (0.914,1.031) lag 0
Cold, 24-h avg: RR 0.970 (0.905,1.038) lag 0
Warm, 24-h avg: RR 0.982 (0.885,1.090) lag 0
1-h max SOz
All yr RR 0.981 (0.958,1.027) lag 3
Cold, 24-h avg: RR 0.972 (0.945,1.000) lag 3
Warm, 24-h avg: RR 1.003 (0.979,1.056) lag 3
Thompson et al. (2001)
Belfast, Northern Ireland
Period of Study:
1993-1995
Hospital admissions/ED Visits Warm Season
Outcome(s): Asthma
ICD 9 Code(s): NR
Age groups analyzed:
Children
Study design: Time-series
N: 1,044
Statistical analyses: Followed
APHEA protocol, Poisson
regression analysis
Covariates: Season, long-
term trend, temperature, day
ofwk, holiday
Season:
Warm (May-Oct); Cold (Nov-
Apr)
Statistical package: Stata
Lag: 0-3
SO2 (ppb):
Mean: 12.60;
SD: 10.60;
IQR: 6.0,16.0
Cold Season
SO2 (ppb):
Mean: 20.40;
SD: 17.90;
IQR: 11.0,24.0
Copollutants:
PM10 (r = 0.66)
N02 (r = 0.82)
NOx (r = 0.83)
NO (r = 0.76)
03 (r = -0.58)
CO (r = 0.64)
Benzene
(r = 0.80)
This study found weak, positive associations for SO2 and adverse respiratory
outcomes in asthmatic children.
SO2 Increment: Per doubling (ppb)
Lag 0 RR 1.07 (1.03,1.11)
Lag 0-1 RR 1.09 (1.04,1.15)
Lag 0-2 RR 1.08 (1.02,1.15)
Lag 0-3 RR 1.08 (1.01,1.15)
Warm only Lag 0-1 RR 1.11 (1.04,1.19)
Cold only Lag 0-1 RR 1.07 (1.00,1.15)
Adjusted for Benzene Lag 0-1 RR 0.99 (0.90,1.09)
Tobias et al. (1999)
Barcelona, Spain
Period of Study:
1986-1989
ED Visits
Outcome(s): Asthma
ICD9: NR
Age groups analyzed: > 14
Study design: Time-series
Statistical analyses: Poisson
regression, followed APHEA
protocol
Covariates: Temperature,
humidity, long-term trend,
season, day ofwk
Statistical package: NR
Lag: NR
24-h avg SO2
(jg/m3
Non-epidemic
Days: 85.8 (62.4)
Epidemic Days:
116.3 (79.3)
Copollutants:
BS NO2O3
The study failed to find a significant association between SO2 and asthma ED
visits.
p x 104 (SE x 104) using Std Poisson
Without modeling asthma epidemics: 3.99 (4.14)
Modeling epidemics with 1 dummy variable: 1.64 (2.76)
Modeling epidemics with 6 dummy variables: 1.53 (2.75)
Modeling each epidemic with dummy variable: 2.20 (2.65)
p x 104 (SE x 104) using Autoregressive Poisson
Without modeling asthma epidemics: 6.99 (14.37)
Modeling epidemics with 1 dummy variable: 1.68 (2.77)
Modeling epidemics with 6 dummy variables: 1.72 (2.75)
Modeling each epidemic with dummy variable: 2.85 (2.89)
F-47
-------
Study
Methods Pollutant Data
Findings
Vigotti et al. (1996)
Milan, Italy
Period of Study:
1980-1989
Hospital Admissions
Outcomes
(ICD 9 codes): Respiratory
disease (460-519).
Age groups analyzed:
15-64 yrs and >64 yrs
Study design: Time-series
N: >73,000
Statistical analyses: APHEA
protocol
Covariates:
Season: Cold season (Oct. to
Mar) and Warm season
(Apr to Sep)
Lag: 0, cumulative 4 day
(0-3)
24-h avg:
117.7 (jg/m3
Range: 3.0, 827.E
5th: 15.0
25th
50th
75th
95th
34.0
65.5
162.5
376.3
Winter:248.6
Range: 30.6,
827.8
5th: 78.8
25th
50th
75th
95th
138.5
216.0
327.8
527.0
Summer:30.5
Range: 3.0,113.E
5th: 9.1
25th
50th
75th
95th
18.5
27.8
39.2
62.7
# of monitors: 4;
r = 0.89, 0.91
Copollutants:
TSP (r = 0.63)
The effect of single day or cumulative day exposure to SO2 was more
pronounced during the cool mos. Interaction between seasons was not
significant. SO2 did not interact with TSP. No differences were noted between
age groups.
There were increased, but not significant (borderline), risks for increased
hospital admissions based on an increment change in SO2 of 125 |jg/m3 in
the winter.
Increment: 100 |jg/m3
All respiratory
15-64 yrs:
All yr round: RR 1.05 (1.00,1.10) lag 0
Warm: RR 1.04 (0.98,1.11) lag 0
Cool: RR 1.06 (1.00,1.13) lagO
>64 yrs:
All yr: RR 1.04 (1.00,1.09) lag 0
Warm: RR 1.02 (0.96,1.08) lagO
Cool: RR 1.05 (1.00,1.11) lag 0
Period of Study
1988-1990
Walters etal. (1994) Hospital Admissions
Birmingham, United Kingdom Outcome(s) (ICD9):
Asthma (493) and acute
respiratory conditions
(466, 480-486, 490-496)
Study design:Time-series
Statistical analyses:
Least squares regression
Covariates: Temperature,
pressure, humidity
Lag: 3 day moving avg.
SO2 24-h avg
(|jg/m3)
All yr: 39.06
Max: 126.3
Spring: 42.9
Summer: 37.8
Autumn: 40.9
Winter: 34.2
Copollutants: BS
In 2-pollutant models BS remained significant but SO2 was no longer
associated significantly with admission.
A100 (jg/m3 increment in SO2 might result in four (0-7) more asthma
admissions and 15.5 (6-25) move respiratory admissions/day. Spring and
autumn did not show associations with admissions for asthma or respiratory.
Increment of 100 |jg/m3
Asthma:
Summer: 1.4% (-10, 39) lag 0
Winter: 2.7% (-0.8, 6.1) lag 0
All respiratory:
Summer: 5.9% (1.1, 10.6) lag 0; (p < 0.02)
Winter: 18% (8.8, 26.8) lag 0; (p < 0.0002)
F-48
-------
Study
Methods Pollutant Data
Findings
AUSTRALIA/NEW ZEALAND
Barnett et al. (2005)
Multicity, Australia/New
Zealand; (Auckland,
Brisbane, Canberra,
Christchurch, Melbourne,
Perth, Sydney)
Period of Study:
Jan 1998-Dec 2001
Hospital Admissions
Outcomes (ICD 9/ICD 10):
All respiratory
(460-519/J00-J99 excluding
J95.4-J95.9, R09.1, R09.8),
asthma (493/J45, J46, J44.8),
COPD (490-492, 494-
496/J40-J44, J47, J67),
pneumonia with bronchitis
(466, 480-486/J12-17, J18.0
j18.1 J18.8J18.9 J20J21)
Age groups analyzed: <1,1-4,
5-14
Study design: Case-crossover
Statistical analyses:
Conditional logistic
regression, random effects
meta-analysis
Covariates: Temperature,
current-previous day
temperature, relative humidity,
pressure, extremes of hot and
cold, day of wk, holiday, day
after holiday
Season: Cool, May-Oct;
Warm, Nov-Apr
Statistical package: SAS
Lag: 0-1 days
24-h avg (ppb)
(range):
Auckland: 4.3
(0, 24.3)
Brisbane: 1.8
(0, 8.2)
Canberra: NA
Christchurch: 2.8
(0, 11.9)
Melbourne: NA
Perth: NA
Sydney: 0.9
(0, 3.9)
Daily 1-h max
(range):
Auckland: NA
Brisbane: 7.6
(0, 46.5)
Canberra: NA
Christchurch: 10.1
(0.1,42.1)
Melbourne: NA
Perth: NA
Sydney: 3.7
(0.1,20.2)
Increased hospital admissions were significantly associated with SO2 for
acute bronchitis, pneumonia, and respiratory diseases. In multipollutant
models the impacts of particulate matter and NO2 were isolated.
There were seasonal impacts on pneumonia and acute bronchitis admissions
in the 1- to 4-yr-old age group for SO2.
Increment: 5.4 ppb
(1-h max IQR)
Pneumonia and acute bronchitis:
Oyrs: No analysis
1-4yrs: 6.9% (2.3,11.7) lag 0-1
Respiratory:
Oyrs: 3.2% (0.3,6.3) lag 0-1
1-4 yrs: 2.7% (0.6, 4.8) lag 0-1
5-14 yrs: No analysis
Asthma:
0 yrs: No analysis (poor diagnosis)
1-4 yrs: No analysis
5-14 yrs: No analysis
Lam (2007)
Australia (New South Wales;
Sydney)
Period of Study:
2001-2002
ED Visits
Outcome(s): Fever,
gastroenteritis, asthma/other
respiratory problems
Study design: Time-series
Statistical Analysis: Auto
Regression Integrated Moving
Average (ARIMA) statistical
modeling
Statistical package: SPSS
Age groups analyzed: < 6
Covariates: NR
Lag(s): NR
24-h avg (ppm):
0.35 (0.19)
Range: 0.10, 0.90
1-h max (ppm):
0.38 (0.20)
Range: 0.10,1.80
Copollutants:
PM10
PM2.5
N02
03
Bivariate correlations resulted in ARIMA models for fever and NO2 max,
gastroenteritis and O3 avg and NO2 max; and respiratory problems and O3
max. Neither NO2 nor O3 was significantly associated with any of the
childhood illnesses analyzed.
SO2 was not significantly correlated with fever, gastroenteritis, or respiratory
problems; therefore, SO2 was not included in the ARIMA models.
F-49
-------
Study
Methods Pollutant Data
Findings
Petroeschevsky et al. (2001)
Brisbane, Australia
Period of Study:
1987-1994
Days: 2922
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory (460-519); Asthma
(493)
Age groups analyzed: 0-4,
5-14,15-64, 65+, all ages
Study design: Time-series
N: 33,710 (respiratory),
13,246 (asthma)
Statistical analyses: APHEA
protocol, Poisson regression,
GEE
Covariates: Temperature,
humidity, season, infectious
disease, day of wk, holiday
Season: Summer, Autumn,
Winter, Spring, All yr
Dose-response investigated?
Yes
Statistical package: SAS
Lag: Single: 1,2,3 day
Cumulative: 0-2, 0-4
Mean: 24-h avg:
Overall: 4.1 ppb
Summer: 3.9 ppb
Autumn: 4.2 ppb
Winter: 4.8 ppb
Spring: 3.7 ppb
Mean: 1-h max
Overall: 9.2 ppb
Summer: 7.8 ppb
Autumn: 9.3 ppb
Winter: 11.3 ppb
Spring: 8.4 ppb
# of stations: 3
Copollutants:
BSP
03
N02
SO2 was highly correlated with max daily ER admissions for respiratory
conditions. The highest association was observed in the winter followed by
autumn, spring, and summer. For asthma, the highest association was
observed in the winter and autumn.
No statistically significant contributions for respiratory admissions were
reported for the age group 5-14 yr olds for any pollutant.
Increment: 0 ppb
Respiratory:
0-4 yrs 24-h avg 1.224 (1.087,1.377) lag 0-4
5-14 yrs 1 -h max 1.049 (0.986,1.116) lag 0-4
15-64yrs 24-h avg 1.033 (0.895,1.118) lag 1
65+ yrs 24-h avg 1.121 (1.019,1.234) lag 0
All ages 24-h avg 1.080 (1.030,1.131) lag 1
Asthma:
0-14 yrs 24-h avg 1.080 (0.971,1.201) lag 0
15-64 yrs 1-h max 0.941 (0.900, 0.984) lag 0
All ages 24-h avg 0.941 (0.876,1.011) lag 2
LATIN AMERICA
Braga* et al. (1999)
Sao Paulo, Brazil
Period of Study:
10/1992-10/1993
Hospital Admissions
Outcome(s) (ICD9): All
respiratory (466, 480-486,
491-492, 496)
Age groups analyzed:
<13 yrs
Study design:Time-series
N: 68,918
# of Hospitals:
112 Statistical analyses:
Multiple linear regression
models (least squares). Also
used Poisson regression
techniques. GLM and GAM
using LOESS for smoothing.
Covariates: Season,
temperature, humidity, day of
wk,
Statistical package: SPSS, S-
Plus
Lag: 1,2,3,4,5,6,7 moving
avgs
24-h avg 22.40
(9.90) (jg/m3
Min: 6.4
Max: 69.6
# of monitors: 13
Copollutants:
PM10 (r = 0.73)
CO (r = 0.62)
N02 (r = 0.53)
03
SO2 did not show a correlation with respiratory hospital admissions with any
lag structure.
Increment: 22.4 |jg/m3
2 (-0.04, 0.28) lag 0
8 (-0.00, 0.37) lag 0-1
9 (-0.01, 0.39) lag 0-2
8 (-0.04, 0.40) lag 0-3
8 (-0.05, 0.42) lag 0-4
2 (-0.13, 0.36) lag 0-5
'8 (-0.18, 0.35) lag 0-6
F-50
-------
Study
Methods
Pollutant Data
Findings
Braga* et al. (2001)
Sao Paulo, Brazil
Period of Study:
1/93-11/97
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory admissions (460-
519)
Age groups analyzed: 0-19, s
2,3-5,6-13,14-19
Study design: Time-series
Statistical analyses: Poisson
regression with GAM
Covariates: Long-term trend,
season, temperature, relative
humidity, day ofwk, holiday
Statistical package: S-Plus
4.5
Lag: 0-6 moving avg
SO2 Mean:
21.4 ug/m3;
SD: 11.2
IQR: 14.4 (jg/m3
Range: 1.6, 76.1
# of stations: 5-6
Copollutants:
PM10 (r = 0.61)
N02 (r = 0.54)
CO (r = 0.47)
03 (r = 0.17)
Children < 2 yrs were most susceptible to the effect of each pollutant.
Pneumonia and bronchopneumonia were the main cause of hospital
admissions (71%) in the < 2-yr-old group. Bronchitis/asthma were more
important for the intermediate age groups. However, in all age groups the
largest increase in admissions was caused by chronic disease in tonsils and
adenoids.
Multipollutant models rendered all pollutants except PM10 and SO2 from
significance. The effect of PM10 stayed relatively unchanged while SO2 was
reduced; however, it remained significant.
Increment: |jg/m3 (IQR)
All respiratory admissions: < 2 yrs: 5.9% (4.5, 7.4);
3-5 yrs: 1.6% (-1.3, 4.4); 6-13 yrs: 0.6% (—2.2, 3.5);
14-19 yrs: 1.3% (-3.2, 5.8); All ages 4.5% (3.3, 5.8)
Farhat* et al. (2005)
Sao Paulo, Brazil
Period of Study:
Aug 1996-Aug 1997
Hospital Admissions/ED Visits
Outcome(s) (ICD9): Lower
Respiratory Disease (466,
480-5)
Age groups analyzed: < 13
Study design: Time-series
N: 4,534
# of Hospitals: 1
Statistical analyses:
1) Poisson regression and 2)
GAM - no mention of more
stringent criteria
Covariates: Long-term trends,
seasonality, temperature,
humidity
Statistical package:
S-Plus
Lag: 0-7 days, 2,3,4 day
moving avg
24-h avg:
Mean: 23.7 |jg/m3
SD: 10.0
Range: 3.4, 75.2
IQR: 12.5
# of Stations: 6
Copollutants:
PM10 (r = 0.69)
N02 (r = 0.66)
CO (r = 0.49)
03 (r = 0.28)
This study reports a significant effect of air pollution on respiratory morbidity,
though several pollutants were associated with increased respiratory events,
making it difficult to isolate a single agent as the main atmospheric
contaminant.
Increment: 12.5 |jg/m3 (IQR)
Single-pollutant models (estimated from graphs):
Pneumonia: -21% (4.8, 37)
Asthma: -12% (-10, 38)
Pneumonia multipollutant models:
Adjusted for: PM1013.3 (-5.7, 32.3) 6-day avg;
NO216.5 (-1.6, 34.6) 6-day avg; CO 18.4 (0.5, 36.2) 6-day avg;
O318.4 (0.5, 36.2) 6-day avg
Multipollutant model: 13.3 (-5.9, 32.6) 6-day avg
Asthma multipollutant models:
Adjusted for: PM10 3.8 (-23.3, 31.0) 2-day avg;
NO2 -1.2 (-27.4, 25.0) 2-day avg; CO 6.2 (-18.8, 31.2) 2-day avg;
O3 9.4 (-14.6, 33.5) 2-day avg
Multipollutant model: -0.5 (-27.7, 26.6) 2-day avg
Gouveia and Fletcher (2000)
Sao Paulo, Brazil
Period of Study:
11/92-9/94
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory; Pneumonia (480-
486); asthma or bronchitis
(466, 490, 491, 493)
Age groups analyzed:
< 1; < 5 yrs
Study design: Time-series
Statistical analyses: Poisson
regression Covariates: Long-
term trend, season,
temperature, relative humidity,
day ofwk, holiday, strikes in
public transport or health
services
Season:
Cool (May-Oct),
Warm (Nov-Apr)
Statistical package: SAS
Lag: 0,1,2 days
24-h avg:
Mean: 18.3 ug/m3
SD: 9.0
Range: 3.2, 61.1
5th: 7.6
25th: 11.9
50th: 16.6
75th: 22.2
95th: 35.8
# of stations: 4
Copollutants:
PM10 (r = 0.72)
N02 (r = 0.37)
CO (r = 0.65)
03 (r = 0.08)
Current ambient air pollution concentrations have short-term adverse effects
on children's respiratory morbidity assessed through admissions to hospitals.
Increment: 27.1 |jg/m3
(90th-10th)
All Respiratory:
<5 yrs RR 1.038 (0.983,1.096) lag 1
< 5 yrs Cool RR 1.06 (0.99,1.11) (estimated from graph)
< 5 yrs Warm RR 0.98 (0.89,1.07) (estimated from graph)
Pneumonia:
<5 yrs RR 1.024 (0.961,1.091) lag 1
< 1 yr RR 1.071 (0.998,1.149) lagO
Asthma:
<5 yrs RR 1.106 (0.981,1.247) lag 2
F-51
-------
Study
Methods Pollutant Data
Findings
llabaca et al. (1999)
Santiago, Chile
Period of Study:
2/1/95-8/31/96
Days: 578
ED Visits
Outcome(s) (ICD9): Upper
respiratory illness (460-465,
487);
Lower respiratory illness (466,
480-486, 490-494, 496,
519.1, 033.9); Pneumonia
(480-486)
Age groups analyzed: < 15
Study design: Time-series
# of Hospitals: 1
Statistical analyses: Poisson
regression
Covariates: Long-term trend,
season, day ofwk,
temperature, humidity,
influenza epidemic
Season:
Warm (Sep-Apr), Cool (May-
Aug)
Lag: 0-3 days
24-h avg SO2
(|jg/m3)
Warm:
Mean: 14.9
Median: 13.2
SD: 8.8
Range: 1.9, 60.2
5th: 5.6
95th: 32.0
Cool:
Mean: 31.8
Median: 28.2
SD: 18.4
Range: 5.6, 92.1
5th: 9.4
95th: 75.2
# of stations: 4
Copollutants:
Warm:
NO2 (r = 0.6556)
03(r= 0.1835)
PM10 (r = 0.6687)
PM2.s(r= 0.5764)
Cool:
N02 (r = 0.7440)
03(r= 0.1252)
PM10 (r = 0.7337)
PM2.s(r= 0.6874)
SO2 was related to the number of respiratory ED visits, but because of the
high correlation between contaminants, it is difficult to establish independent
health effects. These results support the fact that exposure to air pollution
mixtures may decrease immune functions and increase the risk for
respiratory infections among children.
Increment: IQR
All respiratory:
Cool: Lag 2 IQR: RR 1.0289 (1.0151,1.0428);
Lag 3 IQR: RR 1.0374 (1.0236,1.0513);
Lag avg 7 IQR: RR 1.0230 (1.0086,1.0377)
Warm: Lag 2 IQR: RR 1.0029 (0.9860,1.0200)
Lag 3 IQR: RR 1.0108 (0.9937,1.0282)
Lag avg 7 IQR: RR 1.0108 (0.9756,1.0473)
Upper respiratory:
Cool: Lag 2 IQR: RR 1.0584 (1.0394,1.0778)
Lag 3 IQR: RR 1.0513 (1.0324,1.0706)
Lag avg 7 IQR: RR 1.0316 (1.0120,1.0515)
Warm: Lag 2 IQR: RR 1.0061 (0.9850,1.0277)
Lag 3 IQR: RR 1.0130 (0.9916,1.0349)
Lag avg 7 IQR: RR 0.9815 (0.9390,1.0260)
Pneumonia:
Cool: Lag 2 IQR: RR 1.0164 (0.9757,1.0587)
Lag 3 IQR: RR 1.0342 (0.9938,1.0762)
Lag avg 7 IQR: RR 1.0291 (0.9850,1.0751)
Warm: Lag 2 IQR: RR 1.1010 (1.0404,1.1653)
Lag 3 IQR: RR 1.0248 (0.9669,1.0862)
Lag avg 7 IQR: RR 1.2151 (1.0771,1.3709)
Lin etal. (1999)
Sao Paulo, Brazil
Period of Study:
May 1991-Apr 1993
Days: 621
ED Visits
Outcome(s): Respiratory
disease, Upper respiratory
illness, Lower respiratory
illness, Wheezing
ICD 9Code(s): NR
Age groups analyzed: < 13
Study design: Time-series
# of Hospitals: 1
Statistical analyses: Gaussian
and Poisson regression
Covariates: Long-term trend,
seasonality, day ofwk,
temperature, humidity
Statistical package: NR
Lag: 5-day lagged moving
avgs
SO2 (jg/m3:
Mean: 20
SD: 8
Range: 4, 60
Number of
stations: 3
Copollutants:
N02 (r = 0.38)
CO (r = 0.56)
PM10 (r = 0.73)
03 (r = 0.21)
The results of this study demonstrate a significant association between the
increase in emergency visits for all respiratory illness, especially URI, and
SO2 levels.
Increment: 10 |jg/m3
All respiratory illness:
SO2 alone RR 1.079 (1.052,1.107) 5-day moving avg
S02 + PM10 + Os + N02 + CO RR 0.938 (0.900, 0.977)
Lower respiratory illness:
SO2 alone RR 1.052 (0.984,1.125) 5-day moving avg
S02 + PM10 + 03 + N02 + CO RR 0.872 (0.783, 0.971)
Upper respiratory illness:
SO2 alone RR 1.075 (1.044,1.107) 5-day moving avg
S02 + PM10 + 03 + N02 + CO RR 0.951 (0.906, 0.999)
Wheezing:
SO2 alone RR 1.034 (0.975,1.096) 5-day moving avg
S02 + PM10 + 03 + N02 + CO RR 0.908 (0.824,1.002)
F-52
-------
Study
Methods Pollutant Data
Findings
Martins* et al. (2002)
Sao Paulo, Brazil
Period of Study:
5/96-9/98
ED Visits
Outcome(s) (ICD10): Chronic
Lower Respiratory Disease
(CLRD) (J40-J47); includes
chronic bronchitis,
emphysema, other COPDs,
asthma, bronchiectasia
Age groups analyzed: >64
Study design: Time-series
N: 712
# of Hospitals: 1
Catchment area: 13,163 total
ER visits
Statistical analyses: Poisson
regression and GAM - no
mention of more stringent
criteria
Covariates: Weekdays, time,
min temperature, relative
humidity, daily number of non-
respiratory emergency room
visits made by elderly
Statistical package: S-Plus
Lag: 2-7 days and 3 day
moving avgs
SO2 24-h avg
(|jg/m3): 18.7,
SD: 10.6
Range: 2.0, 75.2
IQR: 15.1 (jg/m3
# of stations: 13
Copollutants:
03 (r = 0.28)
N02 (r = 0.67)
PM10 (r = 0.72)
CO (r = 0.51)
The results of the study show a significant association between SO2 and
CLRD among the elderly.
Increment: IQR ofpg/m3
Percent increase: 17.5
(5.0, 23.0) lag 3-day moving avg (estimated from graph)
Single-pollutant model
p = 0.0140 (0.0056)
Multipollutant model (with O3)
p = 0.0104 (0.0059)
Agarwal et al. (2006)
Safdarjung area of south Deli
Period of Study:
2000-2003
Hospital Admissions
Outcome(s) (ICD9): COPD,
asthma and emphysema
Study design: time-series
Statistical Analysis:
Performed Kruskal-Wallis one
way analysis of variance by
rank, chi-square analysis.
Statistical package: SPSS
Age groups analyzed: all
Covariates: Temperature-min
and maximum, relative
humidity at 0830 and 1730 h
and wind speed
N: NR
# Hospitals: 1
Lag:none
Mean, SD
Quarter 1:
16.7.5.5
Quarter 2:
13.6.2.6
Quarter 3:
12.8,3.1
Quarter 4:
14.3,2.8
Copollutants:
NO2SPM RSPM
SO2 was found to be in "low" category the entire time, so no analysis could
be performed
F-53
-------
Study
Methods Pollutant Data
Findings
Chew et al. (1999)
Singapore
Period of Study:
1990-1994
Hospital Admissions/ED Visits 24-h avg: 38.1
Outcome(s) (ICD 9): Asthma 'J3'm
(493)
Age groups analyzed: 3-12,
13-21
Study design: Time-series
N: 23,000
# of Hospitals: 2
Statistical analyses: Linear
regression, GLM
Covariates: variables that
were significantly associated
with ER visits were retained in
the model
Statistical package:
SAS/STAT, SAS/ETS 6.08
Lag: 1, 2 days avgs
SD: 21.8
Range: 3.0,141.0
# of Stations: 15
Copollutants:
NO2O3TSP
SO2 was positively correlated to daily ER visits and hospitalization for asthma
in children (3-12 yrs), but not adolescents. The association of ER visits with
SO2 persisted after standardization for meteorological and temporal
variables. An adjusted increase in 2.9 ER visits for every 20 |jg/m3 increase
in ambient SO2 levels with a lag of 1 was observed.
The increased number of ER visits/day for each quartile are listed below:
Q1: < 9; Q2:10-12; Q3:13-16; Q4:> 16
Categorical analysis (via ANOVA)
p-value and Pearson correlation coefficient (r) using continuous data
comparing daily air pollutant levels and daily number of ER visits
Age Group: 3-12 13-21
Lag 0 r= 0.04 r = 0.05
Lag 1 r = 0.10 r= 0.06
Lag 2 r= 0.08 r= 0.07
p < 0.001 p = 0.086
p < 0.001 p = 0.016
p < 0.001 p = 0.019
Hwang and Chan (2002)
Taiwan
Period of Study:
1998
ED Visits
Outcome(s) (ICD 9): Lower
Respiratory Disease (LRD)
(466, 480-6) including acute
bronchitis, acute bronchiolits,
pneumonia
Age groups analyzed:
0-14,15-64, a 65, all ages
Study design: Time-series
Catchment area: Clinic
records from 50 communities
Statistical analyses: Linear
regression, GLM
Covariates: temperature, dew
point temperature, season,
day ofwk, holiday
Lag: 0,1,2 days and avgs
24-h avg: 5.4 ppb,
SD: 3.0
Range: 1.5,16.9
Copollutants:
NO2PM10O3CO
No correlations for
individual-
pollutants.
Colinearity of pollutants prevented use of multipollutant models
Increment: 10% change in SO2 (natural avg) which is equivalent to 2.4 ppb.
NOTE: The percent change is for the rate of clinic use NOT for relative risk
for adverse effect.
Increased clinic visits for lower respiratory disease (LRD) by age group
0-14 yrs:
Lag 0 0.5% (0.3, 0.6)
15-64 yrs:
Lag 0 0.7% (0.5, 0.8)
a65 yrs:
Lag 0 0.8% (0.6,1.1)
All ages:
Lag 0 0.5% (0.4, 0.7)
Ko et al. (2007b)
Hong Kong
Period of Study:
2000-2005
Hospital Admissions
Outcome(s) (ICD9): Asthma
Study design: Retrospective
ecological study
Statistical Analysis:
Generalized additive models
with Poisson distribution.
Age groups analyzed: All
Covariates:
N: 69,716
# Hospitals: 15
Lag: 0-5 days
Mean, SD (|jg/m3) SO2 had a non-significant effect on respiratory admissions.
Whole yr:
18.8,13.1
<20 °C: 18.0,10.0
>20 °C: 19.1,14.1
Copollutants:
N02 (r =0.573)
PM10 (r =0.430)
PM2.s(r =0.482)
03 (r =0.123)
Relative Risk (95% CI)
Lag 0
Lag 1
Lag 2
Lag 3
Lag 4
Lag 5
Lag 0,1
Lag 0,2
Lag 0,3
Lag 0-4
Lag 0-5
004 (0.998
000 (0.994
999 (0.993
002 (0.998
004 (0.997
997 (0.990
011)
007)
006)
008)
010)
003)
1.003 (0.996,1.011)
1.003 (0.994,1.011)
1.004 (0.994,1.014)
1.007 (0.996,1.017)
1.004 (0.993,1.016)
F-54
-------
Study
Methods Pollutant Data
Findings
Ko et al. (2007a)
Hong Kong
Period of Study:
2000-2004
Hospital Admissions
Outcome(s) (ICD9): COPD
Study design: Retrospective
ecological study
Statistical Analysis: Poisson
distribution
Age groups analyzed: All
ages
Covariates: Autocorrelation
and overdispersion were
corrected
N: 119,225
15.0 (jg/rn3
SD: 11.6
Copollutants:
NO2 PM10 O3 PM2.5
Positive association with hospital admission for acute exacerbations of
COPD.
Relative Risk (95% CI)
LagO: 1.007 (1.001,1.014)
Lag 1:0.991 (0.981,1.001)
Lag 2: 0.992 (0.985,1.000)
Lag 3:1.006 (0.999,1.013)
Lag 4:1.004 (0.998,1.011)
Lag 5:1.004 (0.997,1.010)
Lag 0-1: 0.998 (0.991
Lag 0-2: 0.993 (0.985
Lag 0-3: 0.998 (0.989
Lag 0-4:1.001 (0.991
Lag 0-5:1.004(0.994
006)
001)
007)
010)
014)
# Hospitals: 15
Lag: 0-5 days
Lee* et al. (2002)
Seoul, Korea
Period of Study:
12/1/97-12/31/99
Days: 822
Hospital Admissions
Outcomes (ICD 10): Asthma
(J45-J46)
Age groups analyzed: < 15
Study design: Time-series
N: 6,436
Statistical analyses: Poisson
regression, log link with GAM
Covariates: Time, day of wk,
temperature, humidity
Season: Spring (Mar-May),
Summer (Jun-Aug),
Fall (Sep-Nov),
Winter (Dec-Feb)
Lag: 0-2 days cumulative
24-h avg SO2
(PPb)
Mean: 7.7
SD: 3.3
5th: 3.7
25th
50th
75th
95th
5.1
7.0
9.5
14.3
# of stations: 27
Copollutants:
N02 (r = 0.723)
03 (r = -0.301)
CO (r = 0.812)
PM10 (r = 0.585)
This study reinforces the possible role of SO2 on asthma attacks, although it
should be interpreted with caution because the effect estimates are close to
the null and because results in the multipollutant models are inconsistent.
Increment: 14.6 ppb (IQR)
Asthma:
SO2RRI.H (1.06,1.17) lag 0-2
SO2 + PM10 RR 1.08 (1.02,1.14) lag 0-2
S02 + N02 RR 0.95 (0.88,1.03) lag 0-2
SO2 + O3 RR 1.12 (1.06,1.17) lag 0-2
SO2 + CO RR 0.99 (0.92,1.07) lag 0-2
S02 + 03 + CO + PM10 + N02 RR 0.949 (0.868,1.033)
Lee et al. (2006)
Hong Kong, China
Period of Study:
1997-2002
Days: 2,191
Hospital Admissions
Outcome(s) (ICD 9): Asthma
(493)
Age groups analyzed: s18
Study design: Time-series
N: 26,663
Statistical analyses: Semi-
parametric Poisson
regression with GAM (similar
toAPHEA2)
Covariates: Long-term trend,
temperature, relative humidity,
influenza, day of wk, holiday
Statistical package: SAS 8.02
Lag: 0-5 days
SO2 24-h avg:
17.7 ug/m3,
SD: 10.7
IQR: 11.1 (jg/m3
25th
50th
75th
10.6
15.2
21.7
# of stations:
9-10
Copollutants:
PM10 (r = 0.37)
PM2.5 (r = 0.47)
N02 (r = 0.49)
03 (r = -0.17)
Absence of an association of SO2 with asthma admissions was attributed to
low ambient SO2 levels during the study period due to restrictions on sulfur
content in fuel.
Increment: 11.1 (jg/m3 (IQR)
Asthma:
Single-pollutant model
Lag 0-1.57% (-2.87,-0.26)
Lag 1 -1.77% (-3.06,-0.46)
Lag 2-1.15% (-2.42,0.14)
Lag 3 0.82% (-0.45,2.11)
Lag 4 1.40% (0.13, 2.69)
Lag 5 1.46% (0.19, 2.74)
Multipollutant model-including PM, NO2, and Os
0.81% (-0.75, 2.4) lag 5
Other lags NR
F-55
-------
Study
Methods Pollutant Data
Findings
Lee et al. (2007)
Kaohsiung, Taiwan
Period of Study:
1996-2003
Hospital Admissions
Outcome(s) (ICD9): COPD
(490-492, 494, and 496)
identified by records from the
National Health Insurance
(NHI) program
Study design: Case-crossover
Statistical Analysis:
Conditional logistic regression
Statistical package: SAS
Age groups analyzed: All
ages
Covariates: Adjustment for
temperature and humidity
N: 25,108
# Hospitals: 63
Lag: Cumulative lag up to 2
days
24-h avg (ppb):
9.49
Range: 0.92,
31.33
Copollutants:
PM10NO2CO 03
All pollutants, except SO2, were significantly associated with COPD hospital
admissions on warm days, while on cold days all pollutants were found to be
significantly associated. In two pollutant models, CO and O3 were
significantly associated with each of the other pollutants on warm days, and
on cool days, only NO2 was significantly associated with all pollutants.
Odds Ratio (95% CI)
Single-pollutant model (per 5.79 ppb SO2):
>25°C: 1.024 (0.973,1.077); <25 °C: 1.190(1.093,1.295)
Co-pollutant model (per 5.79 ppb SO2):
>25°C:
SO2 + PM10:1.002 (0.951,1.054)
SO2 + NO2: 0.979 (0.926,1.034)
SO2 +CO: 0.929 (0.876,0.985)
SO2 + O3:1.057 (1.004,1.113)
< 25 °C:
SO2 + PM10:1.043 (0.952,1.143)
SO2 + NO2: 0.767 (0.689,0.855)
SO2 + CO: 1.004 (0.915,1.103)
SO2 + O3:1.198 (1.100,1.304)
Tanaka et al. (1998)
Kushiro, Japan
Period of Study:
1992-1993
ED Visits
Outcome(s): Asthma
Age groups analyzed:
15-79
Study design: Time-series
N: 102
# of Hospitals: 1
Statistical analyses: Poisson
regression
Covariates: temperature,
vapor pressure, barometric
pressure, relative humidity,
wind velocity, wind direction at
maximal velocity
Statistical package: NR
SO2 24-h avg
3.2 (2.4) ppb in fog
3.7 (1.9) ppb in fog
free days
Max SO2 24-h avg
< 11 ppb
Copollutants:
NO N02
SPM (TSP) 03
The results reveal that ED visits by atopic subjects increased on low SO2
days. This observation is inconsistent with most air pollution epidemiology, as
high levels of air pollutants have conventionally been linked with asthma
exacerbation.
Increment: 5 ppb
Nonatopic:
OR 1.18 (0.96, 1.46)
Atopic:
OR 0.78 (0.66, 0.93)
Tsai et al. (2006)
Kaohsiung, Taiwan
Period of Study:
1996-2003
Days: 2922
Hospital Admissions
Outcome(s) (ICD 9): Asthma
(493)
Study design:
Case-crossover
N: 17,682
Statistical analyses:
conditional logistic regression
Covariates: Temperature,
humidity
Season:
Warm (a 25 °C);
Cool (< 25 °C)
Statistical package: SAS
Lag: 0-2 days Cumulative
SO2 24-h avg:
9.49 ppb
Range: 0.92,
31.33
25th: 6.37
50th: 8.94
75th: 12.16
# of stations: 6
Copollutants:
PM10NO2O3CO
Positive associations were observed between air pollutants and hospital
admissions for stroke. In single-pollutant models SO2 was not associated with
either PIH or IS. The season did not affect these associations. SO2 was also
not significant in
2-pollutant models.
Increment: 5.79 ppb (IQR)
Seasonality
Single-pollutant model: >25 °C 1.018 (0.956,1.083) lag 0-2;
< 25 °C 1.187 (1.073,1.314) lag 0-2
Dual-pollutant model:
Adjusted for PM10: >25 °C 0.993 (0.932, 1.058) lag 0-2;
< 25 °C 1.027 (0.921,1.146) lag 0-2
Adjusted for CO: >25 °C 0.910 (0.847, 0.978) lag 0-2;
< 25 °C 1.036 (1.027,1.046) lag 0-2
Adjusted for NO2: >25 °C 0.967 (0.903,1.035) lag 0-2;
< 25 °C 0.735 (0.646, 0.835) lag 0-2
Adjusted for O3: >25 °C 1.055 (0.990,1.123) lag 0-2;
< 25 °C 1.195 (1.080,1.323) lag 0-2
F-56
-------
Study
Methods Pollutant Data
Findings
Wong et al. (1999)
Hong Kong, China
Period of Study:
1994-1995
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory admissions (460-
6, 471-8, 480-7, 490-6);
Asthma (493), COPD (490-
496), Pneumonia (480-7)
Age groups analyzed:
0-4, 5-64, a 65, all ages
# of Hospitals: 12
Study design:Time-series
Statistical analyses: Poisson
regression (followed APHEA
protocol)
Covariates: Trend, season,
day ofwk, holiday,
temperature, humidity
Statistical package: SAS 8.02
Lag: days 0-3 cumulative
Median 24-h avg
S02:
17.05 (jg/m3
Range: 2.74,
68.49
25th: 12.45
75th: 25.01
# of stations: 7
Copollutants:
O3SO2 PM10
Adverse respiratory effects of SO2 were noted at low concentrations. Results
for respiratory outcomes were attributed to the elderly population. This was
also true for the other pollutants. Therefore, it is difficult to be certain that the
effects were due mainly to SO2.
Pair-wise comparisons in multipollutant models showed significant
interactions of PM2.5, NO2, and O3.
Increment = 10 |jg/m3
Overall increase in admissions: 1.013 (1.004,1.021) lag 0
Respiratory relative risks (RR): 0-4 yrs: 1.005 (0.991,1.018) lag 0;
5-64 yrs: 1.008 (0.996,1.021) lag 0; >65 yrs: 1.023 (1.012,1.036) lag 0
Asthma: 1.017 (0.998,1.036) lag 0
COPD: 1.023 (1.011,1.035) lagO
Pneumonia: 0.990 (0.977,1.004) lag 4
Wong et al. (2001b)
Hong Kong, China
Period of Study:
1993-1994
Hospital Admissions
Outcome(s) (ICD 9): Asthma
(493)
Age groups analyzed: s 15
N: 1,217
# of Hospitals: 1
Study design: Time-series
Statistical analyses: Poisson
regression (followed APHEA
protocol)
Covariates: Season,
temperature, humidity
Season:
Summer (Jun-Aug), Autumn
(Sep-Nov),
Winter (Dec-Feb),
Spring (Mar-May)
Lag: 0,1, 2, 3, 4, 5 days; and
cumulative 0-2 and 0-3 days.
24-h avg SO2
Mean: 12.2 |jg/m3
SD: 12.9
Range: 0, 98
(jg/m3
Autumn: 10.6 (9.6)
Winter: 10.0 (7.5)
Spring: 9.6 (8.8)
Summer: 18.5
(19.5)
# of stations: 9
Copollutants:
PM10NO2
SO2 levels were found to be the highest during the summer. There were
consistent and statistically significant associations between asthma
admission and increased daily levels of SO2. No associations were noted in
the spring or winter. No significant associations were found between hospital
admissions and day of the wk, humidity, temperature or atmospheric
pressure.
Total admissions were limited to one hospital.
Increment: 10 |jg/m3
Asthma:
All yr: RR 1.06 p = 0.004
Autumn: NR
Winter: NR
Spring: NR
Summer: NR
F-57
-------
Study
Methods Pollutant Data
Findings
Wong* et al. (2002a)
London England and Hong
Kong
Period of Study:
London:
1992-1994
Hong Kong:
1995-1997
Days: 1,096
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory admissions (460-
519); asthma (493)
Age groups analyzed:
15-64, 65+, all ages
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson regression
with GAM
Covariates: Long-term trend,
season, influenza, day of wk,
holiday, temperature,
humidity, thunderstorms
Season:
Cool, Oct-Mar;
Warm: Apr-Sep
Dose-Response
Investigated?: Yes
Statistical package: S-Plus
Lag: 0,1, 2, 3, 4 days, 0-1
cum. avg.
24-h avg SO2
(jg/m3
Hong Kong:
Mean: 17.7
Warm: 18.3
Cool: 17.2
SD: 12.3
Range: 1.1, 90.0
10th
50th
90th
6.2
14.5
32.8
London:
Mean: 23.7
Warm: 22.2
Cool: 25.3
SD: 12.3
Range: 6.2,113.6
10th
50th
90th
13.2
20.6
38.1
Copollutants:
Hong Kong :
PM10 (r = 0.30)
NO2 (r = 0.37)
03 (r = -0.18)
London:
PM10 (r = 0.64)
N02 (r = 0.71)
03 (r = -0.25)
Similar non-statistically significant associations between asthma hospital
admissions and SO2 were found in both cities. The association between
respiratory hospital admissions and SO2 showed significance in the cold
season in Hong Kong and on an all yr basis. Respiratory hospital admissions
were not significantly associated with SO2 in Britain.
In the 2-pollutant model the association between respiratory hospital
admission and SO2 in London was insignificant, and remained insignificant
after adjusted for the second pollutants.
In Hong Kong, the positive association of SO2 was most affected by NO2,
losing statistical significance. The positive association remained robust when
adjusted for O3, and a slight decrease in association after adjusted for PM2.5.
Increment: 10 |jg/m3
Asthma, 15-64 yrs
Hong Kong: ER -0.1 (-2.4, 2.2) lag 0-1; ER -1.5 (-3.4, 0.5) lag;
Warm: ER 1.5 (-1.5, 4.6) lag 0-1; Cool: ER -2.0 (-5.4,1.4) lag 0-1
London: ER 0.7 (-1.0, 2.5) lag 0-1; ER 2.1 (0.7, 3.6) lag 3
Warm: ER -1.4 (-4.7,1.9) lag 0-1; Cool: ER 1.6 (-0.5, 3.8) lag 0-1
Respiratory 65+ yrs
Hong Kong: ER 1.8 (0.9, 2.6) lag 0-1; ER 1.7 (1.0, 2.4) lag 0
Warm: ER 1.1 (0.0, 2.2) lag 0-1; Cool: ER 2.7 (1.4, 4.0) lag 0-1
+03 ER 1.9 (1.1, 2.8) lag 0-1; +PM2.5 ER 1.2 (0.3, 2.2) lag 0-1;
+NO2ERO.3 (-0.7,1.4) lag 0-1
London: ER 0.2 (-0.6,1.1) lag 0-1; ER 1.2 (0.5, 2.0) lag 3;
Warm: ER 1.3 (-0.5, 3.1) lag 0-1; Cool: ER -0.3 (-1.3, 0.8) lag 0-1
+03 ER 0.5 (-0.4,1.5) lag 0-1; +PM2.5 ER 1.2 (0.3, 2.2) lag 0-1
+NO2ERO.5 (-0.7,1.7)lag 0-1
Yang and Chen (2007)
Taipei, Taiwan
Period of Study:
1996-2003
Hospital Admissions
Outcome(s) (ICD9): COPD
(490-492, 494, and 496)
identified by records from the
National Health Insurance
(NHI) program
Study design: Case-crossover
Statistical Analysis:
Conditional logistic regression
Statistical package: SAS
Age groups analyzed: All
ages
Covariates: Adjustments for
weather variables, day of the
wk, seasonality, and long-term
time trends
N: 46,491
# Hospitals: 47
Lag: Cumulative lag up to 2
days
24-h avg (ppb):
4.33
Range: 0.15,
17.82
25th
50th
75th
2.67
3.90
5.46
Copollutants:
PM10NO2CO 03
In single-pollutant models, all pollutants, except SO2, significantly associated
with COPD hospital admissions on warm days (>20 °C). On cold days (< 20
°C), only SO2 was significantly associated with COPD hospital admissions. In
multi-pollutant models, NO2 and O3 were significantly associated with each
pollutant on warm days.
Odds Ratio (95% CI),
Single-pollutant model (per 2.79 ppb SO2): a 20 °C: 1.006 (0.970,1.043);
< 20 °C: 1.071 (1.015,1.129)
Odds Ratio (95% CI),
Co-pollutant model (per 2.79 ppb):
>20°C:
SO2 + PM10: 0.909 (0.872,0.949)
SO2 + NO2: 0.835(0.798, 0.873)
SO2 +CO: 0.920(0.884, 0.958)
S02 + 03: 0.978(0.943,1.015)
< 20 °C:
SO2 + PM10:1.067 (0.997,1.141)
SO2 + NO21147 (1.072,1.227)
SO2 + CO: 1.140 (1.066,1.219)
SO2 + O3:1.064(1.009,1.123)
"Default GAM
APHEA: Air Pollution and Health: a European Approach
F-58
-------
Table F-3. Short-term exposure to SO2 and cardiovascular morbidity in field/panel studies.
Study
Methods
Pollutant Data
Findings
UNITED STATES
Dockery et al.
(2005)
Boston, MA
Period of Study:
Jul 1995-Jul 2002
Cohort study of 203 cardiac patients with
implanted cardioverter defibrillators.
Patients were followed for an avg of 3.1 yrs
from 1995-2002 to assess the role of air
pollution on the incidence ofventricular
arrhythmias. The association of arrhythmic
episode-days and air pollutions analyzed
with logistic regression using GEE with
random effects. Model adjusted for patient,
season, min temperature, mean humidity,
day of the wk, and previous arrhythmia
within 3 days. Only effects of 2-day running
mean of air pollution concentration reported.
48-h avg SO2;
Median: 4.9 ppb
25th%: 3.3 ppb
75%: 7.4 ppb
95%: 12.8 ppb
Copollutants:
PM2.5 BC S042"
PN NO2CO 03
No statistically significant association between any of the air pollutant and
ventricular arrhythmias when all events were considered. However,
ventricular arrhythmias within 3 days of a prior event were statistically
significant with SO2, PM2.5, BC, NO2, CO, and marginally with SO42-, but
not with O3 or PN. CO, NO2, BC, and PM2.5 correlated, thus it was
impossible to differentiate the independent effects. Since the increased
risk ofventricular tachyarrhythmia was associated with air pollution
observed among patients with a recent tachyarrhythmia, it was suggested
that air pollution acts in combination with cardiac electrical instability to
increase risk of arrhythmia.
For IQR (4.0 ppb) increase in 48-h mean SO2
All events: OR = 1.04(0.94,1.14), p = 0.28
Prior arrhythmia event: < 3 Days: 1.30 (95% CI: 1.06,1.61), p = 0.013
Prior arrhythmia event: >3 Days: 0.98 (0.87,1.11) p = 0.78
Gold et al. (2000)
Boston, MA
Period of Study:
June-Sep 1997
Panel study on 21 active Boston residents
aged 53-87 yrs to investigate the
association between short-term changes in
ambient air pollution and short-term
changes in cardiovascular function.
Participants observed up to 12 times from
June to Sep 1997 (163 observations made
in total). Protocol involved 25 mins per wk of
continuous ECG monitoring, that included
5 mins of rest, 5 mins of standing, 5 mins of
exercise outdoors, 5 mins of recovery, and
20 cycles of slow breathing. Fixed effects
models adjusted for time-varying covariates
and individuals traits.
24-h avg 3.2 ppb
Range: 0,12.6
ppb
IQR: 3.0 ppb
Copollutants:
PM2.5 PM10-2.5
03 NO2CO
In single-pollutant models, 24-h mean SO2 associated with reduced heart
rate in the first rest period but not overall. Associations weaker for shorter
averaging periods. Association between SO2 and heart rate not significant
with the multipollutant model (SO2 and PM2.5). SO2 not associated with
r-MSSD.
Heart rate, first rest period, mean 66.3 bpm
Single-pollutant model: Estimated effect (SE) -1.0 (0.5);
% mean 1.5, p = 0.03
Heart rate, first rest period, mean 66.3 bpm
Multipollutant model (PM2.5 and SO2):
SO2 estimated effect (SE) -0.8 (0.5); % mean 1.2, p = 0.09
PM2.5 estimated effect (SE) -1.6 (0.7); % mean 2.5, p = 0.03
Overall heart rate, mean 74.9 bpm
Single-pollutant model: Estimated effect (SE) -0.5 (0.5), p = 0.30
Multipollutant model: SO2 estimated effect (SE) -0.2 (0.5), p = 0.6 PM2.5;
Estimated effect (SE) -1.9 (0.7), p = 0.01% mean 2.6
Liao et al. (2004)
Three locations in
U.S.: Forsyth
County, NC;
Jackson, MS;
Minneapolis, MN
Period of Study:
1996-1998
Cross-sectional study of 6,784 cohort
members of the Atherosclerosis Risk in
Communities Study. Participants were 45-64
yrs of age; baseline clinical examinations
conducted from 1987-1989. HRV data
collected from 1996-1998. Air pollutants
obtained form EPA AIRS for this same
period. Resting, supine, 5-min beat-to-beat
RR interval data were collected over a 4-h
period. Multivariable linear regression
models used to assess associations
between pollutants measured 1-3 days prior
to HRV measurements. Models controlled
forage, ethnicity-center, sex, education,
current smoking, BMI, heart rate, use of
cardiovascular medication, hypertension,
prevalent coronary heart disease, and
diabetes.
Mean (SD) SO2 Significant interaction between SO2 and prevalence of coronary heart
measured 1 day disease for low-frequency power analyses. SO2 inversely associated with
prior to HRV SD of normal R-R intervals and low-frequency power and positively
measurement associated with heart rate. SO2 association with low-frequency power
was 4 (4) ppb stronger among those with history of coronary heart disease. Effect size
Copollutants' 'D'V'10 'ar9er ^an ^or 9aseous pollutants.
PM10 O3 Log-transformed low-frequency power effect estimate and SE per 1 SD
CO NO2 increment (4 ppb) SO2 lag 1 day:
Log transformed high-frequency power -0.024 (SE 0.016)
Standard deviation of normal R-R intervals -0.532 (SE 0.270), p < 0.05
Heart rate: 0.295 (SE 0.130), p < 0.05
Prevalent CHD: -0.122 (SE 0.056), p < 0.01
No prevalent CHD -0.012 (SE 0.016)
F-59
-------
Study
Methods
Pollutant Data
Findings
Liao et al. (2005)
United States
Period of Study:
1996-1998
Cross-sectional survey 10,208 participants
(avg age 54 yrs) from Atherosclerosis Risk
in Communities (ARIC) study cohort to
assess the association between criteria air
pollutants and hemostatic and inflammatory
markers. 57% of participants were female
and 66% male. Used hemostatis/ inflamma-
tion variables collected during the baseline
examination and air pollution data 1 -3 days
prior to the event. Used multiple linear
regression models that controlled for age,
sex, ethnicity-center, education, smoking,
drinking status, BMI, history of chronic
respiratory disease, humidity, seasons,
cloud cover, and temperature. Also history
of CVD and diabetes if not effect modifier in
a particular model.
SO2 mean (SD)
0.0005 (0.004)
ppm
Q1-3: 0.005
(0.003) ppm
Q4: 0.006
(0.005) ppm
Copollutants:
PM10CO N02 03
Significant curvilinear association between SO2 witfactor Vlll-C, WBC,
and serum albumin. Curvilinear association indicated threshold effect
Results shown in graph.
Luttmann-Gibson
et al. (2006)
Steubenville, OH
Period of Study:
2000
Conducted a panel study during the summer 24-h avg (ppb):
and fall of 2000, which consisted of 32
subjects 54-90 yrs old living in Steubenville,
OH. Used linear mixed models, fixed effects
of pollution, age, gender, race, obesity,
season, time of day, apparent temperature,
and a first order autoregressive process for
within-subject residuals to examine the
relation between air pollution and log-
transformed HRV parameters and heart
rate.
4.1
Copollutants:
PM2.5 SO42"
EC N02
03
Increasing concentrations of PM2.5 and SO42- in the previous day were
both found to be associated with reduced HRV. No association was
observed between increasing SO2 concentrations in the previous day and
HRV.
% Change (95% CI) (per 4.3 ppb SO2)
Standard Deviation of Normal RR Intervals: (SSDN) 0.7 (-1.0, 2.5)
Differences Between Adjacent RR Intervals: (r-MSSD) 0.5 (
High-Frequency Power (HF) 1.7 (-4.9, 8.7);
Low-Frequency Power (LF) 4.9 (-1.4,11.5);
Heart Rate (HR) 0.3 (-0.2, 0.8)
-2.8, 4.0)
Metzgeret al.
(2007)
Atlanta, GA
Period of Study:
1993-2002
Collected information on 518 patients (6287 15.5 ppb (±16.4) Little evidence of associations between ambient air quality measurements
event-days) for ventricular tachyarrhythemic
events over 10-yr period. Used GEE
analysis, a case-crossover analysis, and a
sensitivity analysis stratified on subject
Copollutants:
PM10O3
NO2CO PM2.5
and ventricular tachyarrhythmic events.
Odds ratio (95% CI)
All events: 1.002 (0.968-1.037)
Events resulting in cardiac pacing or defibrillation: 0.988 (0.936-1.042)
Events resulting in defibrillation: 1.004 (0.911-1.105)
Primary GEE model: 1.002 (0.968-1.037)
Controlling formin temperature: 1.010 (0.976-1.046)
Using an unconstrained distributed Lag: 0.996 (0.952-1.083)
Warm Season: 1.029 (0.989-1.116)
Cold Season: 0.986 (0.956-1.023)
Greater Boston
area, MA
Period of Study
Nov 2000-Oct
2003
Park et al. (2005b) Cross-sectional study of effect of ambient
air pollutants on heart rate variability (HRV)
in 497 men who were in the Normative
Aging Study and examined from Nov 2000
and Oct 2003. HRV measured between
0600 and 1300 h after resting for 5 mins. 4-
h, 24-h, and 48-h moving avgs of air pollu-
tion matched to time of ECG measurement.
Linear regression models included: age,
BMI, fasting blood glucose, cigarette
smoking, use of cardiac medications, room
temp, season, and the lagged moving avg of
apparent temp corresponding to the moving
avg period for the air pollutant. Mean arterial
blood pressure (MAP) and apparent
temperature also included. Assessed
modifying effects of hypertension, IHD,
diabetes or use of cardiac/antihypertensive
meds.
24-h avg SO2
4.9 ppb
SD: 3.4
Range: 0.95,
24.7 ppb
Copollutants:
PM2.5 PNC
BC N02
03 CO
No significant association between HRV and SO2 for any of the averaging
periods, but positive relationship.
4-h moving avg SO2: (per 1 SD, 3.4 ppb SO2)
Log10 SDNN: 2.3 (-1.7, 6.4)
Log10 HF:5.6 (-4.9,17.3)
Log10 LF: 2.2 (-5.9,11.1)
Log10 (LF:HF): -3.2 (-10.1,4.2)
F-60
-------
Study
Methods
Pollutant Data
Findings
Peters et al.
(2000a)
Eastern Massa-
chusetts, U.S.
Period of Study:
1995-1997
Pilot study to test hypothesis that patients
with implanted cardioverter defibrillators
would experience potentially life-threatening
arrhythmias associated with air pollution
episodes. Records detected arrhythmias
and therapeutic interventions downloaded
from the implanted defibrillator. Mean age of
patients 62.2 yrs. 100 patients followed for
over 3 yrs for 63,628 person-days. 33
patients with any discharges and 6 patients
with 10 or more events. Data analyzed by
logistic regression models using fixed
effects models with individual intercepts for
each patient. Model controlled for trend,
season, meteorologic conditions, and day of
week. Evaluated air pollutants on same day,
lags 1, 2, and 3 days, and 5-day mean.
24-h avg SO2:
7 ppb
Median: 5 ppb
Max: 87 ppb
Copollutants:
PM10 PM2.5
BC CO
O3 NO2
No association between increased defibrillator discharges and SO2.
33 patients with at least 1 defibrillator discharge
Odds Ratio (95% CI)
Lag 0 0.76 (0.48,1.21); Lag 1 0.91 (0.60, 1.37)
Lag 2 0.89 (0.59,1.34); Lag 3 1.09 (0.78,1.52)
5-day mean 0.85 (0.50,1.43)
6 patients with at least 10 discharges
Odds Ratio (95% CI)
Lag 0 0.72 (0.40,1.31); Lag 1 0.77 (0.44, 1.37)
Lag 2 1.01 (0.63,1.61); Lag 3 1.08 (0.72, 1.62)
5-day mean 0.75 (0.38,1.47)
Peters et al. (2001) Case cross over Study design used to
investigate association between air pollution
and risk of acute myocardial infarctions in
772 patients (mean age 61.6 yrs) with Ml as
part of the determinants of myocardial
infarction onset study. For each subject, one
case period was matched to 3 control
periods, 24 h apart. Used conditional logistic
regression models that controlled for
season, day of wk, temperature, and relative
humidity.
Greater Boston
area, MA
Period of Study:
Jan 1995-May
1996
24-h avg SO2:
7 ppb
SD: 7 ppb
1-h avg SO2:
7 ppb
SD: 10 ppb
Copollutants:
PM2.5, PM10,
PM10-2.5, BC
03, CO, N02
SO2 not statistically associated with risk of onset of Ml. Limitation of study
is only 1 air pollution monitoring site available.
OR for 2-h avg SO2 and 24-h avg SO2 estimated jointly:
2 h per 2 ppb increase SO2
Unadjusted: 1.00 (0.87,1.14)
Adjusted: 0.96 (0.83,1.12)
24 h per 2 ppb increase
Unadjusted: 0.92 (0.71,1.20)
Adjusted: 0.91 (0.67,1.23)
Rich et al. (2005)
Boston, MA
Period of Study:
Jul 1995-Jul 2002
Case cross-over design used to evaluate
association between ventricular arrhythmias
detected by implantable cardioverter
defibrillators and air pollution. Same study
population as Dockery et al. (2005): 203
patients with ICD and residential zip codes
within 40 km of central particle monitoring
site. Analyses conducted on 84 subjects
with confirmed ventricular arrhythmias
during the follow-up. Case periods defined
by time of each confirmed arrhythmic event.
Control periods (3-4 per case) selected by
matching on weekday and hour of the day
within the same calendar mo. Used
conditional logistic regression that controlled
for temperature, dew point, barometric
pressure, and a frailty term for each subject.
ORs presented for IQR increase in mean
concentration and averaging time. Moving
avg of concentrations considered: lags 0-2,
0-6, 0-23, and 0-47 h.
1-h avg SO2:
Median: 4.3 ppb
25th %: 2.6
75th %: 7.5
Max: 71.6
24-h avg SO2:
Median: 4.8
25th %: 3.2
75th %: 7.3
Max: 31.4
Copollutants:
PM2.5 BC
N02 CO 03
An IQR increase in the 24-h moving avg SO2 (4.1 ppb) marginally
associated with a 9% increased risk of ventricular arrhythmia and an
increased risk with 48-h moving avg. There was no risk associated with
24-h moving avg after controlling for PM2.5 cases that had a prior
ventricular arrhythmia within 72-h had greater risk associated with SO2
compared to those without a recent event, suggesting that risk is greater
among cases with more irritable or unstable myocardium.
Odds ratios
Single-pollutant model: 0-2-h lag (per 4.7 ppb) 1.07 (0.97,1.18);
0-6-h lag (per 4.5 ppb) 1.09 (0.98, 1.20);
0-23-h lag (per 4.1 ppb) 1.09 (0.97,1.22);
0-47-h lag (per 4.0 ppb) 1.17 (1.02,1.34)
2-pollutant model: SO2 and PM2.5 per 4.1 ppb SO2:1.00 (0.84,1.20);
S02 and 03 per 4.1 ppb S02:1.12 (0.99,1.27)
Per 4.1 ppb increase SO2
Prior arrhythmia event < 3 Days: 1.20 (1.01,1.44)
Prior arrhythmia event >3 Days: 0.96 (0.83,1.10)
Rich et al. (2006b)
Boston, MA
Period of Study:
June 1995-
December 2002
Case-crossover study consisting of 203
individuals with implantable cardioverter
defibrillators (ICDs) implanted between Jun
1995 and Dec 1999. Used conditional
logistic regression, which included variables
for mean pollutant concentration in the hour
of the arrhythmia, and natural splines for
mean temperature, dew point, and
barometric pressure in the 24 h before the
arrhythmia. The regression analyses were
run for each pollutant individually to
examine the association between increasing
pollutant levels and paroxysmal atrial
fibrillation episodes (PAF).
Max 24-h avg
(ppb): 31.4
Max 1-h max
(ppb): 71.6
Copollutants:
PM2.5BC
NO2CO 03
O3 was significantly associated with PAF in the hour preceding the
arrhythmia, but the effect was not significant when analyzing the
preceding 24-h. Increasing levels of PM2.5, NO2, and BC resulted in non-
significant positive associations with PAF. SO2 was not associated with
PAF.
Odds Ratio (per 4.9 ppb SO2): 1.02 (0.81,1.28) lag 0 h
Odds Ratio (per 4.1 ppb SO2): 0.99 (0.71,1.39) lag 0-23-h
F-61
-------
Study
Methods
Pollutant Data
Findings
Rich et al. (2006a)
St. Louis, Missouri
Period of Study:
5/9/2001-
12/31/2002
Case-crossover design study of 56 patients
with implantable cardioverter defibrillators.
Subjects ranged from 20 to 88 yrs (mean
63). Case period defined by time of
confirmed ventricular arrhythmia. Control
periods matched on weekday and hour of
the day within the same calendar mo. Used
conditional logistic regression model that
included mean of the previous 24-h
temperature, relative humidity, barometric
pressure, mean pollutant concentration in
the 24 h before the arrhythmia. Model also
included a frailty term for each subject.
599 days
25"1 %: 2 ppb
50"1 %: 4 ppb
75"1 %: 7 ppb
Daily IQR: 5 ppb
Case/control IQR:
5 ppb
Copollutants:
PM2.5, EC, OC,
N02, CO, 03
Statistically significant increase in risk of ventricular arrhythmias
associated with each 5 ppm increase in 24-h moving avg SO2.
OR for ventricular arrhythmia associated with IQR increase
6-h moving avg SO2 per 4 ppb: 1.04 (95% CI: 0.96,1.12)
12-h moving avg SO2 per 5 ppb: 1.17 (95% CI: 1.04,1.30)
24-h moving avg SO2 per 5 ppb: 1.24 (95% CI: 1.07,1.44)
48-h moving avg SO2 per 4 ppb: 1.15 (95% CI: 1.00,1.34)
Sarnat et al. (2006) Panel study consisting of 32, non-smoking
Steubenville, OH
Period of Study:
(June 4-Aug 18,
Sep 25-Dec 15)
2000
older adults approximately 53-90 yrs old.
Electrocardiograms (ECGs) and
questionnaires regarding symptoms were
administered on a weekly basis. Used a
logistic mixed effects regression to examine
the association between increasing air
pollutant concentrations and
supraventricular ectopy (SVE) and
ventricular ectopy (VE).
1-day avg SO2
24-h avg (ppb):
10.4(8.3)
Range: 1.8, 58.3
5-day moving avg
24-h avg (ppb):
10.7(5.5)
Range: 2.4, 31.3
Copollutants: 5-day moving avg:
PM2.5, EC, 03, SVE: 1.04(0.78,1.39)
N02,S042-,C0 VE: 1.28 (0.85, 1.92)
PM2.5 was significantly associated with SVE, whereas, SO42" and O3 were
marginally associated in models including 5-day moving avg pollutant
concentrations. However, no pollutants were found to be associated with
VE in similar models. Overall, subjects that reported previous
cardiovascular conditions (e.g., myocardial infarction and hypertension)
were found to be more susceptible to SVE due to increasing air pollutant
concentrations.
Odds Ratio (per 5.4 ppb SO2)
Schwartz et al.
(2005)
Boston, MA
Period of Study: 12
wks during the
summer of 1999
Panel study of 28 subjects (aged 61-89 yrs)
to examine association between
summertime air pollution and HRV. Subjects
examined once a wk up to 12 wks and HRV
measured for approximately 30 mins.
Analyses used hierarchical models that
controlled for baseline medical condition,
smoking history, day ofwk and hour of day,
indicator variable for whether subjects had
taken their medication before they came,
temperature and time trend.
24-h avg SOz
25th %:
0.017 ppm
50th %:
0.020 ppm
75th %: 0.54 ppm
Copollutants:
O3 NO2
CO PM2.5 BC
No significant association with SO2
Percentage change in HRV associated with IQR (0.523 ppm) increase in
S02
1-h avg S02: SDNN (ms) 0.4 (-1.3 to 2.1); RMSSD (ms) 1.4 (-2.6 to 5.5);
PNN50 (ms) 3.8 (-12.1 to 22.5)
24-h avg S02: SDNN (ms) 0.4 (-4.2 to 5.1);
RMSSD (ms) -0.3 (-1.3 to 0.8); PNN50 (%) -0.2 (20.9 to 17.6);
LFHFR 2.9 (-4.9 to 11.4)
Sullivan et al.
(2005)
King County,
Washington
Period of Study:
1988-1994
Case-crossover study of 5,793 confirmed
cases of acute Ml. Data was analyzed usinc
simple descriptive analyses and Pearson's
correlation coefficient.
9 ppb
Range: 0-38 ppb
Copollutants:
PM2.5 PM10CO
Increases in SO2 were not associated with Ml after adjusting for relative
humidity and temperature
Odds Ratio (95% CI) (per 10 ppb SO2)
Averaging time: 1-h: 0.97 (0.94,1.01); 2 h: 0.98 (0.95,1.01);
4 h: 0.99 (0.96,1.03); 24 h: 1.0 (0.95,1.06)
Wheeler et al.
(2006)
Atlanta, Georgia
Period of Study:
19992000
30 individuals with myocardial infarction or Mean: 1.9 ppb
COPD were administered a questionnaire Conollutants-
and an HRV protocol. Linear mixed-effect p^5 q3
models were used to analyze the data. qq '
No association with SO2
Rich etal. (2004)
Vancouver, British
Columbia, Canada
Period of Study:
Feb-Dec 2000
Case-crossover analysis used to investigate
association between air pollution and
cardiac arrhythmia in 34 patients (aged 15-
85 yrs, mean 62) with implantable
cardioverter defibrillators. Study included
only patients who experienced at least 1
ICD discharge during the study period.
Control days were 7 days before and 7 days
after day of ICD discharge. Conditional
logistic regression analyses were stratified
by individual.
24-h avg: 2.6 ppb
SD: 1.3 ppb
IQR: 1.6 ppb
Copollutants:
PM2.5, EC, OC
S042", PM10, CO
NO2,03
No statistically significant association between SO2 and implantable
cardioverter defibrillator discharges. However, when an analysis was
stratified by season, OR for SO2 were higher in the summer compared to
winter.
No quantitative results provided. Results shown in graph.
F-62
-------
Study
Methods
Pollutant Data
Findings
Vedal et al. (2004)
Vancouver, British
Columbia, Canada
Period of Study:
1997-2000
Retrospective, longitudinal panel study of
50 patients, aged 12-77 yrs with implantable
cardioverter defibrillators. Total of
40,328 person-days over 4-yr period. GEE
used to assess associations between short
term increases in air pollutants and
implantable cardioverter defibrillator
discharges. Models controlled for temporal
trends, meteorology, and serial
autocorrelation.
24-h mean (SD)
S02: 2.4 (1.2) ppb
Range: 0.3, 8.1
ppb
Median: 2.2 ppb
25th: 1.5
75th: 3.1
Copollutants:
PM10 O3
N02 CO
Concluded that in general no consistent effect of air pollution on cardiac
arrhythmias in this population. There were no statistically significant
associations between SO2 and cardiac arrhythmias at any lag day, but
positive associations at lag 2. When analysis was restricted to only
patients who had at least 2 arrhythmias per yr over their period of
observation (N: 16), a positive and significant association was seen with
SO2 at 2 days lag. When analysis was restricted to patients averaging 3
or more arrhythmias per yr (N: 13), there was no significant association,
but a positive association was seen at 2 days lag.
When stratified by season, SO2 effects were in the in the positive
direction in the winter, but in the negative direction in the summer.
Authors noted results may be due to chance because of multiple
comparisons or SO2 may be surrogate for some other factor.
Summer analysis: significant negative association with SO2 at lag days 2
and 3 (data not shown). When stratified to patients with 2 or more
arrhythmia event-days per yr, significant negative associations observed
with SO2 at lag of 3 days. Winter analysis: significant positive effect of
SO2 at 3 days lag (data not shown). If restricted to patients with at least 2
arrhythmias peryr, a significant positive association was seen at lags 2
and 3 days. When restricted to patients with 3 or more arrhythmia event
days peryr, positive associations observed for SO2 at lags of 2 and 3
days.
No quantitative results, but % change in arrhythmia event-day rate for
each SD increase in pollution concentration on log scale provided in
figures.
Berger et al.
(2000)
Erfurt, Germany
Period of Study:
Oct 2000-Apr 2001
Prospective panel study of 57 non-smoking 24-h avg (SD)
men, of which 74% are ex-smokers, with (|jg/m3): 4.1 (1.8)
coronary heart disease aged 52-76 yrs old. Ranae- 3 0 11 7
Subjects underwent 24-h electrocardiogram '
(ECG) recordings and analysis once every 4 Copollutants:
wks. Associations analyzed using Poisson Ultrafine Particles
and linear regression modeling, for Accumulation
supraventricular and ventricular tachycardia, Mode Particles
respectively, adjusting for trend, weekday, PM2.5 PM10
and meteorologic data. NO2CO NO
UFP, ACP, PM2.5, and NO2 associated with increased risk for
supraventricular tachycardia and ventricular tachycardia at almost all
lags. The majority of statistically significant associations was observed in
the previous 24-71 -h and with the 5-day moving avg. Associations were
not observed for increasing concentrations of SO2.
Relative Risk (per 1.5 |jg/m3 SO2)
Supraventricular Extrasystoles:
0.92 (0.77,1.09) lag 0. 0.98 (0.86,1.12) lag 0-23-h
1.04 (0.93,1.16) lag 24-47 h. 1.14 (0.98,1.34) lag 48-71-h
0.95 (0.83,1.09) lag 72-95-h. 1.01 (0.80, 1.27) lag 5-d avg
Ventricular Extrasystoles:
-2.1 (-6.1, 2.1) lag 0. -1.8 (-6.1, 2.7) lag 0 - 23-h
-0.1 (-4.4, 4.4) lag 24-47 h. 4.5 (-0.4, 9.5) lag 48-71-h
-2.2 (-6.4, 2.3) lag 72-95-h. -1.2 (-7.5, 5.5) lag 5-d avg
Henrotin et al.
Bi-directional case-crossover design to
Mean: 6.9 |jg/m3
SO2 not significantly associated with occurrence of strokes
(2007)
examine association between air pollutant
SD: 7.5
Odds Ratio (95% CI)
Ischaemic stroke:
Dijon, France
and ischaemic stroke onset
(2,078 cases).
Min: 0
Max: 65
Period of Study:
Copollutants:
SOx, O3, CO
PM10
DO: 0.978 (0.868,1.103). D-1: 0.978 (0.863,1.108)
1994-2004
D-2:1.015 (0.902,1.143). D-3: 1.003 (0.892,1.127)
Hemorrhagic stroke:
DO: 1.099 (0.815,1.483). D-1:1.014(0.747,1.376)
D-2: 0.961 (0.712,1.297). D-3: 0.954(0.729,1.248)
F-63
-------
Study
Methods
Pollutant Data
Findings
Ibald-Mulli et al.
(2001)
Augsburg,
Germany
Period of Study:
1984-85,1987-88
Retrospective analysis of 2,607 subjects
(25-64 yrs, subset of the participants of first
and second MONICA survey who had valid
electrocardiograms recordings in both
surveys and blood pressure
measurements). Used regression models
for repeated measures that controlled for
age, current smoking, and cardiovascular
medication, BMI, total and high density
lipoprotein cholesterol, temp, RH, and
barometric pressure.
24-h avg SO2
(pg/m3)
1984-1985:
Mean: 60.2
SD: 47.4
Range: 13.0,
238.2
follow up 1987-
1988
Mean: 23.8
SD: 12.3
Range: 5.6, 71.1
Copollutants:
TSP, CO
SO2 and TSP associated with increases in systolic blood pressure. In the
multipollutant model with TSP, the effect of TPS remained significant, but
the SO2 effect was substantially reduced. No clear association between
SO2 and CO and diastolic blood pressure was observed.
Same day concentrations: mean change in systolic blood pressure per
5th to 95th percentile increase in SO2 (per 80 (jg/m3)
Same day concentrations (per 80 (jg/m3):
Men (N: 1339): 0.96 (0.07,1.85); Women (N: 1268): 0.96 (-0.46,1.49);
Men and women: 0.74 (0.08,1.40)
5-day avgs; mean change in systolic blood pressure per 5th to 95th
percentile increase in SO2 (per 75 (jg/m3): Men: 0.97 (0.09,1.85);
Women: 1.23 (0.23, 2.22); Men and women: 1.07 (0.41,1.73)
2-pollutant model: Men and women: 0.23 (-0.50, 0.96)
Peters et al. (1999) Retrospective analysis on subsample of
Augsburg,
Germany
Period of Study:
Winter 1984-1985
Winter 1987-1988
2,681 subjects (25-64 yrs) of the MONICA
cohort who had valid electrocardiogram
readings from both surveys and no acute
infections. GEE for clusters used to assess
association between heart rate and air
pollution. Analyses adjusted for
temperature, relative humidity, and air
pressure.
24-h avg SO2
(pg/m3)
Winter 1984-85
Outside episode:
Mean: 48.1
SD: 23.1
Range: 13,103
Winter 1984-85
During episode:
Mean: 200.3
SD: 26.6
Range: 160, 238
Winter: 1987-88
Mean: 23.6
SD: 12.2
Range: 6, 71
Copollutants:
CO, TSP
Increases in SO2 concentrations associated with increases in heart rate
Mean change in heart rate per 5th to 95th percentile SO2
Same day concentrations (per 80 |jg/m3 SO2)
Men: 1.02 (0.41,1.63)
Women: 1.07 (0.41,1.73)
Men and women: 1.04 (0.60,1.49)
5-day avg (per 75 |jg/m3 SO2)
Men: 1.29 (0.68,1.90)
Women: 1.26 (0.57,1.95)
Men and women: 1.28 (0.82,1.74)
Ruidavets et al.
(2005)
Toulouse, France
Period of Study:
1995-1997
Cross-sectional survey of 863 randomly
chosen adults (35-65 yrs) living in Toulouse
(MONICA center) to examine the
relationship between resting heart rate and
air pollution. Resting heart rate was
measured twice in a sitting position after a
five minute rest. Used polytomous logistic
regression models with quintiles of RHR.
Final model controlled for sex, physical
activity, systolic blood pressure,
cardiovascular drug use, CRP, relative
humidity, and season mos.
Mean SO2:13.3
(7.5) |jg/m3
Range: 1.3,
47.7 (jg/m3
Copollutants:
NO2,03
Marginally significant association between SO2 and RHR in Q5 compared
with Q1. No associations with SO2 at 1, 2, or 3 days lag.
OR based on daily levels of SO2.
OR for resting heart rate = 1.19 (95% CI: 1.02,1.39) in 5th quintile (>75
bpm) compared to first quintile (< 60 bpm) for 5 |jg/m3 increase in SO2
same day 0 am-12 pm.
OR for resting heart rate 1.14 (95% CI: 1.01 to 1.30) in 5th quintile (>75
bpm) compared to first quintile (< 60 bpm) for 5(jg/m3 increase in SO2
same day 12 am-12 pm
Not-significant associations not listed
LATIN AMERICA
Holguin et al.
(2003)
Mexico City,
Mexico
Period of Study:
Feb 8 to Apr 30,
2000
Panel study of 34 nursing home residents
(60-96 yrs) to assess association between
heart rate variability and air pollution. Heart
24-h avg SO2
(PPb)
Mean: 24
rate variability measured every alternate day g^. ^
for 3 mos. Thirteen of the subjects had
hypertension. Used GEE models that
controlled for age and avg heart rate during
HRV measurement.
Range: 6, 85
Copollutants:
Indoor PM2.5
Outdoor PM2.5
03, N02, CO
SO2 not related to heart rate variability on the same day or lag 1 day
Change in HRV per 10 ppb
Beta Coefficient (95% CI)
HRV-HF -0.003 (-0.035, 0.035)
HRV-LF -0.004 (-0.004, 0.003)
HRV-LF/HF 0.012 (-0.060, 0.082)
F-64
-------
Table F-4. Short-term exposure to SO2 and emergency department visits and hospital
admissions for cardiovascular diseases.
Methods
Pollutant Data
Findings
UN TED STATES
Gwynn et al. (2000) Hospital Admissions
New York (Buffalo; Outcome(s) (ICD9): Respiratory (466, 480-486),
Rochester) Circulatory (401-405, 410-417), Total (minus 800)
Period of Study: ^om Stat,ewldce p'an™9 £^search
May 1988-Oct 1990 c°°Peratlve sVstem (SPARCS)
Study design: Time-series
Statistical Analysis: Loess fits of temperature and
relative humidity
Age groups analyzed: all ages
Covariates: adjustments for weather
Lag(s): 0, 3
24-h avg (ppb): 12.2
Range: 1.63, 37.7
Copollutants:
H*S042" PM10
Filled PM10
OsCO NO2C0H
p = 0.000245 (0.000917)
t = 0.27
Relative Risk (per 7 ppb SO2)
1.002
Koken et al. (2003) Outcome(s) (ICD9):
Denver, U.S.
Period of Study:
Jul and Aug,
1993-1997
N: 310 days
Acute Ml 410.00-410.92;
Atherosclerosis 14.00-414.05;
Pulmonary Heart Failure 416.0-416.9; Dysrhythmia
427.0-427.9; CHF 428.0.
Discharge data from Agency for Healthcare
Research and Quality (AHRQ) database.
Age group analyzed: 65+ yrs
Study population: 60,000
Covariates: seasonal adjustment not needed.
Adjustment for temperature, dew point temperature
made.
Study design: Time-series
Statistical analysis: GLMs to analyze frequency of
admissions as a function of exposure. GEEs to
estimate parameters in Poisson regression models,
adjusting foroverdispersion.
Lag(s): 0-4 day
SO2 24-h avg (ppb)
Mean (SD): 5.7 (2.94)
Min: 0.4
25th: 3.8
50th: 5.3
75th: 7.2
Max: 18.9
Copollutants:
03 (r = -0.10)
CO (r = 0.21)
PM10 (r = 0.36)
N02 (r = 0.46)
Effects were reported as percent change in
hospitalizations based on an increment of 3.4 ppb.
Single-pollutant model
Dysrhythmia
8.9% (—0.34,18.93) lag 0, adjusted for gender but
not temperature
SO2 was found to be associated with cardiac
dysrhythmia but not other outcomes. No association
was observed for PM or NO2 with the outcomes.
Low et al. (2006)
New York City, NY
Period of
Study:1995-2003
N: 3,287 days
Outcome(s) (ICD): Ischemic stroke 433-434;
Undetermined stroke 436; monitored intake in 11-
hospitals (ER or clinic visits). Excluded stroke
patients admitted for rehabilitation.
Study design: Time-series
Statistical Analysis: Autoregressive integrated
moving avg (ARIMA) models
Software package: SAS
SO2 24-h avg (ppm)
Mean (SD): 0.01098
(0.009124)
Min: 0
25th: 0.005
Median: 0.009
75th: 0.014
Max: 0.096
Copollutants::
PM10 (0.042)
N02 (0.33)
CO (0.303)
Pollen (0.085)
At the highest concentration of SO2 (96 ppb) in New
York city over the study period the expected increase
in strokes would be 0.857 visits on the day of the
event.
Each 1000 ppb (1 ppm) SO2 would produce an
additional 8.878 visits (SE 4.471)
(p = 0.0471) for stroke.
F-65
-------
Study
Methods
Pollutant Data
Findings
Metzgeret al.
(2004)
Atlanta, GA
Period of Study:
Jan 1993-Aug 31
2000
N: 4 yrs
Outcome(s): IHD 410-414; AMI 410; Dysrhythmias
427; cardiac arrest 427.5; congestive heart failure
428; peripheral and cerebrovascular disease 433-
437, 440,
443-444, 451-453; atherosclerosis 440; stroke 436.
ED visits from billing records.
N: 4,407,535 visits, 37 CVD visits/days
# Hospitals: 31
Age groups analyzed: adults a 19, elderly 56+
Statistical Analysis: Poisson regression, GLM.
Sensitivity analyses using GEE and GAM (strict
convergence criteria)
Covariates: long-term trends, mean and dew point
temp, relative humidity (cubic splines)
Statistical Software: SAS
Season: Warm (Apr 15-Oct 14), Cool (Oct 15-Apr
14)
Lag(s): 0-3 days
SO21-h max (ppb)
Median: 11.0
10th-90th Range: 2.0 to 39 ppb
Copollutants:
PM10 (0.20)
03 (0.19)
N02 (0.34)
CO (0.26)
PM2.5 (0.17)
PM10-2.5 (0.21)
Ultrafine (0.24)
Multipollutant models used. All
models specified a priori.
Results presented for RR of an incremental increase
in S02of 20 ppb (a priori lag 3 day moving avg).
All CVD: 1.007 (0.993,1.022)
Dysrhythmia: 1.001 (0.975,1.028)
CHF: 0.992 (0.961,1.025)
IHD: 1.007 (0.981,1.033)
PERI: 1.028 (0.999,1.059)
Finger wounds 1.007 (0.998,1.026)
Single day lag models presented graphically.
No multipollutant models run for SO2 since
association was not observed in single-pollutant
models.
Michaud et al.
(2004)
Hilo, Hawaii
Period of Study:
1997-2001
N: 1385 days
Outcome(s) (ICD9):
Cardiac 410-414, 425-429,
Emergency visits, primary diagnosis.
Study design: Time-series
Statistical Analysis: Exponential regression,
autocorrelation assessed by regressing square root
of number of ED visits on covariates (Durbin-
Watson statistic). Newey-West procedure also
conducted for assessment of autocorrelation.
Covariates: Temperature, humidity, interaction
between SO2 and PM
SO2 (all hourly measurements)
(PPb)
Mean (SD): 1.92 (12.2)
Min: 0
Max: 447
Daily SO2
(12am-6am) (ppb)
Mean (SD): 1.97 (7.12)
Min: 0
Max: 108.5
Effects were presented as relative risk based on an
increment of 10 ppb and the 24-h avg SO2
concentration.
Cardiac
0.92 (0.85,1.00) lag 3
No associations of cardiac ER visits with vog (SO2-
acidic aerosols) observed.
Lag(s): 1-3 days
Copollutants: PM
Moolgavkar (2000a)
Cook County IL,
Los Angeles
County, CA,
Maricopa County,
AZ
Period of Study:
1987-1995
Outcome(s) (ICD9): CVD
390-429; Cerebrovascular disease 430-448.
Hospital admissions from CA department of health
database.
Age groups analyzed: 20-64, 65+ yrs
Study design: Time-series
N: 118 CVD admissions/days
# Hospitals: NR
Statistical analysis: Poisson regression, GAM
Covariates: adjustment for day of wk, long-term
temporal trends, relative humidity, temperature
Statistical package: SPLUS
Lag: 0-5 days
SO2 24-h avg (ppb)
Cook County:
Min: 0.5; Q1: 4
Median: 6; Q3: 8
Max: 36
LA County:
Min: 0; Q1:1
Median: 2; Q3: 4
Max: 16
Maricopa County:
Min: 0; Q1: 0.5
Median: 2; Q3: 4
Max: 14
Copollutants:
PM10 (0.11,0.42)
PM2.5 (0.42) (LA only)
CO (0.35, 0.78)
N02 (0.02, 0.74)
03 (-0.37,0.01)
Results reported for percent change in hospital
admissions per 10 ppb increase in SO2. T statistic in
parentheses.
CVD, 65+:
Cook County: 4.0 (6.1), lag 0
3.1 (4.5), lag 0, 2-pollutant model (CO)
1.0(1.4), lag 0, 2-pollutant model (NO2)
LA County: 14.4 (15.2), lag 0
-2.5 (-1.6), lag 0, 2-pollutant model (CO)
7.7 (5.7), lag 0, 2-pollutant model (NO2)
Maricopa County: 7.4 (4.5), lag 0
3.0 (1.8), lag 0, 2-pollutant model (CO)
3.9 (1.5), lag 0, 2-pollutant model (SO2)
Cerebrovascular Disease, 65+:
Cook County: 3.1 (3.3)
LA County: 6.5 (4.9)
Lags 1-5 also presented. Effect size generally
diminished with increasing lag time. Increase in
hospital admissions (10.3 for CVD and 9.0 for
cerebrovascular) also observed for the
20-64 age group.
F-66
-------
Study
Methods
Pollutant Data
Findings
Moolgavkar (2003b)
Cook County IL,
Los Angeles
County, CA,
Maricopa County,
AZ
Period of Study:
1987-1995
Outcome(s) (ICD9): CVD See original analysis
390-429; Cerebrovascular disease 430-448 was not (Moolgavkar, 2000) above,
considered in the reanalysis. Hospital admissions
from CA department of health database.
Age groups analyzed: 20-64, 65+ yrs
Study design: Time-series
N: 118 CVD admissions/day
# Hospitals: NR
Statistical analysis: Poisson regression, GAM with
strict convergence criteria (10-8), GLM using
natural splines
Covariates: adjustment for day of wk, long-term
temporal trends, relative humidity, temperature
Statistical package: SPLUS
Lag: 0-5 days
Use of stringent criteria in GAM did not alter results
substantially. However, increased smoothing of
temporal trends attenuated results for all gases and
effect size diminished with increasing lag.
Results reported for incremental increase of 10 ppb
SO2. Estimated coefficient and T statistic in
parentheses.
GLM with 100 df (LA County)
13.67 (11.82), lag 0
6.44 (5.23), lag 1
0.23 (0.18), lag 2
Morris et al. (1995)
U.S. (Chicago,
Detroit, Houston,
LA, Milwaukee,
NYC, Philadelphia)
Period of Study:
1986-1989
N: 4 yrs
Outcome(s) (ICD9): CHF 428. Daily Medicare
hospital admission records.
Study design: Time-series
Statistical analyses: GLM, negative binomial
distribution
Age groups analyzed: a 65 yrs
Covariates: temperature, indicator variables for mo
to adjust for weather effects and seasonal trends,
day ofwk, yr
Statistical software: S-PLUS
Lag(s): 0-7 days
SO21-h max (ppm)
Mean (SD):
LA: 0.010 (0.005)
Chicago: 0.025 (0.011)
Philadelphia: 0.029 (0.015)
New York: 0.032 (0.015)
Detroit: 0.025 (0.013)
Houston: 0.018 (0.009)
Milwaukee: 0.017 (0.013)
Copollutants:
N02 03 CO
Correlations of SO2 with other
pollutants strong.
Multipollutant models run.
Results reported for RR of admission for CHF
associated with an incremental increase in SO2 of
0.05 ppm.
CHF:
LA: 1.60 (1.41,1.82)
Chicago: 1.05(1.00,1.10)
Philadelphia: 1.01 (0.96,1.06)
New York: 1.04(1.01,1.08)
Detroit: 1.00 (0.95,1.06)
Houston: 1.07 (0.97,1.17)
Milwaukee: 1.07 (0.99,1.15)
RR diminished in multipollutant (4 copollutants)
models for all cities.
Peel et al. (2007)
Atlanta, GA
Period of Study:
Jan 1993-Aug 2000
Outcome(s) (ICD9): IHD 410-414; dysrhythmia 427;
CHF 428; peripheral vascular and cerebrovascular
disease
433-437, 440, 443, 444, 451-453. Computerized
billing records for ED visits.
Comorbid conditions: Hypertension
401-405; diabetes 250; dysrhythmia 427, CHF 428;
atherosclerosis 440; COPD 491, 492, 496;
pneumonia 480-486; upper respiratory infection
460-465, 466.0; asthma 493, 786.09.
# Hospitals: 31
N: 4,407,535 visits
Study design: case-crossover and time-series. CVD
outcomes among susceptible groups with comorbid
conditions.
Statistical analyses: Conditional logistic regression
and Poisson GLM.
Covariates: cubic splines for temperature and
humidity included in models. Time independent
variables controlled through design.
Statistical Software: SAS
Lag(s): 3 day avg, lagged 0-2 day
SO21-h max (ppb)
Mean (SD): 16.5 (17.1)
10th: 2
90th: 39
Copollutants:
PM10 03 NO2CO
Results expressed as OR for association of CVD
admissions with a 20 ppb incremental increase in
S02.
Case-Crossover:
All CVD: 1.009 (0.995,1.024), 0-2 avg
IHD: 1.013 (0.988,1.039), 0-2 avg
Dysrhythmia: 1.003 (0.975,1.031), 0-2 avg
Peripheral and Cerebrovascular: 1.024 (0.993,
1.055), 0-2 avg
CHF: 0.993 (0.961,1.026), 0-2 avg
Time-series:
Odds Ratio (95% CI) (per 20 ppb SO2)
All cardiovascular disease: 1.007 (0.993,1.022)
Ischemic heart disease: 1.007 (0.981,1.003)
Dysrhythmia: 1.001 (0.975,1.028)
Peripheral and cerebrovascular disease: 1.028
(0.999,1.059)
Congestive heart failure:0.992 (0.961,1.025)
Effect modification by comorbid conditions was not
observed.
F-67
-------
Study
Methods
Pollutant Data
Findings
Schwartz and
Morris (1995)
Detroit, Ml
Period of Study:
1986-1989
Outcome(s) (ICD9): IHD 410-414; CHF 428;
Dysrhythmia 427. Medicare data, diagnosis at
discharge.
Study design: Time-series
Statistical analysis: Poisson regression, GAM
Age groups analyzed: 65+ yrs
Covariates: adjustments for long-term patterns,
temperature, humidity, days of the wk, holidays,
viral infections, etc.
Lag(s): 0-3, cumulative up to 3 days
SO2 24-h avg (ppb):
Mean: 25.4
IQR: 18 ppb
Q2:15
Q3: 33
# Stations: 6
Copollutants:
PM10 (0.42)
CO (0.23)
03 (0.15)
Effects were expressed as relative risk based on an
increment of 18 ppb.
IHD
1.014 (1.003,1.026) lag 0, single-pollutant
1.009 (0.994,1.023), 2-pollutant model with PM10
CHF
1.002 (0.978,1.017), single-pollutant model
Risks for dysrhythmia were NR for SO2.
Schwartz (1997)
Tuscon, AZ
Period of Study:
Jan1988-Dec1990
Outcome(s) (ICD9): CVD 390-429. Ascertained
from hospital discharge records.
Study design: Time-series
Statistical analysis: Poisson regression, GAM
Age groups analyzed: 65+
Covariates: long-term and seasonal trends, day of
the wk, temperature, dew point,
Statistical software: S-PLUS
SO2 24-h avg (ppb)
Mean: 4.6 ppb
IQR: 3.9 ppb
10th: 0.7
Q2: 2.0
Median: 3.4
Q3: 5.9
90th: 10.1
Copollutants:
PM10 (0.095) N02 (0.482)
CO (0.395) 03 (-0.271)
Results were expressed as percent change based
on an increment of 3.9 ppb.
0.14% (-1.3%, 1.6)
No other statistically significant associations for
cardiovascular outcomes were observed.
Tolbert et al. (2007) ED Visits
Atlanta, GA
Period of Study:
1993-2004
Outcome(s) (ICD9): Cardiovascular (410-414, 427,
428, 433-437, 440, 443-445, 451-453); Respiratory
(493, 786.07, 786.09, 491, 492, 496, 460-465, 477,
480-486,466.1,466.11,466.19).
Study design: Time-series
Statistical Analysis: Poisson Generalized Linear
Model (GLM).
Statistical package: SAS
Age groups analyzed: all ages
Covariates: adjustment for day-of-wk, hospital
entry, holidays, time, temperature, dew point
temperature
# Hospitals: 41
N: 238,360 (Cardiovascular);
1,072,429 (Respiratory)
Lag(s): 3-day moving avg
1-h max (ppb): 14.9
Range: 1.0,149.0
Copollutants:
PM10
PM2.5
03
N02
CO
Sulfate
Total Carbon
Organic Carbon
EC
Water-Soluble Metals
Oxygenated Hydrocarbons
Relative Risk (95% CI) (per 16.0 ppb SO2)
1.003 (0.994,1.011)
F-68
-------
Study
Methods
Pollutant Data
Findings
Wellenius et al.
(2005b)
Birmingham,
Chicago, Cleveland,
Detroit,
Minneapolis, New
Haven, Pittsburgh,
Seattle
Period of Study:
Jan 1986-Nov 1999
(varies slightly
depending on city)
Outcome(s) IS, primary diagnosis of acute but ill-
defined cerebrovascular disease or occlusion of the
cerebral arteries; HS, primary diagnosis of
intracerebral hemorrhage. ICD codes not provided.
Hospital admissions ascertained from the Centers
for Medicare and Medicaid Services. Cases
determined from discharge data were admitted from
the ER to the hospital.
N IS: 155,503
NHS: 19,314
Study design: Time-stratified case-crossover.
Control days chosen such that they fell in same mo
and same day of wk. Design controls for
seasonality, time trends, chronic and other slowly
varying potential confounders.
Statistical Analysis: 2-stage hierarchical model
(random effects), conditional logistic regression for
city effects in the first stage
Software package: SAS
Lag(s): 0-2, unconstrained distributed lags
SO2 24 h avg (ppb)
10th: 2.17
25th: 3.57
Median: 6.22
75th: 10.26
90th: 16.17
SO2 data not available for
Birmingham, AL
Copollutants:
PM10 (0.39)
CO
N02
Results reported for percent increase in stroke
admissions for an incremental increase in SO2
equivalent to one IQR (6.69).
Ischemic Stroke: 1.35 (0.43, 2.29), lag 0
Hemorrhagic Stroke: 0.68 (-1.77, 3.19)
Multipollutant models not run.
Wellenius et al.
(2005a)
Allegheny County,
PA (near Pittsburgh)
Period of Study:
Jan 1987-Nov 1999
Outcome(s): CHF 428. Cases are Medicare
patients admitted from ER with discharge of CHF
Study design: Case-crossover, control exposures
same mo and day ofwk, controlling for season by
design.
Statistical Analysis: Conditional logistic regression
N: 55,019 admissions, including repeat admissions,
86% admitted s 5 times
Age groups analyzed: 65+ yrs (Medicare recipients)
Covariates: Temperature and pressure. Effect
modification by age, gender, secondary diagnosis
arrhythmias, atrial fibrillation, COPD, hypertension,
type 2 diabetes, AMI within 30 days, angina
pectoris, IHD, acute respiratory infection.
Statistical software: SAS
Lag(s): 0-3
SO2 24-h avg (ppb)
Mean (SD): 14.78
5th: 3.98
25th: 7.70
Median: 12.24
75th: 18.98
95th: 33.93
# Stations: 10
Copollutants:
PM10 (0.51)
CO (0.54)
N02 (0.52)
03 (-0.19)
Effects were reported as percent change based on
an increment of 11 ppb.
CHF, single-pollutant models:
2.36 (1.05, 3.69) lag 0, or
2.14 (0.95, 3.35) laq 0 after adjusted to an increment
of 10 ppb.
CHF, 2-pollutant models:
1.35 (-0.27, 2.99), S02with PM10
0.10 (-1.35,1.57), S02 with CO
0.68 (-0.82, 2.21), S02 with NO2
2.02 (0.68, 3.37), S02 with 03
Burnett et al.
(1997b)
Metropolitan
Toronto (East York,
Etobicoke, North
York, Scarborough,
Toronto, York),
Canada
Period of Study:
1992-1994, 388
days, summers only
Outcome(s) (ICD9): IHD 410-414; Cardiac
Dysrhythmias 427; Heart failure 428. All Cardiac
410-414, 427, 428. Obtained from hospital
discharge data.
Population: 2.6 million residents
Study design: Time-series
Age groups analyzed: All
# Hospitals: NR
Statistical analysis: relative risk regression models,
GAMs.
Covariates: adjusted for long- term trends, seasonal
and subseasonal variation, day of the wk,
temperature, dew point
Season: summer only
Dose response: figures presented
Lag: 1-4 days
SO2 daily 1-h max (ppb):
Mean: 7.9
CV: 64
Min: 0
25th percentile: 4
50th percentile: 7
75th percentile: 11
Max: 26
# of Stations: 4-6
(Results are reported for
additional metrics including 24-
h avg and daytime avg (day)
Copollutants:
H- (0.45) S04 (0.42)
TP (0.55) FP (0.49)
CP (0.44) COH (0.50)
03 (0.18) N02 (0.46)
CO (0.37)
Effects were expressed as relative risk based on an
increment of 7.00 ppb (IQR). T ratio in parentheses.
All cardiac disease
Single-pollutant model
1.041 (2.66), daily max over 4 days, lag 0
Multipollutant model w/ SO2, O3, NO2
Of 7.72 excess hospital admissions, 2.8% attributed
to S02.
Objective of study was to evaluate the role of particle
size and chemistry on cardiac and respiratory
diseases.
F-69
-------
Study
Methods
Pollutant Data
Findings
Burnett et al. (1999)
Metropolitan
Toronto (East York,
Etobicoke, North
York, Scarborough,
Toronto, York),
Canada
Period of Study:
1980-1994
N: 14 yrs
Outcome(s) (ICD9): IHD 410-414; Cardiac
Dysrhythmias 427; Heart failure 428; All cardiac
410-414, 427, 428; Cerebrovascular Disease
obtained from hospital discharge data 430-438;
Peripheral Circulation Disease 440-459.
Population: 2.13-2.42 million residents
Study design: Time-series
Statistical analysis: GAMs to estimate log RR per
unit changes, stepwise regression used to select
min number of air pollutants in multipollutant
models.
Covariates: long-term trends, seasonal variation,
day of wk, temperature, and humidity.
Statistical package: S-PLUS
Lag(s): 0-2 days
SO2 daily avg (ppb)
Mean: 5.35
5th percentile: 0
25th percentile: 1
50th percentile: 4
75th percentile: 8
95th percentile: 17
Max: 57
Multiple day avgs used in
models
Copollutants:
PM2.5 (0.50) PM10-2.5 (0.38)
PM10 (0.52) CO (0.55)
S02 (0.55) 03 (-0.04)
Effects were reported as % change based on an
increment of 5.35 ppb.
Single-pollutant model
Dysrhythmias 0.8% (-0.3,1.9)
Cerebrovascular 0.04% (-0.7, 0.8)
CHF 1.93% (0.9, 2.9)
IHD 2.32% (1.6,3.1)
Attributed percent increase in admissions for SO2
was determined from multipollutant models.
IHD
Attributed percent increase: 0.95%
Authors note SO2 effects could be largely explained
by other variables in the pollution mix as
demonstrated by the multipollutant model.
Fung et al. (2005)
Windsor, Ontario,
Canada
Period of Study:
Apr 1995-Jan 2000
Outcome(s) (ICD9): CHF 428; IHD 410-414;
dysrhythmias 427 and all cardiac. Hospital
admissions from Ontario Health Insurance Plan
records.
Study design: Time-series
Statistical analysis: GLM
N: 11,632 cardiac admission, 4.4/day for 65+ age
group
Age groups analyzed: 65+, < 65 yrs
Statistical Software: SPLUS
Lag(s): lag 0, 2, 3 day avg
SO21-h max (ppb)
Mean (SD): 27.5 (16.5)
Min: 0
Max: 129
IQR: 19.3 ppb
Copollutants:
CO (0.16) 03 (-0.02)
PM10 (0.22) N02 (0.22)
Effects were expressed as percent change of cardiac
disease hospital admissions based on an increment
of 19.3 ppb.
Single-pollutant model:
< 65 yrs: 2.3% (-1.8, 6.6) lag 0;
3.9% (-1.5, 9.6) lag 0-1; 3.4% (-3.0,10.1) lag 0-2
a 65 yrs: 2.6% (0.0, 5.3) lag 0;
4.0% (0.6,1.6) lag 0-1; 5.6% (1.5, 9.9) lag 0-2
Inclusion of particulate matter and adjustment for
meteorological variables did not change the
association between SO2 and cardiac hospitalization.
Stieb et al. (2000)
Saint John, New
Brunswick Canada
Period of Study:
July 1992-Mar 1996
Outcome(s): Angina pectoris; Ml;
dysrhythmia/conduction disturbance; CHF; All
Cardiac. ED Visits collected prospectively.
Study design: Time-series
Statistical analyses: Poisson regression, GAM
N: 19,821 ER visits
# Hospitals: 2
Lag(s): 1-8 days
SO2 24-h avg (ppb)
Mean (SD): 6.7 (5.6)
95th: 18
Max: 60
SO21-h max (ppb)
Mean (SD): 23.8 (21.0)
95th: 62
Max: 161
Copollutants:
CO (0.31) H2S (-0.01)
03 (-0.02) N02 (0.41)
PM10 (0.36) PM2.5 (0.31)
H+ (-0.24) S04 (0.26)
COH (0.31)
Results reported for percent change in admissions
based on a single-pollutant model for incremental
increase in NO2 equivalent to one IQR (8.9 ppb).
Cardiac visits (p-value in parentheses):
4.9 (0.002), 1 day avg, lag 8, all yr
2.8 (0.067), 5 day avg, lag 6, May-Sept
Multi-pollutant models:
4.9, (1.7, 8.2), 1 day avg, lag 8, all yr (O3)
Lags 0-10 presented graphically.
All but lag 8 in single-pollutant model approximately
F-70
-------
Study
Methods
Pollutant Data
Findings
Villeneuve et al.
(2006a)
Edmonton, Canada
Period of Study:
Apr 1992-Mar 2002
Outcome(s) (ICD9): Acute ischemic stroke 434,
436; hemorrhagic stroke 430, 432; transient
ischemic attack (TIA) 435; Other 433, 437, 438. ED
visits supplied by Capital Health.
N: 12,422
Stroke Visits Catchment area: 1.5 million people
Study design: Case-crossover, exposure index time
compared to referent time. Time independent
variables controlled in the design. Index and
referent day matched by day of wk.
Statistical analysis: conditional logistic regression,
stratified by season and gender.
Covariates: temperature and humidity
Statistical software: SAS
Season: Warm: Apr-Sep; Cool: Oct-Mar.
Lag(s): 0,1, 3 day avg
SO2 24 h avg ppb:
All yr
Mean (SD): 2.6 (1.9)
Median: 2.0
25th: 1.0
75th: 4.0
IQR: 3.0
Summer
Mean (SD): 2.1 (1.6)
Median: 2.0
25th: 1.0
75th: 3.0
IQR: 2
Winter
Mean (SD): 3.1 (2.0)
Median: 3.0
25th: 2.0
75th: 4.0
IQR: 2.0
Correlation between SO2 and
other pollutants (all yr):
N02 (0.42) CO (0.41)
03 (-0.25) PM2.5 (0.22)
PM10 (0.19)
Effects were reported as odds ratios based on an
increment of 3 ppb.
Acute Ischemic stroke, a 65 yrs
All yr OR 1.05 (0.99,1.11), lag 0
Warm OR 1.11 (1.01,1.22), lag 0
Cold OR 1.00 (0.93,1.09), lag 0
Effect stronger among males
Hemorrhagic stroke, a 65 yrs
All yr: 0.98 (0.90,1.06), lag 0
Cold: 0.94 (0.84, 1.05), lag 0
Warm: 1.03 (0.90, 1.17)
Effect stronger among males
Transient Cerebral Ischemic Attack, a 65 yrs
All yr: 1.06 (1.00,1.12), lag 0
Cold: 1.03 (0.95, 1.11), lag 0
Warm OR 1.11 (1.02,1.22), lag 0
2-pollutant models presented graphically. Association
of SO2 with Acute Ischemic stroke diminished with
inclusion of CO and NO2.
Anderson et al.
(2001)
West Midlands
conurbation, UK
Period of Study:
1994-1996
N: 832 days
Outcome(s) (ICD9): All CVD 390-459; cardiac
disease 390-429; IHD 410-414; stroke 430-438.
Emergency admissions counted.
Catchment area: 2.3 million
Age groups analyzed: 0-14,15-64, a 65.
Study design: Time-series, APHEA 2 methods
Statistical analyses: GAMs for modeling non-liner
dependence of some variables.
Covariates: Adjusted for effects of seasonal
patterns, temperature and humidity, influenza
episodes, day of wk and holidays.
Software package: S-PLUS
Season: Interaction by warm and cool season
investigated.
Lag(s): 0-3 days
SO2 24-h avg (ppb)
Mean (SD): 7.2 (4.7)
Min: 1.9
10th: 3.3
Median: 5.8
90th: 12.3
Max: 59.8
# of Stations: 5 sites
Copollutants:
PM10 (0.55) PM2.5 (0.52)
PM2.5-10 (0.31) BS (0.50)
S04 (0.19) N02 (0.52)
03 (-0.22)
Results reported for % change in admissions,
increment = 9 ppb (10th-90th).
All CVD all ages
-0.4 (-2.2,1.5), mean lags 0 + 1
Cardiac all ages:
0.7 (-1.3, 2.8), mean lags 0 + 1
IHD a 65 yrs
1.5 (-2.5, 5.6), mean lags 0 + 1
Stroke a 65 yrs
-5.1 (-9.6, -0.4), mean lags 0 + 1
F-71
-------
Study
Methods
Pollutant Data
Findings
Atkinson et al.
(1999b)
London, England
Period of Study:
1992-1994
N:1,096 days
Outcome(s) (ICD9):
All CVD 390-459; IHD 410-414. Emergency
admissions obtained from the Hospital Episode
Statistics (HES) database (complaints).
Ages groups analyzed: 0-14 yrs, 15-64 yrs, 0-64
yrs, 65+ yrs, 65-74 yrs, 75+ yrs
Study design: Time-series, hospital admission
counts
N: 189,109 CVD admissions
Catchment area: 7 million residing in 1,600 Km2
area ofThames basin.
Statistical analyses: APHEAprotocol, Poisson
regression
Covariates: adjusted long-term seasonal patterns,
day ofwk, influenza, temperature, humidity
(compared alternative methods for modeling
meteorological including linear, quadradic, piece-
wise, spline)
Season: warm season Apr-Sep, cool season
remaining mos, interactions between season
investigated
Dose response investigated: yes, bubble charts
presented
Statistical package: SAS
Lag: 0-3
Dose response: Bubble plots presented
SO2 24 h avg (ppb):
Mean: 21.2
SD: 7.8
Min: 7.4
10th: 13
Median: 19.8
90th: 31
Max: 82.2
10th-90th percentile: 11.2
# of Stations: 3, results
averaged across stations
Copollutants:
PM10 CO S02 03 BS
Correlations of SO2 with CO,
NO2, O3, BS ranged from 0.5-
0.6
Correlation ofS02 with O3
negative
Results reported for % change in admissions,
increment 10th-90th percentile (11.2 ppb).
All CVD, all ages
1.57 (0.22, 2.93), lag 0
All CVD, 0-64yrs
2.44 (0.3, 4.63), lag 0
All CVD, 65+
1.72 (0.15,3.32), lag 0
IHD, 0-64 yrs
-2.03 (-5.35,0.91), lag 2
IHD, 65+
3.10 (0.61, 5.65), lag 0
Effect size and significance diminished in models
containing SO2 and BS.
Ballester et al.
(2001)
Valencia, Spain
Period of Study:
1992-1996
Outcome(s) (ICD9):
All CVD 390-459;
heart diseases 410-414, 427, 428; cerebrovascular
diseases 430-438. Admissions from city registry -
discharge codes used.
Study design: Time-series
N: 1080 CVD admissions
# of Hospitals: 2
Catchment area: 376,681 inhabitants of Urban
Valencia
Statistical analyses: Poisson regression, GAM,
APHEA/ Spanish EMECAM protocol. Both linear
and nonparametric model, including a loess term
was fitted, departure from linearity assess by
comparing deviance of both models.
Covariates: long-term trend and seasonality,
temperature and humidity, weekdays, flu, special
events, air pollution.
Season:
Hot season May to Oct;
Cold season Nov to Apr
Statistical package: SAS
Lag: 0-4
24 h avg (|jg/m3):
Mean: 25.6
SD: NR
Min: 4.4
Max: 68.4
Median: 25
# of Stations: 14 manual, 5
automatic
Correlation among stations:
0.3-0.62 for BS, 0.46-0.78 for
gaseous pollutants
Copollutants:
CO (0.74)
N02 (0.22)
03 (-0.35)
BS (0.63)
2-pollutant models used to
adjust for copollutants.
Results expressed as relative risk, increment of
10 (jg/m3.
All CVD
1.0302(1.0042,1.0568), lag 2
Heart disease
1.0357(1.0012,1.0714), lag 2
Cerebrovascular disease
1.0378 (0.9844 to 1.0940), lag 5
Digestive diseases
1.0234(0.9958,1.0518), lag 1
All CVD, hottest semester
1.050 (1.010,1.092), lag 2
Effect size for all CVD and cerebrovascular disease
diminished in 2-pollutant models.
F-72
-------
Study
Methods
Pollutant Data
Findings
Bal tester et al.
(2006)
Multicity, Spain:
(Barcelona, Bilbao,
Castellon, Gijon,
Granada, Huelva,
Madrid, Oviedo,
Seville, Valencia,
Zaragoza)
Period of Study:
1995/1996-1999
N:1,096 days
Outcome(s) (ICD9):
All CVD 390-459;
Heart diseases 410-414,427,428. Emergency
admission from hospital records. Discharge data
used.
Study design: Time-series, meta-analysis to pool
cities
N: Daily mean admissions reported by city.
Statistical analyses: Poisson regression and GAM,
with stringent convergence criteria, meta-analysis
with random effect model. Tested linearity by
modeling pollutant in linear and non-linear way
(spline smoothing). Linear model provided best
results 55% of time but used in all cases to facilitate
comparability.
Covariates: temperature, humidity and influenza,
day of wk unusual events, seasonal variation and
trend of the series
Season: Hot: May to Oct; Cold: Nov to Apr
Statistical package: S-PLUS
Lag: 0-3
SO2 24-h avg (|jg/m2)
Mean, 10th, 90th
Barcelona: 15.5, 6.6, 27.9
Bilbao: 18.6,10.2,29.3
Cartagena: 27.1,14.6, 40.8
Castellon: 7.7, 3.8,12.7
Gijon: 29.4,10.3,52.4
Granada: 19.1, 8.8, 31.5
Huelva: 11.9, 4.5, 22.6
Madrid: 21.8, 8.7, 41.8
Oviedo: 40.9,16.3, 75.5
Pamplona: 7.6,1.8, 17.0
Seville: 9.6, 5.6,14.6
Valencia: 16.6, 9.4, 24.4
Vigo: 9.3, 2.6,18.2
Zaragoza: 9.3, 2.0,19.9
# of stations: depends on the
city
Correlation among stations:
Correlations between SO2
stations within cities poor.
Copollutants:
CO (0.58) 03 (-0.03)
N02 (0.46) BS (0.24)
TSP (0.31) PM10 (0.46)
Correlations reported are the
median for all cities.
Results reported for % change in admissions,
increment 10 (|jg/m3).
All cardiovascular
1.33% (0.21,2.46) lag 0-1
Heart diseases
1.72% (0.50,2.95) lag 0-1
Single day lags presented graphically. Effect size
decreased with increasing lag.
Multi-pollutant results presented graphically. Control
for CO and particulates diminished SO2 effects.
D'lppoliti et al.
(2003)
Rome, Italy
Period of Study:
Jan 1995- June
1997
Outcome(s) (ICD): AMI 410 (first episode).
Computerized hospital admission data.
Study design: case-crossover, time stratified,
control days within same mo falling on the same
day.
Statistical analyses: conditional logistic regression,
examined homogeneity across co-morbidity
categories
N: 6531 cases
Age groups analyzed: 18-64 yrs,
65-74 yrd, a 75
Season: Cool: Oct-Mar;
Warm: Apr-Sep
Lag(s): 0-4 day, 0-2 day cum avg
Dose Response: OR for increasing quartiles
presented and p-value for trend.
SO2 24 h avg (|jg/m3)
All yr:
Mean (SD): 9.5 (6.0)
25th: 5.4
50th: 8.2
75th: 12.6
IQR: 7.2
Cold season:
Mean (SD): 12.7 (6.5)
Warm Season:
Mean (SD): 88.3 (15.4)
# Stations: 5
Copollutants:
TSP (0.29) N02 (0.37)
CO (0.56)
Results reported as odds ratios for increment equal
to one IQR (7.2 (jg/m3).
AMI
Quartile I (referent)
Quartile II: 0.987 (0.894,1.089), lag 0-2
Quartile III: 1.008 (0.892,1.140), lag 0-2
Quartile IV: 1.144 (0.991,1.321), lag 0-2
Results at various lags NR for SO2.
Llorca et al. (2005)
Torrelavega, Spain
Period of Study:
1992-1995
Outcome(s) (ICD): CVD (called cardiac in paper)
390-459. Emergency admissions, excluding
nonresidents. Obtained admissions records from
hospital admin office.
Study design: Time-series
Statistical analyses: Poisson regression, APHEA
protocol
Covariates: rainfall, temperature, wind speed
direction
N: 18,137 admissions
Statistical software: STATA
Lag(s): NR
SO2 24 h avg |jg/m3:
Mean (SD): 13.3 (16.7)
Copollutants:
TSP (-0.40)
N02 (0.588)
SH2 (0.957)
NO (0.544)
Multipolutant models.
Results expressed as rate ratios.
Increment = 100 |jg/m3.
Cardiac admissions, single-pollutant model
0.94 (0.84,1.05)
Five-pollutant model
1.09 (0.83,1.42)
All cardiorespiratory admissions, single-pollutant
model
RR 0.98 (0.89,1.07)
Five-pollutant model
0.98 (0.80,1.21)
F-73
-------
Study
Methods
Pollutant Data
Findings
Poloniecki et al.
(1997)
London, UK
Period of Study:
Apr, 1987-Mar
1994, 7 yrs
Outcome(s): All CVD 390-459; Ml 410; Angina
pectoris 413; other IHD 414; ARR 427; congestive
heart failure 428; cerebrovascular disease 430-438.
Hospital Episode Statistics (HES) data on
emergency hospital admissions.
Study design: Time-series
N: 373, 556 CVD admissions
Statistical analyses: Poisson regression with GAM,
APHEA protocol
Covariates: long-term trends, seasonal variation,
day of wk, influenza, temperature and humidity.
Season: Warm, Apr-Sep;
Cool, Oct-Mar
Lag: 0-1
S0224 h avg ppb:
Min: 0
10%: 2
Median: 6
90%: 21
Max: 1
Copollutants:
BS
CO 24 h avg
NO2 24 h avg
03 8 h
Correlations between
pollutants high but not
specified
Effects were expressed as relative risk based on an
increment of 19 ppb
(10th-90th percentile).
Single-pollutant models (lag 0-1):
Ml: 1.0326 (1.0133,1.0511)
Angina: 1.0133 (0.9907,1.0383)
IHD: 0.9944 (0.9651,1.0239)
ARR: 1.0181 (1.0000,1.0448)
CHF: 1.0057 (0.9846,1.0258)
Cerebrovascular: 1.0019 (0.9837,1.0189)
All circulatory: 1.0248 (1.0062,1.0444)
Ml, 2-pollutant models, cool season:
1.0399(1.0171,1.0628), SO2 only
1.0285(1.0019,1.0571), S02with NO2
1.0380(1.0057,1.0704), SO2/CO
1.0285(1.0019,1.0552), SO2/BS
1.0476(1.0209,1.0742), S02 with 03
In the warm season no significant associations were
observed in 2-pollutant models.
Prescott et al.
Outcome(s) (ICD9): Cardiac and cerebral ischemia
NO2 24 h avg ppb
Results reported as % increase in admissions,
(1998)
410-414,426-429,
Mean (SD): 8.3 (5.6)
Range: 1-50
increment 10 ppb.
Edinburgh, UK
434-440. Extracted from Scottish record linkage
system.
All CVD, <65 yrs
Period of Study:
Oct 1992-Jun 1995
Study design: Time-series
90th-10th
Percentile = 12 ppb
4.9 (-1.0,11.1), 3 day moving avg
Statistical Analysis: Poisson, log linear regression
models
Copollutants:
All CVD, >65 yrs
Age groups analyzed: < 65, 65+ yrs
O3, 24 h avg PM, 24 h avg
NO2, 24 h avg CO, 24 h avg
-3.7 (-12.4, 5.9), 3 day moving avg
Covariates: Seasonal and weekday variation,
Correlations NR.
temperature, and wind speed.
Lag(s): 0,1, 3 day moving avg
Sunyeretal. (2003)
Europe
(Birmingham,
London, Milan,
Paris, Rome,
Stockholm, the
Netherlands)
Period of Study:
1990-1996
Hospital Admissions
Outcome(s) (ICD9): Cardiovascular diseases (390-
429); IHD (410-413); stroke (430-438)
Study design: Time-series
Statistical Analysis: Poisson autoregression with
GAM
Age groups analyzed: all ages
Covariates: trend, seasonal patterns,
meteorological factors
Lag(s): 0 +1
24-h median (|jg/m3):
Birmingham: 19
London:21
Milan: 18
Netherlands: 9
Paris: 15
Rome: 9
Stockholm: 5
Copollutants:
PM10BS NO2O3
% increase in Hospital Admissions (95% CI) (per 10
(jg/m3 SO2)
Cardiovascular: All ages: 0.7 (0.3, 1.1);
>65:0.7 (0.3,1.2)
IHD: <65: 0.6 (0.2,1.1); >65:1.2(0.8,1.6)
IHD after adjustment for CO, NO2, BS < PMm
<65:0.7(0.1,1.3); >65:-1.4 (-8.0, 6.0)
Stroke: >65: 0.0 (-0.5, 0.5)
Yallop et al. (2007)
London, England
Period of Study:
Jan 1998-Oct 2001
N: >1400 days
Outcome(s): Acute pain in Sickle Cell Disease
(HbSS, HbSC, HbS/30, thalassaemia, HbS/3+).
Admitted to hospital for at least one night.
Study design: Time-series
Statistical analyses: Cross-correlation function
N: 1047 admissions
Covariates: No adjustment made in analysis,
discussion includes statement that the effects of
weather variables and copollutants are inter-
related.
Statistical package: SPSS
Lag(s): 0-2 days
Dose response: quartile analysis, graphs
presented, ANOVA comparing means across
quartiles.
NR
Copollutants:
03
CO
NO
N02
PM10
Daily avg used for all
copollutants.
No association for SO2
F-74
-------
Study
Methods
Pollutant Data
Findings
Jalaludin et al.
(2006)
Sydney, Australia
Period of Study:
Jan 1997-Dec 2001
Outcome(s) (ICD9): All CVD
390-459; cardiac disease
390-429; IHD 410-413; and cerebrovascular
disease or stroke 430-438; Emergency room
attendances obtained from health department data.
Age groups included: 65+
Study design: Time-series, multicity APHEA2
Protocol.
Statistical Analysis: GAM (with appropriate
convergence criteria) and GLM Models. Only GLM
presented.
SO2 24 h avg avg (ppb)
Mean (SD): 1.07 (0.58)
Min: 0.09
25th: 0.64
Median: 1.01
75th: 1.39
Max: 3.94
IQR: 0.75
# of Stations: 14
Lag: 0-3
Covariates: Daily avg temperature and daily
relative, humidity, long-term trends, seasonality,
weather, day ofwk, public school holidays, outliers
and influenza epidemics.
Dose response: quartile analysis
Season: Separate analyses for warm (Nov-Apr) and
cool periods (May-Oct).
Copollutants:
BS (0.21)
PM10 (0.37)
03 (0.454)
N02 (0.52)
CO (0.46)
Effects were presented as percent change based on
an increment of 0.75 ppb.
Single-pollutant model:
All CVD, all yr: 1.33% (0.24, 2.43) lag 0
Cardiac: 1.62% (0.33,2.93) lagO
IHD: 1.12% (-0.84, 3.12) lag 0
Stroke:-1.41% (-3.67, 0.90) lag 0
Cool Season
All cardiovascular: 2.15% (0.84, 3.46) lag 0
Cardiac: 2.48% (0.94, 4.04) lag 0
IHD: 2.49% (0.13, 4.91) lag 0
Stroke:-0.19% (-2.90, 2.60) lag 0
Warm Season
All cardiovascular: 0.06% (-1.48,1.62) lag 0
Cardiac: 0.38% (-1.37, 2.16) lag 0
IHD: -0.47% (-3.08, 2.22) lag 0
Stroke: -2.74% (-5.92, 0.55) lag 0
Results for lags 0-3 presented. In general, effect size
diminished with increasing lag.
Effects of SO2 on all CVD were diminished with
inclusion of PM and CO (graphically presented.)
Petroeschevsky et
al. (2001)
Outcome(s) (ICD9): CVD 390-459. Hospital
admissions, non-residents excluded.
Brisbane, Australia Study design: Time-series
Period of Study: Statistical analyses: Poisson regression, APHEA
Jan 1987-Dec 1994 protocol, linear regression and GEEs
N: 2,922 days Age groups analyzed: 15-64, 65+
Covariates: temperature, humidity, rainfall. Long-
term trends, season, flu, day ofwk, holidays.
Dose response: quintile analysis.
Statistical software: SAS
Lag(s): lag 0-4, 3 day avg, 5 day avg
SO2 24-h avg (pphm):
Summer: mean, min, max
0.39,0.0,1.63
Fall: mean, min, max
0.42,0.01,3.55
Winter: mean, min, max
0.48, 0.0, 2.08
Spring: mean, min, max 0.37,
0.0, 6.02
Overall: mean, min, max
0.41,0.0,3.55
SO21-h max (pphm):
Summer: Mean, min, max
0.78,0.0,5.5
Fall: Mean, min, max
0.93, 0.05, 5.95
Winter: Mean, min, max
1.13,0.0,6.68
Spring: Mean, min, max 0.84,
0.0,6.01
Overall: Mean, min, max
0.92, 0.0, 6.68
Copollutants:BSP O3 NO2
Correlation between pollutants
NR.
Effects were expressed as relative risk based on an
increment of 10 ppb and the 24-h avg SO2
concentrations.
All CVD
15 to a 65 yrs: 1.028 (0.987,1.070) lag 0
15 to 64yrs: 1.081 (1.010,1.158) lag 0
>65 yrs: 1.038 (0.988,1.091) lag 1
Non-significant increasing risk for CVD in those 15-
64 by quintile of SO2 concentration observed.
F-75
-------
Study
Methods
Pollutant Data
Findings
ASIA
Chan et al. (2006)
Outcome(s) (ICD9): Cerebrovascular disease 430-
SO2 24-h avg (ppb):
Results reported for OR for association of
Taipai, Taiwan
437; stroke 430-434; hemorrhagic stroke 430-432;
Mean: 4.3
emergency department admissions with an IQR
ischemic stroke 433-434. Emergency admission
SD: 2.4
Min: 0.4
Max: 17.1
increase in SO2 (3.1 ppb)
Period of Study:
Apr1997-Dec2002
data collected from National Taiwan University
Hospital.
Cerebrovascular:
1.008 (0.969,1.047), lag 0
N: 2,090 days
Ages groups analyzed: age >50 included in study
IQR: 3.1 ppb
Stroke:
Study design: Time-series
# of Stations: 16
0.991 (0.916,1.066), lag 0
Ischemic stroke:
N: 7341 Cerebrovascular admissions among those
Correlation among stations:
1.044 (0.966,1.125), lag 0
>50 yrs old
NR.
Hemorrhagic stroke:
Catchment area:
Copollutants:
0.918(0.815,1.021), lagO
Statistical analyses: Poisson regression, GAMs
PM10 (0.59) PM2.5 (0.51)
CO (0.63) N02 (0.64)
03 (0.51)
No significant associations for SO2 reported. Lag 0
used to adjust for non-linear relation between
shown but similar null results were obtained for laqs
confounders and ER admissions.
0-3.
Covariates: time trend variables: yr, mo, and day of
2-pollutant models to adjust for copollutants but not
wk, daily temperature difference, and dew point
for SO2, which was not associated with health
temperature. outcomes.
Linearity: investigated graphically by using the
LOESS smoother.
Lag: 0-3, cumulative lag up to 3 days
Chang et al. (2005)
Taipei, Taiwan
Period of Study:
1997-2001
N: 5 yrs
Outcome(s) (ICD9): CVD 410-429.
Daily clinic visits or hospital admission from
computerized records of National Health Insurance.
Discharge data.
Source populatioN: 2.64 million
N: 40.8 admissions/day; 74,509/5 yrs
# Hospitals: 41
Study design: Case-crossover, referent day 1 wk
before or after index day
Statistical analyses: Conditional logistic regression.
Covariates: same day temperature and humidity.
Season: warm/cool (stratified by temperature
cutpoint of 20 °C)
Lag(s): 0-2 days
SO2 24-h avg (ppb)
Mean: 4.32
Min: 0.15
25th: 2.74
Median: 3.95
75th: 5.49
Max: 14.57
IQR: 2.75
# of Stations: 6
Copollutants:
CO 03 N02
PM10 Correlations NR.
2-pollutant models to adjust for
copollutants.
Effects were expressed as odds ratios based on an
increment of 2.75 ppb.
Warm (> 20 °C) 0.967 (0.940, 0.995)
Cool (< 20 °C) 1.015 (0.965,1.069)
In 2-pollutant models with (PM10, NO2, CO, or O3) the
effect of SO2 was attenuated for both temperature
ranges such that it was negatively associated with
CVD.
>20 °C: 0.874 (0.77, 0.880), W/PM10
< 20 °C: 0.986 (0.928,1.048), w/ PM10
> 20 °C: 0.826 (0.798, 0.854), w/ NO2
< 20 °C: 0.922 (0.865, 0.984), w/ NO2
>20 °C: 0.903 (0.876, 0.931), w/CO
<20 °C: 0.960 (0.901,1.022), w/CO
>20 °C: 0.953 (0.926, 0.981), w/03
<20 °C: 1.014(0.963,1.067), w/03
Lee et al. (2003b)
Seoul, Korea
Period of Study:
Dec 1997-Dec
1999, 822-days,
N: 184 days in
Outcome(s) (ICD10): IHD: Angina pectoris 120;
Acute or subsequent Ml 121-123; other acute IHD
124. Electronic medical insurance data used.
Study design: Time-series
Statistical methods: Poisson regression, GAM with
strict convergence criteria.
Age groups analyzed: all ages, 64+
Covariates: Long-term trends LOESS smooth,
temperature, humidity, day of wk.
Season: Presented results for summer (Jun, Jul,
Aug) and entire period.
Lag(s): 0-6
SO2 24 h avg (ppb):
5th: 3.7
10th: 5.1
Median: 7.0
75th: 9.5
95th: 14.3
Mean (SD): 7.7 (3.3)
IQR: 4.4
Copollutants:
All yr
N02 (0.72) 03 (-0.30)
CO (0.81) PM10 (0.59)
Warm season
N02 (0.79) 03 (-0.56)
CO (0.41) PM10 (0.61)
2-pollutant models.
Results reported for RR of IHD hospital admission
for an incremental increase in SO2 equivalent to one
IQR (4.4 ppb).
Single-pollutant model:
Entire season- IHD
All ages 0.96 (0.92, 0.99) lag 3
a 64 yrs 0.95 (0.90, 1.01) lag 3
Summer- IHD
All ages 1.09 (0.96, 1.24) lag 3
a 64 yrs 1.32 (1.08, 1.62) lag 3
2-pollutant model:
Entire season; SO2 and PM10
a 64 yrs 0.98 (0.94, 1.03) lag 3
F-76
-------
Study
Methods
Pollutant Data
Findings
Tsai et al. (2003a)
Kaohsiung, Taiwan
Period of Study:
1997-2000
Outcome(s) (ICD9): All cerebrovascular
430-438; SHS 430;PIH 431-432; IS 433-435; Other
436-438. Ascertained from National Health
Insurance Program computerized admissions
records.
Study design: Case-crossover
Statistical Analysis: conditional logistic regression.
N: 23,179 stroke admissions
# Hospitals: 63
Statistical software: SAS
Season: warm (a 20 °C); cool (< 20 °C)
Lag(s): 0-2, cumulative lag up to 2 previous days
SO2 (ppb)
Min: 1.25
25th: 6.83
Median: 9.76
75th: 13.00
Max: 26.80
Mean: 10.08
# Station: 6
Copollutants:
PM10SO2
CO 03
Results reported as OR for the association of
admissions with an incremental increase of SO2
equivalent to the IQR of 6.2 ppb
PIH admissions: Warm: 1.06 (0.95,1.18), lag 0-2;
Cool: 0.85 (0.58, 1.26), lag 0-2
IS admissions: Warm: 1.06 (1.00,1.13), lag 0-2;
Cool: 1.11 (0.83, 1.48), lag 0-2
2-pollutant models: PIH 0.91 (0.80,1.03) w/ NO2
IS 0.93 (0.87, 1.00)w/NO2;
PIH 0.94 (0.83, 1.06), w/CO
IS 0.94 (0.88,1.02), w/CO
PIH 1.08 (0.96, 1.20) w/ 03
IS 1.08(1.01,1.15) w/03
PIH 0.99 (0.88, 1.11) w/PM
IS 1.01 (0.95-1.08) w/PM
Wong et al. (1999)
Hong Kong, China
Period of Study:
1994-1995
Outcome(s) (ICD9): CVD:
410-417, 420-438, 440-444; CHF 428; IHD 410-
414; Cerebrovascular Disease 430-438. Hospital
admissions through ER departments via Hospital
Authority (discharge data).
Study design: Time-series
Statistical analyses: Poisson regression, APHEA
protocol
# Hospitals: 12
Covariates: daily temperature, relative humidity day
ofwk, holidays, influenza, long-term trends (yrand
seasonality variables). Interaction of pollutants with
cold season examined.
Season: cold (Dec-Mar)
Lag(s): 0-3 days
SO2 24-h avg (|jg/m3)
Mean: 20.2
IQR: 10
Copollutants:
PM10
S02
03
Results reported for RR associated with incremental
increase in NO2 equal to 10 |jg/m3.
All CVD, All ages
1.016 (1.006,1.026) lag 0-1
All CVD, 5-65 yrs
1.004 (0.989,1.020) lag 0-1
All CVD, >65 yrs
1.021 (1.010,1.032) lag 0-1
CHF
1.036 (1.013,1.059) lagO
IHD
1.010 (0.995,1.025) lag 0-1
Cerebrovascular
0.990 (0.978,1.002) lag 3
2-pollutant model results not presented for SO2
Wong et al. (2002a)
Hong Kong, London
Period of Study:
1995-1997 (Hong
Kong),
1992-1994
(London)
Outcome(s) (ICD9): Cardiac disease 390-429; IHD
410-414. Patients admitted to hospitals from
emergency departments, out patient departments
or directly to inpatient wards.
Statistical Analysis: Poisson regression, GAMs
Covariates: smooth functions of time, temperature,
humidity (up to 3 days before admission) day ofwk,
holidays and unusual events.
Statistical software:
S-PLUS
Season: warm/cold
Lag(s): 0-3, cumulative 0-1
SO2 24-h avg (|jg/m3)
Hong Kong
Mean, all yr: 17.7 (12.3)
Mean, warm: 18.3
Mean, cold: 17.2
Min: 1.1
10th: 6.2
50th: 14.5
90th: 32.8
Max: 90
London
Mean, all yr: 23.7 (12.3)
Mean, warm: 22.2
Mean, cold: 25.3
Min 6.2
10th: 13.2
50th: 20.6
90th: 38.1
Max: 113.6
Copollutants:
Hong Kong
N02 (0.37) PM10 (0.30)
03 (-0.18)
London
N02 (0.71) PM10 (0.64)
03 (-0.25)
Effects expressed as % change, increment was
10 (jg/m3
Cardiac (all ages)
Hong Kong: All yr: 2.1% (1.3, 2.8) lag 0-1
Warm: 1.0% (0.0, 2.0) lag 0-1
Cold: 1.9% (1.2,2.7) lag 0-1
London: All yr: 1.6% (1.0, 2.2) lag 0-1
Warm: 0.6% (-0.6,1.7) lag 0-1
Cold: 1.9% (1.2,2.7) lag 0-1
IHD (all ages)
Hong Kong: All yr: 0.1% (-1.1,1.2) lag 0-1
Warm:-0.6% (-2.0,0.8) lag 0-1
Cold: 1.0% (-0.8,2.8) lag 0-1
London: All yr: 1.7% (0.8, 2.6) lag 0-1
Warm: 1.0% (-0.6, 2.6) lag 0-1
Cold: 2.0% (0.9, 3.1) lag 0-1
Multipollutant model
Cardiac (all ages)
Hong Kong: SO2 alone 2.1% (1.3, 2.8)
S02with N021.4% (0.4,2.3)
S02 with 03 2.1% (1.4,2.9)
S02with PM10 2.0% (1.1,2.8)
London: S021.6% (1.0,2.2)
S02with N021.4% (0.6,2.3)
S02with 031.6% (0.9,2.2)
S02with PM10 2.2% (1.2, 3.2)
F-77
-------
Study
Methods
Pollutant Data
Findings
Yang et al. (2004b)
Kaohsiung, Taiwan
Period of Study:
1997-2000
Outcome(s) (ICD9): All CVD: 410-429 * (All CVD
typically defined to include ICD9 codes 390-459)
N: 29,661
Study design: Case-crossover
Statistical Analysis: Poisson Time-series regression
models, APHEA protocol
# of Hospitals: 63
Season: Authors indicate not considered because
the Taiwanese climate is tropical with no apparent
seasonal cycle
Covariates: Stratified by warm (a 25°) and cold (<
25°) days, temperature, and humidity
measurements included in the model
Statistical package: SAS
Lag: 0-2 days
SO2 24-h avg (ppb)
Min: 1.25
25%: 6.83
50%: 9.76
75%: 13.00
Max: 26.80
Mean: 10.08
# of Stations: 6
Correlation among stations:
NR
Copollutants:
PM10 CO S02 03 8
2-pollutant models used to
adjust for copollutants
Correlations NR
OR's for the association of one IQR (17.08 ppb)
increase in SO2 with daily counts of CVD hospital
admissions are reported
All CVD (ICD9: 410-429)
One-pollutant model: a 25°: 0.999 (0.954,1.047);
<25°: 1.187(1.092, 1.291)
2-pollutant models:
Adjusted for PM10: > 25°: 0.961 (0.917,1.008)
<25°: 1.048(0.960, 1.145)
Adjusted for N02:a 25°: 0.921 (0.875, 0.969)
<25°: 0.711 (0.641,0.789)
Adjusted for CO: a 25°: 0.831 (0.785, 0.879)
<25°: 0.996 (0.910, 1.089)
Adjusted for 63: a 25°: 1.034 (0.987,1.084)
<25°: 1.194(1.098, 1.299)
Hosseinpoor et al. Outcome(s) (ICD9): Angina pectoris 413. Primary
(2005)
Tehran, Iran
Period of Study:
Mar 1996-Mar
2001, 5 yrs
discharge diagnosis from registry databases or
records.
Study design: Time-series
Statistical methods: Poisson regression
# Hospitals: 25
Covariates: Long-term trends, seasonality,
temperature, humidity, holiday, post-holiday, day of
wk.
Lag(s): 0-3
SO2 24-h avg (|jg/m3)
Mean (SD): 73.74 (33.30)
Min: 0.30
25th: 48.23
Median: 74.05
75th: 98.64
Max: 499.26
Copollutants:
N02 CO 03 PM10
Correlations NR
Results reported for relative risk in hospital
admissions per increment of 10 |jg/m3 SO2.
Angina
0.99995 (0.99397,1.00507), lag 1
In a multipollutant model only CO (lag 1) was
significantly associated with angina pectoris related
hospital admissions.
Table F-5. Short-term exposure to SO2 and mortality.
Study
Methods
Pollutant Data
Outcome
Findings
METAANALYSIS
Stieb et al. (2002),
(reanalysis 2003)
The lags and multiday
averaging used varied
24-h avg ranged from 0.7 ppb
(San Bernardino) to 75 ppb
All cause
Single-pollutant model
(29 estimates): 1.0% (0.6,1.3)
Meta-analysis of estimates
from various countries.
Meta-analysis of time-series
study results.
(Shenyang)
"Representative"
concentration: 9.4 ppb
Copollutants: PM10, O3, NO2,
CO
Multipollutant model estimates
(10 estimates): 0.9% (0.3,1.4)
UNITED STATES
Chock et al. (2000)
Lags: 0,1, 2, 3
Mean NR
All cause; age < 75 yrs; age
All cause:
Pittsburgh, PA
Period of Study:
1989-1991
Poisson GLM. Time-series
study. Numerous results
Copollutants: PM10, O3, NO2,
CO; 2-, 5-, and 6-pollutant
models
75+ yrs
Age 0-75 yrs:
Lag 1:0.7% (-0.7,2.2)
Age 75+ yrs:
Lag 1: -0.2% (-1.6,1.3)
F-78
-------
Study
Methods
Pollutant Data
Outcome
Findings
De Leon et al. (2003)
New York City
Period of Study:
1985-1994
Lags:0 or 1
Poisson GAM with
Stringent convergence
Criteria; Poisson GLM.
Time-series study.
24-h avg: 14.64 ppb
Copollutants: PM10, O3, NO2,
CO; 2-pollutant models
Circulatory and cancer with
and without contributing
respiratory causes
Gaseous pollutants results
were given only in figures.
Circulatory:
Age < 75 yrs: -2%
Age 75+ yrs: -2%
Dockery et al. (1992)
Lag: 1
24-h avg:
All cause
All cause:
St. Louis, MO and
Eastern Tennessee
Period of Study:
1985-1986
Poisson with GEE.
Time-series study.
St. Louis: 8.9 ppb
Eastern Tennessee:
5.1 ppb
Copollutants: PM10, PM2.5,
S042", H+, 03, N02
St. Louis, MO: 0.8% (-1.7, 3.2)
Eastern Tennessee: 0.4%
(-0.4,1.1)
Gamble (1998)
Dallas, TX
Period of Study:
1990-1994
Lag: 0
Poisson GLM.
Time-series study.
24-h avg: 3 ppb
Copollutants: PM10, O3, NO2,
CO; 2-pollutant models
All cause; respiratory;
cardiovascular
All cause: -0.8% (-3.8, 2.4)
Respiratory:-1.0% (-5.8, 4.1)
Cardiovascular: -0.5% (-11.4,
11.8)
Gwynn et al. (2000)
Buffalo, NY
May 1998-Oct 1990
Lag: 0,1,2,3
Poisson GAM with
Default convergence criteria.
Time-series study.
24-h avg: 12.2 ppb
Copollutants: PM10, CoH,
S042", 03, N02, CO, H+
All cause; respiratory;
circulatory
All cause:
Lag 0:-0.1% (-1.8,1.7)
Circulatory: Lag 3:1.3% (-2.9,
5.6)
Respiratory: Laq 0: 6.4%
(-2.5,16.2)
Kelsall et al. (1997)
Philadelphia, PA
Period of Study:
1974-1988
Lag: 0
(AIC presented for 0 through
5)
Poisson GAM.
24-h avg: 17.3 ppb
Copollutants: TSP, CO, NO2,
03
All cause; respiratory;
cardiovascular
All cause:
Single-pollutant:
0.8% (0.3,1.4)
With all other pollutants:
0.8% (0.1,1.6)
Kinney and Ozkaynak (1991)
Lag: 1
24-h avg: 15 ppb
All cause; respiratory;
All cause:
Los Angeles County, CA
Period of Study:
1970-1979
OLS (ordinary least squares)
on high-pass filtered variables.
Time-series study.
Copollutants: KM (particle
optical reflectance), Ox, NO2,
CO; multipollutant models
circulatory
Exhaustive multipollutant
model: 0.0% (-1.1,1.2)
Klemm and Mason (2000)
Lag: 0-1
1-h max: 18.7 ppb
All cause; respiratory;
All cause - Age 65+ yrs:
Atlanta, GA
Period of Study:
Aug 1998-Jul 1999
Poisson GLM using quarterly,
monthly, or biweekly knots for
temporal smoothing. Time-
series study.
Copollutants: PM2.5,
PM10-2.5, EC, OC, S042", NOs-
03, N02, CO
cardiovascular; cancer; other;
age < 65 yrs; age 65+ yrs
Quarterly knots:
4.7% (-2.6, 12.5)
Monthly knots:
3.4% (-4.1,11.5)
Bi-weekly knots:
1.0% (-6.7, 9.3)
Klemm et al. (2004)
Lags: 2-day avg (avg of lag 0
1-h max (ppb): 19.4 (13.42)
All-cause
a 65 yrs old
Georgia (Fulton; DeKalb
counties)
Period of Study:
1998-2000
and lag 1)
Poisson with GLM. Time-
series study.
Copollutants: PM2.5, Coarse
mass, O3, NO2, CO, Acid,
Ultrafine surface area,
Ultrafine count, EC, Organic
carbon, SO4,Oxygenated
hydrocarbons, Nonmethane
hydrocarbons, NO3
Quarterly knots (SE)
p = 0.00115 (0.00092) t = 1.24
Monthly knots (SE)
p = 0.00084 (0.00096) t = 0.87
Biweekly knots (SE)
p = 0.00024 (0.00101) t = 0.24
Lipfert et al. (2000a)
Lag: 0-1
24-h avg: ~8 ppb
All cause; respiratory;
All-cause:
Seven counties in
Philadelphia, PA area
Period of Study:
May 1992-Sep 1995
Linear with 19-day weighted
avg Shumway filters. Time-
series study. Numerous
results.
Copollutants: PM10, PM2.5,
PM10-2.5, SO42", other PM
indices, O3, NO2, CO; 2-
pollutant models
cardiovascular; all ages; age
65+yrs; age
< 65 yrs; various subregional
boundaries
Philadelphia:
0.7% (p > 0.05)
F-79
-------
Study
Lippmann et al. (2000);
reanalysis Ito, (2003)
Detroit, Ml
Period of Study:
1985-1990
1992-1994
Mar et al. (2000; 2003)
Phoenix, AZ.
Period of Study:
1995-1997
Moolgavkar et al. (1995)
Philadelphia, PA
Period of Study:
1973-1988
Moolgavkar (2000b; 2003b)
Cook County, IL; Los Angeles
County, CA; and Maricopa
County, AZ
Period of Study:
1987-1995
Moolgavkar (2003a)
Cook County, IL and Los
Angeles County, CA
Period of Study:
1987-1995
Samet et al. (2000a; 2000b);
reanalysis Dominici et al.
(2003)
90 U.S. cities (58 U.S.cities
with SO2 data)
Period of Study: 1987-1994
Schwartz (1991)
Detroit, Ml
Period of Study:
1973-1982
Methods
Lags: 0,1,2, 3,0-1,0-2,0-3
Poisson GAM, reanalyzed with
stringent convergence criteria;
Poisson GLM. Numerical SO2
risk estimates were not
presented in the re-analysis.
Time-series study.
Lag: 1
Poisson GLM. Time-series
study.
Lags: 0,1, 2, 3, 4, 5
Poisson GAM with default
convergence criteria in the
original Moolgavkar (2000);
GAM with stringent
convergence criteria and GLM
with natural splines in the 2003
re-analysis. The 2000 analysis
presented total death risk
estimates only in figures.
Lags: 0,1, 2, 3, 4, 5
Poisson GAM with default
convergence criteria. Time-
series study.
Lags: 0,1, 2
Poisson GAM, reanalyzed with
stringent convergence criteria;
Poisson GLM. Time-series
study.
Lags: 0,1, 0-1
Poisson GEE. Time-series
study.
Pollutant Data
24-h avg:
1985-1990:9.8 ppb
1992-1994: 7 ppb
Copollutants: PM10, PM2.5,
PMio-2.5, S042", H+, 03, N02,
CO;
2-pollutant models
24-h avg:
Spring: 16.8 ppb
Summer: 15.7 ppb
Fall: 17.8 ppb
Winter: 25.4 ppb
Copollutants: TSP, O3;
2-pollutant models
24-h avg ranged from 0.4 ppb
(Riverside) to 14.2 ppb
(Pittsburgh)
Copollutants: PM10, O3, NO2,
CO; multipollutant models
24-h avg: 12 ppb
Copollutants: TSP (predicted
from extinction coefficient); 2-
pollutant models
Outcome
All cause; respiratory;
circulatory; cause-specific
All cause
All cause; cardiopulmonary
All cause
Findings
Poisson GAM:
All cause:
1985-1990:
Lag 1:0.5% (-1.5,2.4)
1992-1994:
Lag 1:1.1% (-1.4,3.6)
Poisson GAM:
All cause:
Lag 0:11.2% (-1.5, 25.6)
Poisson GLM:
Cardiovascular:
Lag 1:7.4% (-13.1, 32.6)
All yr: 1.3% (0.8,1.8)
Spring: 1.7% (0.6, 2.9)
Summer: 0.9% (-0.7, 2.5)
Fall: 1.3% (0.0,2.6)
Winter: 2.0% (0.9, 3.0)
GLM (re-analysis):
Cook County:
All-cause:
Lag 1:2.6% (1.4,3.8)
Cardiovascular:
Lag 1:2.9% (1.0, 4.8)
Los Angeles: Cardiovascular:
Lag 1:5.9% (3.0,9.0)
All cause: Cook County
Single-pollutant:
Lag 1:2.6% (1.5,3.7)
With PM10:
Lag 1:1.9% (0.6,3.2)
Los Angeles
Single-pollutant:
Lag 1:6.9% (5.4,8.4)
With PM2.5:
Lag 1:7.6% (3.4,12.0)
Posterior means:
All cause:
Single-pollutant:
Lag 1:0.6% (0.3,1.0)
With PM10 and NO2:
Lag 1:0.4% (-0.6,1.4)
Poisson regression coefficient
Single-pollutant:
Lag 1:0.863 (SE = 0.323)
With TSP:
Lag 1:0.230 (SE = 0.489)
(Though SO2 levels were
reported in ppb, these
coefficients must have been
forS02 in ppm.)
Lags: 0 for all cause; 0,1, 2, 3,
4 for cardiovascular
Poisson GAM with default
convergence criteria (only
cardiovascular deaths were
reanalyzed in 2003). Time-
series study.
24-h avg: 3.1 ppb
Copollutants: PM2.5, PM10,
PMio-2.5, CO, NO2, O3, and
selected trace elements, ions,
EC, OC, TOC, and factor
analysis components
cause, cardiovascular
24-h avg median: Cardiovascular;
Cook County: 6 ppb cerebrovascular;
Los Angeles: 2 ppb COPD
Maricopa County: 2 ppb
Copollutants: PM2.5, PM10,03,
NO2, CO; 2- and 3-pollutant
models
24-h avg median: All cause; cardiovascular
Cook County: 6 ppb
Los Angeles: 2 ppb
Copollutants: PM2.5, PM10,03,
NO2, CO; 2-pollutant models
F-80
-------
Study
Methods
Pollutant Data
Outcome
Findings
Schwartz (2000)
Philadelphia, PA
Period of Study:
1974-1988
Lag: 0
Poisson GAM model in 15
winter and 15 summer periods.
The second stage regressed
the TSP and SO2 risk
estimates on SO2/TSP
relationships.
24-h avg summer mean
declined from 20 ppb in 1974
to 9 ppb in 1988; winter mean
declined from 35 ppb in 1974
to 17 in 1988
Copollutants: TSP, extinction
coefficient
All cause
Single-pollutant:
2.3% (1.6, 3.0)
With TSP:
0.4% (-2.2, 3.1)
Schwartz (2004)
14 U.S. cities that had daily
PM10 data
Period of Study:
1986-1993
Lag: 1
Case-crossover design,
estimating PM10 risks by
matching by the levels of
gaseous pollutants.
24-h avg median ranged from All cause
2.2 ppb (Spokane, WA) to 39.4
ppb (Pittsburgh, PA)
Copollutants: PM10 risk
estimates computed, matched
by the levels of SO2, CO, NO2,
and O3
SO2 risk estimates not
computed. PM10 risk estimates
showed the largest risk
estimate when matched for
S02.
Burnett et al. (2004)
12 Canadian cities
Period of Study:
1981-1999
Lag: 1
Poisson GLM. Time-series
study.
24-h avg ranged from 1 ppb
(Winnipeg) to 10 ppb (Halifax)
Copollutants: PM2.5, PMio-2.5,
03, N02, CO
All cause
Single-pollutant:
0.7% (0.3,1.2)
With N02:
0.4% (0.0, 0.8)
Burnett et al. (1998a)
11 Canadian cities
Period of Study:
1980-1991
Lags: 0,1,2,0-1,0-2
examined but the best
lag/averaging for each city
chosen
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg ranged from 1 ppb
(Winnipeg) to 11 ppb
(Hamilton)
Copollutants: O3, NO2, CO
All cause
Single-pollutant:
3.4% (2.0, 4.7)
With all gaseous pollutants:
2.6% (1.3, 3.9)
Burnett et al. (1998b)
Toronto
Period of Study:
1980-1994
Lags: 0,1, 0-1
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 5 ppb
Copollutants: O3, NO2, CO,
TSP, COH, estimated PM10,
estimated PM2.5
All cause
Single-pollutant:
LagO: 1.0% (0.3,1.8)
With CO:
Lag 0: 0.6% (
-0.4,1.5)
Goldberg et al. (2003)
Montreal, Quebec
Period of Study:
1984-1993
Lags: 0,1, 0-2
Poisson GLM with natural
splines. Time-series study.
24-h avg: 6 ppb Congestive heart failure (CHF) CHF as underlying cause of
Copollutants: PM25, coefficient as und<*9 fus.® °f death death: La9 1: "°-1% I"8'9'
of haze, S042-, 03, N02, CO versus those classlfied as
9.6)
having CHF 1 yr prior to death Hgv|ng mF 1 yr pr|Qr ,Q
death: Lag 1: 5.4% (1.3, 9.5)
Vedal et al. (2003)
Vancouver, British Columbia
Period of Study:
1994-1996
Lags: 0,1, 2
Poisson GAM with stringent
convergence criteria. Time-
series study. By season.
24-h avg: 3 ppb
Copollutants: PM10, O3, NO2,
CO
All cause; respiratory;
cardiovascular
Results presented in figures
only.
All cause:
Summer: Lag 0: -3%
Winter: Lag 1: -1%
Villeneuve et al. (2003)
Vancouver, British Columbia
Period of Study:
1986-1999
Lags: 0,1, 0-2
Poisson GLM with natural
splines. Time-series study.
24-h avg: 5 ppb
Copollutants: PM2.5, PM10,
PMio-2.5, TSP, coefficient of
haze, S042-, 03, NO2, CO
All cause; respiratory;
cardiovascular; cancer; SES
All yr:
All cause:
Lag 1:1.7% (-1.1, 4.5)
Cardiovascular:
Lag 1:1.1% (-3.1,5.4)
Respiratory:
Lag 1:8.3% (0.6,16.6)
Anderson et al. (1996)
London, England
Period of Study:
1987-1992
Lag: 1
Poisson GLM. Time-series
study.
24-h avg: 11 ppb
Copollutants: BS, O3, NO2;
2-pollutant models
All cause; respiratory;
cardiovascular
All cause: 1.0% (0.0, 2.0)
Respiratory: 1.7% (-1.3, 4.9)
Cardiovascular: 0.2% (-1.4,
1.8)
F-81
-------
Study
Methods
Pollutant Data
Outcome
Findings
Anderson et al. (2001)
West Midlands region,
England
Period of Study:
1994-1996
Lag: 0-1
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 7.2 ppb
Copollutants: PM10, PM2.5,
PMio-2.5, BS, S042-, 03,
N02, CO
All cause; respiratory;
cardiovascular
All cause: -0.2% (-2.5, 2.1)
Respiratory: -2.2% (-7.4, 3.2)
Cardiovascular: -0.2% (-3.5,
3.1)
Ballester et al. (2002)
13 Spanish cities
Period of Study:
1990-1996
Lags: 0-1 for 24-h avg SO2; 0
for 1-h max SO2
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg SO2 ranged from
2.8 ppb (Sevilla) to 15.6 ppb
(Oviedo)
Copollutants: TSP, BS, PM10
All cause, cardiovascular,
respiratory
All cause:
Lag 0-1: 1.4% (0.2, 2.7)
Cardiovascular:
Lag 0-1:1.4% (-0.4, 3.3)
Respiratory:
Lag 0-1: 3.5% (1.0, 6.0)
Biggeri et al. (2005)
8 Italian cities
Period of study variable
between 1990-1999
Lag: 0-1
Poisson GLM. Time-series
study.
24-h avg ranged from 2 ppb
(Verona) to 14 ppb (Milan)
Copollutants: O3, NO2, CO,
PM10
All cause; respiratory;
cardiovascular
All cause: 4.1% (1.1, 7.3)
Respiratory: 7.4% (-3.6,19.6)
Cardiovascular: 4.9% (0.4,
9.7)
Bremneret al. (1999)
London, England
Period of Study:
1992-1994
Lags: Selected best from 0,1,
2, 3, (all cause); 0,1, 2, 3, 0-1,
0-2, 0-3 (respiratory,
cardiovascular)
Poisson GLM. Time-series
study.
24-h avg: 7 ppb
Copollutants: BS, PM10, O3,
NO2, CO; 2-pollutant models
All cause; respiratory;
cardiovascular; all cancer; all
others; all ages; age specific
(0-64, 65+, 65-74, 75+ yrs)
All cause: Lag 1:1.6% (-0.5,
3.7)
Respiratory: Lag 2: 4.8%
(-0.2,10.0)
Cardiovascular Lag 1:
1.3% (-1.7, 4.3)
Clancy et al. (2002)
Dublin, Ireland
Period of Study:'
1984-1996
NA
Comparing standardized
mortality rates for 72 mos
before and after the ban on
coal sales in Sep 1990.
24-h avg:
1984-1990:11.7 ppb
1990-1996:7.7 ppb
Copollutants: BS
All cause, cardiovascular, and
respiratory
BS mean declined by a larger
percentage (70%) than SO2
(34%) between the two
periods.
All cause death rates reduced
by 5.7% (4, 7); respiratory
deaths by 15.5% (12,19);
cardiovascular deaths by
10.3% (8,13).
Dab et al. (1996)
Paris, France
Period of Study:
1987-1992
Lag: 1 24-h avg: 11.2 ppb
Poisson autoregressive. Time- ^ max' ^ ® P'3'3
series study. Copollutants: BS, PM13, O3,
N02, CO
Respiratory
Lag 1:2.3% (-0.9, 5.5)
Diaz et al. (1999)
Madrid, Spain
Period of Study:
1990-1992
Lag: 1
24-h avg: Levels NR.
Autoregressive OLS Copollutants: TSP, O3, NO2,
regression. Time-series study. CO
All cause; respiratory;
cardiovascular
Only significant regression
coefficients were shown, but
description of the table was
not clear enough to derive risk
estimates.
Fischer etal. (2003)
The Netherlands
Period of Study:
1986-1994
Lags: 0-6
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg median: 3.5 ppb
Copollutants: PM10, BS, O3,
N02, CO
All-cause, cardiovascular,
COPD, and pneumonia in age
groups < 45, 45-64, 65-74,
75+
Cardiovascular: Age < 45 yrs:
4.3% (-4.6, 13.9); Age 45-64
yrs: -0.5% (-3.6, 2.7); Age 65-
74 yrs: 1.6% (-0.8, 4.2); Age
75+ yrs: 2.8% (1.3, 4.3)
F-82
-------
Study Methods Pollutant Data Outcome Findings
Garcia-Aymerich et al. (2000)
Selected best averaged lag
Levels NR.
All cause; respiratory;
All cause:
Barcelona, Spain
Poisson GLM. Time-series
study.
Copollutants: BS, O3, NO2
cardiovascular; general
population; patients with
General population:
Lag 0-3: 4.4% (2.3, 6.5);
Period of Study:
COPD
COPD patients:
1985-1989
Lag 0-2:2.6% (-5.0, 10.7)
Respiratory:
General population: Lag 0-1:
3.5% (-0.6, 7.8)
COPD patients:
Lag 0-2:2.3% (-8.9, 15.0)
Cardiovascular:
General population: Lag 0-3:
5.1% (2.3, 8.0)
COPD patients: Lag 0-2: 2.0%
(-11.5,17.5)
Hoek et al. (1997)
Lag: 1
24-h avg median:
All cause
Single-pollutant:1.5% (0.0, 3.0)
Rotterdam, the Netherlands
Poisson GAM with default
7.7 ppb
With TSP and 03: 0.5% (-1.2,
Period of Study:
1983-1991
convergence criteria. Time-
series study.
Copollutants: TSP, BS, Fe, O3,
CO
2.3)
Hoek et al. (2001; reanalysis Lag: 1, 0-6
2003)
The Netherlands: Entire
country, four urban areas
Period of Study:
1986-1994
24-h avg median: 3.5 ppb in
Poisson GAM, reanalyzed with
stringent convergence criteria;
Poisson GLM.
Time-series study.
the Netherlands; 5.6 ppb in the cardiovascular
four major cities
Copollutants: PM10, BS, SO42-,
NOs-, 03, N02, CO;
2-pollutant models
I cause; COPD; pneumonia; Poisson GLM:
All cause:
Lag 1:1.3% (0.7,1.9)
Lag 0-6:1.8% (0.9,2.7) With
BS: 1.1% (-0.3, 2.4)
Cardiovascular:
Lag 0-6: 2.7% (1.3, 4.1)
COPD:
Lag 0-6: 3.6% (-0.3, 7.7)
Pneumonia:
Lag 0-6: 6.6% (1.2,12.2)
Hoek et al. (2001); reanalysis
Hoek (2003)
The Netherlands
Period of Study:
1986-1994
Lag: 0-6
Poisson GAM, reanalyzed with
stringent convergence criteria;
Poisson GLM. Time-series
study.
24-h avg median: 3.5 ppb in
the Netherlands; 5.6 ppb in the
four major cities
Copollutants: PM10, O3, NO2,
CO
Total cardiovascular;
myocardial infarction;
arrhythmia; heart failure;
cerebrovascular; thrombosis-
related
Poisson GLM:
Total cardiovascular: 2.7%
(1.3,4.1)
Myocardial infarctioN: 0.8%
(-1.2,2.8)
Arrhythmia: 2.3% (-3.9, 8.8)
Heart failure: 7.1% (2.6,11.7)
Cerebrovascular: 4.4%
(1.4,7.5)
Thrombosis-related: 9.6% (3.1,
16.6)
Katsouyanni et al. (1997) "Besf lag variable across 24-h avg median of the Allcause All cities: 1.1% (0.9, 1.4)
12 European cities cltles from 0 to 3 ™edlan across clt'es was Western cities: 2.0% (1.2, 2i
Pnissnnantnrenressive Time 14 ppb, ranging from 5 ppb
Period of Studys vary by city, " (Bratislava) to 26 ppb Central eastern cities: 0.5%
ranging from 1977 to 1992 series study. (Cracow) (-0.4,1.4)
Copollutants: BS, PM10
F-83
-------
Study
Methods
Pollutant Data
Outcome
Findings
Keatinge and Donaldson
(2006)
London, United Kingdom
Period of Study:
1991-2002
Lags: Mean of 0, -1, -2
Graphic analysis and GAM.
Time-series study.
24-h avg: Levels NR
Copollutants: 03, PM10
All-cause
Relative Risk for a 106
Increase in Mortality (per 10
ppb SO2)
SO2 + Temp: 3.1 (0.6, 5.5)
SO2 + Temp + Acclim.: 2.2
(-0.1,4.6)
SO2 + Temp + Acclim. +
Acclim. x Temp: 2.5 (0.2, 4.8)
SO2 + Temp + Acclim. +
Acclim. x Temp + Sun:
2.3 (-0.03, 4.5)
SO2 + Temp + Acclim. +
Acclim. x Temp + Sun + Wind:
1.6 (-0.7,3.8)
SO2 + Temp + Acclim. +
Acclim. x Temp + Sun + Wind
+ Abs. Humidity: 1.7 (-0.6,
3.9)
SO2+ Temp + Acclim. +
Acclim. x Temp + Sun + Wind
+ Abs. Humidity + Rain: 1.8
(-0.4, 4.1)
SO2 + Temp. + Abs. Humidity:
2.5 (0.03, 4.9)
Kotesovec etal. (2000)
Northern Bohemia, Czech
Republic
Period of Study:
1982-1994
Lags: 0,1,2, 3, 4,5,6,0-6
Poisson GLM, time-series
study
24-h avg: 34.9 ppb All cause, cardiovascular (only All cause:
Copollutants: TSP age = < 65 presented), cancer Lag 1: 0.1% (-0.1, 0.4)
Le Tertre et al. (2002) Lags: 0-1
Bordeaux , Le Havre, Lille, Poisson GAM with default
Lyon, Marseille, Paris, Rouen, convergence criteria. Time-
Strasbourg, France series study.
Period of Study varies by city,
ranging from 1990-1995
24-h avg ranged from 3 ppb
(Bordeaux) to 9 ppb (Rouen)
Copollutants: BS, O3, NO2
All cause; respiratory;
cardiovascular
8-city pooled estimates:
All cause:
2.0% (1.2, 2.9)
Respiratory: 3.2% (0.1, 6.3)
Cardiovascular: 3.0% (1.5,
4.5)
Michelozzi et al. (1998)
Rome, Italy
Period of Study:
1992-1995
Lags: 0,1, 2, 3, 4
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 5.7 ppb
Copollutants: PM13, NO2, O3,
CO
All-cause
Lag 1: -2.0% (-4.4,0.5);
(negative estimates at all lags
examined)
Peters et al. (2000b)
NE Bavaria, Germany 1982-
1994
Coal basin in Czech Republic
Period of Study:
1993-1994
Lags: 0,1, 2, 3
Poisson GLM. Time-series
study.
24-h avg:
Czech Republic: 35 ppb
Bavaria, Germany: 14 ppb
Copollutants: TSP, PM10, O3,
N02, CO
All cause; respiratory;
cardiovascular; cancer
Czech Republic:
All cause:
Lag 1:0.8% (-0.2,1.8)
Bavaria, Germany:
All cause:
Lag 1:0.3% (-0.3,0.9)
Ponka et al. (1998)
Helsinki, Finland
Period of Study:
1987-1993
Lags: 0,1,2,3, 4,5,6,7
Poisson GLM. Time-series
study.
24-h avg median:
3.5 ppb
Copollutants: TSP, PM10, O3,
N02
All cause; cardiovascular; age
< 65 yrs, age 65+ yrs
No risk estimate presented for
SO2. PM10 and O3 were
reported to have stronger
associations.
F-84
-------
Study
Methods
Pollutant Data
Outcome
Findings
Prescott et al. (1998)
Edinburgh, Scotland
Period of Study:
1992-1995
Lag: 0
Poisson GLM. Time-series
study.
24-h avg: 1981-1995:15 ppb
1992-1995: 8 ppb
Copollutants: BS, PM10, O3,
NO2, CO; 2-pollutant models
All cause; respiratory;
cardiovascular; all ages; age <
65 yrs; age 65+ yrs
Results presented as figures
only. Essentially no
associations in all categories.
Very wide confidence intervals.
Rahlenbeck and Kahl (1996)
East Berlin, Germany
Period of Study:
1981-1989
Lags: 0,1, 2, 3, 4, 5
OLS, with log ofS02, Time-
series study.
24-h avg: 61.9 ppb
"SP" (beta absorption)
All cause
Single-pollutant:
Lag 1:4.4% (0,8.7);
With SP:
Lag 1:2.9% (-2.7,8.5)
Roemer and van Wijinen
(2001)
Amsterdam, the Netherlands
Period of Study:'
1987-1998
Lags: 1, 2, 0-6
Poisson GAM with default
convergence criteria (only one
smoother). Time-series study.
24-h avg:
Background sites: 3.1 ppb
Traffic sites: 4.2 ppb
Copollutants: BS, PM10, O3,
N02, CO
All cause
Total population using
background sites:
Lag 1:2.6% (-0.6,5.8)
Traffic population using
background sites:
Lag 1:0.6% (-6.9,8.6)
Total population using traffic
sites: Lag 1:2.4% (-0.3, 5.1)
Saez et al. (1999)
Lags: 0-1
Levels NR.
Asthma mortality; age 2-45 yrs
RR = 1.9 (0.7, 4.4)
Barcelona, Spain
Period of Study:
1986-1989
Poisson with GEE. Time-
series study.
Copollutants: BS, O3, NO2,
Saez et al. (2002)
Seven Spanish cities
Variable periods of study
between 1991 and 1996
Lags: 0-3
Poisson GAM with default
convergence criteria. Time-
series study.
Values for SO2 NR.
Copollutants: O3, PM, NO2,
CO
All cause; respiratory;
cardiovascular
Risk estimates for SO2 was
NR. Including SO2 in
regression model did not
appear to reduce NO2 risk
estimates.
Spix and Wichman (1996)
Lags: 0,1, 0-3
24-h avg: 15 ppb
All-cause
Lag 1:0.8% (0.2,1.4)
Koln, Germany
Poisson GLM. Time-series
1-h max: 32 ppb
Period of Study:
1977-1985
study.
Copollutants: TSP, PM7, NO2
Sunyeretal. (2002)
Barcelona, Spain
Period of Study:'
1986-1995
Lags: 0-2
Conditional logistic (case-
crossover)
24-h avg median:
6.6 ppb
Copollutants: PM10, BS, NO2,
03, CO, pollen
All cause, respiratory, and
cardiovascular mortality in a
cohort of patients with severe
asthma
Odds ratio: Patients with 1
asthma admission:
All cause: 14.8% (-19.8,64.4)
Patients with more than 1
asthma adm: All cause:
50.4% (-48.6, 340.4)
Patients with more than 1
asthma or COPDadm: All
cause: 20.2% (-17.5, 75.0)
NO2 and O3 were more
strongly associated with
outcomes than SO2.
Sunyeretal. (1996)
Barcelona, Spain
Period of Study:
1985-1991
Selected best single-day lag 24-h avg median:
Autoregressive Poisson. Time- Summer: 13 ppb
series study. Winter: 16 ppb
Copollutants: BS, NO2, O3
All cause; respiratory;
cardiovascular; all ages; age
70+ yrs
All yr, all ages: All cause:
Lag 1:3.5% (1.9,5.1)
Respiratory:
Lag 0: 3.5% (-0.2, 5.0)
Cardiovascular:
Lag 1:2.2% (0.5,3.9)
Verhoeff et al. (1996)
Amsterdam, the Netherlands
Period of Study:
1986-1992
Lags: 0,1, 2
Poisson GLM. Time-series
study.
24-h avg: 4.5 ppb
Copollutants: BS, PM10, O3,
CO; multipollutant models
I cause; all ages; age 65+
yrs
Single-pollutant:
Lag 1:1.4% (-1.4, 4.2)
With BS:
-3.7% (-8.1, 0.9)
F-85
-------
Study
Methods
Pollutant Data
Outcome
Findings
Zeghnoun et al. (2001)
Rouen and Le Havre, France
Period of Study:
1990-1995
Lags: 0,1,2,3,0-3
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: Rouen: 10 ppb
Le Havre: 12 ppb
Copollutants: NO2, BS, PM13,
03
All cause; respiratory;
cardiovascular
All cause:
Rouen: Lag 1: 2.3% (-1.1, 5.9)
Le Havre: Lag 1:1.1% (-0.3,
2.5)
Zmirou et al. (1996)
Lyon, France
Period of Study:
1985-1990
Lags: Selected best from 0,1,
2,3
Poisson GLM. Time-series
study.
24-h avg: 16 ppb
Copollutants: PM13, NO2, O3
All cause; respiratory;
cardiovascular; digestive
All cause:
Lag 0:3.4% (1.4,5.4)
Respiratory:
Lag 3: 2.8% (0.9, 4.8)
Cardiovascular:
Lag 0-3: 4.5% (2.0, 7.0)
Zmirou et al. (1998)
10 European cities
Period of Studys vary by city,
ranging from 1985-1992
Lags: 0,1, 2, 3,0-1,
0-2, 0-3 (best lag selected for
each city)
Poisson GLM. Time-series
study.
24-h avg: Cold Season: Respiratory; cardiovascular
Ranged from 12 ppb (London)
to 87 ppb (Milan) ppb
Warm Season: Ranged from 5
ppb (Bratislava) to 21 ppb
(Cracow) in warm season
Copollutants: BS, TSP, NO2,
03
Western cities:
Respiratory: 2.8% (1.7, 4.0)
Cardiovascular: 2.3% (0.9,
3.7)
Central eastern cities:
Respiratory: 0.6% (-1.1, 2.3)
Cardiovascular: 0.6% (0.0,
1.1)
Simpson et al. (1997)
Brisbane, Australia
Period of Study:
1987-1993
LATIN AMERICA
Lag: 0
Autoregressive Poisson with
GEE. Time-series study.
24-h avg: 4.2 ppb
1-h max: 9.6 ppb
Copollutants: PM10, bsp, 03,
N02, CO
All cause; respiratory;
cardiovascular
All cause:
All yr: Lag 0: -2.8% (-2.7, 8.6)
Summer:
Lag 0:2.8% (-8.3,15.2)
Winter:
Lag 0: 2.8% (-3.9, 9.8)
Borja-Aburto et al. (1998)
SW Mexico City
Period of Study:
1993-1995
Lags: 0,1, 2, 3, 4, 5, and
multiday avg.
Poisson GAM with default
convergence criteria (only one
smoother). Time-series study.
24-h avg: 5.6 ppb
Copollutants: PM2.5, O3, NO2;
2-pollutant models
All cause; respiratory; SO2 risk estimates NR. PM2.5
cardiovascular; other; all ages; and O3 were associated with
age >65 yrs mortality.
Borja-Aburto et al. (1997)
Mexico City
Period of Study:
1990-1992
Lags: 0,1, 2
Poisson iteratively weighted
and filtered least-squares
method. Time-series study.
24-h avg median:
5.3 ppb
Copollutants:
TSP, O3CO;
2-pollutant models
All cause; respiratory;
cardiovascular; all ages; age <
5 yrs; age >65 yrs
All-cause:
Lag 0:0.2% (-1.1,1.5)
Cardiovascular:
Lag 0:0.7% (-1.6,3.0)
Respiratory:
Lag 0:-1.0% (-5.0, 3.2)
Cakmak et al. (2007b)
7 Chilean urban centers
Period of Study:
1997-2003
Lags: 0,1, 2, 3, 4, 5, 0-5
Poisson GLM with random
effects between cities. Time-
series study.
24-h avg ranged from 9.12 ppb
(Las Condes) to 64.06 ppb
(Independencia)
Population-weighted avg
concentration: 14.08 ppb
Copollutants: PM10, O3, CO
All cause; respiratory;
cardiovascular; all ages; age
< 65 yrs; age
65-74 yrs; age 75-84 yrs; age
85+ yrs
All cause: All ages:
Single-pollutant:
Lag 1:4.0% (2.4, 5.6)
Lag 0-5: 6.5% (4.5, 8.5)
Multipollutant:
Lag 1:3.2% (1.3,5.1)
< 65 yrs:
Lag 0-5: 3.0% (0.6, 5.5)
65-74 yrs:
Lag 0-5:5.1% (1.2,9.1
75-84 yrs:
Lag 0-5: 7.8% (4.1,11.6)
85+ yrs: 7.8% (4.2,11.5)
Warm Season:
Lag 0-5: 7.2% (4.1,10.3)
Cool Season: Lag 0-5: 3.0% (
0.4, 6.5)
F-86
-------
Study
Methods
Pollutant Data
Outcome
Findings
Cifuentes et al. (2000)
Santiago, Chile
Period of Study:
1988-1966
Lags: 1-2
Poisson GAM with default
convergence criteria;
Poisson GLM. Time-series
study.
24-h avg: 18.1 ppb
Copollutants: PM2.5, PMio-2.5,
CO, N02, 03
All cause
Poisson GLM: Single-pollutant:
Lag 1-2:0.2% (-0.9, 1.3)
With other pollutants: Lag 1-2:
-0.6% (-1.7, 0.5)
Conceigao et al. (2001)
Sao Paulo, Brazil
Period of Study:
1994-1997
Lag: 2
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 7.4 ppb
Copollutants: PM10, CO, O3
Child mortality (age under 5
yrs)
Single-pollutant:
Lag 2:17.0% (7.0,28.0);
With all other pollutants:
Lag 2:13.7% (-1.1, 30.8)
Loomis et al. (1999)
Mexico City
Period of Study:
1993-1995
Lags: 0,1, 2, 3, 4, 5, 3-5
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 5.6 ppb
Copollutants: PM2.5, O3
Infant mortality
SO2 risk estimates NR. PM2.5
and O3 were associated with
mortality.
Ostro et al. (1996)
Lag: 0
1-h max: 60 ppb
All cause
Lag 0:0.7% (-0.3,1.7)
Santiago, Chile
Period of Study:
1989-1991
OLS, Poisson. Time-series
study.
Copollutants:
PM10,03, NO2;
2-pollutant models
Pereira et al. (1998)
Sao Paulo, Brazil
Period of Study:
1991-1992
Lag: 0
Poisson GLM. Time-series
study.
24-h avg: 6.6 ppb
Copollutants: PM10, O3, NO2,
CO
Intrauterine mortality
Single-pollutant model: 11.5%
(-0.3, 24.7)
With other pollutants: 8.6%
(-8.7, 29.3)
Saldiva et al. (1994)
Lags: 0-2
24-h avg: 6.0 ppb
Respiratory; age < 5 yrs
-1.0% (-47.1, 45.1)
Sao Paulo, Brazil
Period of Study:
OLS of raw or transformed
data. Time-series study.
Copollutants: PM10, O3, NO2,
CO; multipollutant models
1990-1991
Saldiva et al. (1995)
Lag: 0-1
24-h avg: 6.5 ppb
All cause; age 65+yrs
Single-pollutant: 8.5%
Sao Paulo, Brazil
Period of Study:
1990-1991
OLS; Poisson with GEE. Time-
series study.
Copollutants: PM10, O3, NO2,
CO; 2-pollutant models
(1.3,15.6)
With other pollutants:
-3.1% (-13.0,6.9)
1 ASIA 1
Ha et al. (2003)
Seoul, Korea
Period of Study:
1995-1999
Lag: 0
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 11.1 ppb
Copollutants: PM10, O3, NO2,
CO
All cause; respiratory;
postneonatal (1 mo to 1 yr);
age
2-64 yrs; age 65+
All cause:
Postneonates: 11.3% (4.0,
19.1)
Age 65+ yrs: 3.2% (3.1, 3.3)
Hong et al. (2002)
Lag: 2
24-h avg (ppb): 12.1 (7.4)
Stroke
% increase (per 5.7 ppb SO2)
Seoul, Korea
Period of Study:
1995-1998
GAM with default convergence Copollutants: PM10, NO2
criteria. Time-series study. CO, O3
2.9% (0.8, 5.0) lag 2
Stratified by PM10
(Median: 47.4 (jg/m3)
Med: 3.8%
Hong et al. (2002)
Lag: 2
24-h avg: 12.1 ppb
Acute stroke mortality
5.2% (1.4, 9.0)
Seoul, Korea
Period of Study:
1995-1998
Poisson GAM with default
convergence criteria. Time-
series study.
Copollutants: PM10, O3, NO2,
CO
F-87
-------
Study
Methods
Pollutant Data
Outcome
Findings
Kwon et al. (2001)
Seoul, Korea
Period of Study:
1994-1998
Lag: 0
Poisson GAM with default
convergence criteria; case-
crossover analysis using
conditional logistic regression.
24-h avg: 13.4 ppb
Copollutants: PM10, O3, NO2,
CO
Mortality in a cohort of patients
with congestive heart failure
Odds ratio in general
population: 1.0% (-0.1, 2.1)
Congestive heart failure
cohort: 6.9% (-3.4,18.3)
Lee et al. (2000)
Seoul, Korea
Period of Study:
2000-2004
Lag: 1
GAM with stringent
convergence criteria. Time-
series study.
24-h avg (ppb): 5.20 (2.17)
Copollutants: PM10, CO, NO2,
03
Non-accidental
% Increase (per 3.06 ppb SO2)
2.7(1.8,3.5) lag 1
Lee et al. (1999)
Seoul and Ulsan, Korea
Period of Study:
1991-1995
Lags: 0-2
Poisson with GEE. Time-
series study.
1-h max: Seoul: 26 ppb
Ulsan: 31 ppb
Copollutants: TSP, O3
All cause
Seoul: 1.5% (1.1,1.9)
Ulsan: 1.0% (-0.2,2.2)
Lee and Schwartz (1999)
Seoul, Korea
Period of Study:
1991-1995
Lags: 0-2
Conditional logistic regression.
Case-crossover with
bidirectional control sampling.
1-h max: 26 ppb
Copollutants: TSP, O3
All cause
Two controls, ± 1 wk:
0.3% (-0.5, 1.0)
Four controls, ± 2 wks:
1.0% (0.3,1.6)
Lee et al. (2007)
7 Korean cities
Period of Study:
1991-1997
Lags: 0-1
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg SO2 ranged from
12.1 ppb (Kwangju) to 31.4
ppb (Taegu)
Copollutants: TSP, NO2,03,
CO
All cause
Single-pollutant:
Lag 0-1 : 0.6% (0.3, 0.8)
Multipollutant:
Lag 0-1 : 0.6% (0.2, 0.9)
Non-accidental, Mean % change (per 10 |jg/m3
cardiovascular, stroke, cardiac, SO2)
respiratory, cardiopulmonary Non-accidental:
All Ages: 0.01 (-0.46,0.47)
<65:-0.55 (-1.33,0.23)
>65:0.22 (-0.32,0.76)
Cardiovascular:
All Ages: 0.20 (-0.45,0.86)
<65:-0.63 (-1.96,0.72)
>65:0.41 (-0.31,1.14)
Stroke:
All Ages:-0.27 (-1.04,0.51)
<65:-1.35 (-3.01,0.33)
>65:0.01 (-0.87,0.88)
Cardiac:
All Ages: 0.88 (-0.22,1.99)
<65:0.29 (-2.11,2.75)
>65:1.01 (-0.18,2.21)
Respiratory:
All Ages: 1.13 (-0.28, 2.56)
<65:-0.59 (-4.24,3.19)
>65:1.36 (-0.05,2.80)
Cardiopulmonary:
All Ages: 0.29 (-0.33,0.92)
< 65: -0.80 (-2.07, 0.49)
>65:0.53 (-0.15,1.20)
The levels of SO2 in these All-cause The estimates were found to
Asian cities were generally be heterogeneous across 11
higher than those in the U.S. studies. Random-Effects
or Canadian cities, with more Estimate: 1.49% (95% CI:
than half of these studies 0.86, 2.13); Fixed-Effects
reporting the mean SO2 levels Estimate: 1.01% (95% CI:
higher than 10 ppb. 0.73,1.28).
Copollutants considered varied
across studies.
Qian et al. (2007) Lag: 0 24-h avg (|jg/m3): 44.1 (25.3)
Wuhan, China Poisson GAM with stringent Copollutants:
Period of Study ¦ convergence criteria. Time- PM10
2000-2004
series study. NO2
03
HEI (2004) The lags and multi-day
East As,an c,ties averaging used ,n varied
Meta-analysis of time-series
study results
F-88
-------
Study
Methods
Pollutant Data
Outcome
Findings
Tsai et al. (2003b)
Kaohsiung, Taiwan
Period of Study:
1994-2000
Lags: 0-2
Conditional logistic regression.
Case-crossover analysis.
24-h avg: 11.2 ppb
Copollutants: PM10, NO2, O3,
CO
All cause; respiratory;
cardiovascular; tropical area
Odds ratios:
All cause: 1.1% (-4.4, 6.8)
Respiratory:
3.5% (-17.6, 29.9)
Cardiovascular:
2.4% (-9.1, 15.4)
Venners et al. (2003)
Chonqing, China
Period of Study:
1995
Lags: 0,1, 2, 3, 4, 5
Poisson GLM, time-series
study
24-h avg: 74.5 ppb
CopollutantsPMzs
All cause, cardiovascular,
respiratory, cancer, and other
All cause:
Lag 2:1.1% (-0.1,2.4)
Cardiovascular:
Lag 2: 2.8% (0.4, 5.2)
Respiratory:
Lag 2: 3.0% (0.4, 5.7)
Wong et al. (2001a)
Hong Kong
Period of Study:
1995-1997
Lags: 0,1, 2
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg:
Warm Season: 6.4 ppb
Cool Season: 6.0 ppb
Copollutants: PM10, O3, NO2;
2-pollutant models
All cause; respiratory;
cardiovascular
All cause: Lag 1: 3.2% (1.1,
5.3)
Respiratory: Lag 0: 5.3% (2.2,
8.6)
Cardiovascular: Lag 1: 4.3%
(1.1,7.5)
Wong et al. (2002b)
Hong Kong
Period of Study:
1995-1998
Lags: 0,1, 2, 0-1, 0-2
Poisson GLM. Time-series
study.
24-h avg: 29 ppb
Copollutants: PM10, O3, NO2;
2-pollutant models
Respiratory; cardiovascular;
COPD; pneumonia and
influenza; ischemic heart
disease; cerebrovascular
Respiratory: Lag 0-1: 2.6%
(0.2,5.1)
Cardiovascular: Lag 0-1:1.2%
(-1.0,3.5)
Yang et al. (2004a)
Taipei, Taiwan
Period of Study:
1994-1998
Lags: 0-2
Conditional logistic regression.
Case-crossover analysis.
24-h avg: 5.5 ppb
Copollutants: PM10, NO2, O3,
CO
All cause; respiratory;
cardiovascular; subtropical
area
Odds ratios:
All cause: -0.5% (-7.0, 6.6)
Respiratory:-1.8% (-23.1,
25.3);
Cardiovascular: -3.4% (-15.2,
10.0)
F-89
-------
Table F-6. Long-term exposure to SO2 and respiratory morbidity.
Study
Methods
Pollutant Data
Findings
UNITED STATES AND CANADA
Dockery et al. (1989)
Kingston-Harriman,
TN; Portage, Wl; St.
Louis, MO;
Steubenville, OH;
Topeka, KS;
Watertown, MA
Period of Study:
1980-1981 school yr
Cross-sectional assessment of the association between
air pollution and chronic respiratory health of 5,422 (10-
12 yrs) white children examined in the 1980-1981 school
yr. Children were part of the cohort of children in the Six
Cities Study of Air pollution and Health. Symptoms were
analyzed using logistic regression that included sex, age,
indicators of parental education, maternal smoking,
indicator for gas stove, and an indicator for city.
Respiratory symptoms investigated were bronchitis,
chronic cough, chest illness, persistent wheeze, asthma.
The logarithm of pulmonary function was fitted to a
multiple linear regression model that included sex, sex-
specific log of height, age, indicators of parental
education, maternal smoking, a gas stove indicator, and
city indicator. Annual means of the 24 h avg air pollutant
concentration for the 12 mos preceding the examination
of each child was calculated for each city.
Daily mean
concentrations,
averaging hourly
concentrations for each
day with at least
18 hourly values
Portage: 4.2 ppb
Topeka: 3.5
Watertown: 10.5
Kingston: 6.5
St. Louis: 13.5
Steubenville: 27.8
No significant associations between SO2 and any
pulmonary function measurements. No significant
association between SO2 and symptoms.
Relative odds and 95% CI between most/least polluted
cities:
Bronchitis: 1.5 (0.4, 5.8)
Chronic cough: 1.8 (0.3,12.5)
Chest illness: 1.5 (0.4, 5.9)
Persistent wheeze: 0.9 (0.4,1.9)
Asthma: 0.6 (0.3, 1.2)
Reference symptoms:
Hay fever: 0.6 (0.2,1.7)
Ear ache: 1.2 (0.3, 5.3)
Nonrespiratory illness: 1.0 (0.6,1.5)
Analysis stratified by asthma or persistent wheeze
bronchitis
No wheeze or asthma 1.5 (0.5, 4.3)
Yes wheeze or asthma 2.0 (0.3,14.3)
Chronic cough
No wheeze or asthma 2.4 (0.5,11.7)
Yes wheeze or asthma 1.9 (0.1, 44.1)
Chest illness
No wheeze or asthma 1.5 (0.4, 5.6)
Yes wheeze or asthma 1.9 (0.3,13.0)
Dockery et al. (1996) Study of the respiratory health effects of acid aerosols in S02avg 4.8 ppm
18 sites in U.S.
6 sites in Canada
Period of Study:
1988-1991
13,369 white children aged 8 to 12 yrs old from 24
communities in the U.S. and Canada between 1988 and
1991. Information was gathered by questionnaire and a
pulmonary function.
SD 3.5
Range 0.2,12.9
With the exception of the gaseous acids (nitrous and
nitric acid), none of the particulate or gaseous
pollutants, including SO2, were associated with
increased asthma or any asthmatic symptoms. Stronger
associations with particulate pollutants were observed
for bronchitis and bronchitic symptoms.
Odds Ratio (95% CI) for 12.7 ppb range of SO2
pollution
Asthma 1.05 (057,1.93)
Attacks of Wheeze 1.07 (0.75,1.55)
Persistent Wheeze 1.19 (0.80,1.79)
Any asthmatic symptoms 1.16 (0.80,1.68)
Bronchitis 1.56 (0.95, 2.56)
Chronic cough 1.02 (0.66,1.58)
Chronic phlegm 1.55 (1.01, 2.37)
Any Bronchitic symptoms 1.29 (0.98,1.71)
F-90
-------
Study
Methods
Pollutant Data
Findings
Euleret al. (1987)
California, USA
Cross-sectional study of 7,445 (25 yrs or older) Seventh- None provided
Day Adventists who lived in their 1977 residential areas
(Los Angeles and it border counties, San Francisco, and
San Diego) for at least 10 yrs to determine the effect of
long-term cumulative exposure to ambient levels of TSP
and SO2 on COPD symptoms. Study population is
subgroup of NCI-funded ASHMOG study that enrolled
36,805 Seventh-Day Adventists in 1974. Each
participant's cumulative exposure to the pollutant
exceeding 4 different threshold levels were estimated
using moly residence ZIP code histories and interpolated
dosages from state monitoring stations. Participants
completed a questionnaire on respiratory symptoms,
smoking history, occupational history, and residence
history.
Study reported that SO2 exposure was not associated
with symptoms of COPD until concentrations exceeded
4 ppm. The correlation coefficient of SO2 (above 4 ppm)
with TSP (above 200 (jg/m3) the highest exposure
levels for these two pollutants was 0.30; thus, the
authors believed that it was possible to separate the
effects of SO2 from TSP. Multiple regressions used in
the analysis. No significant effect at exposures levels
below 4 ppm or above 8 ppm.
Relative risk estimate (based on 1,003 cases)
SO2 exposure above 2 ppm during 11 yrs of study
2000 h/yr: 1.09 1000 h/yr: 1.04
500 h/yr: 1.03
SO2 exposure above 4 ppm
500 h/yr: 1.18 250 h/yr: 1.09
100 h/yr: 1.03
SO2 above 8 ppm
60 h/yr: 1.07 30 h/yr: 1.03
15 h/yr: 1.02
SO2 above 14 ppm
10 h/yr: 1.03 5 h/yr: 1.01
1 h/yr: 1.00
Goss et al. (2004)
U.S. nationwide
Period of Study:
1999-2000
Cohort study of 18,491 cystic fibrosis patients over 6 yrs
of age who were enrolled in the Cystic Fibrosis
Foundation National Patient Registry in 1999 and 2000.
Mean age of patients was 18.4 yrs; 92% had pancreatic
insufficiency. Air pollution from the Aerometric
Information Retrieval System linked with patient's home
ZIP code. Air pollutants studied included O3, NO2, SO2,
CO, PM10, and PM2.5. Health endpoints of interest were
pulmonary exacerbations, lung function, and mortality.
However, study did not have enough power to assess
the outcome of mortality. Logistic regression and
polytomous regression models that adjusted for sex,
age, weight, race, airway colonization, pancreatic
function, and insurance status were used.
Mean (SD): 4.91
(2.6) ppb
Median: 4.3 ppb
IQR: 2.7-5.9 ppb
With the single-pollutant model, no significant
association between SO2 and pulmonary
exacerbations.
Odds ratio per 10 ppb increase in SO2:
0.83 (95% CI: 0.71,1.01), p = 0.068
No clear association between pulmonary function and
SO2. No effect estimates provided.
McDonnell et al.
(1999)
California, U.S.
Period of Study:
1973-1992
Prospective study (over 15 yrs) of 3,091 nonsmokers
aged 27-87 yrs that evaluated the association between
long-term ambient O3 exposure and the development of
adult-onset asthma. Cohort consisted of nonsmoking,
non-Hispanic white, California Seventh Day Adventists
who were enrolled in 1977 in the AHSMOG study.
Logistic regression used to assess the association
between the 1973-1992 mean 8-h avg ambient O3
concentration and the 1977-1992 incidence of doctor-told
asthma. Levels of PM10, NO2, and S04were measured
but no effect estimates were given.
Mean: SO2 6.8 |jg/m3
Range: 0.0-10.2 |jg/m3
Correlation coefficient
r = 0.25 with O3
No significant positive association between SO2 and
asthma for males or females. Addition of a second
pollutant to the O3 model for the male subjects, did not
result in a decrease of more than 10% in the magnitude
of the regression coefficient for O3, and for the females
addition did not cause the coefficient for O3 to become
significantly positive
Schwartz (1989) Cross-sectional study using data from the Second
United States National Health and Nutrition Examination Survey
(NHANES II) to examine the relation between air
Period of Study: Feb pollution and lung function growth in 4,300 children and
1976-Feb 1980 youths 6-24 yrs old. A two-staged analysis was
performed that consisted of (1) regression equations
including factors known to affect lung function and (2) a
regression of the residuals ofthe first regression on air
pollution.
Annual percentiles
(ppm):
10th
25th
50th
75th
90th
0.0060
0.0106
0.0131
0.0159
0.0193
The study did not find an association between SO2 and
any ofthe lung function growth measurements (i.e.,
FVC, FEV1, and Peak flow).
F-91
-------
Study
Methods
Pollutant Data
Findings
Ackermann-Liebrich
et al. (1997)
8 communities in
Switzerland
Aarau, Basel, Davos,
Geneva, Lugano,
Montana, Payerne,
and Wald
Period of Study:
1991-1993
Cross-sectional population based study of 9,651 adults Avg SO2 in 1991 (|jg/m3)
(18-60 yrs) in 8 areas in Switzerland (SAPALDIA), to Mean: 11.7
evaluate the effect of long-term exposure of air pollutants SD: 7.1
on lung function. Examined the effects of SO2, NO2, O3, Range: 2.5, 25.5
TSP, and PM10. Participants were given a medical exam
that included questionnaire data, lung function tests, skin
prick testing, and end-expiratory CO concentration.
Subjects had to reside in the area for at least 3 yrs to be
in the study.
Mean values of SO2, PM10, and NO2 were significantly
associated with reduction in pulmonary function. SO2
was correlated with Pm3o (r = 0.78), PM10 (r = 0.93) and
NO2 (r = 0.86). Authors stated that the association with
SO2 disappeared after controlling for PM10 but no data
was shown.
Regression coefficients and 95% CI in healthy never
smokers (per 10 |jg/m3 increase in annual avg SO2)
FVC: -0.0325 (-0.0390, -0.0260)
FEV1: -0.0125 (-0.0192,-0.0058)
Braun-Fahrlander
et al. (1997)
10 communities in
Switzerland
Anieres, Bern, Biel,
Geneva, Langnau,
Lugano, Montana,
Payerne, Rheintal,
Zurich
Period of Study:
1992-1993
Annual avg SO2 (|jg/m3)
Cross-sectional study of 4,470 children (6-15 yrs) living
in 10 different communities in Switzerland to determine |_uaano- 23
the effects of long term exposure to PM10, NO2, SO2, and Qeneva'.
O3 on respiratory and allergic symptoms and illnesses. Zurich' 16
Part of the Swiss Study on Childhood Allergy and ger^. ^
Respiratory Symptoms with Respect to Air Pollution . ¦ ,
(SCARPOL). J™
Rheintal: 8
Langnau: NA
Payerne: 3
MontaN: 2
This study reported that the annual mean SO2, PM10,
and NO2 were positively and significantly associated
with prevalence rates of chronic cough, nocturnal dry
cough, and bronchitis and conjunctivitis symptoms.
Strongest association found with PM10. However, there
was no significant association between SO2 and
asthma or allergic rhinitis.
Adjusted relative odds between the most/least polluted
community 2-23 pg/m2 (0.8, 8.8 ppb)
Chronic cough: 1.57 (1.02, 2.42)
Nocturnal dry cough: 1.66 (1.16, 2.38)
Bronchitis: 1.48 (0.98, 2.24)
Wheeze: 0.88 (0.54, 1.44)
Asthma (ever): 0.74 (0.45,1.21)
Sneezing during pollen Season: 1.07 (0.67,1.70)
Hay fever: 0.84 (0.55,1.29)
Conjunctivitis symptoms: 1.74 (1.22, 2.46)
Diarrhea: 1.02 (0.75,1.39)
Charpin et al. (1999)
Etang de Berre area
of France: Aries,
Istres, Port de Bouc,
Rognac-Velaux,
Salon de Provence,
Sausset, Vitrolles
Period of Study:
Jan-Feb 1993
Cross-sectional cohort study of 2,073 children (10-11
yrs) from 7 communities in France (some with the
highest photochemical exposures in France) to test the
hypothesis that atopy is greater in towns with higher
photochemical pollution levels. Mean levels of SO2, NO2,
and Oswere measured for 2 mos in 1993. Children
tested for atopy based on skin prick test (house dust
mite, cat dander, grass pollen, cypress pollen, and
Alternaria). To be eligible for the study, subjects must
have resided in current town for at least 3 yrs.
Questionnaire filled out by parents that included
questions on SES passive smoking at home. Two-mo
mean level of air pollutants used in logistic regression
analysis.
24-h avg (SD)
SO2 ((jg/m3)
Aries: 29.7 (15.5)
Istres: 23.8 (12.7)
Port de Bouc: 32.3
(24.5)
Rognanc-Velaux:
39.5(21.8)
Salon de Provence:
17.3(11.6)
Sausset: 29.0 (28.7)
Vitrolles : 57.4 (32.0)
Study did not demonstrate any association between air
pollution and atopic status of the children living in the
seven communities, some with high photochemical
exposures. A limitation of study is that authors did not
consider short-term variation in air pollution and did not
have any indoor air pollution measurements.
Frischer et al. (1999) Longitudinal cohort study of 1150 children (mean age
Nine communities in 7 to lnvestl9ate the lon9-term effecte of°30n lun9
Austria
Period of Study:
1994-1996
growth. Children were followed for 3 yrs and lung
function was recorded biannually, before and after
summertime. The dependant variables were change in
FVC, FEV1, and MEF50. The 9 sites were selected to
represent a broad range of O3 exposures. GEE models
adjusted for baseline function, atopy, gender, site,
environmental tobacco smoke exposure, season, and
change in height. Other pollutants studied included PM10,
SO2, and NO2.
Annual mean SO2 (ppb)
in 1994
Amstetten: 3.75
St. Valentin: 3.00
Krems: 3.75
Heidenreichstein: 4.13
Ganserndorf: 5.63
Mistelbach: 5.25
Wiesmath: 6.00
Bruck: 4.88
Pollau: 2.25
No consistent association observed between lung
function and SO2, NO2 and PM10. A negative effect
estimate was observed during the summer and a
positive estimate during the winter.
Change in lung function (per ppb SO2):
FEV1 (mL/day):
Summer: -0.018 (0.004), p < 0.001
Winter: 0.003 (0.001), p < 0.001
FVC (mL/day):
Summer: -0.009 (0.004), p = 0.02
Winter: 0.002 (0.001), p = 0.03
MEF50 (mL/s/day):
Summer: -0.059 (0.010), p < 0.001
Winter: 0.003 (0.003), p = 0.26
F-92
-------
Study
Methods
Pollutant Data
Findings
Frischer et al. (2001) Cross-sectional cohort study of 877 children (mean age 1/4-h avg SO2:
Nine communities in " 2 yrs) llvlng insites with different 03 exposures. 3Q d megn 2 7Q b
. . ¦ Urinary eosinophil protein U-EPX) measured as a marker ,qR ~ , ,
of eosinophil activation. U-EPX determined from a single
Period of Study: spot urine sample analyzed with linear regression
Sep-Oct 1997 models.
No significant association between SO2 and U-EPX
Regression coefficient and SE -10.57 (0.25) per ppb
S02
Frye et al. (2003)
Zerbst, Hettstedt,
Bitterfeld.East
Germany
Period of Study:
1992-93,1995-1996
1998-1999
Three consecutive cross-sectional surveys of children
(11-14 yrs) from three communities in East Germany.
Parents of 3,155 children completed a questionnaire on
symptoms. Lung function tests performed on
2,493 children. Study excluded children if they lived for
less than 2 yrs in current home and if their previous
home was more than 2 km away. The log-transformed
lung function parameters were used as the response
variables in a linear regression analysis that controlled
for sex, height, season of examination, lung function
equipment, parental education, parental atopy, and
environmental tobacco smoke. Used avg of annual
means of pollutants 2 yrs preceding each survey.
Used avg of annual
means of pollutants 2 yrs
preceding health
measurement
High of 113 (jg/m3 (in
Bitterfeld) to a low of
6 (jg/m3. (Pollution
values only described in
figure)
The annual mean TSP declined from 79 to 25 |jg/m3
and SO2 from 113 to 6 |jg/m3 and the mean FVC and
FEV1 increased from 1992-1993 to 1998-1999. Study
concluded that reduction of air pollution in a short time
period may improve children's lung function.
Percent change of lung function for a 100-(jg/m3
decrease in SO2 2 yrs before the investigation
(N: 1,911)
FVC: 4.9 (0.7, 9.3)
FEV1: 3.0 (-1.1, 7.2)
FEV1/FVC: -1.5 (-3.0, 0.1)
Garcia-Marcos et al. A total of 340 children (10-11 yrs) living in and attending Annual mean SO2
(1999)
Cartagena, Spain
Period of Study:
winter 1992
schools within a polluted and a relatively nonpolluted (|jg/m3) Polluted areas
area were included in this study which aimed to establish 75 |jg/m3
the relative contribution socioeconomic status, parental Nonpolluted areas:
smoking, and air pollution on asthma symptoms, 20 |jg/m3
spirometry, and bronchodilator response. Parents
completed questionnaire on respiratory symptoms and
risk factors including, living in polluted area, maternal
smoking, paternal smoking, number of people living in
the house, proximity to heavy traffic roads. Spirometry
was performed before and after an inhaled 0.2 mg
fenoterol was delivered to determine bronchodilator
response. Bronchodilator response was considered
positive if the FVC after fenoterol was increased by at
least 10% or PEF by 12%. Logistic regression included
as independent variables all the risk factors.
This study found that living in the polluted areas
reduced the risk of a positive bronchodilator response
(RR = 0.61, p = 004).
Gokirmak et al.
(2003)
Malatya, Turkey
SO2 cone ranged from
106.6 to 639.2 ppm in
9 apricot farms.
Mean cone around
sulfurization chamber:
Study on occupational exposure to SO2 in apricot
sulfurization workers that investigated the role of
oxidative stress resulting exposure to high
concentrations of SO2 on bronchoconstriction. Forty
workers (mean age: 28 yrs, range 16-60 yrs) who have
been working in apricot sulfurization for 20-25 days each oX?,, ,,r
yr and 20 controls (mean age: 29 yrs, range 17-42) who
had no SO2 exposure participated in the study. Activities
of antioxidant enzymes (glutathione peroxidase (GSH-
Px), superoxide dismutase (SOD) and catalase)
malondialdehyde (MDA) concentrations (marker of lipid
peroxidation), and pulmonary function test measured in
subjects.
SOD, GSH-Px, and catalase activities were lower and
malondialdehyde concentrations were higher in the
apricot sulfurization workers compared to controls.
Pulmonary function decreased after SO2 exposure
among the apricot sulfurization workers. Authors
concluded that occupational exposure to high
concentrations of SO2 enhances oxidative stress and
that lipid peroxidation may be a mechanism of SO2
induced bronchoconstriction.
Apricot sulfurization workers vs. controls
Mean (SD)
SOD (U/mL): 2.2 (0.6) vs. 3.2 (0.7) U/m,
p< 0.0001
Glutathione peroxidase (U/mL): 0.6 (0.3) vs. 1.1 (0.3),
p< 0.0001
Catalase (L/L): 107.6 (27.4) vs. 152.6 (14.3),
p< 0.0001
MDA (nmol/L): 4.1 (0.9) vs. 1.9 (5.3), p < 0.0001
Before vs. after SO2 exposure among apricot
sulfurization workers
Mean (SD)
FVC (% predicted) 88 (17) vs. 84 (16), p < 0.001
FEV1 (% predicted) 98 (14) vs. 87 (14), p < 0.001
FEV1/FVC: 92 (7) vs. 86 (9), p < 0.001
FEF25-75% (% predicted) 108 (19) vs. 87 (23), p <
0.001
F-93
-------
Study
Methods
Pollutant Data
Findings
Heinrich et al. (2002) Three cross-sectional surveys of children (5-14 yrs) from
Reunified Germany
Bitterfeld, Hettstedt,
Zerbst
Period of Study:
1992-1993,1995-
1996,1998-1999
3 areas that were formerly part of East Germany to
investigate the impact of declines in TSP and SO2 on
prevalence of nonallelic respiratory disorders in
children. Study excluded children if they lived for less
than 2 yrs in current home and if their previous home
was more than 2 km away. GEE used for analysis.
SO2 concentration in
(jg/m3
Y r /Zerbst/Bitterf/ Hettst
1991/78/113/84
1992/58/75/46
1993/42/60/49
1994/29/35/38
1995/21/30/26
1996252425
1997/13/13/13
1998/8/9/6
Study found that SO2 exposure was significantly
associated with prevalence of bronchitis, frequent
colds, and febrile infections. While results are reported
as risk for an increase in air pollutant, the respiratory
health of children improved with declines in TSP and
SO2. Authors concluded that exposure to combustion-
derived air pollution is causally related to nonallelic
respiratory health in children.
Odds ratio and 95% CI: (per 100 |jg/m3 in 2 yr mean
S02)
All children:
Bronchitis: 2.72 (1.74, 4.23)
Otitis media: 1.42 (0.94, 2.15)
Sinusitis: 2.26 (0.85, 6.04)
Frequent colds: 1.81 (1.23,2.68)
Febrile infections: 1.76 (1.02, 3.03)
Cough in morning: 1.10 (0.73,1.64)
Shortness of breath: 1.31 (0.84,2.03)
Children without indoor exposures (living in damp
houses with visible molds, ETS in the home, gas
cooking emissions, and contact with cats):
Bronchitis: 4.26 (2.15, 8.46)
Otitis media: 1.43 (0.73, 2.81)
Sinusitis: 2.95 (0.52,16.6)
Frequent colds: 2.29 (1.15, 4.54)
Febrile infections: 1.75 (0.78, 3.91)
Cough in morning: 1.00 (0.38, 2.64)
Shortness of breath: 2.07 (0.90, 4.75)
Herbarth et al.
(2001)
East Germany
Period of Study:
1993-1997
Meta-analysis of three cross-sectional studies: (1) Study Avg lifetime exposure
on Airway Diseases and Allergies among Kindergarten burden of SO2 (pg/m3)
Children (KIGA), (2) the Leipzig Infection, Airway KIGA' 142
Disease and Allergy Study on School starters (LISS), LISS 48
and (3) KIGA-IND, which was based on the KIGA design ugg. p 47
but conducted in 3 differentially polluted industrial areas. KIGA IND" 59
A total of 3,816 children participated in the three studies.
Analysis of data from parent-completed questionnaires to
determine the effect of life time exposure to SO2 and
TSP on the occurrence of acute bronchitis. Total lifetime
exposure burden corresponds to the exposure duration
from birth to time of the study. The LISS study was
divided in to LISS-U for the urban area and LISS-R for
the rural area. Logistic regression analysis used that
adjusted for predisposition in the family (mother or father
with bronchitis), ETS, smoking during pregnancy or in
the presence of the pregnant women.
This study found the highest bronchitis prevalence in
the KIGA cohort and the lowest in the LISS cohort,
which is consistent with the SO2 concentrations in these
cohorts. Study found a correlative link between SO2 and
bronchitis (R = 0.96, p < 0.001) but not TSP (R = 0.59).
Results of study suggest that SO2 may be a more
important factor than TSP in the occurrence of
bronchitis in these study areas.
Odds ratio for bronchitis adjusted for parental
predisposition, smoking, and lifetime exposure to SO2
and TSP (2-pollutant model).
S02: 3.51 (2.56, 4.82)
TSP: 0.72 (0.49,1.04)
Hirsch et al. (1999)
Dresden, Germany
Cross sectional study to relate the prevalence of
respiratory and allergic diseases in childhood to
measurements of outdoor air pollutants. 5,421 children
ages 5-7 yrs and 9-11 yrs were evaluated by
questionnaires, skin-prick testing, venipuncture for (lg)E,
lung function, and bronchial challenge test.
Mean (|jg/m3): 48.3
Range: 29.0-69.3
25-75 percentile 42.7-
54.3
SOxwas positively associated with current morning
cough but not with bronchitis.
Prevalence odds ratio (95% CI) for symptoms within
past 12 mos, +10 |jg/m3:
Wheeze: Atopic 103 (0.79,1.35) |jg/m3
Nonatopic 1.36 (1.01,1.84)
Morning Cough: Atopic 1.22 (0.92,1.61)
Nonatopic 1.32 (1.07,1.63)
Prevalence odds ratio (95% CI) for doctor's diagnosis,
+10 (jg/m3:
Asthma: Atopic 1.07 (0.79,1.45)
Nonatopic 1.35 (1.00,1.82)
Bronchitis: Atopic 1.04 (0.87,1.25)
Nonatopic 0.99 (0.88,1.12)
F-94
-------
Study
Methods
Pollutant Data
Findings
Horak et al. (2002)
Eight communities in
Austria
Period of Study:
1994-1997
Longitudinal cohort study that continued the work of Seasonal avg SO2
Frischeretal. (1999) by adding one more yr of data and |jg/m3:
analyzing the effects of PM10 in addition to SO2, NO2, Winter:
and O3. At the beginning of the study 975 children (mean Mean: 16.8
age 8.11 yrs) were recruited for the study but only 80.6% Range: 7.5, 37.4
of the children performed all 6 lung function tests (twice a 3ummer
yr). The difference for each lung function parameter ^ean . g g UQ/m3
between two subsequent measures was divided by the Ranae- 3 -| ^ 7
days between measurements and presents as difference '
per day (dpd) for that parameter. 860 children were
included in the GEE analysis that controlled for sex,
atopy, passive smoking, initial height, height difference,
site, and initial lung function.
Moderate correlation between PM10 and SO2 in the
winter (r = 0.52). In a one-pollutant model for SO2, long
term seasonal mean concentration of SO2 was had a
positive association with FVC dpd and FEV1 dpd in the
winter, but no effect on MEF25-75 dpd. In a two-pollutant
model with PM10, wintertime SO2 had a positive
association with FEV1 dpd.
Single-pollutant model
FVC dpd:
Summer: 0.009, p = .336; Winter: 0.006, p = .009
FEV1 dpd: Summer: 0.005, p = 0.576;
Winter: 0.005, p = 0.013
MEF25-75: Summer: 0.015, p = 0.483;
Winter: 0.003, p = 0.637
Two-pollutant model:
SO2 + PM10
FVC dpd: Summer: 0.008, p = 0.395;
Winter: 0.004, p = 0.225
FEV1 dpd: Summer: 0.010 (0.271); Winter: 0.007
(0.025)
MEF25-75 dpd: Summer: 0.037, p = 0.086;
Winter: 0.007, p = 0.429
Jedrychowski et al.
(1999)
Krakow, Poland
Period of Study:
1995 (Mar-Jun) and
1997 (Mar-Jun)
Annual avg:
City Center (|jg/m3):
43.87 (32.69)
Control Area (|jg/m3):
Cohort prospective study consisting of 1,001
preadolescent children (9 yrs old) from two areas of
Krakow, Poland. The study examined lung function
growth using FVC and FEV1 measurements taken in
1995 and then again two yrs later, 1997. Used a two-
stage analysis that consisted of (1) multivariate linear 31.77 (21.93)
regression analyses to determine body variables that are
significant predictors of lung function growth, and then
(2) multivariate logistic regression to examine the relation
between air pollution and lung function growth.
The study did not provide individual estimates for SO2.
Koksal et al. (2003)
Malatya, Turkey
Study on occupational exposure to high concentrations
of SO2 on respiratory symptoms and pulmonary function
on apricot sulfurization workers. Apricot sulfurization
workers (N: 69) from 15 apricot farms who have been
working in sulfurization of apricots for 20-25 days a yr
during each summer were recruited for the study.
Subjects rated symptoms (itchy eyes, runny nose, stuffy
nose, itchy or scratchy throat, cough, shortness of
breath, phlegm, chest pain, and fever) before during and
1-h after each exposure.
SO2 cone ranged from SO2 exposure at high concentrations increased
106.6 to 721.0 ppm symptoms of itchy eyes, shortness of breath, cough,
running and/or stuffy nose, and itchy or scratchy throat
during exposure (p < 0.05). Inhalation of high
concentrations of SO2 for 1-h caused significant
decreases in pulmonary function. Difference in
pulmonary function measured before and after
exposure:
FVC (L) 0.16 (0.42), p< 0.05
FEV1 (L) 0.39 (0.36), p< 0.001
FEV1/FVC: 5.22 (6.75), p < 0.001
PEF (L/s) 1.39 (1.06), p< 0.001
FEF25-75%(L/s) 0.82 (0.70), p < 0.001
F-95
-------
Study
Methods Pollutant Data
Findings
Kopp et al. (2000)
Longitudinal cohort study of 797 children (mean age Mean SO2 (95% CI) ppb
Lower FVC and FEV1 increases observed in children
Ten communities in
Austria and SW
Germany
8.2 yrs) from 2nd and 3rd grades of 10 schools in Austria Apr-Sep 1994
exposed to high ambient O3 levels vs. those exposed to
and SW Germany to assess the effects of ambient Cbon Amstetten: 3.7 (0.7, 3.9)
lower levels in the summer. This study found no effect
lung function in children over a 2-summer period. Study St Valentin: 2.6 (1.5, 5.2)
of SO2 and PM10 on FVC increase during the summer of
also examined the association between avg daily lung Krems: 3.7 (0.7, 7.5)
1995 and winter 1994/1995, however, SO2 was
growth and SO2, NO2, and PM10. Each child performed 4 Villingen: 0.7 (0 , 3.0)
negatively associated with FVC during the summer of
lung function tests during spring 1994 and summer 1995. Heindenreichstein: 3.7,
1994.
ISAAC questionnaire used for respiratory history. Linear (0.7, 7.5)
Change in FVC (per ppb SO2)
Summer 1994: -0.044, p = 0.006
Winter 1994/95: 0.007, p = 0.243
Summer 1995: 0.045, p = 0.028
regression models used to assess effect of air pollutants Ganserndorf: 3.7 (0.7,
on FVC and FEV1, which were surrogates of lung growth. 11.2)
Mistelbach: 3.7 (0.7, 7.5)
Wiesmath: 6.3 (3.4, 9.4)
Bruck: 1.5 (0.7, 4.1)
Freudenstadt: 0.7 (0,
3.0)
Oct 1994-Mar 1995
Amstetten: 3.7 (0.7, 7.5)
St Valentin: 3.0 (1.1, 9.4)
Krems: 3.7 (0.7,11.0)
Villingen: 1.9 (0 , 3.0)
Heindenreichstein: 3.7
(0.7,15.0)
Ganserndorf: 3.7 (0.7,
22.5)
Mistelbach: 3.7 (0.7,
22.5)
Wiesmath: 2.23 (0.7,
10.1)
Bruck: 15 (1.1, 7.9)
Freudenstadt: 1.57 (0.4,
5.3)
Apr-Sep 1995
Amstetten: 3.7 (0.7, 3.8)
St Valentin: 2.6 (1.1, 6.8)
Krems: 3.7 (0.5, 3.8)
Villingen: 0.7 (0 , 2.6)
Heindenreichstein: 0.7
(0.5, 0.9)
Ganserndorf: 3.7 (0.7,
7.5)
Mistelbach: 3.7 (0.7, 7.5)
Wiesmath: 7.5 (0.7,
14.9)
Bruck: 3.7 (0.4, 4.9)
Freudenstadt: 0.7 (0,
3.4)
F-96
-------
Study
Methods
Pollutant Data
Findings
Kramer etal. (1999)
East and West
Germany
Period of Study:
1991 to 1995
Repeated cross-sectional studies between 1991 and
1995 on 7-yr-old children in East Germany and between
1991 and 1994 in West Germany. Comparison of
prevalence of airway diseases and allergies in East and
West Germany during the first five yrs after reunification.
A total of 19,090 children participated in the study.
Logistic regression used to assess the effect of SO2 and
TSP on airway diseases and allergies. Analysis
performed on 14,144 children with information on all
covariates of interest.
East Germany 2-yr avg
concentration ranged
from 45 to 240 |jg/m3
West Germany 2-yr avg
concentration ranged
from 18-33
All infectious airway diseases and irritation of the airway
was associated with either SO2 or TSP in East
Germany in 1991. The decrease of pollution between
1991 and 1995 had a favorable effect on the prevalence
of these illnesses. SO2 was significantly associated with
more than 5 colds in the last 12 mos, tonsillitis, dry
cough in the last 12 mos, and frequent cough in 1991-
1995.
Odds ratio and 95% CI: (per 200 |jg/m3 SO2) in East
Germany areas, 1991-1995 for children living at least 2
yrs in the areas, adjusted for time trend:
Infectious airway diseases:
Pneumonia ever diagnosed: 1.17 (0.85,1.62)
Bronchitis ever diagnosed: 0.85 (0.68,1.05)
35 colds in last 12 mos: 1.55 (1.18, 2.04)
Tonsillitis in the last 12 mos: 1.89 (1.49, 2.39)
Dry cough in the last 12 mos: 1.46 (1.12,1.91)
Frequent cough ever: 2.51 (1.79. 3.53)
Allergic diseases and symptoms:
Irritated eyes in the last 12 mos: 1.06 (0.66,1.70)
Irritated nose in the last 12 mos: 1.26 (0.96,1.66)
Wheezing ever diagnosed: 0.68 (0.46,1.01)
Bronchial asthma ever diagnosed: 2.73 (1.24, 6.04)
Hay fever ever diagnosed: 0.60 (0.24,1.52)
Eczema ever diagnosed: 0.87 (0.65,1.18)
Allergy ever diagnosed: 0.93 (0.67,1.29)
Liebhart et al. (2007)
Poland (Bialystok,
Bydgoszcz, Gdansk,
Krakow, Lublin,
Lodz, Poznan,
Rabka, Warszawa,
Wroclaw, Zabrze)
Period of Study:
1998-1999
The Polish Multicentre Study of Epidemiology of Allergic Range (|jg/m3):
Diseases (PMSEAD), which consisted of a cohort of 4 n 35 0
16,238 individuals aged 3-80 yrs old from 33 areas in 11
regions of Poland. Asthma diagnosis was determined
through household questionnaires. Conducted
multivariate and univariate logistic regression analyses to
examine the prevalence of and risk factors for asthma.
In multivariate logistic regression models, BS was found
to be a significant risk factor for asthma for both
children and adults. SO2 was found to be a significant
risk factor for asthma in both children and adults, but
only in a univariate logistic regression.
Adjusted Odds Ratio (95% CI)
Univariate logistic regression
Children: 1.34 (1.04,1.72)
Adults: 1.19 (1.02,1.38)
Multivariate logistic regression
Children: 1.20 (0.91,1.59)
Adults: 1.01 (0.85,1.20)
Kohlhammer et al.
(2007)
Hettstedt, Germany
Period of Study:
1992-1999
Three repeated cross-sectional studies of 5,360 children
ages 5-14 examining health impacts (lifetime
pneumonia) of social and environmental factors
No relationship between SO2 and pneumonia was
observed.
F-97
-------
Study
Methods
Pollutant Data
Findings
Penard-Morand et al.
(2006)
Six communities in
France: Bordeaux,
Clermont-Ferrand,
Creteil, Marseille,
Strasbourg and
Reims
Period of Study: Mar
1999-Oct 2000
Cross-sectional study of 4,901 children Estimated 3-yr avg
(9-11 yrs) form 108 randomly selected schools in 6 cities concentrations at 108
to assess the association between long-term exposure to schools
background air pollution (N02, S02, PM10, 03) and atopy Lqw CQnc; 4 6 ,3
rnp>KM r^+/*\r\ / r\i r\c* A rtk /r»in +/*\ /^rt irt ' ^
and respiratory outcomes. Analysis restricted to children
who had lived at least the last 3 yrs in their house at the
time of the examination. Analysis used three yr avgd air
pollutant concentrations at the children's schools.
Parents completed questionnaire on respiratory and
allergic disorders (asthma, allergic rhinitis (AR), and
atopic dermatitis) and children underwent examination
that included a skin prick test to assess allergic
sensitization, evidence of visible flexural dermatitis and
measure of exercise-induced bronchial reactivity (EIB).
(Range: 1.3, 7.4),
High cone: 9.6 |jg/m3
(range 7.7,13.7)
Increased concentrations of SO2 were significantly
associated with an increased risk of EIB, lifetime
asthma and lifetime AR. Past yr wheeze and asthma
were also associated with SO2. In a two-pollutant model
with PM10, significant associations were observed
between SO2 and EIB and past yr wheeze.
Odds ratio and 95% CI (per 5 |jg/m3 SO2)
EIB: 1.39 (1.15,1.66), p< 0.001
Flexural dermatitis: 0.86 (0.73,1.02), p < 0.10
Past yr wheeze: 1.23 (1.0,1.51), p< 0.05
Past yr asthma: 1.28 (1.00,1.65), p < 0.01
Pastyrrhinoconjunctivitis: 1.05 (0.89, 1.24)
Past yr atopic dermatitis: 1.01 (0.86,1.18)
Lifetime asthma: 1.19 (1.00,1.41), p < 0.10
Lifetime allergic rhinitis: 1.16 (1.01,1.32),
p < 0.05
Lifetime atopic dermatitis: 0.93 (0.82,1.05)
Two-pollutant model with PM10
EIB: 1.46 (1.12,1.90)
Past yr wheeze: 1.45 (1.09,1.93)
Pikhart et al. (2001) Part of the small-area variation in air pollution and health Mean SO2 (pg/m3)
Czech Republic,
Poland,
Period of Study:
1993-1994
(SAVIAH) study to assess long-term effects of air
pollution on respiratory outcomes. Analysis on data from
two centers of the multicenter study: Prague, Czech
Republic, and Poznan, Poland. Both cities had wide
variation in air pollution levels. Parents/guardians of
6,959 children (7-10 yrs) completed a questionnaire
about the socioeconomic situation of the family, type of
housing, family history of atopy, parental smoking, family
composition, and health of the child. SO2 was measured
at 80 sites in Poznan and 50 sites in Prague during 2-wk
campaigns. From these data GIS was used to estimate
pollutant concentrations at a small area level. Logistic
regression used to assess effect of air pollution on the
prevalence of respiratory outcomes.
Prague: 83.9
Range: 65.8-96.6
PoznaN: 79.7
Range: 44.2-140.2
SO2 levels (mean of home and school) were associated
with the prevalence of wheezing/whistling in the past
12 mos. There was a marginal association between
S02and lifetime prevalence of wheezing and physician
diagnosed asthma. Fully adjusted model controlled for
age, gender, maternal education, number of siblings,
dampness at home, heating and cooking on gas,
maternal smoking, and family history of atopy and
center. Authors noted SO2 is strongly spatially
correlated with particles in the Czech Republic and
probably Poland, so SO2 may be proxy for exposure to
other pollutants. Not other pollutants measured in study.
Odds ratio (per 50 (jg/m3) SO2
Wheezing/whistling in past 12 mos: 1.32 (1.10,1.57)
Wheezing/whistling ever: 1.13(0.99,1.30)
Asthma ever diagnosed by doctor: 1.39 (1.01,1.92)
Dry cough at night: 1.06 (0.89,1.27)
Ramadour et al. Cross-sectional cohort study of 2,445 children (age 13-
(2000) 14 yrs) who had lived for at least 3 yrs in their current
Seven towns in SE res'c'ence t° compare the levels of O3, SO2, and NO2 to
prance the prevalence rates of rhinitis, asthma, and asthmatic
symptoms. Some of the communities had the heaviest
Period of Study: Jan- photochemical exposure in France. Subjects completed
Feb 1993 ISAAC survey of asthma and respiratory symptoms.
Analysis conducted with logistic regression models that
controlled for family history of asthma, personal history of
early -life respiratory diseases, and SES. Also performed
simple univariate linear regressions.
Mean (SD) |jg/m3 ofS02
during 2-mo period
Port de Bouc: 32.3
(24.5)
Istres: 23.8 (12.7)
Sausset: 29.0 (28.7)
Rognanc-Velaux: 39.5
(21.8)
Vitrolles: 57.4 (32.0)
Aries: 29.7 (15.5)
SaloN: 17.3(11.6)
Study found no relationship between mean levels of
SO2, NO3, or O3 and rhinitis ever, 12-mo rhinitis,
rhinoconjunctivitis, and hay fever or asthmatic
symptoms. Simple regression analyses of respiratory
outcomes vs. mean SO2 levels in the 7 towns indicated
that nocturnal dry cough was associated with mean SO2
levels (r= 0.891). Potential confounding across towns.
Soyseth et al.
(1995b)
Ardal and Laerdal,
Norway
Period of Study:
winter seasons
1989-92
Median SO2
37.1 uq/m3ataqes
0-12 mos
37.9 (jg/m3 at
Cross-sectional study of 529 children
(aged 7-13 yrs) to determine whether exposure to SO2
during infancy is related to the prevalence of bronchial
hyperresponsiveness (BHR). A sulfur dioxide emitting
aluminum smelter is present in Ardal, but there is no air
polluting industry in Laerdal. Parents filled out
questionnaire regarding family history of asthma, type of ages 13-36 mos
housing, respiratory symptoms and parent's smoking
habits. Spirometry was performed on each child and
bronchial hyperactivity was determined by methacholine
challenge or reversibility test. Skin prick test done to
assess atopy. Also examined, the effects of fluoride.
This study found that the risk of BHR was associated
with SO2 exposure at 0-12 mos
Odds ratio for BHR (per 10 |jg/m3 SO2) for various ages
at exposure
0-12 mos: 1.62 (1.11,2.35)
13-36 mos: 1.40 (0.90,2.21)
37-72 mos: 1.19 (0.77,1.82)
73-108 mos: 1.19(0.63,2.22)
F-98
-------
Study
Methods
Pollutant Data
Findings
Studnicka et al.
(1997)
Austria (8 nonurban
communities)
Period of Study:
1991-1993
Longitudinal study of 843 children 7 yrs old from 8
nonurban Austrian communities. A logistic regression
was used to examine the association between SO2
concentrations and asthma and respiratory symptoms by
comparing low, regular, and high SO2 communities with
very low SO2 communities.
Range:
Jan. 1991 -Dec. 1993
(ppb): 6.0 (Krems), 12.0
(Mistel. and Gans)
S02was significantly associated with bronchial asthma
in the last 12 mos and positively associated with parent-
reported "ever asthma" when comparing low SO2
concentration communities with very low SO2
communities.
Adjusted Prevalence Odds Ratio
Wheeze last 12 mos:
Low: 0.68. Regular: 0.88. High: 0.42
Cough apart from colds last 12 mos:
Low: 0.75. Regular: 0.85. High: 0.72
Bronchitis last 12 mos:
Low: 0.21. Regular: 0.45. High: 0.56
Bronchial asthma last 12 mos:
Low: 2.35. Regular: 0.22. High
Parent-reported "ever asthma"
Low: 1.70. Regular: 0.23. High
0.33
0.67
von Mutius et al. The effects of high to moderate levels of air pollution
(1995) (SO2, NOx, and PM) on the incidence of upper
Leipzig [iast respiratory were investigated in 1,500 schoolchildren (9-
Ormanv ^ yrs) 'n Lsipzig, East Germany. Logistic regression
models controlled for paternal education, passive smoke
Period of Study: Oct exposure, number of siblings, temperature, and humidity.
1991 -Jul 1992
During winter mos, SO2
daily max concentrations
ranged from 40-1283
(jg/m3.
During high pollution
period, avg
concentration of SO2
was 188 (jg/m3 and
during low pollution avg
was 57 (jg/m3.
The daily mean values of SO2 and NOx were
significantly associated with increased risk of
developing upper respiratory illnesses during the high
concentration period. In the low concentration period,
only NOx daily mean values were associated with
increased risks. In a two-pollutant model with PM,
similar estimates to the single-pollutant model were
obtained, thus collinearity of data may not account for
the effects of high mean concentrations ofS02.
Odds ratio and 95% CI: (did not indicate per what level
of SO2 increase)
Daily mean SO2: High period: 1.72 (1.19, 2.49);
Low period: 1.40 (0.95, 2.07)
Daily max SO2: High period: 1.26 (0.80,1.96);
Low period: 0.99 (0.66,1.47)
F-99
-------
Study
Methods
Pollutant Data
Findings
Sole et al. (2007)
Sao Paulo, Brazil
(Sao Paulo West
(SPW), Sao Paulo
South (SPS), Santo
Andre (SA), Curitiba
(CR), Porto Alegre
(PoA)
Cohort of 16,209 adolescents (13-14 yrs old) from the 21 NR
centers involved in the International Study of Asthma and
Allergies in Childhood (ISAAC). Each participant was
given a questionnaire to identify various allergy-related
symptoms that occurred in the last 12 mos. The
relationship between affirmative answer to a question,
socioeconomic status, and air pollutants was analyzed
by the Spearman correlation coefficient. The location
with the lowest level of a specific air pollutant was
defined as the reference and the risk of an affirmative
answer to a question was presented as an odds ratio for
each location.
In the analysis of the risk of allergy-related symptoms
due to SO2 levels in relation to the center with the
lowest annual mean SO2 concentrations SPW was
significantly associated with every symptom. Other
significant associations were observed in SA for current
wheezing; in CR for rhinitis and rhinoconjunctivitis; and
in PoA for current wheezing, nighttime cough, rhinitis,
and eczema.
Odds Ratio (95% Cl)-Reference Center: Sao Paulo
South (SPS)
Current Wheezing: SPW: 1.21 (1.08,1.38);
SA: 1.31 (1.16,1.48); CR: 1.02(0.90,1.15)
PoA: 1.68 (0.85,1.10)
Severe Asthma: SPW: 2.01 (1.56, 2.60);
SA: 1.04 (0.78, 1.40); CR: 1.08 (0.81,1.42);
PoA: 1.01 (1.29,2.20)
Nighttime Cough: SPW: 1.14(1.03,1.26);
SA: 0.94 (0.85, 1.04); CR: 0.93 (0.84,1.02);
PoA: 1.25 (0.91,1.12)
Rhinitis Last Yr: SPW: 1.14 (1.02,1.27);
SA: 1.05 (0.94,1.18); CR: 1.71 (1.54,1.90);
PoA: 1.36(1.12,1.40)
Rhinoconjunctivitis: SPW: 1.78 (1.55, 2.04);
SA: 1.15 (0.99,1.33); CR: 1.50 (1.31,1.72);
PoA: 1.48(1.18,1.57)
Severe Rhinitis: SPW: 1.50 (1.32,1.71);
SA: 1.08 (0.94,1.24); CR: 1.52 (1.34,1.73);
PoA: 0.97 (1.30,1.69)
Eczema: SPW: 1.40 (1.17,1.68); SA: 1.00 (0.83,1.21);
CR: 0.88 (0.72,1.06); PoA: 1.40 (0.80,1.18)
Flexural Eczema: SPW: 2.00 (1.58, 2.52);
SA: 0.95 (0.73, 1.24); CR: 1.02 (0.79,1.31);
PoA: 2.41 (1.09,1.80)
Severe Eczema: SPW: 2.58 (1.94, 3.44);
SA: 0.92 (0.65, 1.30); CR: 0.71 (0.50,1.02);
PoA: NR (1.80, 3.22)
Ho et al. (2007)
Taiwan
Period of Study:
1995-1996
Survey of 69,367 children ages 12-15 by questionnaire.
The max likelihood estimation was carried out with
Fisher's scoring algorithm and GEE.
NR
SO2 not significant in both genders. However, SO2
showed a reversal effect on monthly asthma attack rate.
(Authors state that this reversal effect could be caused
by the interaction of sulfur dioxide with the lowest 5%
monthly temperature avg)
Hwang et al. (2005)
Taiwan
Period of Study:
2001
A cross-sectional study consisting of 32,672 Taiwanese 2000 (ppb): 3.53 (2.00)
school children aged 6-15 yrs old. Using a modified
Chinese version of the International Study of Asthma and
Allergies in Childhood (ISAAC) questionnaire collected
information on each participant's health, environmental
exposures, and other variables. A two-stage hierarchical
model consisting of logistic and linear regression
analyses was used to account for, in the first stage,
variation among subjects, and, in the second stage,
variation among municipalities.
Increased annual levels of NOx, CO, and Oswere
associated with an increased risk of childhood asthma
levels. In both single- and co-pollutant models SO2 was
not found to be associated with the risk of asthma.
Odds Ratio (95% CI) (per 10 ppb SO2)
Single-pollutant model
0.874 (0.729,1.054)
Two-pollutant model
NOx+ SO2:0.724 (0.545, 0.963)
CO+ SO2: 0.689 (0.542,0.875)
SO2 + O3: 0.826 (0.674,1.014)
F-100
-------
Study
Methods
Pollutant Data
Findings
Peters et al. (1996)
Hong Kong (Kwai
Tsing; Southern)
Period of Study:
1989-1991
Cohort of 3,521 children from two districts in Hong Kong Annual avg (|jg/m3)
with good and poor air quality prior to the 1990 Southern
legislation to reduce fuel sulfur levels. Analyses ^ggg
consisted of multivariate methods using logistic ^ggg
regression along with generalized estimating equations ^gg^
(GEE) to examine the effect of legislation implemented to
reduce fuel sulfur levels on respiratory health. Kwai Tsing
11
7
1989
1990
1991
111
67
23
SO2 emissions were reduced by 80% after institution of
the legislation. The study does not provide effect
estimates for individual pollutants.
Wang et al. (1999)
Taiwan (Kaohsiung;
Pintong)
Period of Study:
1995-1996
Across-sectional study consisting of 165,173 high school Median:
students aged 11-16 yrs old residing in the communities ^ggg /DDm\. q q-|3
of Kaohsiung and Pintong in Taiwan from Oct 1995 to
Jun 1996. Used a video and questionnaire developed by
the International Study of Asthma and Allergies in
Childhood (ISAAC). The association between air
pollution and asthma was examined using logistic
regression. In addition, the study performed a multiple
logistic regression to examine the independent effects of
risk factors of asthma after adjusting for age, sex,
parents' education, and area of residence. The multiple
logistic regression included pollutant concentrations to
examine the combined effect.
In the univariate analysis, increasing concentrations of
TSP, SO2, NO2, CO, 63, and airborne dust were all
found to be significantly associated with asthma. These
univariate estimates are associated with concentrations
above a cutoff (i.e., the median concentrations of each
pollutant). In the multivariate analysis increasing
concentrations of TSP, N02, CO, 03, and airborne dust
were significantly associated with asthma.
Odds Ratio (95% CI) (per 0.013 ppm SO2)
Univariate analysis
>0.013 ppm: 1.05 (1.02,1.09)
Adjusted Odds Ratio (95% CI)
Multivariate analysis
0.98 (0.95,1.02)
Dubnov et al. (2007) Cohort of 1,492 schoolchildren (7-14 yrs old) living near
Israel
(Hadera, Pardes-
Hanna)
Period of Study:
1996 and 1999
a major coal-fired power station. Subjects underwent
pulmonary function tests (PFT) for forced vital capacity
(FVC) and forced expiratory volume during the first
second (FEV1) to examine the association between
pulmonary function and long-term exposure to air
pollution. Using stepwise multiple regression (SMR) and
ordinary leas squares regression (OLS) examined the
multiplicative effect of NOx and SO2 on pulmonary
function.
1996 and 1999 avg (SD) Using an integrated concentration value (ICV), which
(ppm): equals the product of NOx concentration and SO2
12.9 (11.3) concentration when both concentrations individually
exceed the half-hour reference level, found significant
associations between exposure to air pollution and
decrements in pulmonary function.
All Children:
NOXxS02
AFVC (%)
p = -0.004, p < 0.001
AFEV1 (%)
p = -0.004, p < 0.001
Children in zone of highest concentration of air
pollution:
NOXxS02
AFVC (%)
p = -0.005, p < 0.001
AFEVi(%)
p = -0.005, p < 0.001
Houssaini et al.
(2007)
Morocco
Cross-sectional study of 1,318 children with a mean age Annual Avg:
of 12 yrs. Used a questionnaire and medical 2000-2001
diagnosis/reporting for asthma, and evaluated using 2001-2002
Student's t-test, Chi-square, odds ratios, and Cochran- 2002-2003
Armitage tests. 2003-2004
60.2 (jg/m3
50.2 (jg/m3
49.6 (jg/m3
36.8 (jg/m3
Significant prevalence for respiratory diseases, asthma,
and infectious disease, when combined with TSP.
F-101
-------
Table F-7. Long-term exposure to SO2 and lung cancer incidence and mortality.
Study
Methods
Pollutant Data
Findings
UNITED STATES
Abbey et al. (1999)
Three California air
basins: San
Francisco, South
Coast (Los Angeles
and eastward), San
Diego
Period of Study:
1977-1992
Prospective cohort study of 6,338 nonsmoking non-
Hispanic white adult members of the Adventist Health
Study followed for all cause, cardiopulmonary,
nonmalignant respiratory, and lung cancer mortality
Participants were aged 27-95 yrs at enrollment in
1977.1,628 (989 females, 639 males) mortality
events followed through 1992. All results were
stratified by gender. Used Cox proportional hazards
analysis, adjusting for age at enrollment, past
smoking, environmental tobacco smoke exposure,
alcohol use, education, occupation, and body mass
index. Analyzed mortality from all natural causes,
cardiopulmonary, nonmalignant respiratory, and lung
cancer.
Mean SO2 Levels:
24-h avg SO2: 5.6 ppb
Copollutants:
PM10
S04
03
N02
Lung cancer mortality showed large risk estimates
for most of the pollutants in either or both sexes, but
the number of lung cancer deaths in this cohort was
very small (12 for female and 18 for male) Generally
wide confidence intervals (relative to other U.S.
cohort studies).
Adjusted Mortality Relative Risk
(95% CI) (per 3.72 ppb S02)
Lung Cancer
Males: 1.99 (1.24,3.20)
Females: 3.01 (1.88, 4.84)
Beeson et al.
(1998)
Three California air
basins: San Diego,
San Francisco,
South Coast (Los
Angeles and
eastward)
Period of Study:
1977-1992
Prospective cohort study of 6,338 nonsmoking non- 24-h avg SO2: 5.6 ppb
Hispanic white adult members of the Adventist Health
Study aged 27-95 yrs at time of enrollment.
36 (20 females, 16 males) histologically confirmed
lung cancers were diagnosed through 1992.
Extensive exposure assessment, with assignment of
individual long-term exposures to O3, PM10, SO42",
and SO2, was a unique strength of this study. All
results were stratified by gender. Used Cox
proportional hazards analysis, adjusting for age at
enrollment, past smoking, education, and alcohol use.
Lung cancer incidence relative risk:
Male: RR = 3.72 (95%CI: 1.91, 7.28);
Female: RR = 2.78 (95%CI: 1.51, 5.12) per 5 ppb
increase in SO2
Case number very small (16 for male, 20 for female).
Krewski et al. Re-analysis and sensitivity analysis of Dockery et al.
(2000) (1993) Harvard Six Cities study.
Mean SO2 Levels: 24-h avg
SO2 ranged from 1.6
(Topeka) to 24.0
(Steubenville) ppb
Copollutants: Fine Particles,
Sulfates
SO2 showed positive associations with lung cancer
deaths (1.03 (95% CI: 0.91,1.16), but in this dataset,
SO2 was highly correlated with PM2.5 (r = 0.85),
sulfate (r = 0.85), and NO2
(r = 0.84)
Beelen et al. (2008)
The Netherlands
Period of Study:
1987-1996.
Cohort study on diet and cancer with 120,852
subjects who were followed from 1987 to 1996. BS,
NO2, SO2, and PM2.5 and traffic-exposure estimates
were analyzed. Cox regression model adjusted for
age, sex, smoking, and area-level socioeconomic
status.
Mean SO2 Levels:
Mean: 4.8 ppb, with a range
of 1.5 to 11.8 ppb.
Copollutants:
PM2.5 BS N02
Traffic intensity on the nearest road was not
associated with exposure SO2. Background SO2
levels were not associated with lung cancer mortality.
Adjusted RR
(per 20 (jg/m3 SO2)
1.00 (0.79,1.26)
Filleul et al. (2005)
Seven French cities
Period of Study:
1975-2001
Cohort study of 14,284 adults who resided in 24
areas from seven French cities when enrolled in the
PAARC survey (air pollution and chronic respiratory
diseases) in 1974. Daily measurements of SO2, TSP,
BS, NO2, and NO were made in 24 areas for three yrs
(1974-1976). Cox proportional hazards models
adjusted for smoking, educational level, BMI, and
occupational exposure. Models were run before and
after exclusion of six area monitors influenced by
local traffic as determined by the NO/NO2 ratio >3.
Mean SO2 Levels:
24-h avg SO2 ranged from
17 mg/m3 ("Area 3° in Lille)
to 85 mg/m3 ("Area 3" in
Marseille) in the 24 areas in
seven cities during 1974-
1976. Median levels during
1990-1997 ranged from 8.5
mg/m3 (Bordeaux) to 23.4
mg/m3 (Rouen) in the five
cities where data were
available.
Copollutants: TSP
BS N02 NO
The authors noted that inclusion of air monitoring
data from stations directly influenced by local traffic
could overestimate the mean population exposure
and bias the results. It should be noted that the table
describing air pollution levels in Filleul et al.'s report
indicates that the SO2 levels in these French cities
declined markedly from 1974-1976 and 1990-1997
period, by a factor of 2 to 3, depending on the city,
whereas NO2 levels between the two periods were
variable, increased in some cities, and decreased in
others. These changes in air pollution levels over the
study period complicate interpretation of reported
risk estimates.
Relative Risk (95% CI) for lung cancer mortality (per
10 mg/m3 multi-year average)
All 24 areas: 0.99 (0.92,1.07)
18 areas:1.00 (0.91,1.11)
F-102
-------
Study
Methods
Pollutant Data
Findings
Nafstad et al.
(2003)
Oslo, Norway
Period of Study:
1972-1998
Retrospective study associating cardiovascular risk Estimated for each person
factors to a national cancer register among 16,209
men ages 10-49 yrs. Survival analyses and Cox
proportional hazards regression were used to
estimate associations.
each yr from 1974 to 1998
Five-yr median avg levels
SO2 participants home
address, 1974-1978: 9.4
(jg/m3 (range 0.2 to 55.8)
Median levels within the
quartiles:
2.5 (jg/m3. 6.2 |jg/m3
14.7 (jg/m3.31.3 |jg/m3
Copollutants:
NOx
Adjusted risk ratios (95% CI) of developing lung
cancer:
Model 1:
0-9.99 (jg/m3: Ref
10-19.99 |jg/m3:1.05 (0.81,1.35)
20-29.99 (jg/m3: 0.95 (0.72,1.27)
30+ [jg/m3:1.06 (0.79,1.43)
Model 2: Per 10 pg/m3:1.01 (0.94,1.08)
Adjusted risk ratios (95% CI) of developing non-lung
cancer:
Model 1: 0-9.99 |jg/m3: Ref.
10-19.99 |jg/m3:1.07 (0.96,1.19)
20-29.99 (jg/m3: 0.90 (0.80,1.02)
30+ |jg/m3: 0.98 (0.86,1.10)
Model 2: Per 10 |jg/m3: 0.99 (0.96,1.02)
Nafstad et al.
(2004)
Oslo, Norway
Period of Study:
1972-1998
Cohort study of 16,209 Norwegian men 40-49 yrs of
age living in Oslo, Norway, in 1972-1973. Data from
the Norwegian Death Register were linked with esti-
mates of avg yearly air pollution levels at the partici-
pants' home addresses from 1974 to 1998. NOx,
rather than N02was used. Exposure estimates for
NOx and SO2 were constructed using models based
on the subject's address, emission data for industry,
heating, and traffic, and measured concentrations.
Addresses linked to 50 of the busiest streets were
given an additional exposure based on estimates of
annual avg daily traffic. Cox proportional-hazards
regression was used to estimate associations
between exposure and total and cause-specific
mortality, adjusting for age strata, education,
occupation, smoking, physical activity level, and risk
groups for cardiovascular diseases
Mean SO2 Levels:
The yearly avg of 24-h avg
SO2 were reduced with a
factor of 7 during the study
period from 5.6 ppb in 1974
to 0.8 ppb in 1995.
Copollutants:
NOx
SO2 did not show any associations with lung cancer,
e.g., 1.00 (0.93,1.08) per 10 |jg/m3 increase
mortality in SO2. No association was also observed
when including SO2 in the model as a categorical
variable. Note the very low levels of SO2.
Nyberg et al. (2000) Case-control study of men 40-70 yrs, with 1,042 Annual levels computed for Little effect of SOx in any time window, but highest
Stockholm County,
Sweden
Period of Study:
Jan 1,1985-Dec
31,1990
cases of lung cancer and 1,274 controls, to evaluate
the suitability of an indicator of air pollution from
heating.
each yr between 1950 and
1990, but not provided
herein
NOx/N02
correlations in early yrs.
SOx RR (CI 95%) from heating (per 10 (jg/m3) for
30-yr avg
< 41.30 (jg/m3:1
>41.30 to < 52.75:1.06 (0.83,1.35)
>52.75 to < 67.14: 0.98 (0.77,1.24)
>67.14 to < 78.20: 0.90 (0.68,1.19)
>78.20:1.00 (0.73,1.37)
SOx RR (CI 95%) from heating (per 10 (jg/m3)
10-yravg
< 66.20 (jg/m3:1
>66.20 to < 87.60:1.16(0.91,1.47)
> 87.60 to < 110.30: 1.00 (0.79,1.27)
>110.30 to < 129.10:0.92 (0.70,1.21)
> 129.10:1.21 (0.89,1.66)
for
F-103
-------
Table F-8. Long-term exposure to SO2 and prenatal and neonatal outcomes.
Study
Methods
Pollutant Data
Findings
UNITED STATES 1
Bell etal. (2007)
Connecticut and
Massachusetts
Period of Study:
1999-2002
Outcome(s): LBW
Study design: Case-control
N: 358,504 live singleton births
Statistical Analysis: Linear models and logistic
regression
Gestational exposure (ppb)
Mean: 4.7
SD: 1.2
IQR: 1.6
Copollutants:
N02
No relationship between gestational exposure to
SO2 and birth weight. First trimester exposure to
S02was associated with low birth weight. No
statistical difference n the effect estimates of SO2 for
infants of black and white mothers.
Increment: 1.6 ppb (IQR)
Covariates: Gestational length, prenatal care, type of
delivery, child's sex, birth order, weather, yr, and
mother's race, education, marital status, age, and
tobacco use.
CO
PM10
PM2.5
Change in birth weight:
Entire pregnancy: -0.9 g (-4.4, 2.6)
Black mother: 1.2 (-6.5, 8.8)
White mother: -1.4 (-5.1, 2.3)
1st trimester: -3.7 to -3.3 grams
LBW: OR 1.003 (0.961,1.046)
Gilboa et al. (2005) Outcome(s): Selected birth defects
Study design: Case-control
N: 4,570 cases and 3,667 controls
Statistical Analysis: Logistic regression
Covariates: Maternal education, maternal
race/ethnicity, season of conception, plurality,
maternal age, maternal illness
Statistical package: SAS vs. 8.2
Seven Texas
Counties
Period of Study:
1997-2000
Levels NR
Copollutants
PM10
03
N02
CO
When the fourth quartile of exposure was compared
with the first, SO2 was associated with increased
risk of isolated ventricular septal defects. Inverse
associations were noted for SO2 and risk of isolated
atrial septal defects and multiple endocardial
cushion defects.
Aortic artery and valve defects: < 1.3 ppb: 1.00;
1.3 to < 1.9: NA; 1.9 to < 2.7:1.06 (0.34, 3.29);
>2.7:0.83 (0.26,2.68)
Atrial septal defects: < 1.3 ppb: 1.00;
1.3 to < 1.9:1.22 (0.79,1.88);
1.9 to <2.7: 0.76 (0.47,1.23);
>2.7:0.42 (0.22,0.78)
Pulmonary artery and valve defects: < 1.3 ppb: 1.00;
.3 to < 1.9:0.63 (0.23,1.74);
1.9 to <2.7: 0.93 (0.36,2.38);
>2.7:1.07 (0.43,2.69)
Ventricular septal defects: < 1.3 ppb: 1.00;
1.3 to < 1.9:1.02 (0.68,1.53);
1.9 to <2.7:1.13 (0.76,1.68);
>2.7:2.16(1.51,3.09)
Conotruncal defects: < 1.3 ppb: 1.00;
1.3 to < 1.9: 0.71 (0.46,1.09);
1.9 to <2.7: 0.71 (0.46,1.09);
>2.7:0.58 (0.37,0.91)
Endocardial cushion and mitral valve defects:
< 1.3 ppb: 1.00; 1.3 to < 1.9: 0.89 (0.50,1.61);
1.9 to <2.7: 0.89 (0.49,1.62);
>2.7:1.18 (0.68,2.06)
Cleft lip with or without cleft palate: < 1.3 ppb: 1.00;
1.3 to < 1.9: 0.79 (0.52,1.20);
1.9 to <2.7: 0.95 (0.64,1.43);
>2.7:0.75 (0.49,1.15)
Cleft palate: < 1.3 ppb: 1.00;
1.3 to < 1.9: 0.89 (0.40,1.97);
1.9 to <2.7:1.49 (0.72,3.06);
>2.7:1.22 (0.56,2.66)
F-104
-------
Study
Methods
Pollutant Data
Findings
Lipfert et al. (2000c) Mortality
United States Outcome(s): SIDS Study design: Cohort
Period of Study: Statistical Analysis: Three logistic regression
1990 analyses to examine the relation between annual avg
air pollutant values and various infant mortality
endpoints (i.e., all causes, SIDS, respiratory, and
other causes
Statistical package: NR
Age groups analyzed: 0-1
Covariates: Altitude, degree days (°F), median
income (U.S. $), population density
Lag(s): N/A
Mean (SD) SO2: 4.57 (2.6) ppb
Range: 0.3, 9.9
Copollutants:
PM10 (r = 0.04)
CO (r = -0.15)
S042- (r = 0.36)
NSPM10 (r = -0.13)
In a model that included states that lacked data on
maternal education and smoking along with various
personal and ecological variables, SO2 was not
found to be a significant predictor of SIDS mortality
in infants with birth weights >2,500 g.
p = -0.0118 (0.0094)
Mean Risk (95% CI)
0.95 (0.87,1.03)
Maisonet et al.
(2001)
6 Northeastern
cities of U.S.
Period of Study:
1994-1996
Outcome(s): Term LBW
Study design: Case-control
N: 89,557 live singleton births
Statistical Analysis: Logistic regression models linear
regression models
Covariates: Maternal age, race, season of the yr,
smoking and alcohol use during pregnancy, firstborn,
gender, marital status, and previous terminations,
prenatal care (ordinal variable), weight gain, and
gestational age
Stratified by race/ethnicity
Statistical package: STATA
Exposure distribution
(< 25th, 25th to < 50th, 50th to
< 75th, 75th to < 95th, a 95th)
First trimester: < 7.09, 7.090 to
8.906,8.907 to 11.969,11.970
to 18.447, >18.448
Second trimester: < 6.596,
6.596 to 8.896, 8.897 to
11.959,11.960 to 18.275, >
18.276
Third trimester: < 5.810,
5.810 to 8.453,
8.454 to 11.777,11.778 to
18.134, >18.135
Copollutants:
CO
PM10
This study provides evidence of an increased risk
for term LBW in relation to increased ambient air
levels of SO2 at concentrations well below the
established standards. Higher risk estimates among
whites when stratified by race/ethnicity
First trimester:
< 25th: Referent
25th-50th: 1.04 (0.88,1.23)
50th-75th: 1.04 (0.94,1.15)
75th-95th: 0.98 (0.81,1.17)
> 95th: 0.88 (0.73,1.07)
Increment (10 ppm): 0.98 (0.93,1.03)
Second trimester:
25th-50th: 1.18(1.12,1.25)
50th-75th: 1.12(1.07,1.17)
75th-95th: 1.13 (1.05,1.22)
> 95th: 0.87 (0.80,0.95)
Increment (10 ppm): 1.01 (0.93,1.10)
Third trimester:
25th-50th: 1.04 (0.92,1.18)
50th-75th: 1.02 (0.87,1.18)
75th-95th: 1.04 (0.84,1.28)
> 95th: 1.06 (0.76,1.47)
Increment (10 ppm): 1.01 (0.86,1.20)
Sagiv et al. (2005) Outcome(s): Pre-term birth
4 Pennsylvania
counties
Period of Study:
1997-2001
Study design: Time-series
N: 187,997 births
Study design: Poisson-regression models
Covariates: Long-term trends, copollutants,
temperature, dew point temperature, and day of wk.
Lag: Daily lags ranging from 1-7 days
Mean SO2 Levels:
6-wk Mean: 7.9 ± 3.5 ppb
Range: 0.8,17
Median: 8.1
Daily Mean: 7.9 ± 6.2
Range: 0, 54.1
Median: 6.4
Copollutants:
PM10; r= 0.46
CO
N02
This study found an increased risk for preterm
delivery during the Iast6wks of pregnancy with
exposure to SO2.
Increment: 15 ppb
Mean: 6-wk SO2:
RR= 1.15 (1.00,1.32)
< 4.9 ppb: Referent
4.9 to 8.1 ppb: 1.02 (0.97,1.06)
8.1 to 10.6 ppb: 1.04 (0.98,1.10)
10.6 to 17.0 ppb: 1.06 (0.99,1.14)
Mean: Daily SO2:
RR= 1.07 (0.99,1.15) lag 3
Dales et al. (2004)
12 Canadian cities
Period of Study:
1984-1999
Outcome(s): SIDS
Study design: Time-series
N: 1556 SIDS deaths
Mean SO2 Levels:
24-h avg: 5.51 ppb
IQR: 4.92
Copollutants:
Statistical Analysis: Random effects regression model CO NO2 O3 PM10
Covariates: Temperature, humidity, barometric PM2.5PM10-2.5
pressure, season
Lag: 0-5 days
SIDS was associated with air pollution, with the
effects of SO2 seeming to be independent of
sociodemographic factors, temporal trends, and
weather.
Increment: 4.92 ppb (IQR)
Increase in SIDS incidence: 8.49%;
p = 0.0079 lag 1
F-105
-------
Study
Methods
Pollutant Data
Findings
Dales et al. (2006)
11 Canadian cities
Period of Study:
1986-2000
Outcome(s): Hospitalization for respiratory disease in
the neonatal period
Study design: Time-series
N: 9,542
Statistical Analysis: Random effects regression
model; Poisson using fixed- or random-effects model
Covariates: Fayofwk, temperature, humidity,
pressure
Lag: 0-5 days
Statistical package:
S-PLUS vs. 6.2
Mean SO2 Levels:
24-h avg: 4.3 ppb
IQR: 3.8
Copollutants:
N02 (r = 0.20, 0.67)
CO (r = 0.19, 0.66)
03 (r = -0.41, 0.13)
PM10 (r = -0.09, 0.61)
S04
This study detected a significant association for
respiratory disease among neonates and gaseous
air pollutants.
Increment: 3.8 ppb (IQR)
Increase in neonatal respiratory
hospital admissions:
S02 alone: 2.06% (1.04,3.08)
Multipollutant model: 1.66% (0.63, 2.69)
Multipollutant model restricted to days with PM10
measures: 1.41% (0.35, 2.47)
Dugandzic et al.
(2006)
Nova Scotia,
Canada
Period of Study:
1988-2000
Outcome(s): Term LBW
Study design: Retrospective cohort study
N: 74,284 term, singleton births
Statistical Analysis: Logistic regression models
Covariates: Maternal age, parity, prior fetal death,
prior neonatal death, and prior low birth weight infant,
smoking during pregnancy, neighborhood family
income, infant gender, gestational age, weight
change, andyr of birth.
Statistical package:
SAS vs. 8.0
Mean: SO210 ppb
Median: 10
25th%: 7
75th%: 14
Max: 38
Copollutants:
03
PM10
In the analyses unadjusted for birth yr, first trimester
exposures in the highest quartile for SO2 associated
with increased risk of LBW. After adjusting for birth
yr, RR attenuated and not statistically significant.
There was a linear concentration-response effect
with increasing levels of SO2 during the first
trimester.
First Trimester
25th-50th: 0.96 (0.73,1.28)
51 st-75th: 1.18(0.88,1.58)
>75th: 1.36 (1.04,1.78)
Increment (7 ppb): 1.20 (1.05,1.38)
Second Trimester
25th-50th: 1.12 (0.86,1.46)
51 st-75th: 1.13(0.85,1.50)
>75th: 1.04 (0.79,1.37)
Increment (7 ppb): 0.99 (0.87,1.13)
Third Trimester
25th-50th: 1.04 (0.80,1.34)
51 st-75th: 0.85 (0.63,1.15)
>75th: 0.88 (0.67,1.15)
Increment (7 ppb): 0.93 (0.81,1.06)
Liu etal. (2003)
Vancouver, Canada
Period of Study:
1985-1998
Outcomes: Preterm birth, LBW, IUGR
Study design: Case-control
N: 229,085 singleton live births
Statistical Analysis: Multiple logistic regressions
Covariates: Maternal age, parity, infant sex,
gestational age or birth weight and season of birth
24-h avg: 4.9 ppb,
5th: 1.5
25th
50th
75th
95th
2.8
4.3
6.3
10.5
100th: 30.5
1-h max: 13.4 ppb,
5th: 4.3
25th
50th
75th
95th
7.8
11.7
16.8
28.3
100th: 128.5
Copollutants:
N02 (r = 0.61)
CO (r = 0.64)
03 (r = -0.35)
LBW and IUGR were associated with maternal
exposure to SO2 during the first mo of pregnancy
and preterm birth was associated with SO2 during
the last mo. These results were robust to adjustment
for copollutants.
Increment: 5 ppb
Low birth weight
First mo: OR 1.11 (1.01,1.22)
Last mo: OR 0.98 (0.89,1.08)
Preterm birth
First mo: OR 0.95 (0.88,1.03)
Last mo: OR 1.09 (1.01,1.19)
IUGR
First mo: OR 1.07 (1.01,1.13)
Last mo: OR 1.00 (0.94,1.06)
First trimester: OR 1.07 (1.00,1.14)
Second trimester: 0.98 (0.91,1.04)
Third trimester: 1.03 (0.96,1.10)
F-106
-------
Study
Methods
Pollutant Data
Findings
Liu et al. (2006)
Calgary, Edmonton
and Montreal,
Canada
Period of Study:
1986-2000
Outcome(s): IUGR
Study design: Case-control
N: 386,202 singleton live births
Statistical Analysis: Multiple logistic regression
Covariates: Maternal age, parity, infant sex, season of
birth, city of residence
24-h avg: 3.9 ppb, 25% 2.0
PPb
50% 3.0 ppb
75% 5.0 ppb
95% 10.0 ppb
1-h max: 10.8 ppb, 25% 5.0
ppb
50% 8.6 ppb
75% 14.0 ppb
95% 28.0 ppb
Copolllutants:
N02 (r = 0.34)
CO (r = 0.21)
03 (r = -0.30)
PM2.5 (r = 0.44)
IUGR did not increase with maternal exposure to
SO2. Risk decreased during first 3 mos.
Increment: 3.0 ppb
ORs estimated from graph:
1st mo: OR -0.966 (0.94,0.99)
2nd mo: OR -0.97 (0.95, 0.995)
3rd mo: OR -0.97 (0.95, 0.995)
1st trimester: OR -0 .96 (0.93, 0.99)
Bobak (2000)
Czech Republic
Period of Study:
1991
Outcomes: LBW, preterm birth
Study design: Case-control
N: 108,173 live singleton births
Statistical Analysis: Logistic regression
Covariates: Temperature, humidity, day of wk,
season, residential area, maternal age, gender
Statistical package: STATA
Mean trimester exposures
25th
50th
75th
17.5 (jg/m3
32.0 (jg/m3
55.5 (jg/m3
Copollutants:
TSP; r = 0.68 0.73
NOx;r = 0.53, 0.63
LBW and preterm birth were associated with
maternal exposure to SO2, though the association
between SO2 and LBW was explained to a large
extent by low gestational age.
Increment: 50 |jg/m3
LBW (adjusted for sex, parity, maternal age group,
education, marital status, and nationality, and mo of
birth):
1st trimester: 1.20 (1.11,1.30)
2nd trimester: 1.14 (1.06,1.22)
3rd trimester: 1.14(1.06,1.23)
LBW (also adjusted for gestational age):
1st trimester: 1.01 (0.88,1.17)
2nd trimester: 0.95 (0.82,1.10)
3rd trimester: 0.97 (0.85,1.10)
Preterm birth (AOR):
1st trimester: 1.27 (1.16,1.39)
2nd trimester: 1.25 (1.14,1.38)
3rd trimester: 1.24 (1.13,1.36)
Reduction in mean birth weight:
1st trimester: 11.4 g (5.9,16.9)
Mohorovic (2004) Outcomes: LBW and preterm delivery
Labin, Istra, Croatia Study design: Cross-sectional
N: 704 births
Statistical Analysis: Multiple correlation analyses,
factor analyses, chi-square
Statistical package: DBASE IV, SPSS
Period of Study
1987-1989
Monthly ground levels of SO2:
Range: 34.1, 252.9 |jg/m3
The results show an association between SO2
exposure at the end of the first and second mo of
pregnancy and a negative correlation between
length of gestations and lower birth weight of
newborns.
Correlation coefficients:
1st mo: Gestation length: -0.09, p = 0.008
Birthweight: -0.08, p = 0.016
2nd mo: Gestation length: -0.08, p = 0.016
Birthweight: -0.07, p = 0.026
3rd mo: Gestation length: -0.04, p = 0.147
Birthweight: -0.04, p = 0.135
6th mo: Gestation length: -0.02, p = 0.266
Birthweight: -0.04, p = 0.151
Whole pregnancy:
Gestation length: -0.09, p = 0.007
Birthweight: -0.04, p = 0.153
Weekly avg during whole pregnancy:
Gestation length: -0.05, p = 0.086
Birthweight: -0.06, p = 0.069
F-107
-------
Study
Methods
Pollutant Data
Findings
Pereira et al. (1998) Outcome(s): Intrauterine mortality
Sao Paulo, Brazil
Period of Study:
1991-1992
Study design: Time-series
Statistical Analysis: Poisson regression models
Covariates: Mo, day of wk, min daily temperature,
relative humidity
Lag: 2 to 14 days
24-h avg SO2:
18.90 (8.53) mg/m3
Range: 3.80, 59.70
Copollutants:
PM10 (r = 0.45)
N02 (r = 0.41)
03 (r = 0.17)
CO (r = 0.24)
SO2 exhibited a marginal association with
intrauterine mortality, but only when Poisson
regression was employed. A concentration-
response relationship was found.
Estimated regression coefficients and standard
errors:
S02 alone: 0.0038 (0.0020)
S02 + N02 + CO + PM10 + 03: 0.0029 (0.0031)
LATIN AMERICA
Gouveia et al.
(2004)
Sao Paulo, Brazil
Period of Study:
1997
Outcome(s): LBW
Study design: Time-series
N: 179,460 live singleton births
Statistical Analysis: Logistic and linear regression with
GAM
Covariates: Gender, gestational age, maternal age,
maternal education, antenatal care, parity, delivery
method
Statistical package: S-Plus 2000
Annual Mean: SO2 ((jg/m3)
Mean: 19.6
SD: 10.3
Range: 3.4, 56.9
Jan-Mar: 22.3 (7.7)
Apr-June: 28.1 (10.1)
Jul-Aug: 17.9 (8.7)
Oct-Dec: 10.3(3.9)
Copollutants:
PM10
CO
N02
03
First and second trimester exposures to SO2 had a
significant association with birth weight, though in
different directions. When air pollutants were
divided into quartiles and the lowest quartile was
used as the referent exposure category, SO2
during the second trimester was marginally
associated with low birth weight.
Increment: 10 |jg/m3
Reduction in birth weight
First trimester: -24.2 g (-55.5, 7.1)
Second trimester: 33.7 g (1.6, 65.8)
Third trimester: 9.7 g (-25.6, 44.9)
First trimester:
2nd: 0.902 (0.843, 0.966)
3rd: 0.911 (0.819, 1.013)
4th: 0.906 (0.793,1.036)
Second trimester:
2nd: 0.986 (0.922,1.053)
3rd: 1.005 (0.904,1.117)
4th: 1.017 (0.883,1.173)
Third trimester:
2nd: 1.203 (0.861,1.68)
3rd: 1.225 (0.872, 1.722)
4th: 1.145 (0.749,1.752)
Ha et al. (2001)
Seoul, Korea
Period of Study:
1996-1997
Outcome(s): LBW
Study design: Time-series
N: 276,763
Statistical Analysis: Multiple regression, GAM
Covariates: Gestational age, maternal age, parental
education level, infant's birth order, gender
24-h avg:
1st trimester:
25th: 10.0 ppb
50th: 13.2 ppb
75th: 16.2 ppb
3rd trimester:
25th: 8.4 ppb
50th: 12.2 ppb
75th: 16.3 ppb
Copollutants:
CO (r = 0.83) N02 (r = 0.70)
TSP (r = 0.67) 03 (r = -0.29)
Ambient SO2 concentrations during the first
trimester of pregnancy were associated with LBW
Increment: 1st trimester: 6.2 ppb;
3rd trimester: 7.9 ppb
1st trimester: RR 1.06 (1.02,1.10)
3rd trimester: RR 0.93 (0.88, 0.98)
Reduction in birth weight (1st trimester): 8.06 g
(5.59,10.53)
Lee et al. (2003a)
Seoul, Korea
Period of Study:
1996-1998
Outcome(s): Term LBW
Study design: Time series
N: 388,105 full-term singleton births
Statistical Analysis: GAM
Covariates: Infant sex, birth order, maternal age,
parental education level, time trend, and gestational
age.
Avg concentration (ppb)
Mean: 12.1
SD: 7.4
Range: 3, 46
25th
50th
75th
6.8
9.8
15.6
Copollutants:
PM10; r= 0.78, 0.85
CO; r = 0.79, 0.86
NO2; r = 0.75, 0.76
Second trimester exposures to SO2 as well as
during the entire pregnancy were associated with
LBW. Reduction in birth weight was 14.6 g for IQR
increase in SO2 in the second trimester. When the
exposure for each mo of pregnancy was evaluated
separately, SO2 exposure during 3 to 5 mos of
pregnancy associated with LBW. Increment: 8.8 ppb
(IQR)
First trimester: 1.02 (0.99,1.06)
Second trimester: 1.06 (1.02,1.11)
Third trimester: 0.96 (0.91,1.00)
All trimesters: 1.14 (1.04,1.24)
F-108
-------
Study
Methods
Pollutant Data
Findings
Leem et al. (2006) Outcome(s): Preterm delivery
Incheon, Korea
Period of Study:
2001-2002
Study design: Time series
N: 52,113 singleton births
Statistical Analysis: Log-binomial regression
Covariates: Maternal age, parity, sex, season,
maternal education, paternal education
Mean: SO2 concentrations by
trimester:
1st trimester:
Min: 7.86 |jg/m3
25th: 17.61. 50th: 22.74
75th: 45.85. Max: 103.96
3rd trimester:
Min: 6.55 |jg/m3
25th: 17.03. 50th: 25.62
75th: 46.53. Max: 103.15
Copollutants:
N02 (r = 0.54) CO (r = 0.31)
PM10 (r = 0.13)
This study found the highest SO2 concentrations
during the first trimester to be significantly
associated with elevated risks of preterm delivery.
1st trimester: 7.86 to 17.61 |jg/m3: referent
17.62 to 22.74:1.13 (0.99,1.28)
22.75 to 45.85:1.13 (0.98,1.30)
45.86 to 103.96:1.21 (1.04,1.42)
3rd trimester: 6.55 to 17.03 |jg/m3: referent
17.04 to 25.62: 0.87 (0.76,1.01)
25.63 to 46.53: 0.97 (0.83,1.13)
46.54 to 103.15:1.11 (0.94,1.31)
Lin et al. (2004a)
Kaohsiung and
Taipei, Taiwan
Period of Study:
1995-1997
Outcome(s): LBW
Study design: Case-control
N: 92,288 live births
Statistical Analysis: Multiple logisistic regression
Covariates: Gestational period, gender, birth order,
maternal age, maternal education, season of birth
24-h avg:
Kaohsiung:
Range: 10.07, 25.36 ppb
Taipei:
Range: 5.65, 9.33 ppb
Copollutants:
CO NO2O3PM10
Few women living in Taipei were exposed to high
levels of SO2. In Kaohsiung, almost all women were
exposed to high levels of SO2. Women living in
Kaohsiung had significantly higher risk of term LBW
compared with women living in Taipei.
OR for Kaoshiung births (compared to Taipei births)
All births: OR: 1.13 (1.03,1.24)
Female births only: OR: 1.14 (1.01,1.28)
Lin et al. (2004b)
Kaohsiung and
Taipei, Taiwan
Period of Study:
1995-1997
Outcome(s): Term LBW
Study design: Cohort
N: 92,288 live births
Statistical Analysis: Multiple logisistic regression
Covariates: Gestational period, gender, birth order,
maternal age, maternal education, season of birth
24-h avg:
Kaohsiung:
Range: 10.07, 25.36 ppb
Taipei:
Range: 5.65, 9.33 ppb
Copollutants:
CO
N02
03
PM10
This study found a 26% higher risk of term LBW
delivery for mothers exposed to mean SO2
concentrations exceeding 11.4 ppb during the entire
pregnancy, as compared with mothers exposed to
mean concentrations less than 7.1 ppb. Trimester
specific analysis showed a significant association
only for the third trimester. Lowest quartile of
exposure = referent
Entire pregnancy: 25th-75th: 1.16 (1.02,1.33)
>75th: 1.26 (1.04,1.53)
1st trimester: 25th-75th: 1.02 (0.90,1.16)
>75th: 1.11 (0.94,1.33)
2nd trimester: 25th-75th: 1.09 (0.96,1.24)
>75th: 1.17 (0.99,1.37)
3rd trimester:25th-75th: 1.13 (0.99,1.28)
>75th: 1.20(1.01,1.41)
Wang et al. (1997) Outcome(s): Term LBW
Four residential
areas: Congwen,
Dongcheng,
Xicheng, Xuanwu
Beijing, China
Period of Study:
1988-1991
Study design: Cohort study
N: 74,671 first parity live births
Statistical Analysis: Multiple linear regression and
logistic regression
Covariates: Gestational age, residence, yr of birth,
maternal age, and infant gender.
Mean pollution concentrations
provided in graph
TSP; r = 0.92
Exposure-response relationship between SO2 during
the third trimester of pregnancy and low birth weight.
3rd trimester: 9 to 18 |jg/m3 (reference)
18 to 55:1.09 (0.94,1.26)
55 to 146:1.12 (0.97,1.29)
146 to 239:1.16 (1.01,1.34)
239 to 308:1.39 (1.22,1.60)
SO2 as continuous variable: Odds ratio per 100
(jg/m3:1.11 (1.06,1.16)
F-109
-------
Study
Methods
Pollutant Data
Findings
Xu et al. (1995)
Four residential
areas:
Congwen,
Dongchen, Xichen,
Xuanwu
Beijing, China
Period of Study:
Outcome(s): Preterm delivery
Study design: Prospective cohort study
N: 25,370 singleton first live births
Statistical Analysis: Multiple linear and logistic
regression
Covariates: Temperature, humidity, day of wk,
season, residential area, maternal age, and gender of
child.
2 monitors forSCfe
Dongcheng andXicheng
Dongcheng annual mean: 108
|jg/m3SD: 141 (jg/m3)
Xicheng annual mean:
93 (jg/m3 (SD: 122 (jg/m3)
Copollutants:
TSP
Exposure response relationship between quartiles of
S02and crude incidence rates of preterm birth.
Dose dependent relationship between SO2 and
gestational age. The estimated reduced length of
gestation was 0.075 wks or 12.6 h per 100/m3
increase in SO2. When TSP and SO2 included in a
multipollutant model, the effect of SO2 was reduced
by 32%. Effect on gestational age (wk) per
100 (jg/m3 regression coef and SE for lagged
moving avg of SO2.
lagO
lag 6
lag 7
lag
-0.016(0.021). lag 1:-0.022 (0.021)
-0.067 (0.024), p < 0.01
-0.075 (0.024) ,p< 0.01
-0.075 (0.025), p < 0.01
OR for each quartile of SO2
1st: 1.00.2nd: 1.70(1.15,2.52)
3rd: 1.74 (1.03, 2.92). 4th: 1.58 (0.87, 2.86)
Adjusted OR for preterm delivery:
1.21 (1.01,1.46) per In |jg/m3 increase in SO2
Yang et al. (2003a) Outcome(s): Term LBW
Kaohsiung, Taiwan Study design: Cohort
N: 13,396 first parity singleton live births
Statistical Analysis: Multiple linear regression
Covariates: Maternal age, season, marital status,
maternal education, gender
Statistical package: SAS
Period of Study
1995-1997
Mean: trimester exposure
(|jg/m3)
1st trimester
33rd: 26.02
67th: 36.07
2nd trimester
33rd: 25.76
67th: 35.63
3rd trimester
33rd: 25.39
67th: 36.96
Copollutants:
PM10 (r = 0.45, 0.46)
A significant exposure-response relationship
between maternal exposures to SO2 and birth
weight was found first trimester of pregnancy.
Reduction in birth weight:
1st trimester:
33rd-67th: 3.68 g (-12.45,19.21)
>67th: 18.11 g (1.88, 34.34)
Continuous: 0.52 g (0.09, 2.63)
2nd trimester:
33rd-67th: 1.78 g (-17.91,14.35)
>67th: 13.53 g (-2.62, 29.68)
Continuous: 0.19 g (-0.78,1.8)
3rd trimester:
33rd-67th: 0.43 g (-16.56,15.70)
>67th: 1.97 g (-18.24,14.30)
Continuous: 0.03 g (-1.21,1.37)
F-110
-------
Table F-9. Long-term exposure to SO2 and mortality.
Study Pollutant Data Methods Findings
UNITED STATES
Abbey et al. (1999) 24-h avg SO2: 5.6 ppb
Three California air
basins: San Diego,
San Francisco, South
Coast (Los Angeles
and eastward)
Period of Study: 1977-
1992
Prospective cohort study of 6,338 nonsmoking
non-Hispanic white adult members of the
Adventist Health Study followed for all cause,
cardiopulmonary, nonmalignant respiratory, and
lung cancer mortality. Participants were aged 27-
95 yrsat enrollment in 1977.1,628 (989 females,
639 males) mortality events followed through
1992. All results were stratified by gender. Used
Cox proportional hazards analysis, adjusting for
age at enrollment, past smoking, environmental
tobacco smoke exposure, alcohol use, educa-
tion, occupation, and body mass index. Analyzed
mortality from all natural causes,
cardiopulmonary, nonmalignant respiratory, and
lung cancer.
SO2 was not associated with total RR = 1.07 (95%
CI: 0.92,1.24) for male and 1.00 (95% CI: 0.88,
1.14) for female per 5 ppb increase in multiyear avg
SO2, cardiopulmonary, or respiratory mortality for
either sex. Lung cancer mortality showed large risk
estimates for most of the pollutants in either or both
sexes, but the number of lung cancer deaths in this
cohort was very small (12 for female and 18 for
male) Generally wide confidence intervals (relative
to other U.S. cohort studies).
Dockery et al. (1993) 24-h avg SO2 ranged from 1
and reanalysis by (Topeka) to 24.0
Krewski et al. (2000) (Steubenville) ppb.
Harriman, TN; Portage,
Wl;
St. Louis, MO;
Steubenville, OH;
Topeka, KS;
Watertown, MA;
Period of Study: 1974-
1991.
.6 A prospective cohort study to study the effects of
air pollution with main focus on PM components
in six U.S. cities, which were chosen based on
the levels of air pollution (Portage, Wl, the least
polluted to Steubenville, OH, the most polluted).
Cox proportional hazards regression was
conducted with data from a 14-to-16-yr follow-up
of 8,111 adults in the six cities, adjusting for
smoking, sex, BMI, occupational exposures, etc.
PM2.5 and sulfate were associated with these
causes of deaths.
SO2 result presented only graphically. Fine particles
and sulfate showed better fit than SO2.
In the reanalysis by Krewski et al (2000), SO2
showed positive associations with total RR = 1.05
(95% CI: 1.02,1.09) per 5 ppb increase in the avg
SO2 over the study period, cardiopulmonary (1.05
(95% CI: 1.00,1.10)), and lung cancer deaths (1.03
(95% CI: 0.91,1.16)), but in this dataset, S02was
highly correlated with PM2.5 (r = 0.85), sulfate (r =
0.85), and NO2 (r= 0.84)
F-111
-------
Study
Pollutant Data
Methods
Findings
Multiyear avg of 24-h avg 9.3 Re-analysis of Pope etal. (1995) study,
ppb. Extensive sensitivity analysis with ecological
covariates and spatial models to account of
spatial pattern in the ACS data.
Krewski et al. (2000);
Jerrett et al. (2003);
Krewski et al. (2003);
Re-analysis/sensitivity
analysis of Pope et al.
(1995) study.
Lipfertetal. SO2 mean levels NR.
(2006)
32 Veterans
Administration
hospitals nationwide
in the U.S.
Period of Study: 1976-
1996
Cohort study of approximately 50,000 U.S.
veterans (all males) diagnosed with
hypertension. Mean age at recruitment was 51
yrs. Exposure to O3 during four periods (1960-
1974,1975-1981,1982-1988,1989-1996)
associated with mortality over three periods
(1976-1981,1982-1988,1989-1996). Long-term
exposures to TSP, PM15, PM10, PM2.5, PM15-2.5,
S042-, NO2, and CO also analyzed. Used Cox
proportional hazards regression, adjusting for
race, smoking, age, systolic and diastolic blood
pressure, body mass index, and socioeconomic
factors.
In the Jerret et al. reanalysis the relative risk
estimates for total mortality was 1.06 (95% CI: 1.05,
1.07) per 5 ppb increase in the annual avg SO2. In
the spatial filtering model (this was the model that
resulted in the largest reduction of SO2 risk estimate
when sulfate was included), the SO2 total mortality
risk estimate was 1.07 (95% CI: 1.03,1.11) in the
single-pollutant model and 1.04 (95% CI: 1.02,1.06)
with sulfate in the model. The risk estimates for
PM2.5 and sulfate were diminished when SO2 was
included in the models.
In the sensitivity analysis conducted by Krewski
et al. SO2 was significantly associated with all-cause
mortality in a single-pollutant model: 1.30 (1.23,
1.38). In the spatial analysis, SO2 was found to be
the only gaseous pollutant strongly associated with
all-cause and cardiopulmonary disease mortality:
Relative Risk (95% CI) (per 19.9 |jg/m3 Sulfates)
Spatial Analysis, All-Cause:
Independent Observations:
Sulfates + S02:1.05 (0.98,1.12)
Fine particles + SO2:1.03 (0.95,1.13)
Independent Cities:
Sulfates + S02:1.13 (1.02,1.25)
Fine particles + SO2:1.14 (0.98,1.32)
Regional Adjustment:
Sulfates + S02:1.10 (0.97,1.24)
Fine particles + SO2:1.11 (0.93,1.33)
Spatial Filtering: Sulfates + SO2:1.05 (0.97,1.14)
Cardiopulmonary Disease:
Independent Observations:
Sulfates + S02:1.13 (1.03,1.24)
Fine particles + SO2:1.17 (1.03,1.33) Independent
Cities:
Sulfates + S02:1.18 (1.04,1.34)
Fine particles + SO2:1.25 (1.05,1.49)
Regional Adjustment:
Sulfates + S02:1.12 (0.96,1.32)
Fine particles + SO2:1.23 (0.97,1.55)
Spatial Filtering: Sulfates + SO2:1.10 (0.99,1.22)
Lung Cancer:
Independent Observations:
Sulfates + S02:1.37 (1.08,1.73)
Independent Cities:
Sulfates + S02:1.39 (1.08,1.81)
SO2 andd Pb were considered less thoroughly. The
authors presented only qualitative results for SO2
from the "screening regressions" which indicated
negatively significant risk estimate in the univariate
model and non-significant positive estimate in the
multivariate model.
F-112
-------
Study
Pollutant Data
Methods
Findings
Lipfert et al. (2000b) Mean ofthe 95th percentile of Update of the Lipfert et al. (2000a) study, with
32 Veterans
Administration
hospitals nationwide in
the U.S.
Period of Study: 1976-
2001
the 24-h avg SO2 for 1997-
2001 period: 15.8 ppb.
follow-up period extended to 2001. Study
focused on the traffic density data. The county-
level traffic density was derived by dividing
vehicle-km traveled by the county land area.
Because ofthe wide range ofthe traffic density
variable, log-transformed traffic density was used
in their analysis. They reported that traffic density
was a better predictor of mortality than ambient
air pollution variables, with the possible
exception of O3. The log-transformed traffic
density variable was weakly correlated with SO2
(r = 0.32) in this data set.
RR using the 1997-2001 air quality data period: 0.99
(95% CI: 0.97,1.01) per 5 ppb increase; in a single-
pollutant model.
The 2-pollutant model with the traffic density
variable: 0.99 (95% CI: 0.96,1.01) per 5 ppb.
Lipfert et al. (2006b) Mean ofthe 95th percentile of Update ofthe Lipfert et al. (2000a) study,
32 Veterans
Administration
hospitals nationwide in
the U.S.
Period of Study: 1997-
2001
the 24-h avg SO2 for 1999-
2001 period: 16.3 ppb.
examined PM2.5 chemical constituents data. The
analysis used county-level air pollution data for
the period 1999-2001 and cohort mortality data
for 1997-2001.
Traffic density was the most important predictor of
mortality, but associations were also seen for EC, V,
nitrate, and Ni. NO2, O3, and PM10 also showed
positive but weaker associations. The risk estimate
for SO2 was essentially the same as that reported in
the 2006a Lipfert et al. analysis (0.99 (95% CI: 0.96,
1.01) per 5 ppb) in a single-pollutant model.
Multipollutant model results were not presented for
S02.
Pope et al. (1995)
U.S. nationwide
Period of Study: 1982-
1989
Not analyzed/ reported.
Investigated associations between long-term
exposure to PM and the mortality outcomes in
the American Cancer Society cohort. Ambient air
pollution data from 151 U.S. metropolitan areas
in 1981 were linked with individual risk factors in
552,138 adults who resided in these areas when
enrolled in the prospective study in 1982. Death
outcomes were ascertained through 1989. Cox
proportional hazards model adjusted for
smoking, education, BMI, and occupational
exposures. PM2.5 and sulfate were associated
with total, cardiopulmonary, and lung cancer
mortality, but not with mortality for all other
Gaseous pollutants not analyzed.
Pope et al. (2002)
U.S. nationwide
Period of Study: 1982-
1998
24-h avg mean of 118 MSA's
in 1980: 9.7 ppb; mean of
126 MSA's during
1982-1998:6.7 ppb.
Prospective cohort study of approximately
500,000 members of American Cancer Society
cohort enrolled in 1982 and followed through
1998 for all cause, cardiopulmonary, lung cancer,
and all other cause mortality. Age at enrollment
was 30+ yrs. Air pollution concentrations in urban
area of residence at time of enrollment assessed
from 1982 through 1998. Other pollutants
considered include TSP, PM15, PM10, PM2.5, PM15-
2.5, SO42", SO2, NO2, and CO.
PM2.5 was associated with total, cardiopulmonary,
lung cancer mortality, but not with deaths for all
other causes. SO2 was associated with all the
mortality outcomes, including all other causes of
deaths. S02's risk estimate for total mortality was
1.03 (95% CI: 1.02,1.05) per 5 ppb increase (1982-
1998 average).Residential location was known only
at enrollment to study in 1982. Thus, exposure
misclassification possible.
Willis et al. (2003)
Re-analysis/ sensitivity
analysis of Pope et al.
(1995) study.
Multiyear average of
24-h avg using MSA scales:
9.3 ppb; using county scales:
10.7 ppb.
Investigation ofthe effects of geographic scale
over which the air pollution exposures are
averaged. Exposure estimates were averaged
over the county scale, and compared the original
ACS results in which MSA scale avg exposures
were used. Less than half of the cohort used in
the MSA-based study were used in the county
scale based analysis because ofthe limited
availability of sulfate monitors and because of
the loss of subjects from the use of five-digit zip
codes
In the analysis comparing the 2-pollutant model with
sulfate and SO2, they found that, in the MSA-scale
model, the inclusion of SO2 reduced sulfate risk
estimates substantially (>25%), but not substantially
(< 25%) in the county-scale model. In the MSA-level
anlaysis (with 113 MSA's), SO2 relative risk estimate
was 1.04 (95% CI: 1.02,1.06) per 5 ppb increase,
with sulfate in the model. In the county-level anaysis
(91 counties) with sulfate in the model, the
corresponding estimate was smaller (RR = 1.02
[95% CI: 1.00,1.05]). The correlations between
covariates are different between the MSA-level data
and county-level data.
F-113
-------
Study
Pollutant Data
Methods
Findings
CANADA
Finkelstein et al. 24-h avg (ppb): 4.9 (1.0)
(2003)
Ontario, Canada
Period of Study: 1992-
1999
Cohort consisting of 5,228 people >40 yrs old
that were referred for pulmonary function testing
between 1985 and 1999. Within the cohort
identified nonaccidental deaths that occurred
from 1992 through 1999. Used air quality data for
TSP from 1992-1994 and S02 for 1993-1995.
Analyzed the association between TSP or SO2
and socioeconomic status and mortality using a
Cox proportional hazards model stratified by sex
and 5-yr age groups.
Mean SO2 Levels:
Mean: 4.8 ppb, with a range of 1.5 to 11.8 ppb.
Copollutants:
PM2.5
BS
N02
A small area analysis of mortality rates in
electoral ward, with the mean area of 7.4 km2
and the mean population of 5,301 per electoral
ward. Deaths rates were computed for four
successive 4-yr periods from 1982 to 1994. The
number of wards in these four periods ranged
from 118 in the 1994-1998 period to 393 in the
1982-1986 period. Poisson model was fit to
model observed deaths for each ward with a
linear function for pollutant and random intercept,
with and without adjustment for social
deprivation.
Cohort study of 14,284 adults who resided in 24
areas from seven French cities when enrolled in
the PAARC survey (air pollution and chronic
respiratory diseases) in 1974. Daily
measurements of SO2, TSP, BS, NO2, and NO
were made in 24 areas for three yrs (1974-76).
Cox proportional hazards models adjusted for
smoking, educational level, BMI, and
occupational exposure. Models were run before
and after exclusion of six area monitors
influenced by local traffic as determined by the
NO/NO2 ratio >3.
Using the high income-low pollutant level as the
reference the following results were reported for
each of the mortality endpoints:
Relative Risk (95% CI) All causes
High income-high pollutant level: 1.35 (1.05,1.73)
Low income-low pollutant level: 1.64 (1.21, 2.24)
Low income-high pollutant level: 2.40 (1.61, 3.58)
Interaction with age group: 0.97 (0.95, 0.99)
Cardiopulmonary causes
High income-high pollutant level: 1.54 (1.13, 2.10)
Low income-low pollutant level: 2.05 (1.45, 2.91)
Low income-high pollutant level: 3.36 (2.12, 5.32)
Interaction with age group: 0.95 (0.92, 0.97)
Traffic intensity on the nearest road was not
associated with exposure SO2. Background SO2
levels were not associated with mortality.
Adjusted RR
(per 20 (jg/m3 SO2)
All cause: 0.97 (0.90,1.05)
Cardiovascular: 0.94 (0.82,1.06)
Respiratory: 0.88 (0.64,1.22)
They observed associations for both BS and SO2
and mortality outcomes. The estimated effects were
stronger for respiratory illness than other causes of
mortality for the most recent exposure periods and
most recent mortality period (pollution levels were
lower). The adjustment for social deprivation
reduced the risk estimates for both pollutants. The
adjusted risk estimates for SO2 for the pooled
mortality periods using the most recent exposure
windows were: 1.021 (95% CI: 1.018,1.024) for all-
cause; 1.015 (95% CI: 1.011,1.019) for
cardiovascular; and 1.064% (95% CI: 1.056,1.072)
for respiratory causes per 5 ppb increase in SO2.
The risk estimates for the most recent mortality
period using the most recent exposure windows
were larger.
Before exclusion of the six areas, none of the air
pollutants were associated with mortality outcomes.
After exclusion of these areas, analyses showed
associations between total mortality and TSP, BS,
N02, and NO, but not SO2 (1.01 (95% CI: 0.97,
1.06) per 5 ppb multi-yr average). From these
results, the authors noted that inclusion of air
monitoring data from stations directly influenced by
local traffic could overestimate the mean population
exposure and bias the results. It should be noted
that the table describing air pollution levels in Filleul
et al.'s report indicates that the SO2 levels in these
French cities declined markedly between 1974-76
and 1990-1997 period, by a factor of 2 to 3,
depending on the city, whereas NO2 levels between
the two periods were variable, increased in some
cities, and decreased in others. This change in air
pollution levels over the study period complicates
interpretation of reported risk estimates.
Beelen et al. (2008)
The Netherlands
Period of Study: 1987-
1996.
Cohort study on diet and
cancer with 120,852 subjects
followed from 1987 to 1996.
BS, NO2, SO2, and PM2.5 and
traffic-exposure estimates
were analyzed. Cox
regression model adjusted for
age, sex, smoking, and area-
level socioeconomic status.
Elliott etal. (2007)
Great Britain; 1966-
1994 air pollution;
1982-1998 mortality in
four periods.
24-h avg SO2 levels declined
from 41.4 ppb in 1966-1970
to 12.2 ppb in 1990-1994
Filleul et al. (2005)
Seven French cities
Period of Study: 1975-
2001
24-h avg SO2 ranged from 5.9
ppb ("Area 3" in Lille) to 29.7
ppb ("Area 3° in Marseille) in
the 24 areas in seven cities
during 1974-1976. Median
levels during 1990-1997
ranged from 3.0 ppb
(Bordeaux) to 8.2 ppb
(Rouen) in the five cities
where data were available.
F-114
-------
Study
Pollutant Data
Methods
Findings
Lepeule et al. (2006)
Bordeaux, France
Period of Study:
1988-1997
24-h avg (|jg/m3): 10.3 (6.6)
Identified 439 non-accidental deaths and 158
cardiorespiratory deaths from the Personnes
Agees QUID (PAQUID) cohort. Used a Cox
proportional hazards model with time dependent
covariates to examine the association between
BS and sulfur dioxide-strong acidity (SO2-AF)
and non-accidental and cardiorespiratory
mortality.
Relative Risk (per 10 |jg/m3 SO2-AF)
Non-accidental Mortality:
1.03 (0.86,1.24) lag 0. 0.96 (0.88, 1.06) lag 1
0.96 (0.85,1.09) lag 2.1.03 (0.94, 1.12) lag 3
1.17 (0.99,1.39) lag 4.1.16 (0.86,1.55) cumulative
Cardiorespiratory mortality:
0.84(0.65,1.10) lag 0.1.06 (0.94,1.19) lag 1
1.19(1.03,1.37) lag 2.1.19 (1.03, 1.37) lag 3
1.07 (0.95,1.19) lag 4. 0.85 (0.66, 1.10) lag 5
1.15 (0.75,1.77) cumulative
Nafstad et al. (2004)
Oslo, Norway
Period of Study: 1972-
1998.
The yearly averages of 24-h
avg S02were reduced with a
factor of 7 during the study
period from 5.6 ppb in 1974 to
0.8 ppb in 1995.
Cohort study of 16,209 Norwegian men 40-49
yrs of age living in Oslo, Norway, in 1972-1973.
Data from the Norwegian Death Register were
linked with estimates of avg yearly air pollution
levels at the participants' home addresses from
1974 to 1998. NOx, rather than N02was used.
Exposure estimates for NOx and SO2 were
constructed using models based on the subject's
address, emission data for industry, heating, and
traffic, and measured concentrations. Addresses
linked to 50 of the busiest streets were given an
additional exposure based on estimates of
annual avg daily traffic. Cox proportional-hazards
regression was used to estimate associations
between exposure and total and cause-specific
mortality, adjusting for age strata, education,
occupation, smoking, physical activity level, and
risk groups for cardiovascular diseases.
NOx was associated with total, respiratory, lung
cancer, and ischemic heart disease deaths. SO2 did
not show any associations with mortality (e.g., 0.97
(95% CI: 0.94,1.01) per 5 ppb multi-yr average).
The risk estimates presented for categorical levels
of these pollutants showed mostly monotonic
exposure-response relationships for NOx, but not for
SO2. Note the very low levels of SO2.
F-115
-------
References
Abbey DE; Nishino N; McDonnell WF; Burchette RJ; Knutsen SF; Beeson WL; Yang JX. (1999). Long-
term inhalable particles and other air pollutants related to mortality in nonsmokers. Am J Respir
Crit Care Med 159,373-382.
Abe M. (1967). Effects of mixed N02-S02 gas on human pulmonary functions. Effects of air pollution on
the human body. Bull Tokyo Med Dent Univ, 14, 415-433.
Abraham WM; Oliver W, Jr.; Welker MJ; King MM; Wanner A; Sackner MA. (1981). Differences in
airway reactivity in normal and allergic sheep after exposure to sulfur dioxide. J Appl Physiol, 51,
1651-1656.
Abraham WM; Sielczak MW; Delehunt JC; Marchette B; Wanner A. (1986). Impairment of tracheal
mucociliary clearance but not ciliary beat frequency by a combination of low level ozone and
sulfur dioxide in sheep. Eur J Respir Dis, 68, 114-120.
Ackermann-Liebrich U; Leuenberger P; Schwartz J; Schindler C; Monn C; Bolognini G; Bongard JP;
Brandli O; Domenighetti G; Elsasser S; Grize L; Karrer W; Keller R; Keller-Wossidlo H; Kunzli
N; Martin BW; Medici TC; Perruchoud AP; Schoni MH; Tschopp JM; Villiger B; Wuthrich B;
Zellweger JP; Zemp E. (1997). Lung function and long term exposure to air pollutants in
Switzerland. Study on Air Pollution and Lung Diseases in Adults (SAPALDIA) Team. Am J
Respir Crit Care Med, 155, 122-129.
Adamkiewicz G; Ebelt S; Syring M; Slater J; Speizer FE; Schwartz J; Suh H; Gold DR. (2004).
Association between air pollution exposure and exhaled nitric oxide in an elderly population.
Thorax, 59, 204-209.
Agar A; Kucukatay V; Yargicoglu P; Aktekin B; Kipmen-Korgun S; Gumuslu D; Apaydin C. (2000). The
effect of sulfur dioxide inhalation on visual evoked potentials, antioxidant status, and lipid
peroxidation in alloxan-induced diabetic rats. Arch Environ Contam Toxicol, 39, 257-264.
Agarwal R; Jayaraman G; Anand S; Marimuthu P. (2006). Assessing Respiratory Morbidity Through
Pollution Status and Meteorological Conditions for Delhi. Environ Monit Assess, 114, 489-504.
Aharonson EF; Menkes H; Gurtner G; Swift DL; Proctor DF. (1974). Effect of respiratory airflow rate on
removal of soluble vapors by the nose. J Appl Physiol, 37, 654-657.
Aikawa M; Hiraki T; Yamagami M; Kitase M; Nishikawa Y; Uno I. (2008). Regionality and particularity
of a survey site from the viewpoint of the S02 and S042" concentrations in ambient air in a 250-
kmx250-km region of Japan. Atmos Environ, 42, 1389-1398.
Ainsworth, BE; Haskell WL; Leon AS; Jacobs DR Jr; Montoye HJ; Sallis JF; Paffenbarger RS Jr. (1993).
Compendium of physical activities: classification of energy costs of human physical activities.
Med Sci Sports Exerc 25, 71-80.
Ainsworth N; Fumagalli I; Giorcelli A; Mignanego L; Schenone G; Vietto L. (1996). Assessment of EDU
stem injections as a technique to investigate the response of trees to ambient ozone in field
conditions, Agr Ecosyst Environ, 59 (1-2) pp. 33-42, 1996.
R-1
-------
Amdur MO; Melvin WW, Jr.; Drinker P. (1953). Effects of inhalation of sulphur dioxide by man. Lancet,
265, 758-759.
Amdur MO; McCarthy JF; Gill MW. (1983). Effect of mixing conditions on irritant potency of zinc oxide
and sulfur dioxide. Am Ind Hyg Assoc J, 44, 7-13.
Amdur MO; Chen LC; Guty J; Lam HF; Miller PD. (1988). Speciation and pulmonary effects of acidic
SOx formed on the surface of ultrafine zinc oxide aerosols. Atmos Environ, 22, 557-560.
American Academy of Pediatrics Committee on Environmental Health. (2004). Ambient air pollution:
health hazards to children. Pediatrics, 114, 1699-1707.
Andersen IB; Lundqvist GR; Jensen PL; Proctor DF. (1974). Human response to controlled levels of
sulfur dioxide. Arch Environ Health, 28, 31-39.
Anderson HR; Ponce de Leon A; Bland JM; Bower JS; Strachan DP. (1996). Air pollution and daily
mortality in London: 1987-92. British Medical Journal, 312, 665-669.
Anderson HR; Spix C; Medina S; Schouten JP; Castellsague J; Rossi G; Zmirou D; Touloumi G;
Wojtyniak B; Ponka A; Bacharova L; Schwartz J; Katsouyanni K. (1997). Air pollution and daily
admissions for chronic obstructive pulmonary disease in 6 European cities: results from the
APHEA project. EurRespir J, 10, 1064-1071.
Anderson HR; Ponce de Leon A; Bland JM; Bower JS; Emberlin J; Strachan DP. (1998). Air pollution,
pollens, and daily admissions for asthma in London 1987-92. Thorax, 53, 842-848.
Anderson HR; Bremner SA; Atkinson RW; Harrison RM; Walters S. (2001). Particulate matter and daily
mortality and hospital admissions in the west midlands conurbation of the United Kingdom:
associations with fine and coarse particles, black smoke and sulphate. Occup Environ Med, 58,
504-510.
Andreae MO; Atlas, E; Cachier, H; Cofer III, WR; Harris, GW; Helas, G; Koppmann R; Lacaux, P;
Ward, DE. (1995).Trace gas and aerosol emissions from savanna fires. In (Levine, JS ed.)
Biomass Burning and Global Change. Cambridge, MA; MIT Press
Appel KW; Gilliland A; Eder B. (2005). An operational evaluation of the 2005 release of models-3
CMAQ version 4.5. Washington, DC; National Oceanic and Atmospheric Administration
(NOAA) Air Resources Laboratory, Atmospheric Sciences Modeling Division.
Aranyi C; Vana SC; Thomas PT; Bradof JN; Fenters JD; Graham JA; Miller FJ. (1983). Effects of
subchronic exposure to a mixture of 03, S02, and (NH4)2S04 on host defenses of mice. J Toxicol
Environ Health, 12, 55-71.
Arnold JR; Dennis RL; Tonnesen GS. (2003). Diagnostic evaluation of numerical air quality models with
specialized ambient observations: Testing the Community Multiscale Air Quality modeling
system (CMAQ) at selected SOS 95 ground sites. Atmos Environ, 37, 1185-1198.
Atkinson RW; Anderson HR; Strachan DP; Bland JM; Bremner SA; Ponce de Leon A. (1999a). Short-
term associations between outdoor air pollution and visits to accident and emergency departments
in London for respiratory complaints. Eur Respir J, 13, 257-265.
R-2
-------
Atkinson RW; Bremner SA; Anderson HR; Strachan DP; Bland JM; de Leon AP. (1999b). Short-term
associations between emergency hospital admissions for respiratory and cardiovascular disease
and outdoor air pollution in London. Arch Environ Health, 54, 398-411.
Atkinson RW; Anderson HR; Sunyer J; Ayres J; Baccini M; Vonk JM; Boumghar A; Forastiere F;
Forsberg B; Touloumi G; Schwartz J; Katsouyanni K. (2001). Acute effects of particulate air
pollution on respiratory admissions: results from APHEA 2 project. Am J Respir Crit Care Med,
164, 1860-1866.
ATS (American Thoracic Society). (1985). Guidelines as to what constitutes an adverse respiratory health
effect, with special reference to epidemiologic studies of air pollution. Am Rev Respir Dis, 131,
666-668.
ATS (American Thoracic Society) (1996). Health effects of outdoor air pollution. Committee of the
Environmental and Occupational Health Assembly of the American Thoracic Society.. Am J
Respir Crit Care Med, 153, 3-50
ATS (American Thoracic Society). (2000). What constitutes an adverse health effect of air pollution? Am
J Respir Crit Care Med, 161, 665-673.
Atzori L; Bannenberg G; Corriga A; Lou Y; Lundberg J; Ryrfeldt A; Moldeus P. (1992). Sulfur dioxide-
induced bronchoconstriction via ruthenium red-sensitive activation of sensory nerves.
Respiration, 59, 272-278.
Azoulay-Dupuis E; Bouley G; Blayo MC. (1982). Effects of sulfur dioxide on resistance to bacterial
infection in mice. Environ Res, 29, 312-319.
Bai J; Meng Z. (2005a). Effects of sulfur dioxide on apoptosis-related gene expressions in lungs from
rats. Regul Toxicol Pharmacol, 43, 272-279.
Bai J; Meng Z. (2005b). Expression of apoptosis-related genes in livers from rats exposed to sulfur
dioxide. Toxicology, 216, 253-260.
Ballester F; Tenias JM; Perez-Hoyos S. (2001). Air pollution and emergency hospital admissions for
cardiovascular diseases in Valencia, Spain. J Epidemiol Community Health, 55, 57-65.
Ballester F; Saez M; Perez-Hoyos S; Iniguez C; Gandarillas A; Tobias A; Bellido J; Taracido M; Arribas
F; Daponte A; Alonso E; Canada A; Guillen-Grima F; Cirera L; Perez-Boillos MJ; Saurina C;
Gomez F; Tenias JM. (2002). The EMECAM project: a multicentre study on air pollution and
mortality in Spain: combined results for particulates and for sulfur dioxide. Occup Environ Med,
59, 300-308.
Ballester F; Rodriguez P; Iniguez C; Saez M; Daponte A; Galan I; Taracido M; Arribas F; Bellido J;
Cirarda FB; Canada A; Guillen JJ; Guillen-Grima F; Lopez E; Perez-Hoyos S; Lertxundi A; Toro
S. (2006). Air pollution and cardiovascular admisisons association in Spain: results within the
EMECAS project. J Epidemiol Community Health, 60, 328-336.
Balmes JR; Fine JM; Sheppard D. (1987). Symptomatic bronchoconstriction after short-term inhalation of
sulfur dioxide. Am Rev Respir Dis, 136, 1117-1121.
R-3
-------
Barnes I; Bastian V; Becker KH; Overath RD. (1991). Kinetic studies of the reactions of 10, BrO and
CIO with Dimethylsulfide. Int J Chem Kinet, 23, 579-591.
Barnett AG; Williams GM; Schwartz J; Neller AH; Best TL; Petroeschevsky AL; Simpson RW. (2005).
Air pollution and child respiratory health: a case-crossover study in Australia and New Zealand.
Am J Respir Crit Care Med, 171, 1272-1278.
Barthelemy P; Badier M; Jammes Y. (1988). Interaction between S02 and cold-induced bronchospasm in
anesthetized rabbits. Respir Physiol, 71, 1-10.
Baskurt OK. (1988). Acute hematologic and hemorheologic effects of sulfur dioxide inhalation. Arch
Environ Health, 43, 344-348.
Bates DV; Baker-Anderson M; Sizto R. (1990). Asthma attack periodicity: a study of hospital emergency
visits in Vancouver. Environ Res, 51, 51-70.
Bateson TF; Schwartz J. (2004). Who is sensitive to the effects of particulate air pollution on mortality? A
case-crossover analysis of effect modifiers. Epidemiology, 15, 143-149.
Bedi JF; Folinsbee LJ; Horvath SM; Ebenstein RS. (1979). Human exposure to sulfur dioxide and ozone:
absence of a synergistic effect. Arch Environ Health, 34, 233-239.
Bedi JF; Folinsbee LJ; Horvath SM. (1984). Pulmonary function effects of 1.0 and 2.0 ppm sulfur dioxide
exposure in active young male non-smokers. J Air Pollut Control Assoc, 34, 1117-1121.
Beelen R; Hoek G; van den Brandt PA; Goldbohm RA; Fischer P; Schouten LJ; Jerrett M; Hughes E;
Armstrong B; Brunekreef B. (2008). Long-term effects of traffic-related air pollution on mortality
in a Dutch cohort (NLCS-AIR study). Environ Health Perspect, 116, 196-202.
Beeson WL; Abbey DE; Knutsen SF. (1998). Long-term concentrations of ambient air pollutants and
incident lung cancer in California adults: results from the AHSMOG study.Adventist Health
Study on Smog. Environ Health Perspect, 106, 813-823.
Bell ML; Ebisu K; Belanger K. (2007). Ambient air pollution and low birth weight in Connecticut and
Massachusetts. Environ Health Perspect, 115, 1118-1124.
Ben-Jebria A; Full AP; DeMaria DD; Ball BA; Ultman JS. (1990). Dynamics of sulfur dioxide absorption
in excised porcine tracheae. Environ Res 53: 119-134.
Benner RL; Stedman DH. (1989). Universal sulfur detection by chemiluminescence. Anal Chem, 61,
1268-1271.
Benner RL; Stedman DH. (1990). Field evaluation of the sulfur chemiluminescence detector. Environ Sci
Tech, 24, 1592-1596.
Berger A; Zareba W; Schneider A; Rucker R; Ibald-Mulli A; Cyrys J; Wichmann H-E; Peters, A. (2006)
Runs of ventricular and supraventricular tachycardia triggered by air pollution in patients with
coronary heart disease. J Occup Environ Med 48:1149-1158.
Berkowitz GS; Papiernik E. (1993). Epidemiology of preterm birth. Epidemiol Rev, 15, 414-443.
R-4
-------
Berresheim H; Eisele FL; Tanner DJ; Mclnnes LM; Ramsey-Bell DC; Covert DS. (1993). Atmospheric
sulfur chemistry and cloud condensation nuclei (CCN) concentrations over the northeastern
pacific coast. J Geophys Res 98, 12701-12711.
Berresheim H; Wine PH; Davis DD. (1995). Sulfur in the atmosphere. In Composition, chemistry, and
climate of the atmosphere. Singh, HB (ed.) New York, New York; Van Nostrand Reinhold.
RA; Epstein J; Sheppard D; Nadel JA; Boushey HA. (1983). Sulfur dioxide-induced
bronchoconstriction in freely breathing, exercising, asthmatic subjects. Am Rev Respir Dis, 128,
987-990.
RA; Sheppard D; Epstein J; Tam E; Nadel JA; Boushey HA. (1984). Interaction of sulfur dioxide
and dry cold air in causing bronchoconstriction in asthmatic subjects. J Appl Physiol, 57, 419-
423.
RA; Sheppard D; Geffroy B; Tam E; Nadel JA; Boushey HA. (1985). Effect of 0.25 ppm sulfur
dioxide on airway resistance in freely breathing, heavily exercising, asthmatic subjects. Am Rev
Respir Dis, 131, 659-661.
Biersteker K; de Graaf H; Nass CAG. (1965). Indoor air pollution in Rotterdam homes. Int J Air Water
Pollut, 9, 343-350.
Bigby B; Boushey H. (1993). Effects of nedocromil sodium on the bronchomotor response to sulfur
dioxide in asthmatic patients. J Allergy Clin Immunol, 92, 195-197.
Biggeri A; Baccini M; Bellini P; Terracini B. (2005). Meta-analysis of the Italian studies of short-term
effects of air pollution (MISA), 1990-1999. Int J Occup Environ Health, 11, 107-122.
Biller WF; Feagans TB; Johnson TR; Duggan GM; Paul RA; McCurdy T; Thomas HC. (1981). A general
model for estimating exposure associated with alternative NAAQS. Paper No. 81-18.4 in
Proceedings of the 74th annual meeting of the Air Pollution Control Association, June 21-26,
Philadelphia, PA.
Binkowski FS; Arunachalam S; Adelman Z; Pinto JP. (2007). Examining photolysis rates with a
prototype online photolysis module in CMAQ. J. Appl. Meteorol. Climatol. 46: 1252-1256.
Blanchard P; Brook JR; Brazil P. (2002). Chemical characterization of the organic fraction of atmospheric
aerosol at two sites in Ontario, Canada. J Geophys Res, 107.
Blanquart C; Giuliani I; Houcine O; Jeulin C; Guennou C; Marano F. (1995). In vitro exposure of rabbit
tracheal epithelium to S02: effects on morphology and ciliary beating. Toxicol Vitro, 9, 123-132.
Bobak M. (2000). Outdoor air pollution, low birth weight, and prematurity. Environ Health Perspect, 108,
173-176.
Boezen HM; van der Zee SC; Postma DS; Vonk JM; Gerritsen J; Hoek G; Brunekreef B; Rijcken B;
Schouten JP. (1999). Effects of ambient air pollution on upper and lower respiratory symptoms
and peak expiratory flow in children. Lancet, 353, 859-860.
Boezen HM; Vonk JM; van der Zee SC; Gerritsen J; Hoek G; Brunekreef B; Schouten JP; Postma DS.
(2005). Susceptibility to air pollution in elderly males and females. Eur Respir J, 25, 1018-1024.
R-5
Bethel
Bethel
Bethel
-------
Boezen M; Schouten J; Rijcken B; Vonk J; Gerritsen J; van der Zee S; Hoek G; Brunekreef B; Postma D.
(1998). Peak expiratory flow variability, bronchial responsiveness, and susceptibility to ambient
air pollution in adults. Am J Respir Crit Care Med, 158, 1848-1854.
Borja-Aburto VH; Loomis DP; Bangdiwala SI; Shy CM; Rascon-Pacheco RA. (1997). Ozone, suspended
particulates, and daily mortality in Mexico City. Am J Epidemiol, 145, 258-268.
Boija-Aburto VH; Castillejos M; Gold DR; Bierzwinski S; Loomis D. (1998). Mortality and ambient fine
particles in southwest Mexico City, 1993-1995. Environ Health Perspect, 106, 849-855.
Borsboom GJ; van Pelt W; van Houwelingen HC; van Vianen BG; Schouten JP; Quanjer PH. (1999).
Diurnal variation in lung function in subgroups from two Dutch populations: consequences for
longitudinal analysis. Am J Respir Crit Care Med, 159, 1163-1171.
Boutin-Forzano S; Adel N; Gratecos L; Jullian H; Gamier JM; Ramadour M; Lanteaume A; Hamon M;
Lafay V; Charpin D. (2004). Visits to the emergency room for asthma attacks and short-term
variations in air pollution. A case-crossover study. Respiration, 71, 134-137.
Braga AL; Saldiva PH; Pereira LA; Menezes JJ; Conccicao GM; Lin CA; Zanobetti A; Schwartz J;
Dockery DW. (2001). Health effects of air pollution exposure on children and adolescents in Sao
Paulo, Brazil. Pediatr Pulmonol, 31, 106-113.
Braga ALF; Conccicao GMS; Pereira LAA; Kishi HS; Pereira JCR; Andrade MF; Goncalvcs FLT;
Saldiva PHN; Latorre MRDO. (1999). Air pollution and pediatric respiratory hospital admissions
in Sao Paulo, Brazil. J Environ Med, 1, 95-102.
Brain JD. (1970). The uptake of inhaled gases by the nose. Ann Otol Rhinol Laryngol Med, 79, 529-539.
Brauer M; Brumm J; Vendal S; Petkau AJ. (2002a) Exposure misclassification and threshold
concentrations in time series analyses of air pollution health effects. Risk Anal. 22:1183-1193
Brauer M; Koutrakis P; Spengler JD. (1989). Personal exposures to acidic aerosols and gases. Environ Sci
Tech, 23, 1408-1412.
Brauer M; Koutrakis P; Keeler GJ; Spengler JD. (1991). Indoor and outdoor concentrations of inorganic
acidic aerosols and gases. J Air Waste Manage Assoc, 41, 171-181.
Brauer M; Hirtle RD; Hall AC; Yip TR. (1999). Monitoring personal fine particle exposure with a particle
counter. J Expo Anal Environ Epidemiol, 9, 228-236.
Brauer M; Henderson S; Kirkham T; Lee KS; Rich K; Teschke K. (2002b). Review of the health risks
associated with nitrogen dioxide and sulfur dioxide in indoor air. Report to Health Canada.
Vancouver, British Columbia, Canada: University of British Columbia, School of Occupational
and Environmental Hygiene.
Braun-Fahrlander C; Vuille JC; Sennhauser FH; Neu U; Kunzle T; Grize L; Gassner M; Minder C;
Schindler C; Varonier HS; Wuthrich B; team. S. (1997). Respiratory health and long-term
exposure to air pollutants in Swiss schoolchildren. Am J Respir Crit Care Med, 155, 1042-1049.
R-6
-------
Bremner SA; Anderson HR; Atkinson RW; McMichael AJ; Strachan DP; Bland JM; Bower JS. (1999).
Short-term associations between outdoor air pollution and mortality in London 1992-4. Occup
Environ Med, 56, 237-244.
Brook JR; Sirois A; Clarke JF. (1996). Comparison of dry deposition velocities for S02, HN03, and S042
estimated with two inferential models. Water Air Soil Pollut, 87, 205-218.
Brook RD; Franklin B; Cascio W; Hong Y; Howard G; Lipsett M; Luepker R; Mittleman M; Samet J;
Smith SC, Jr.; Tager I. (2004). Air pollution and cardiovascular disease. A statement for
healthcare professionals from the Expert Panel on Population and Prevention Science of the
American Heart Association. Circulation, 109, 2655-2671.
Buchdahl R; Parker A; Stebbings T; Babiker A. (1996). Association between air pollution and acute
childhood wheezy episodes: prospective observational study. BMJ, 312, 661-665.
Buescher PA; Taylor KP; Davis MH; Bowling JM. (1993). The quality of the new birth certificate data: a
validation study in North Carolina. Am J Public Health, 83, 1163-1165.
Burke JM; Zufall MJ; Ozkaynak H. (2001). A population exposure model for particulate matter: case
study results for PM2.5 in Philadelphia, PA. J Expo Sci Environ Epidemiol, 11, 470-489.
Burnett RT; Brook JR; Yung WT; Dales RE; Krewski D. (1997a). Association between Ozone and
Hospitalization for Respiratory Diseases in 16 Canadian Cities. Environ Res, 72, 24-31.
Burnett RT; Cakmak S; Brook JR; Krewski D. (1997b). The role of particulate size and chemistry in the
association between summertime ambient air pollution and hospitalization for cardiorespiratory
diseases. Environ Health Perspect, 105, 614-620.
Burnett RT; Dales RE; Brook JR; Raizenne ME; Krewski D. (1997c). Association between ambient
carbon monoxide levels and hospitalizations for congestive heart failure in the elderly in 10
Canadian cities. Epidemiology, 8, 162-167.
Burnett RT; Cakmak S; Brook JR. (1998a). The effect of the urban ambient air pollution mix on daily
mortality rates in 11 Canadian cities. Can J Public Health, 89, 152-156.
Burnett RT; Cakmak S; Raizenne ME; Stieb D; Vincent R; Krewski D; Brook JR; Philips O; Ozkaynak
H. (1998b). The association between ambient carbon monoxide levels and daily mortality in
Toronto, Canada. J Air Waste Manag Assoc, 48, 689-700.
Burnett RT; Smith-Doiron M; Stieb D; Cakmak S; Brook JR. (1999). Effects of particulate and gaseous
air pollution on cardiorespiratory hospitalizations. Arch Environ Health, 54, 130-139.
Burnett RT; Brook J; Dann T; Delocla C; Philips O; Cakmak S; Vincent R; Goldberg MS; Krewski D.
(2000). Association between particulate- and gas-phase components of urban air pollution and
daily mortality in eight Canadian cities. Inhal Toxicol, 12 Suppl 4, 15-39.
Burnett RT; Smith-Doiron M; Stieb D; Raizenne ME; Brook JR; Dales RE; Leech JA; Cakmak S;
Krewski D. (2001). Association between ozone and hospitalization for acute respiratory diseases
in children less than 2 years of age. Am J Epidemiol, 153, 444-452.
R-7
-------
Burnett RT; Stieb D; Brook JR; Cakmak S; Dales R; Raizenne M; Vincent R; Dann T. (2004).
Associations between Short-Term Changes in Nitrogen Dioxide and Mortality in Canadian Cities.
Arch Environ Health, 59, 228-236.
Byun D; Schere KL. (2006). Review of the governing equations, computational algorithms, and other
components of the models-3 community multiscale air quality (CMAQ) modeling system. Appl
MechRev, 59, 51-77.
Byun DW; Ching JKS, eds. (1999). Science algorithms of the EPA Models-3 Community Multiscale Air
Quality model (CMAQ) modeling system. Research Triangle Park, NC; Atmospheric modeling
division, National Exposure Research Laboratory, US Environmental Protection Agency.
Cakmak S; Dales RE; Judek S. (2006). Do gender, education, and income modify the effect of air
pollution gases on cardiac disease? J Occup Environ Med, 48, 89-94.
Cakmak S; Dales RE; Judek S. (2007a). Respiratory health effects of air pollution gases: Modification by
education and income. Arch Environ Occup Health 61,5-10.
Cakmak S; Dales RE; Vidal CB. (2007b). Air pollution and mortality in Chile: susceptibility among the
elderly. Environ Health Perspect, 115, 524-527.
Calvert JG; Su F; Bottenheim JW; Strausz OP. (1978). Mechanism of the homogeneous oxidation of
sulfur dioxide in the troposphere. Atmos Environ, 12, 197-226.
Camp DC; Stevens RK; Cobourn WG; Husar RB; Collins JF; Huntzicker JJ; Husar JD; Jaklevic JM;
McKenzie RL; Tanner RL; Tesch JW. (1982). Intercomparison of concentration results from fine
particle sulfur monitors. Atmos Environ, 16, 911-916.
Carter WPL. (1990). A detailed mechanism for the gas-phase atmospheric reactions of organic
compounds. Atmos Envion Part A, 24, 481-518.
Castellsague J; Sunyer J; Saez M; Anto JM. (1995). Short-term association between air pollution and
emergency room visits for asthma in Barcelona. Thorax, 50, 1051-1056.
CDC. (2004). The health consequences of smoking: a report of the Surgeon General. Atlanta, GA: Office
on Smoking and Health; National Center for Chronic Disease Prevention and Health Promotion;
Centers for Disease Control and Prevention; U.S. Department of Health and Human Services.
http://www.cdc.gov/tobacco/data_statistics/sgr/sgr_2004/chapters.htm
Chameides WL; Stelson AW. (1992). Aqueous-phase chemical processes in deliquescent sea-salt
aerosols: a mechanism that couples the atmospheric cycles of S and sea salt. J Geophys Res 97,
20,565-20,580.
Chan C-C; Hung H-F; Fu L-F. (1994). The indoor/outdoor relationship of acid aerosols in Taipei. Sci
Total Environ 153, 267-273
Chan CC; Chuang KJ; Su TC; Lin LY. (2005). Association between nitrogen dioxide and heart rate
variability in a susceptible population. Eur J Cardiovasc Prev Rehabil, 12, 580-586.
Chan CC; Chuang KJ; Chien LC; Chen WJ; Chang WT. (2006). Urban air pollution and emergency
admissions for cerebrovascular diseases in Taipei, Taiwan. Eur Heart J, 27, 1238-1244.
R-8
-------
Chang C-H; Meroney RN. (2003). Concentration and flow distributions in urban street canyons: wind
tunnel and computational data. Journal Wind Eng Ind Aerodyn, 91, 1141-1154.
Chang C-H. (2006). Computational fluid dynamics simulation of concentration distributions from a point
source in the urban street canyons. J Aerosp Eng, 19, 80-86.
Chang CC; Tsai SS; Ho SC; Yang CY. (2005). Air pollution and hospital admissions for cardiovascular
disease in Taipei, Taiwan. Environ Res, 98, 114-119.
Chang LT; Koutrakis P; Catalano PJ; Suh HH. (2000). Hourly personal exposures to fine particles and
gaseous pollutants—results from Baltimore, Maryland. J Air Waste Manag Assoc, 50, 1223-1235.
Chao CY. (2001). Comparison between indoor and outdoor air contaminant levels in residential buildings
from passive sampler study. Build Environ, 36, 999-1007.
Charpin D; Pascal L; Birnbaum J; Armengaud A; Sambuc R; Lanteaume A; Vervloet D. (1999). Gaseous
air pollution and atopy. Clin Exp Allergy, 29, 1474-1480.
Chatfield RB; Crutzen PJ. (1984). Sulfur dioxide in remote oceanic air: cloud transport of reactive
precursors. J Geophys Res, 89, 7111-7132.
Chen LC; Peoples SM; Amdur MO. (1991). Pulmonary effects of sulfur oxides on the surface of copper
oxide aerosol. Am Ind Hyg Associ J, 52, 187-191.
Chen LC; Miller PD; Amdur MO; Gordon T. (1992). Airway hyperresponsiveness in guinea pigs exposed
to acid-coated ultraflne particles. J Toxicol Environ Health, 35, 165-174.
Chen PC; Lai YM; Chan CC; Hwang JS; Yang CY; Wang JD. (1999). Short-term effect of ozone on the
pulmonary function of children in primary school. Environ Health Perspect, 107, 921-925.
Chew FT; Goh DYT; Ooi BC; Saharom R; Hui JKS; Lee BW. (1999). Association of ambient air-
pollution levels with acute asthma exacerbation among children in Singapore. Allergy, 54, 320-
329.
Ching J; Herwehe J; Swall J. (2006). On joint deterministic grid modeling and sub-grid variability
conceptual framework for model evaluation. Atmos Environ, 40, 4935-4945.
Chock DP; Winkler SL. (1994). A comparison of advection algorithms coupled with chemistry. Atmos
Environ, 28, 2659-2675.
Chock DP; Winkler SL; Chen C. (2000). A study of the association between daily mortality and ambient
air pollutant concentrations in Pittsburgh, Pennsylvania. J Air Waste Manag Assoc, 50, 1481-
1500.
Christakos G; Hristopulos DT. (1998). Spatiotemporal environmental health modelling. A tractatus
stochasticus. Boston, MA; Kluwer Academic Publishers.
Christakos G; Kolovos A. (1999). A study of the spatiotemporal health impacts of ozone exposure. J
Expo Sci Environ Epidemiol, 9, 322-335.
Christakos G. (2000). Modern spatiotemporal geostatistics. New York, New York; Oxford University
Press.
R-9
-------
Cifuentes LA; Vega J; Kopfer K; Lave LB. (2000). Effect of the fine fraction of particulate matter versus
the coarse mass and other pollutants on daily mortality in Santiago, Chile. J Air Waste Manag
Assoc, 50, 1287-1298.
Clancy L; Goodman P; Sinclair H; Dockery DW. (2002). Effect of air pollution control on death rates in
Dublin, Ireland: an intervention study. Lancet, 360, 1210-1214.
Clarke RW; Antonini JM; Hemenway DR; Frank R; Kleeberger SR; Jakab GJ. (2000). Inhaled particle-
bound sulfate: effects on pulmonary inflammatory responses and alveolar macrophage function.
Inhal Toxicol, 12, 169-186.
Clegg NA; Toumi R. (1998). Non-sea-salt-sulphate formation in sea-salt aerosol. J Geophys Res, 103,
31,095-31,102.
Clench-Aas J; Bartonova A; Bohler T; Gronskei KE; Sivertsen B; Larssen S. (1999). Air pollution
exposure monitoring and estimating. Part I. Integrated air quality monitoring system. J Environ
Monit, 1, 313-319.
Clewell HJ; Crump KS. (2005). Quantitative Estimates of Risk for Noncancer Endpoints. Risk Anal, 25,
285-289.
Cobourn WG; Husar RB; Husar JD. (1978). Continuous in situ monitoring of ambient particulate sulfur
using flame photometry and thermal analysis. Atmos Environ, 12, 89-98.
Cobourn WG; Husar RB. (1982). Diurnal and seasonal patterns of particulate sulfur and sulfuric acid in
St. Louis, July 1977-June 1978. Atmos Environ, 16, 1441-1450.
Coleridge HM; Coleridge JCG. (1994). Pulmonary Reflexes: Neural Mechanisms of Pulmonary Defense.
Annu Rev Physiol, 56, 69-91.
Conceicao GM; Miraglia SG; Kishi HS; Saldiva PH; Singer JM. (2001). Air pollution and child mortality:
a time-series study in Sao Paulo, Brazil. Environ Health Perspect, 109 Suppl 3, 347-350.
Conner MW; Lam HF; Rogers AE; Fitzgerald S; Amdur MO. (1985). Lung injury in guinea pigs caused
by multiple exposures to submicron zinc oxide mixed with sulfur dioxide in a humidified furnace.
J Toxicol Environ Health, 16, 101-114.
Conner MW; Flood WH; Rogers AE; Amdur MO. (1989). Changes in pulmonary lavage fluid of guinea
pigs exposed to ultrafine zinc oxide with adsorbed sulfuric acid. J Toxicol Environ Health, 26,
223-234.
CFR. (2006) Title 40, Code of Federal Regulations, Part 58, Ambient Air Quality Surveillance. CFR40, Part 58.
Crump KS; Hoel DG; Langley CH; Peto R. (1976). Fundamental carcinogenic processes and their
implications for low dose risk assessment. Cancer Res, 36, 2973-2979.
Crutzen PJ. (1976).The possible importance of OCS for the sulphate layer of the stratosphere. Geophys
Res Lett, 3, 73-76.
R-10
-------
Crutzen PJ; Gidel LT. (1983). A two-dimensional photochemical model of the atmosphere. 2. The
tropospheric budgets of the anthropogenic chlorocarbons, CO, CH4, CH3C1 and the effect of
various NOx sources on tropospheric ozone. J Geophys Res, 88, 6641-6661.
Cuijpers CEJ; Swaen GMH; Wesseling G; Wouters EFM. (1994). Acute respiratory effects of summer
smog in primary school children. Toxicol Lett, 72, 227-235.
DC Cir (U.S. Court of Appeals for the District of Columbia) (1980). Lead Industries v. U.S.
Environmental Protection Agency. DC Cir. 1980 Report #647F2dl 130,1154; Washington, DC;
U.S. Court of Appeals for the District of Columbia
DC Cir (U.S. Court of Appeals for the District of Columbia) (1981). American Petroleum Institute v.
Costle. DC Cir. 1981 Report # 665F2dll76,1186; Washington, DC; U.S. Court of Appeals for
the District of Columbia
D'lppoliti D; Forastiere F; Ancona C; Agabiti N; Fusco D; Michelozzi P; Perucci CA. (2003). Air
pollution and myocardial infarction in Rome: a case-crossover analysis. Epidemiology, 14, 528-
535.
Dab W; Medina S; Quenel P; Le Moullec Y; Le Tertre A; Thelot B; Monteil C; Lameloise P; Pirard P;
Momas I; Ferry R; Festy B. (1996). Short term respiratory health effects of ambient air pollution:
results of the APHEA project in Paris. J Epidemiol Community Health, 50 Suppl 1, s42-46.
Dagnall RM; Thompson KC; West TS. (1967). Molecular-emission in cool flames. Part I. the behaviour
of sulphur species in a hydrogen-nitrogen diffusion flame and in a shielded air-hydrogen flame.
Analyst, 92, 506-512.
Dales R; Burnett RT; Smith-Doiron M; Stieb DM; Brook JR. (2004). Air pollution and sudden infant
death syndrome. Pediatrics, 113, e628-631.
Dales RE; Cakmak S; Doiron MS. (2006). Gaseous air pollutants and hospitalization for respiratory
disease in the neonatal period. Environ Health Perspect, 114, 1751-1754.
de Diego Damia A; Leon Fabregas M; Perpina Tordera M; Compte Torrero L. (1999). Effects of air
pollution and weather conditions on asthma exacerbation. Respiration, 66, 52-58.
De Leon SF; Thurston GD; Ito K. (2003). Contribution of respiratory disease to nonrespiratory mortality
associations with air pollution. Am J Respir Crit Care Med, 167, 1117-1123.
de Paula Santos U; Braga AL; Giorgi DM; Pereira LA; Grupi CJ; Lin CA; Bussacos MA; Zanetta DM; do
Nascimento Saldiva PH; Filho MT. (2005). Effects of air pollution on blood pressure and heart
rate variability: a panel study of vehicular traffic controllers in the city of Sao Paulo, Brazil. Eur
Heart J, 26, 193-200.
Delfino RJ; Gong H, Jr.; Linn WS; Pellizzari ED; Hu Y. (2003a). Asthma symptoms in Hispanic children
and daily ambient exposures to toxic and criteria air pollutants. Environ Health Perspect, 111,
647-656.
R-11
-------
Delfino RJ; Gong H; Linn WS; Hu Y; Pellizzari ED. (2003b). Respiratory symptoms and peak expiratory
flow in children with asthma in relation to volatile organic compounds in exhaled breath and
ambient air. J Expo Anal Environ Epidemiol, 13, 348-363.
Delmas R., P. Loudjani, A. Podaire, and J.C. Menaut. 1991. Biomass burning in Africa: an assessment of
annually burnt biomass. In Global biomass Burning, ed. J.S. Levine, pp. 126-133. Cambridge,
MA: MIT Press.
Demokritou P; Kavouras IG; Ferguson ST; Koutrakis P. (2001). Development and laboratory
performance evaluation of a multipollutant sampler for simultaneous measurements for
particulate and gaseous pollutants. Aerosol Sci Technol, 35, 741-752.
Dentener F; Stevenson D; Cofala J; Mechler R; Amann M; Bergamaschi P; Raes F; Derwent R. (2005).
The impact of air pollutant and methane emission controls on tropospheric ozone and radiative
forcing: CTM calculations forthe period 1990-2030. Atmos Chem Phys, 5, 1731-1755.
Dentener F; Drevet J; Lamarque JF; Bey I; Eickhout B; Fiore AM; Hauglustaine D; Horowitz LW; Krol
M; Kulshrestha UC; Lawrence M; Galy-Lacaux C; Rast S; Shindell D; Stevenson D; Van Noije
T; Atherton C; Bell N; Bergman D; Butler T; Cofala J; Collins B; Doherty R; Ellingsen K;
Galloway J; Gauss M; Montanaro V; Miiller JF; Pitari G; Rodriguez J; Sanderson M; Solman F;
Strahan S; Schultz M; Sudo K; Szopa S; Wild O. (2006a). Nitrogen and sulfur deposition on
regional and global scales: a multi-model evaluation. Global Biogeochem Cycles, 20, GB4003,
doi:4010.1029/2005GB002672.
Dentener F; Stevenson D; Ellingsen K; Van Noije T; Schultz M; Amann M; Atherton C; Bell N;
Bergmann D; Bey I; Bouwman L; Butler T; Cofala J; Collins B; Drevet J; Doherty R; Eickhout
B; Eskes H; Fiore A; Gauss M; Hauglustaine D; Horowitz L; Isaksen IS; Josse B; Lawrence M;
Krol M; Lamarque JF; Montanaro V; Muller JF; Peuch VH; Pitari G; Pyle J; Rast S; Rodriguez I;
Sanderson M; Savage NH; Shindell D; Strahan S; Szopa S; Sudo K; Van Dingenen R; Wild O;
Zeng G. (2006b). The global atmospheric environment for the next generation. Environ Sci
Technol, 40, 3586-3594.
Desqueyroux H; Pujet JC; Prosper M; Le Moullec Y; Momas I. (2002a). Effects of air pollution on adults
with chronic obstructive pulmonary disease. Arch Environ Health, 57, 554-560.
Desqueyroux H; Pujet JC; Prosper M; Squinazi F; Momas I. (2002b). Short-Term Effects of Low-Level
Air Pollution on Respiratory Health of Adults Suffering from Moderate to Severe Asthma.
Environ Res, 89, 29-37.
Devalia JL; Rusznak C; Herdman MJ; Trigg CJ; Tarraf H; Davies RJ. (1994). Effect of nitrogen dioxide
and sulphur dioxide on airway response of mild asthmatic patients to allergen inhalation. Lancet,
344, 1668-1671.
Diaz J; Garcia R; Ribera P; Alberdi JC; Hernandez E; Pajares MS; Otero A. (1999). Modeling of air
pollution and its relationship with mortality and morbidity in Madrid, Spain. Int Arch Occup
Environ Health, 72, 366-376.
Dickerson RR; Huffman GJ; Luke WT; Nunneracker LJ; Pickering KE; Leslie ACD; Lindsey CG; Slinn
WGN; Kelly TJ; et al. (1987). Thunderstorms: An important mechanism in the transport of air
pollutants. Science, 235, 460-464.
R-12
-------
Dietert RR; Etzel RA; Chen D; Halonen M; Holladay SD; Jarabek AM; Landreth K; Peden DB;
Pinkerton K; Smialowicz RJ; Zoetis T. (2000). Workshop to identify critical window of exposure
for children's health: immune and respiratory systems work group summary. Environ Health
Perspect, 108, 483-490.
Dockery DW; Ware JH; Ferris BG, Jr.; Speizer FE; Cook NR; Herman SM. (1982). Change in pulmonary
function in children associated with air pollution episodes. J Air Pollut Control Assoc, 32, 937-
942.
Dockery DW; Speizer FE; Stram DO; Ware JH; Spengler JD; Ferris BG, Jr. (1989). Effects of inhalable
particles on respiratory health of children. Am Rev Respir Dis, 139, 587-594.
Dockery DW; Schwartz J; Spengler JD. (1992). Air pollution and daily mortality: associations with
particulates and acid aerosols. Environ Res, 59, 362-373.
Dockery DW; Pope CA, 3rd; Xu X; Spengler JD; Ware JH; Fay ME; Ferris BG, Jr.; Speizer FE. (1993).
An association between air pollution and mortality in six U.S. cities. N Engl J Med, 329, 1753-
1759.
Dockery DW; Cunningham J; Damokosh AI; Neas LM; Spengler JD; Koutrakis P; Ware JH; Raizenne
M; Speizer FE. (1996). Health effects of acid aerosols on North American children: respiratory
symptoms. Environ Health Perspect, 104, 500-505.
Dockery DW; Luttmann-Gibson H; Rich DQ; Link MS; Mittleman MA; Gold DR; Koutrakis P; Schwartz
JD; Verrier RL. (2005). Association of air pollution with increased incidence of ventricular
tachyarrhythmias recorded by implanted cardioverter defibrillators. Environ Health Perspect, 113,
670-674.
Dodge R; Solomon P; Moyers J; Hayes C. (1985). A longitudinal study of children exposed to sulfur
oxides. Am J Epidemiol, 121, 720-736.
Dominici F; McDermott A; Daniels M; Zeger SL; Samet JM. (2003). Mortality among residents of 90
cities. In: Revised analyses of time-series studies of air pollution and health. Special report.
Boston, MA: Health Effects Institute; pp. 9-24. Available:
http://www.healtheffects.org/Pubs/TimeSeries.pdffl2 May, 2004],
Douglas GJ; Price JF; Page CP. (1994). A method for the long-term exposure of rabbits to environmental
pollutant gases. Eur Respir J, 7, 1516-1526.
Du Z; Meng Z. (2004a). Effects of derivatives of sulfur dioxide on transient outward potassium currents
in acutely isolated hippocampal neurons. Food Chem Toxicol, 42, 1211-1216.
Du Z; Meng Z. (2004b). Modulation of sodium currents in rat dorsal root ganglion neurons by sulfur
dioxide derivatives. Brain Res, 1010, 127-133.
Du Z; Meng Z. (2006). Sulfur dioxide derivatives modulation of high-threshold calcium currents in rat
dorsal root ganglion neurons. Neurosci Lett, 405, 147-152.
Dubnov J; Barchana M; Rishpon S; Leventhal A; Segal I; Carel R; Portnov BA. (2007). Estimating the
effect of air pollution from a coal-fired power station on the development of children's pulmonary
function. Environ Res, 103, 87-98.
R-13
-------
Dubowsky SD; Suh H; Schwartz J; Coull BA; Gold DR. (2006). Diabetes, obesity, and hypertension may
enhance associations between air pollution and markers of systemic inflammation. Environ
Health Perspect, 114, 992-998.
Dugandzic R; Dodds L; Stieb D; Smith-Doiron M. (2006). The association between low level exposures
to ambient air pollution and term low birth weight: a retrospective cohort study. Environ Health,
5,3.
Durham JL; Wilson WE; Bailey EB. (1978). Application of an S02-denuder for continuous measurement
of sulfur in submicrometric aerosols. Atmos Environ, 12, 883-886.
Eckert DJ; Catcheside PG; Smith JH; Frith PA; McEvoy RD. (2004). Hypoxia suppresses symptom
perception in asthma. Am J Respir Crit Care Med, 169, 1224-1230.
Eder B; Yu S. (2006). A performance evaluation of the 2004 release of Models-3 CMAQ. Atmos
Environ, 40, 4811-4824.
Egerton-Warburton L; Allen EB. (2000). Shifts in Arbuscular mycorrhizal communities along an
anthropogenic nitrogen deposition gradient. Ecol Appl, 10, 484-496.
Eggleston HS. (1993). Uncertainties in the estimates of emissions of VOCs from motor cars. Proceedings
of the TNO/EURASAP workshop on the reliability of VOC emission databases, Delft, The
Netherlands.
Elliott P; Shaddick G; Wakefield JC; de Hoogh C; Briggs DJ. (2007). Long-term associations of outdoor
air pollution with mortality in Great Britain. Thorax, 62, 1088-1094.
Etlik O; Tomur A; Kutman MN; Yorukan S; Duman O. (1995). The effects of sulfur dioxide inhalation
and antioxidant vitamins on red blood cell lipoperoxidation. Environ Res, 71, 25-28.
Etlik O; Tomur A; Tuncer M; Ridvanagaoglu AY; Andac O. (1997). Protective effect of antioxidant
vitamins on red blood cell lipoperoxidation induced by S02 inhalation. J Basic Clin Physiol
Pharmacol, 8, 31-43.
Euler GL; Abbey DE; Magie AR; Hodgkin JE. (1987). Chronic obstructive pulmonary disease symptom
effects of long-term cumulative exposure to ambient levels of total suspended particulates and
sulfur dioxide in California Seventh-Day Adventist residents. Arch Environ Health, 42, 213-222.
Fairley D. (1990). The relationship of daily mortality to suspended particulates in Santa Clara County,
1980-1986. Environ Health Perspect, 89, 159-168.
Faraji M; Kimura Y; Allen D. (2005). Comparison of the carbon bond and SAPRC photochemical
mechanisms under conditions relevent to southeast Texas. Project report H12.8HRB. Houston,
TX: Texas Environ Res Consortium (TERC) Available:
http://files.harc.edu/Projects/AirQuality/Projects/H012.2004/8HRB/H128HRBMechanismReport.
pdf [4 October, 2007],
Faraji M; Kimura Y; Allen D. (2008) Comparison of the carbon bond and SAPRC photochemical
mechanisms under conditions relevant to southeast Texas. Atmos Environ 42, 5821-5836
R-14
-------
Farhat SC; Paulo RL; Shimoda TM; Conceicao GM; Lin CA; Braga AL; Warth MP; Saldiva PH. (2005).
Effect of air pollution on pediatric respiratory emergency room visits and hospital admissions.
Braz J Med Biol Res, 38, 227-235.
Farwell SO; Rasmussen RA. (1976). Limitations of the FPD and ECD in atmospheric analysis: a review.
J Chromatogr Sci, 14, 224-234.
Fehsenfeld FC; Ancellet G; Bates TS; Goldstein AH; Hardesty RM; Honrath R; Law KS; Lewis AC;
Leaitch R; McKeen S; Meagher J; Parrish DD; Pszenny AAP; Russell PB; Schlager H; Seinfeld
J; Talbot R; Zbinden R. (2006). International Consortium for Atmospheric Research on Transport
and Transformation (ICARTT): North America to Europe -Overview of the 2004 summer field
study. J Geophys Res (D Atmos), 111, [np].
Field PI; Simmul R; Bell SC; Allen DH; Berend N. (1996). Evidence for opioid modulation and
generation of prostaglandins in sulphur dioxide (SO)2-induced bronchoconstriction. Thorax, 51,
159-163.
Filleul L; Rondeau V; Vandentorren S; Le Moual N; Cantagrel A; Annesi-Maesano I; Charpin D;
Declercq C; Neukirch F; Paris C; Vervloet D; Brochard P; Tessier JF; Kauffmann F; Baldi I.
(2005). Twenty five year mortality and air pollution: results from the French PAARC survey.
Occup Environ Med, 62, 453-460.
Finkelstein MM; Jerrett M; DeLuca P; Finkelstein N; Verma DK; Chapman K; Sears MR. (2003).
Relation between income, air pollution and mortality: a cohort study. CMAJ, 169, 397-402.
Finkelstein MM; Jerrett M; Sears MR. (2005). Environmental inequality and circulatory disease mortality
gradients. J Epidemiol Community Health, 59, 481-487.
Finlayson-Pitts BJ; Pitts JN, Jr. (1999). Chemistry of the upper and lower atmosphere: theory,
experiments and applications. San Diego, CA: Academic Press.
Fiore M; Petruzzi S; DeH'Omo G; Alleva E. (1998). Prenatal sulfur dioxide exposure induces changes in
the behavior of adult male mice during agonistic encounters. Neurotoxicol Teratol, 20, 543-548.
Fischer P; Hoek G; Brunekreef B; Verhoeff A; van Wijnen J. (2003). Air pollution and mortality in the
Netherlands: are the elderly more at risk? Eur Respir J, 21, 34S-38S.
Foley GJ; Georgopoulos PG; Lioy PJ. (2003). Decision points for consulting APEX for air toxics
exposure assessments. (Report#. 37, 392A-399A). Research Triangle Park, NC: Office of Air
Quality Planning and Standards; U.S. Environmental Protection Agency; Available:
http://www.epa.gov/ttnmainl/fera/data/apex/APEXDecisionPoints_121205.pdffl9 September,
2007],
Folinsbee LJ; Bedi JF; Horvath SM. (1985). Pulmonary response to threshold levels of sulfur dioxide (1.0
ppm) and ozone (0.3 ppm). J Appl Physiol, 58, 1783-1787.
Forsberg B; Sternberg N; Falk M; Lundback B; Wall S. (1993). Air pollution levels, meteorological
conditions and asthma symptoms. Eur Respir J, 6, 1109-1115.
R-15
-------
Fox-Rabinovitz MS; Takacs LL; Govindaraju RC. (2002). A variable-resolution stretched-grid general
circulation model and data assimilation system with multiple areas of interest: studying the
anomalous regional climate events of 1998. J Geophys Res, 107D24.
Frank NR; Amdur MO; Worcester J; Whittenberger JL. (1962). Effects of acute controlled exposure to
S02 on respiratory mechanics in healthy male adults. J Appl Physiol, 17, 252-258.
Frank NR; Yoder RE; Brain JD; Yokoyama E. (1969). S02 (35S labeled) absorption by the nose and
mouth under conditions of varying concentration and flow. Arch Environ Health, 18, 315-322.
Freijer JI; Bloemen HJT; de Loos S; Marra M; Rombout PJA; Steentjes GM; Van Veen MP. (1998).
Modelling exposure of the Dutch population to air pollution. Journal of Hazard Mater, 61, 107-
114.
Frischer T; Studnicka M; Gartner C; Tauber E; Horak F; Veiter A; Spengler J; Kuhr J; Urbanek R.
(1999). Lung function growth and ambient ozone: a three-year population study in school
children. Am J Respir Crit Care Med, 160, 390-396.
Frischer T; Studnicka M; Halmerbauer G; Horak F; Gartner C; Tauber E; Koller DY. (2001). Ambient
ozone exposure is associated with eosinophil activation in healthy children. Clin Exp Allergy, 31,
1213-1219.
Fritz GK; Adams SK; McQuaid EL; Klein R; Kopel S; Nassau J; Mansell A. (2007). Symptom
Perception in Pediatric Asthma: Resistive Loading and In Vivo Assessment Compared. Chest,
132, 884.
Frye C; Hoelscher B; Cyrys J; Wjst M; Wichmann HE; Heinrich J. (2003). Association of lung function
with declining ambient air pollution. Environ Health Perspect, 111, 383-387.
Fuentes M; Raftery AE. (2005). Model evaluation and spatial interpolation by Bayesian combination of
observations with outputs from numerical models. Biometrics, 61, 36-45.
Fung KY; Luginaah I; Gorey KM; Webster G. (2005a). Air pollution and daily hospital admissions for
cardiovascular diseases in Windsor, Ontario. Can J Public Health, 96, 29-33.
Fung KY; Luginaah I; Gorey KM; Webster G. (2005b). Air pollution and daily hospitalization rates for
cardiovascular and respiratory diseases in London, Ontario. Int J Environ Stud, 62, 677-685.
Fung KY; Khan S; Krewski D; Chen Y. (2006). Association between air pollution and multiple
respiratory hospitalizations among the elderly in Vancouver, Canada. Inhal Toxicol, 18, 1005-
1011.
Fusco D; Forastiere F; Michelozzi P; Spadea T; Ostro B; Area M; Perucci CA. (2001). Air pollution and
hospital admissions for respiratory conditions in Rome, Italy. Eur Respir J, 17, 1143-1150.
Galan I; Tobias A; Banegas JR; Aranguez E. (2003). Short-term effects of air pollution on daily asthma
emergency room admissions. Eur Respir J, 22, 802-808.
R-16
-------
Gamble JL. (1998). Effects of ambient air pollution on daily mortality: a time series analysis of Dallas,
Texas, 1990-1994., Paper no. 98-MP26.03. Presented at: 91st annual meeting and exhibition of
the Air & Waste Management Association; San Diego, CA. Pittsburgh, PA: Air & Waste
Management Association.
Garcia-Aymerich J; Tobias A; Anto JM; Sunyer J. (2000). Air pollution and mortality in a cohort of
patients with chronic obstructive pulmonary disease: a time series analysis. J Epidemiol
Community Health, 54, 73-74.
Garcia-Marcos L; Guillen JJ; Dinwiddie R; Guillen A; Barbero P. (1999). The relative importance of
socio-economic status, parental smoking and air pollution (S02) on asthma symptoms, spirometry
and bronchodilator response in 11-year-old children. Pediatr Allergy Immunol, 10, 96-100.
Garty BZ; Kosman E; Ganor E; Berger V; Garty L; Wietzen T; Waisman Y; Mimouni M; Waisel Y.
(1998). Emergency room visits of asthmatic children, relation to air pollution, weather, and
airborne allergens. Ann Allergy Asthma Immunol, 81, 563-570.
Gaylor DW; Sheehan DM; Young JF; Mattison DR. (1988). The threshold dose question in teratogenesis.
Teratology, 38, 389-391.
Gee GC; Payne-Sturges DC. (2004). Environmental health disparities: a framework integrating
psychosocial and environmental concepts. Environ Health Perspect, 112, 1645-1653.
Georgopoulos PG; Lioy PJ. (1994). Conceptual and theoretical aspects of human exposure and dose
assessment. J Expo Sci Environ Epidemiol, 4, 253-285.
Georgopoulos PG; Purushothaman V; Chiou R. (1997). Comparative evaluation of methods for
estimating potential human exposure to ozone: photochemical modeling and ambient monitoring.
J Expo Sci Environ Epidemiol, 7, 191-215.
Georgopoulos PG; Wang S-W; Vyas VM; Sun Q; Burke J; Vedantham R; McCurdy T; Ozkaynak H.
(2005). A source-to-dose assessment of population exposures to fine PM and ozone in
Philadelphia, PA, during a summer 1999 episode. J Expo Sci Environ Epidemiol, 15, 439-457.
Georgopoulos PG; Lioy PJ. (2006). From theoretical aspects of human exposure and dose assessment to
computational model implementation: The Modeling Environment for Total Risk studies
(MENTOR). J Toxicol Environ Health B, 9, 457-483.
Gery MW; Whitten GZ; Killus JP; Dodge MC. (1989). A photochemical kinetics mechanism for urban
and regional scale computer modeling. J Geophys Res, 94, 12,925-12,956.
Gidel LT. (1983). Cumulus cloud transport of transient tracers. J Geophys Res, 88, 6587-6599.
Gilboa SM; Mendola P; Olshan AF; Langlois PH; Savitz DA; Loomis D; Herring AH; Fixler DE. (2005).
Relation between ambient air quality and selected birth defects, seven county study, Texas, 1997-
2000. Am J Epidemiol, 162, 238-252.
Gilliam FS; Hockenberry AW; Adams MB. (2006). Effects of atmospheric deposition on the herbaceous
layer of a central Appalachian hardwood forest. J Torrey Bot Soc, 133, 240-254
R-17
-------
Gilliland FD; McConnell R; Peters J; Gong H, Jr. (1999). A theoretical basis for investigation ambient air
pollution and children's respiratory health. Environ Health Perspect, 3, 403-407.
Gilliland FD; Rappaport EB; Berhane K; Islam T; Dubeau L; Gauderman WJ; McConnell R. (2002).
Effects of glutathione S-transferase PI, Ml, and T1 on acute respiratory illness in school
children. Am J Respir Crit Care Med, 166, 346-351.
Ginsberg GL; Foos BP; Firestone MP. (2005). Review and analysis of inhalation dosimetry methods for
application to children's risk assessment. J Toxicol Environ Health A, 68, 573-615.
Gokirmak M; Yildirim Z; Hasanoglu HC; Koksal N; Mehmet N. (2003). The role of oxidative stress in
bronchoconstriction due to occupational sulfur dioxide exposure. Clin Chim Acta, 331, 119-126.
Gold DR; Litonjua A; Schwartz J; Lovett E; Larson A; Nearing B; Allen G; Verrier M; Cherry R; Verrier
R. (2000). Ambient pollution and heart rate variability. Circulation, 101, 1267-1273.
Gold DR; Schwartz J; Litonjua A; Verrier R; Zanobetti A. (2003). Ambient pollution and reduced heart
rate variability. In Revised analyses of time-series studies of air pollution and health. Special
Report. Boston, MA: Health Effects Institute.
Goldberg MS; Burnett RT; Bailar JC, 3rd; Tamblyn R; Ernst P; Flegel K; Brook J; Bonvalot Y; Singh R;
Valois MF; Vincent R. (2001). Identification of persons with cardiorespiratory conditions who
are at risk of dying from the acute effects of ambient air particles. Environ Health Perspect, 109
Suppl 4, 487-494.
Goldberg MS; Burnett RT. (2003). Revised analysis of the Montreal time-series study. In: Revised
analyses of time-series studies of air pollution and health. Special Report. Boston, MA: Health
Effects Institute; pp. 113-132. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdffl8
October, 2004],
Goldberg MS; Burnett RT; Valois MF; Flegel K; Bailar JC, 3rd; Brook J; Vincent R; Radon K. (2003).
Associations between ambient air pollution and daily mortality among persons with congestive
heart failure. Environ Res, 91, 8-20.
Gondal MA. (1997). Laser photoacoustic spectrometer for remote monitoring of atmospheric pollutants.
Appl Opt, 36, 3195-3201.
Gondal MA; Mastromarino J. (2001). Pulsed laser photoacoustic detection of S02 near 225.7 nm. Appl.
Opt 40, 2010-2016.
Gong H; Linn WS; Shamoo DA; Anderson KR; Nugent CA; Clark KW; Lin AE. (1996). Effect of
inhaled salmeterol on sulfur dioxide-induced bronchoconstriction in asthmatic subjects. Chest,
110, 1229-1235.
Gong H, Jr.; Lachenbruch PA; Harber P; Linn WS. (1995). Comparative short-term health responses to
sulfur dioxide exposure and other common stresses in a panel of asthmatics. Toxicol Ind Health,
11,467-487.
Gong H, Jr.; Linn WS; Terrell SL; Anderson KR; Clark KW. (2001). Anti-inflammatory and lung
function effects of montelukast in asthmatic volunteers exposed to sulfur dioxide. Chest, 119,
402-408.
R-18
-------
Goss CH; Newsom SA; Schildcrout JS; Sheppard L; Kaufman JD. (2004). Effect of ambient air pollution
on pulmonary exacerbations and lung function in cystic fibrosis. Am J Respir Crit Care Med, 169,
816-821.
GouveiaN; Fletcher T. (2000). Respiratory diseases in children and outdoor air pollution in Sao Paulo,
Brazil: a time series analysis. Occup Environ Med, 57, 477-483.
GouveiaN; Hajat S; Armstrong B. (2003). Socioeconomic differentials in the temperature-mortality
relationship in Sao Paulo, Brazil. Int J Epidemiol, 32, 390-397.
Gouveia N; Bremner SA; Novaes HM. (2004). Association between ambient air pollution and birth
weight in Sao Paulo, Brazil. J Epidemiol Community Health, 58, 11-17.
Greenhut GK; Ching JKS; Pearson R, Jr.; Repoff TP. (1984). Transport of ozone by turbulence and
clouds in an urban boundary layer. J Geophys Res, 89, 4757-4766.
Greenhut GK. (1986). Transport of ozone between boundary layer and cloud layer by cumulus clouds. J
Geophys Res, 91, 8613-8622.
Grell GA; Emeis S; Stockwell WR; Schoenemeyer T; Forkel R; Michalakes J; Knoche R; Seidl W.
(2000). Application of a multiscale, coupled MM5/chemistry model to the complex terrain of the
VOTALP valley campaign. Atmos Environ, 34, 1435-1453.
Groneberg DA; Quarcoo D; Frossard N; Fischer A. (2004). Neurogenic mechanisms in bronchial
inflammatory diseases. Allergy, 59, 1139-1152.
Gross A; Stockwell WR. (2003). Comparison of the EMEP, RADM2 and RACM mechanisms. J Atmos
Chem, 44, 151-170.
Grunstein MM; Hazucha M; Sorli MH; Milic-Emili J. (1977). Effect of S02 on control of breathing in
anesthetized cats. J Appl Physiol, 43, 844-851.
Gumuslu S; Akbas H; Aliciguzel Y; Agar A; Kucukatay V; Yargicoglu P. (1998). Effects of sulfur
dioxide inhalation on antioxidant enzyme activities in rat erythrocytes. Ind Health, 36, 70-73.
Gumuslu S; Bilmen S; Korgun DK; Yargicoglu P; Agar A. (2001). Age-related changes in antioxidant
enzyme activities and lipid peroxidation in lungs of control and sulfur dioxide exposed rats. Free
Radical Res, 34, 621-627.
Gunnison AF; Benton AW. (1971). Sulfur dioxide: Sulfite. Interaction with mammalian serum and
plasma. Arch Environ Health, 22, 381-388.
Gunnison AF; Palmes ED. (1973). Persistence of plasma S-sulfonates following exposure of rabbits to
sulfite and sulfur dioxide. Toxicol Appl Pharmacol, 24, 266-278.
Gunnison AF; Sellakumar A; Currie D; Snyder EA. (1987). Distribution, metabolism and toxicity of
inhaled sulfur dioxide and endogenously generated sulfite in the respiratory tract of normal and
sulfite oxidase-deficient rats. J Toxicol Environ Health, 21, 141-162.
R-19
-------
Gunnison AF; Sellakumar A; Snyder EA; Currie D. (1988). The effect of inhaled sulfur dioxide and
systemic sulfite on the induction of lung carcinoma in rats by benzo[a]pyrene. Environ Res, 46,
59-73.
Gwynn RC; Burnett RT; Thurston GD. (2000). A time-series analysis of acidic particulate matter and
daily mortality and morbidity in the Buffalo, New York, region. Environ Health Perspect, 108,
125-133.
Ha EH; Hong YC; Lee BE; Woo BH; Schwartz J; Christiani DC. (2001). Is air pollution a risk factor for
low birth weight in Seoul? Epidemiology, 12, 643-648.
Ha EH; Lee JT; Kim H; Hong YC; Lee BE; Park HS; Christiani DC. (2003). Infant susceptibility of
mortality to air pollution in Seoul, South Korea. Pediatrics, 111, 284-290.
Hagen JA; Nafstad P; Skrondal A; Bjorkly S; Magnus P. (2000). Associations between outdoor air
pollutants and hospitalization for respiratory diseases. Epidemiology, 11, 136-140.
Haider SS; Hasan M; Hasan SN; Khan SR; Ali SF. (1981). Regional effects of sulfur dioxide exposure on
the guinea pig brain lipids, lipid peroxidation and lipase activity. Neurotoxicology, 2, 443-450.
Haider SS; Hasan M; Khan NH. (1982). Air pollutant sulfur dioxide-induced alterations on the levels of
lipids, lipid peroxidation and lipase activity in various regions of the rat brain. Pharmacol
Toxicol, 51, 45-50.
Haider SS; Hasan M. (1984). Neurochemical changes by inhalation of environmental pollutants sulfur
dioxide and hydrogen sulfide: degradation of total lipids, elevation of lipid peroxidation and
enzyme activity in discrete regions of the guinea pig brain and spinal cord. Ind Health, 22, 23-31.
Hains JC; Chen LWA; Taubman BF; Doddridge BG; Dickerson RR. (2007). A side-by-side comparison
of filter-based PM2.5 measurements at a suburban site: A closure study. Atmos Environ, 41,
6167-6184.
Hajat S; Haines A; Goubet SA; Atkinson RW; Anderson HR. (1999). Association of air pollution with
daily GP consultations for asthma and other lower respiratory conditions in London. Thorax, 54,
597-605.
Hajat S; Haines A; Atkinson RW; Bremner SA; Anderson HR; Emberlin J. (2001). Association between
air pollution and daily consultations with general practitioners for allergic rhinitis in London,
United Kingdom. Am J Epidemiol, 153, 704-714.
Hajat S; Anderson HR; Atkinson RW; Haines A. (2002). Effects of air pollution on general practitioner
consultations for upper respiratory diseases in London. Occup Environ Med, 59, 294-299.
Hajj AM; Burki NK; Lee LY. (1996). Role of tachykinins in sulfur dioxide-induced bronchoconstriction
in anesthetized guinea pigs. J Appl Physiol, 80, 2044-2050.
Halinen AI; Salonen RO; Pennanen AS; Kosma VM. (2000a). Combined respiratory effects of cold air
with SO(2) orNO(2) in repeated 10-minute exposures of hyperventilating guinea pigs. Inhal
Toxicol, 12, 671-691.
R-20
-------
Halinen AI; Salonen RO; Pennanen AS; Kosma VM. (2000b). Combined respiratory effects of cold air
with S02 or N02 in single 1-hour exposures of hyperventilating guinea pigs. Inhal Toxicol, 12,
693-713.
Hanke M; Umann B; Uecker J; Arnold F; Bunz H. (2003). Atmospheric measurements of gas-phase
HNO3 and S02 using chemical ionization mass spectrometry during the MINATROC field
campaign 2000 on Monte Cimone. Atmos. Chem. Phys. 3: 417-436.
Hansen K; Draaijers GPJ; Ivens WPMF; Gundersen P; van Leeuwen NFM. (1994). Concentration
variations in rain and canopy throughfall collected sequentially during individual rain events.
Atmos Environ, 28, 3195-3205.
Harris MI; Flegal KM; Cowie CC; Eberhardt MS; Goldstein DE; Little RR; Wiedmeyer HM; Byrd-Holt
DD. (1998). Prevalence of diabetes, impaired fasting glucose, and impaired glucose tolerance in
U.S. adults. The Third National Health and Nutrition Examination Survey, 1988-1994. Diabetes
Care, 21,518-524.
Hazucha M; Bates DV. (1975). Combined effect of ozone and sulphur dioxide on human pulmonary
function. Nature, 257, 50-51.
Hedley AJ; Wong CM; Thach TQ; Ma S; Lam TH; Anderson HR. (2002). Cardiorespiratory and all-cause
mortality after restrictions on sulphur content of fuel in Hong Kong: An intervention study.
Lancet, 360, 1646-1652.
Hedley AJ; McGhee SM; Wong CM; Barron B; Chau P; Chau J; Thach TQ; Wong TW; Loh C. (2006).
Air Pollution: Costs and Paths to a Solution. Hong Kong: Civic Exchange.
HEI. (2004). Health effects of outdoor air pollution in developing countries of Asia: a literature review.
HEI Public Health and Air Pollution in Asia Program Special Report #15. Boston, MA: HEI
International Scientific Oversight Committee; Health Effects Institute.
Heinrich J; Hoelscher B; Frye C; Meyer I; Pitz M; Cyrys J; Wjst M; Neas L; Wichmann HE. (2002).
Improved air quality in reunified Germany and decreases in respiratory symptoms. Epidemiology,
13, 394-401.
Heinrich U; Mohr U; Fuhst R; Brockmeyer C. (1989). Investigation of a potential cotumorigenic effect of
the dioxides of nitrogen and sulfur, and of diesel-engine exhaust, on the respiratory tract of Syrian
golden hamsters. Res Rep Health Eff Inst, 1-27.
Henneberger A; Zareba W; Ibald-Mulli A; Ruckerl R; Cyrys J; Couderc JP; Mykins B; Woelke G;
Wichmann HE; Peters A. (2005). Repolarization changes induced by air pollution in ischemic
heart disease patients. Environ Health Perspect, 113, 440-446.
Hennigan CJ; Sandholm S; Kim S; Stickel RE; Huey LG; Weber RJ. (2006). Influence of Ohio River
valley emissions on fine particle sulfate measured from aircraft over large regions of the eastern
United States and Canada during INTEX-NA. J. Geophys. Res. 111(D24S04):
10.1029/2006JD007282.
Henrotin JB; Besancenot JP; Bejot Y; Giroud M. (2007a). Ozone pollution and the incidence of
cerebrovascular accident in Dijon, France from 1994 to 2004. Environnement, Risques et Sante,
6, 421-422.
R-21
-------
Henrotin JB; Besancenot JP; Bejot Y; Giroud M. (2007b). Short-term effects of ozone air pollution on
ischaemic stroke occurrence: a case-crossover analysis from a 10-year population-based study in
Dijon, France. Occup Environ Med, 64, 439-445.
Herbarth O; Fritz G; Krumbiegel P; Diez U; Franck U; Richter M. (2001). Effect of sulfur dioxide and
particulate pollutants on bronchitis in children—a risk analysis. Environ Toxicol, 16, 269-276.
Higgins BG; Francis HC; Yates CJ; Warburton CJ; Fletcher AM; Reid JA; Pickering CA; Woodcock AA.
(1995). Effects of air pollution on symptoms and peak expiratory flow measurements in subjects
with obstructive airways disease. Thorax, 50, 149-155.
Hill AB. (1965). The Environment and Disease: Association or Causation? Proc R Soc Med, 58, 295-300.
Hiller FC. (1991). Health implications of hygroscopic particle growth in the human respiratory tract. J
Aerosol Med, 4, 1-23.
Hiltermann TJN; Stolk J; Van der Zee SC; Brunekreef B; De Bruijne CR; Fischer PH; Ameling CB; Sterk
PJ; Hiemstra PS; Van Bree L. (1998). Asthma severity and susceptibility to air pollution. Eur
Respir J, 11, 686-693.
Hirsch T; Weiland SK; von Mutius E; Safeca AF; Grafe H; Csaplovics E; Duhme H; Keil U; Leupold W.
(1999). Inner city air pollution and respiratory health and atopy in children. Eur Respir J, 14, 669-
677.
Ho WC; Hartley WR; Myers L; Lin MH; Lin YS; Lien CH; Lin RS. (2007). Air pollution, weather, and
associated risk factors related to asthma prevalence and attack rate. Environ Res, 104, 402-409.
Hoek G; Brunekreef B. (1992). Time trends in repeated spirometry in children. Eur Respir J, 5, 553-559.
Hoek G; Brunekreef B. (1993). Acute effects of a winter air pollution episode on pulmonary function and
respiratory symptoms of children. Arch Environ Health, 48, 328-335.
Hoek G; Brunekreef B. (1994). Effects of low-level winter air pollution concentrations on respiratory
health of Dutch children. Environ Res, 64, 136-150.
Hoek G; Brunekreef B. (1995). Effect of photochemical air pollution on acute respiratory symptoms in
children. Am J Respir Crit Care Med, 151, 27-32.
Hoek G; Brunekreef B; Verhoeff A; van Wijnen J; Fischer P. (2000). Daily mortality and air pollution in
The Netherlands. J Air Waste Manag Assoc, 50, 1380-1389.
Hoek G; Brunekreef B; Fischer P; van Wijnen J. (2001). The association between air pollution and heart
failure, arrhythmia, embolism, thrombosis, and other cardiovascular causes of death in a time
series study. Epidemiology, 12, 355-357.
Hoek G; Brunekreef B; Goldbohm S; Fischer P; van den Brandt PA. (2002). Association between
mortality and indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet,
360, 1203-1209.
R-22
-------
Hoek G. (2003). Daily mortality and air pollution in The Netherlands. In Revised analyses of time-series
studies of air pollution and health. Special report, (pp. 133-141). Boston, MA: Health Effects
Institute.
Hoel DG. (1980). Incorporation of background in dose-response models. Fed Proc, 39, 73-75.
Holguin F; Tellez-Rojo MM; Hernandez M; Cortez M; Chow JC; Watson JG; Mannino D; Romieu I.
(2003). Air pollution and heart rate variability among the elderly in Mexico City. Epidemiology,
14, 521-527.
Holland HD. (1978). The chemistry of the atmosphere and oceans. New York, NY: Wiley.
Hong YC; Lee JT; Kim H; Ha EH; Schwartz J; Christiani DC. (2002). Effects of air pollutants on acute
stroke mortality. Environ Health Perspect, 110, 187-191.
Hoppel WA; Caffrey PF. (2005). Oxidation of S(IV) in sea-salt aerosol at high pH: ozone versus aerobic
reaction. J Geophys Res, 110D23202.
Horak F, Jr.; Studnicka M; Gartner C; Spengler JD; Tauber E; Urbanek R; Veiter A; Frischer T. (2002b).
Particulate matter and lung function growth in children: a 3-yr follow-up study in Austrian
schoolchildren [author response]. Eur. Respir. J. 20: 1355.
Horak F, Jr.; Studnicka M; Gartner C; Spengler JD; Tauber E; Urbanek R; Veiter A; Frischer T. (2002a).
Particulate matter and lung function growth in children: a 3-yr follow-up study in Austrian
schoolchildren. Eur Respir J, 19, 838-845.
Horowitz LW; Walters S; Mauzerall DL; Emmons LK; Rasch PJ; Granier C; Tie X; Lamarque J-F;
Schultz MG; Tyndall GS; Orlando JJ; Brasseur GP. (2003). A global simulation of tropospheric
ozone and related tracers: description and evaluation of MOZART, version 2. J Geophys Res,
108D24.
Horstman D; Roger LJ; Kehrl H; Hazucha M. (1986). Airway sensitivity of asthmatics to sulfur dioxide.
Toxicol Ind Health, 2, 289-298.
Horstman DH; Seal E, Jr.; Folinsbee LJ; Ives P; Roger LJ. (1988). The relationship between exposure
duration and sulfur dioxide-induced bronchoconstriction in asthmatic subjects. Am Ind Hyg
Assoc J, 49, 38-47.
Hosseinpoor AR; Forouzanfar MH; Yunesian M; Asghari F; Naieni KH; Farhood D. (2005). Air pollution
and hospitalization due to angina pectoris in Tehran, Iran: a time-series study. Environ Res, 99,
126-131.
Houssaini ASS; Messaouri H; Nasri I; Roth MP; Nejjar C; Benchekroun MN. (2007). Air pollution as a
determinant of asthma among schoolchildren in Mohammedia, Morocco. Int J Environ Health
Res, 17, 243-257.
Huebert BJ; Zhuang L; Howell S; Noone K; Noone B. (1996). Sulfate, nitrate, methanesulfonate,
chloride, ammonium, and sodium measurements from ship, island, and aircraft during the
Atlantic Stratocumulus Transition Experiment/Marine Aerosol Gas Exchange. J Geophys Res,
101,4413-4423.
R-23
-------
Huey LG; Tanner DJ; Slusher DL; Dibb JE; Arimoto R; Chen G; Davis D; Buhr MP; Nowak JB; Mauldin
RL; Eisele FL; Kosciuch E. (2004). CIMS measurements ofHN03 and S02 at the South Pole
during ISCAT 2000. Atmos Environ, 38, 5411-5421.
Huntzicker JJ; Hoffman RS; Ling C-S. (1978). Continuous measurement and speciation of sulfur-
containing aerosols by flame photometry. Atmos Environ, 12, 83-88.
Hwang BF; Lee YL; Lin YC; Jaakkola JJ; Guo YL. (2005). Traffic related air pollution as a determinant
of asthma among Taiwanese school children. Thorax, 60, 467-473.
Hwang JS; Chan CC. (2002). Effects of air pollution on daily clinic visits for lower respiratory tract
illness. Am J Epidemiol, 155, 1-10.
IARC. (1989). Diesel and gasoline engine exhausts and some nitroarenes. In IARC monographs on the
evaluation of carcinogenic risks to humans, v46. Lyon, France: International Agency for
Research on Cancer. World Health Organization.
IARC. (2006). Report of the advisory group to review the amended preamble to the IARC monographs.
In: IARC monographs on the evaluation of carcinogenic risks to humans. Available:
http://monographs.iarc.fr/ENG/Preamble/CurrentPreamble.pdf (21 February, 2008).Lyon, France:
International Agency for Research on Cancer; World Health Organization.
Ibald-Mulli A; Stieber J; Wichmann HE; Koenig W; Peters A. (2001). Effects of air pollution on blood
pressure: a population-based approach. Am J Public Health, 91, 571-577.
ICF Consulting. (2005). Decision points for configuring APEX for air toxics exposure assessment.
Research Triangle Park, NC: ICF Consulting, prepared for Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency.
Ilabaca M; Olaeta I; Campos E; Villaire J; Tellez-Rojo MM; Romieu I. (1999). Association between
levels of fine particulate and emergency visits for pneumonia and other respiratory illnesses
among children in Santiago, Chile. J Air Waste Manag Assoc, 49, 154-163.
Imai M; Yoshida K; Kitabatake M. (1986). Mortality from asthma and chronic bronchitis associated with
changes in sulfur oxides air pollution. Arch Environ Occup Health, 41, 29-35.
IOM (2008). Improving the Presumptive Disability Decision-Making Process for Veterans. Samet James
M; Bodurow CC (Eds.), Washington, DC; Committee on Evaluation of the Presumptive
Disability Decision-Making Process for Veterans; Board on Military and Veterans Health;
Institute of Medicine of the National Academies.
http://books.nap.edu/catalog.php7record id=l 1908
IPCS. (2004). IPCS risk assessment terminology. Part 2: IPCS glossary of key exposure assessment
terminology. IPCS Harmonization Project document #1 Geneva, Switzerland: International
Programme on Chemical Safety (IPCS); World Health Organization (WHO). Available:
http://www.who.int/ipcs/methods/harmonization/areas/ipcsterminologypartsland2.pdffl9 April,
2007],
Ito K. (2003). Associations of particulate matter components with daily mortality and morbidity in
Detroit, Michigan. In: Revised analyses of time-series studies of air pollution and health. Special
report. Boston, MA; Health Effects Institute.
R-24
-------
Ito K; Thurston GD; Silverman RA. (2007). Characterization of PM2.5, gaseous pollutants, and
meteorological interactions in the context of time-series health effects models. J Expo Sci
Environ Epidemiol, 17 Suppl 2, S45-60.
Ito T; Ohyama K; Kusano T; Usuda Y; Nozawa A; Hayashi H; Ohji H; Kitamura H; Kanisawa M. (1997).
Pulmonary endocrine cell hyperplasia and papilloma in rats induced by intratracheal injections of
extract from particulate air pollutants. Exp Toxicol Pathol, 49, 65-70.
Jacob DJ. (1999) Introduction to Atmospheric Chemistry. Princeton, NJ; Princeton University Press.
Jacobson MZ. (2002). Atmospheric pollution: history, science, and regulation. New York, NY:
Cambridge University Press
Jadsri S; Singhasivanon P; Kaewkungwal J; Sithiprasasna R; Siriruttanapruk S; Konchom S. (2006).
Spatio-temporal effects of estimated pollutants released from an industrial estate on the
occurrence of respiratory disease in Maptaphut Municipality, Thailand. Int J Health Geogr 5, 48.
Jaffe DH; Singer ME; Rimm AA. (2003). Air pollution and emergency department visits for asthma
among Ohio Medicaid recipients, 1991-1996. Environ Res, 91, 21-28.
Jakab GJ; Clarke RW; Hemenway DR; Longphre MV; Kleeberger SR; Frank R. (1996). Inhalation of
acid coated carbon black particles impairs alveolar macrophage phagocytosis. Toxicol Lett, 88,
243-248.
Jaklevic JM; Loo BW; Fujita TY. (1981). Automatic particulate sulfur measurements with a dichotomous
sampler and on-line x-ray fluorescence analysis. Environ Sci Tech, 15, 687-690.
Jalaludin B; Morgan G; Lincoln D; Sheppeard V; Simpson R; Corbett S. (2006). Associations between
ambient air pollution and daily emergency department attendances for cardiovascular disease in
the elderly (65+ years), Sydney, Australia. J Expo Sci Environ Epidemiol, 16, 225-237.
Jedrychowski W. (1999). Ambient air pollution and respiratory health in the east Baltic region. Scand J
Work Environ Health, 25 Suppl 3,5-16.
Jenkin ME; Saunders SM; Wagner V; Pilling MJ. (2003).Protocol for the development of the Master
Chemical Mechanism, MCM v3 (Part B): troposheric degradation of aromatic volatile organic
compounds. Atmos Chem Phys 3, 181-193.
Jerrett M; Burnett RT; Goldberg MS; Sears M; Krewski D; Catalan R; Kanaroglou P; Giovis C;
Finkelstein N. (2003a). Spatial analysis for environmental health research: concepts, methods,
and examples. (Special issue on Health and air quality: interpreting science for decision makers.).
J Toxicol Environ Health A, 66, 1783-1810.
Jerrett M; Burnett RT; Willis A; Krewski D; Goldberg MS; DeLuca P; Finkelstein N. (2003b). Spatial
analysis of the air pollution-mortality relationship in the context of ecologic confounders.
(Special issue on Health and air quality: interpreting science for decision makers.). J Toxicol
Environ Health A, 66: 1735-1777.
R-25
-------
Jerrett M; Burnett RT; Brook J; Kanaroglou P; Giovis C; Finkelstein N; Hutchison B. (2004). Do
socioeconomic characteristics modify the short term association between air pollution and
mortality? Evidence from a zonal time series in Hamilton, Canada. J Epidemiol Community
Health, 58, 31-40.
Jerrett M; Finkelstein M. (2005). Geographies of risk in studies linking chronic air pollution exposure to
health outcomes. J Toxicol Environ Health A, 68, 1207-1242.
Jet Propulsion Laboratory. (2003). Chemical kinetics and photochemical data for use in atmospheric
studies. Jet Propulsion Laboratory (JPL) publication no. 02-25. Pasadena, CA; California Institute
of Technology; Jet Propulsion Laboratory. Available: http://jpldataeval.jpl.nasa.gov/pdf/JPL_02-
25_l_intro_rev0.pdf (18 December, 2003).
Jimenez JL; Jayne JT; Shi Q; Kolb CE; Worsnop DR; Yourshaw I; Seinfeld JH; Flagan RC; Zhang X;
Smith KA; Morris JW; Davidovits P. (2003a). Ambient aerosol sampling using the Aerodyne
Aerosol Mass Spectrometer. J Geophys Res 108, 8425, doi:10.1029/2001JD001213, 2003
Jimenez P; Baldasano JM; Dabdub D. (2003b). Comparison of photochemical mechanisms for air quality
modeling. Atmos Environ, 37, 4179-4194.
Johnson CW; Williams WC; Copeland CB; DeVito MJ; Smialowicz RJ. (2000). Sensitivity of the SRBC
PFC assay versus ELISA for detection of immunosuppression by TCDD and TCDD-like
congeners. Toxicology, 156, 1-11.
Johnson T. (2001). A guide to selected algorithms, distributions, and databases used in exposure models
developed by the Office of Air Quality Planning and Standards (DRAFT). Report # CERM report
TR01. Chapel Hill, NC and Piscataway, NJ TRJ Environmental, Inc & Environmental and
Occupational Health Sciences Institute for OAQPS; Office of Research and Development; U.S.
Environmental Protection Agency
Johnson T. (2002). A guide to selected algorithms, distributions, and databases used in exposure models
developed by the Office of Air Quality Planning and Standards [revised draft]. Research Triangle
Park; TRJ Environmental, Inc & Environmental and Occupational Health Sciences Institute for
OAQPS; Office of Research and Development; U.S. Environmental Protection Agency.
http://www.epa.gov/ttn/fera/data/human/report052202.pdf
Jorres R; Magnussen H. (1990). Airways response of asthmatics after a 30 min exposure, at resting
ventilation, to 0.25 ppm N02 or 0.5 ppm S02. Eur Respir J, 3, 132-137.
Junge CE; Chagnon CW; Manson JE. (1961). Stratospheric aerosols. Journal of Meteorol, 18, 81-108.
Just J; Segala C; Sahraoui F; Priol G; Grimfeld A; Neukirch F. (2002). Short-term health effects of
particulate and photochemical air pollution in asthmatic children. Eur Respir J, 20, 899-906.
Kasibhatla P; Chameides WL. (2000). Seasonal modeling of regional ozone pollution in the eastern
United States. Geophys Res Lett, 27, 1415-1418.
Kasper-Giebl A; Krenn S; Puxbaum H. (1999). Laboratory and field measurements of a badge type
passive sampler for the determination of ambient sulfur dioxide concentrations. Anal Bioanal
Chem, 363, 73-76.
R-26
-------
Katsouyanni K; Touloumi G; Spix C; Schwartz J; Balducci F; Medina S; Rossi G; Wojtyniak B; Sunyer
J; Bacharova L; Schouten JP; Ponka A; Anderson HR. (1997). Short term effects of ambient
sulphur dioxide and particulate matter on mortality in 12 European cities: results from time series
data from the APHEA project. Br Med J, 314, 1658-1663.
Kauffmann F. (2004). Post-genome respiratory epidemiology: a multidisciplinary challenge. Eur Respir J,
24, 471-480.
Keatinge WR; Donaldson GC. (2006). Heat acclimatization and sunshine cause false indications of
mortality due to ozone. Environ Res, 100, 387-393.
Keck L; Wittmaack K. (2005). Effect of filter type and temperature on volatilisation losses from
ammonium salts in aerosol matter. Atmos Environ, 39, 4093-4100.
Keck L; Wittmaack K. (2006). Simplified approach to measuring semivolatile inorganic particulate matter
using a denuded cellulose filter without backup filters. Atmos Environ, 40, 7106-7114.
Keene WC; Jacob DJ; Fan S-M. (1996). Reactive chlorine: a potential sink for dimethylsulfide and
hydrocarbons in the marine boundary layer. Atmos Environ, 306, i-iii.
Keene WC; Sander R; Pszenny AAP; Vogt R; Crutzen PJ; Galloway JN. (1998). Aerosol pH in the
marine boundary layer: a review and model evaluation. J Aerosol Sci, 29, 339-356.
Keene WC; Pszenny AAP; Maben JR; Sander R. (2002). Variation of marine aerosol acidity with particle
size. Geophys Res Lett, 297.
Keene WC; Pszenny AAP; Maben JR; Stevenson E; Wall A. (2004). Closure evaluation of size-resolved
aerosol pH in the New England coastal atmosphere during summer. J Geophys Res, 109D23307.
Kehrl HR; Roger LJ; Hazucha MJ; Horstman DH. (1987). Differing response of asthmatics to sulfur
dioxide exposure with continuous and intermittent exercise. Am Rev Respir Dis, 135, 350-355.
Keles N; Ilicali C; Deger K. (1999). The effects of different levels of air pollution on atopy and symptoms
of allergic rhinitis. Am J Rhinol, 13, 185-190.
Kelly FJ; Mudway IS. (2003). Protein oxidation at the air-lung interface. Amino Acids, 25, 375-396.
Kelsall JE; Samet JM; Zeger SL; Xu J. (1997). Air pollution and mortality in Philadelphia, 1974-1988.
Am J Epidemiol, 146, 750-762.
Kesten S; Szalai J; Dzyngel B. (1995). Air quality and the frequency of emergency room visits for
asthma. Ann Allergy Asthma Immunol, 74, 269-273.
Khoury MJ; Davis R; Gwinn M; Lindegren ML; Yoon P. (2005) Do we need genomic research for the
prevention of common diseases with environmental causes?.Am J Epidemiol 161(9):799-805
Kilic D. (2003). The effects of ageing and sulfur dioxide inhalation exposure on visual-evoked potentials,
antioxidant enzyme systems, and lipid-peroxidation levels of the brain and eye. Neurotoxicol
Teratol, 25, 587-598.
R-27
-------
Kimbell JS; Miller FJ. (1999). Regional respiratory-tract absorption of inhaled reactive gases: a modeling
approach. In: (Gardner DE; Crapo JD; McClellan RO, eds.) Toxicology of the Lung 3rd edition.
New York, NY; Taylor and Francis, pp 557-597.
Kindzierski WB; Sembaluk S. (2001). Indoor outdoor relationship of S02 concentrations in a rural and an
urban community of Alberta. Canadian Journal of Civil Engineering, 28, 163-169.
Kindzierski WB; Ranganathan HKS. (2006). Indoor and outdoor S02 in a community near oil sand
extraction and production facilities in northern Alberta. Journal of Environmental Engineering
and Science, 5, S121-S129.
Kinney PL; Ozkaynak H. (1991). Associations of daily mortality and air pollution in Los Angeles
County. Environ Res, 54, 99-120.
Kirkpatrick MB; Sheppard D; Nadel JA; Boushey HA. (1982). Effect of the oronasal breathing route on
sulfur dioxide-induced bronchoconstriction in exercising asthmatic subjects. Am Rev Respir Dis,
125, 627-631.
Kitabatake M; Yoshida K; Kasama K; Murase S; Yuan PF; Manjurul H; Yamauchi T. (1992). Procedure
for evaluating changes in respiratory symptoms of experimentally asthma-induced guinea pigs by
a personal computer. J Toxicol Environ Health, 37, 265-275.
Kitabatake M; Yamamoto H; Yuan PF; Manjurul H; Murase S; Yamauchi T. (1995). Effects of exposure
to N02 or S02 on bronchopulmonary reaction induced by Candida albicans in guinea pigs. J
Toxicol Environ Health, 45, 75-82.
Kittelson DB; McKenzie R; Vermeersch M; Dorman F; Pui D; Linne M; Liu B; Whitby K. (1978). Total
sulfur aerosol concentration with an electrostatically pulsed flame photometric detector system.
Atmos Environ 12, 105-111.
Klein RG; Janowsky I; Schmezer P; Hermann R; Spiegelhalder B; Zeller WJ; Pool BL. (1989). Effect of
long-term inhalation of N-nitroso-dimethylamine (NDMA) and S02/NOx in rats. Exp Pathol, 37,
273-280.
Klemm RJ; Mason RM, Jr. (2000). Aerosol Research and Inhalation Epidemiological Study (ARIES): air
quality and daily mortality statistical modeling—interim results. J Air Waste Manag Assoc, 50,
1433-1439.
Klemm RJ; Mason RM, Jr.; Heilig CM; Neas LM; Dockery DW. (2000). Is daily mortality associated
specifically with fine particles? Data reconstruction and replication of analyses. J Air Waste
Manag Assoc, 50, 1215-1222.
Klemm RJ; Lipfert FW; Wyzga RE; Gust C. (2004). Daily mortality and air pollution in Atlanta: two
years of data from ARIES. Inhal Toxicol, 16 Suppl 1, 131-141.
Klepeis NE; Nelson WC; Ott WR; Robinson JP; Tsang AM; Switzer P; Behar JV; Hern SC; Engelmann
WH. (2001). The National Human Activity Pattern Survey (NHAPS): a resource for assessing
exposure to environmental pollutants. J Expo Anal Environ Epidemiol, 11, 231-252.
R-28
-------
Knorst MM; Kienast K; Riechelmann H; Muller-Quernheim J; Ferlinz R. (1994). Effect of sulfur dioxide
on mucociliary activity and ciliary beat frequency in guinea pig trachea. Int Arch Occup Environ
Health, 65, 325-328.
Ko FW; Tarn W; Wong TW; Chan DP; Tung AH; Lai CK; Hui DS. (2007a). Temporal relationship
between air pollutants and hospital admissions for chronic obstructive pulmonary disease in Hong
Kong. Thorax, 62, 780-785.
Ko FW; Tam W; Wong TW; Lai CK; Wong GW; Leung TF; Ng SS; Hui DS. (2007b). Effects of air
pollution on asthma hospitalization rates in different age groups in Hong Kong. Clin Exp Allergy,
37, 1312-1319.
Koenig JQ; Pierson WE; Horike M; Frank R. (1981). Effects of S02 plus NaCl aerosol combined with
moderate exercise on pulmonary function in asthmatic adolescents. Environ Res, 25, 340-348.
Koenig JQ; Pierson WE; Horike M; Frank R. (1983). A comparison of the pulmonary effects of 0.5 ppm
versus 1.0 ppm sulfur dioxide plus sodium chloride droplets in asthmatic adolescents. J Toxicol
Environ Health, 11, 129-139.
Koenig JQ; Marshall SG; Horike M; Shapiro GG; Furukawa CT; Bierman CW; Pierson WE. (1987). The
effects of albuterol on sulfur dioxide-induced bronchoconstriction in allergic adolescents. J
Allergy Clin Immunol, 79, 54-58.
Koenig JQ; Marshall SG; van Belle G; McManus MS; Bierman CW; Shapiro GG; Furukawa CT; Pierson
WE. (1988). Therapeutic range cromolyn dose-response inhibition and complete obliteration of
S02-induced bronchoconstriction in atopic adolescents. J Allergy Clin Immunol, 81, 897-901.
Koenig JQ; Covert DS; Hanley QS; van Belle G; Pierson WE. (1990). Prior exposure to ozone potentiates
subsequent response to sulfur dioxide in adolescent asthmatic subjects. Am Rev Respir Dis, 141,
377-380.
Koenig JQ; Dumler K; Rebolledo V; Williams PV; Pierson WE. (1992). Theophylline mitigates the
bronchoconstrictor effects of sulfur dioxide in subjects with asthma. J Allergy Clin Immunol, 89,
789-794.
Kohlhammer Y; Rzehak P; Behr J; Wichmann HE; Heinrich J. (2007). High prevalence of pneumonia in
children of a smelter town. Int J Occup Environ Health 13, 167-174
Koken PJ; Piver WT; Ye F; Elixhauser A; Olsen LM; Portier CJ. (2003). Temperature, air pollution, and
hospitalization for cardiovascular diseases among elderly people in Denver. Environ Health
Perspect, 111, 1312-1317.
Koksal N; Yildirim Z; Gokirmak M; Hasanoglu HC; Mehmet N; Avci H. (2003). The role of nitric oxide
and cytokines in asthma-like syndrome induced by sulfur dioxide exposure in agricultural
environment. Clinica Chimica Acta, 336, 115-122.
Kopp MV; Bohnet W; Frischer T; Ulmer C; Studnicka M; Ihorst G; Gardner C; Forster J; Urbanek R;
Kuehr J. (2000). Effects of ambient ozone on lung function in children over a two-summer
period. Eur Respir J, 16, 893-900.
R-29
-------
Korhonen P; Kulmala M; Laaksonen A; Viisanen Y; McGraw R; Seinfeld JH. (1999). Ternary nucleation
of H2S04, NH3, and H20 in the atmosphere. J Geophys Res, 104, 26,349-326,353.
Kotesovec F; Skorkovsky J; Brynda J; Peters A; Heinrich J. (2000). Diurnal mortality and air pollution in
the North Bohemia. Cas Lek Cesk, 139: 6, 166-173.
Koutrakis P; Wolfson JM; Slater JL; Brauer M; Spengler JD; Stevens RK; Stone CL. (1988a). Evaluation
of an annular denuder/filter pack system to collect acidic aerosols and gases. Environ Sci Technol
22, 1463-1468.
Koutrakis P; Wolfson JM; Spengler JD. (1988b). An improved method for measuring aerosol strong
acidity: results from a nine-month study in St. Louis, Missouri and Kingston, Tennessee. Atmos
Environ, 22, 157-162.
Koutrakis P; Suh HH; Sarnat JA; Brown KW; Coull BA; Schwartz J. (2005). Characterization of
particulate and gas exposures of sensitive populations living in Baltimore and Boston. Research
Report no. 131. Boston, MA; Health Effects Institute.
Kramer U; Behrendt H; Dolgner R; Ranft U; Ring J; Wilier H; Schlipkoter HW. (1999). Airway diseases
and allergies in East and West German children during the first 5 years after reunification: time
trends and the impact of sulphur dioxide and total suspended particles. Int J Epidemiol, 28, 865-
873.
Krewski D; Burnett RT; Goldberg MS; Hoover K; Siemiatycki J; Jerrett M; Abrahamowicz M; White
WH. (2000). Reanalysis of the Harvard Six Cities study and the American Cancer Society study
of particulate air pollution and mortality: A special report of the Institute's Particle Epidemiology
Reanalysis Project. Cambridge, MA; Health Effects Institute.
http://pubs.healtheffects.org/view.php?id=6
Krewski D; Burnett RT; Goldberg MS; Hoover BK; Siemiatycki J; Jerrett M; Abrahamowicz M; White
WH. (2003). Overview of the reanalysis of the Harvard Six Cities Study and American Cancer
Society Study of Particulate Air Pollution and Mortality. J Toxicol Environ Health A, 66, 1507-
1551.
Kucukatay V; Agar A; Yargicoglu P; Gumuslu S; Aktekin B. (2003). Changes in somatosensory evoked
potentials, lipid peroxidation, and antioxidant enzymes in experimental diabetes: effect of sulfur
dioxide. Arch Environ Health, 58, 14-22.
Kulle TJ; Sauder LR; Shanty F; Kerr HD; Farrell BP; Miller WR; Milman JH. (1984). Sulfur dioxide and
ammonium sulfate effects on pulmonary function and bronchial reactivity in human subjects. Am
Ind Hyg Assoc J, 45, 156-161.
Kulmala M; Piijola L; Makela JM. (2000). Stable sulphate clusters as a source of new atmospheric
particles. Nature, 404, 66-69.
Kumar N; Russell AG. (1996). Development of a computationally efficient, reactive subgrid-scale plume
model and the impact in the northeastern United States using increasing levels of chemical detail.
J Geophys Res, 101, 16,737-16,744.
Kunzli N. (2005). Unifying susceptibility, exposure, and time: discussion of unifying analytic approaches
and future directions. J Toxicol Environ Health A, 68, 1263-1271.
R-30
-------
Kwon HJ; Cho SH; Nyberg F; Pershagen G. (2001). Effects of ambient air pollution on daily mortality in
a cohort of patients with congestive heart failure. Epidemiology, 12, 413-419.
Lagorio S; Forastiere F; Pistelli R; Iavarone I; Michelozzi P; Fano V; Marconi A; Ziemacki G; Ostro BD.
(2006). Air pollution and lung function among susceptible adult subjects: a panel study. Environ
Health, 5, 11.
Lam HF; Peisch R; Amdur MO. (1982). Changes in lung volumes and diffusing capacity in guinea pigs
exposed to a combination of sulfur dioxide and submicron zinc oxide mixed in a humidified
furnace. Toxicol Appl Pharmacol, 66, 427-433.
Lam LT. (2007). The association between climatic factors and childhood illnesses presented to hospital
emergency among young children. Int J Environ Health Res, 17, 1-8.
Langley-Evans S. (2007). Letter to Annette Iannucci on sulfur oxides concentrations. Loughborough,
United Kingdom: University of Nottingham; January 30, 2007.
Langley-Evans SC; Phillips GJ; Jackson AA. (1996). Sulphur dioxide: a potent glutathione depleting
agent. Comp Biochem Physiol C Pharmacol Toxicol Endocrinol, 114, 89-98.
Langley-Evans SC; Phillips GJ; Jackson AA. (1997). Fetal exposure to low protein maternal diet alters
the susceptibility of young adult rats to sulfur dioxide-induced lung injury. J Nutr, 127, 202-209.
Langstaff JE. (2007). Analysis of uncertainty in ozone population exposure modeling [technical
memorandum]. Research Triangle Park, NC; Office of Air Quality Planning and Standards; U.S.
Environmental Protection Agency.
Lawther PJ. (1955). Effect of inhalation of sulphur dioxide on respiration and pulse-rate in normal
subjects. Lancet, 269, 745-748.
Lawther PJ; Waller RE; Henderson M. (1970). Air pollution and exacerbations of bronchitis. Thorax, 25,
525-539.
Lawther PJ; Macfarlane AJ; Waller RE; Brooks AG. (1975). Pulmonary function and sulphur dioxide,
some preliminary findings. Environ Res, 10, 355-367.
Lazarus SC; Wong HH; Watts MJ; Boushey HA; Lavins BJ; Minkwitz MC. (1997). The leukotriene
receptor antagonist zafirlukast inhibits sulfur dioxide-induced bronchoconstriction in patients
with asthma. Am J Respir Crit Care Med, 156, 1725-1730.
Le Tertre A; Quenel P; Eilstein D; Medina S; Prouvost H; Pascal L; Boumghar A; Saviuc P; Zeghnoun A;
Filleul L; Declercq C; Cassadou S; Le Goaster C. (2002). Short-term effects of air pollution on
mortality in nine French cities: a quantitative summary. Arch Environ Health, 57, 311-319.
Lee BE; Ha EH; Park HS; Kim YJ; Hong YC; Kim H; Lee JT. (2003a). Exposure to air pollution during
different gestational phases contributes to risks of low birth weight. Hum Reprod, 18, 638-643.
Lee IM; Ho CK; Tsai SS; Chiu HF; Yang CY. (2007). Air pollution and hospital admissions for
congestive heart failure in a tropical city: Kaohsiung, Taiwan. Inhal Toxicol, 19, 899-904.
R-31
-------
Lee JT; Shin D; Chung Y. (1999a). Air pollution and daily mortality in Seoul and Ulsan, Korea. Environ
Health Perspect, 107, 149-154.
Lee J; Schwartz J. (1999b). Reanalysis of the effects of air pollution on daily mortality in Seoul, Korea: a
case-crossover design. Environ Health Perspect, 107, 633-636.
Lee JT; Kim H; Hong YC; Kwon HJ; Schwartz J; Christiani DC. (2000). Air pollution and daily mortality
in seven major cities of Korea, 1991-1997. Environ Res, 84, 247-254.
Lee JT; Kim H; Song H; Hong YC; Cho YS; Shin SY; Hyun YJ; Kim YS. (2002). Air pollution and
asthma among children in Seoul, Korea. Epidemiology, 13, 481-484.
Lee JT; Kim H; Cho YS; Hong YC; Ha EH; Park H. (2003b). Air pollution and hospital admissions for
ischemic heart diseases among individuals 64+ years of age residing in Seoul, Korea. Arch
Environ Health, 58, 617-623.
Lee JT; Son JY; Kim H; Kim SY. (2006a). Effect of air pollution on asthma-related hospital admissions
for children by socioeconomic status associated with area of residence. Arch Environ Occup
Health, 61, 123-130.
Lee LY; Widdicombe JG. (2001). Modulation of airway sensitivity to inhaled irritants: role of
inflammatory mediators. Environ Health Perspect, 109, 585-589.
Lee SC; Chan LY; Chiu MY. (1999c). Indoor and outdoor air quality investigation at 14 public places in
Hong Kong. Environ Int, 25, 443-450.
Lee SL; Wong WH; Lau YL. (2006b). Association between air pollution and asthma admission among
children in Hong Kong. Clin Exp Allergy, 36, 1138-1146.
Lee YL; Lin YC; Lee YC; Wang JY; Hsiue TR; Guo YL. (2004). Glutathione S-transferase PI gene
polymorphism and air pollution as interactive risk factors for childhood asthma. Clin Exp
Allergy, 34, 1707-1713.
Leem JH; Kaplan BM; Shim YK; Pohl HR; Gotway CA; Bullard SM; Rogers JF; Smith MM; Tylenda
CA. (2006). Exposures to air pollutants during pregnancy and preterm delivery. Environ Health
Perspect, 114, 905-910.
Lemos M; Lichtenfels AJFC; Amaro E, Jr.; Macchione M; Martins MA; King M; Bohm GM; Saldiva
PHN. (1994). Quantitative pathology of nasal passages in rats exposed to urban levels of air
pollution. Environ Res, 66, 87-95.
Lepeule J; Rondeau V; Filleul L; Dartigues JF. (2006). Survival analysis to estimate association between
short-term mortality and air pollution. Environ Health Perspect, 114, 242-247.
Leung FT, A. J. Colussi, M. R. Hoffmann, and G. C. Toon. (2002). Isotopic fractionation of carbonyl
sulfide in the atmosphere: implications for the source of background stratospheric sulfate aerosol.
Geophys Res Lett, 29, 1474, doi:10.1029/2001GL013955.
R-32
-------
Levine JS; Pinto JP. (1998). The production of CO by biomass burning. In: (Khalil MAK; Pinto JP;
Shearer MJ; eds.) Atmospheric carbon monoxide and its environmental effects: Proceedings of
the international conference; December 1997; Portland, Oregon. Report #EPA/600/R-98/047
Research Triangle Park, NC; Office of Research and Development; U.S. Environmental
Protection Agency; pp. 251-256.
Lewis AJ; Kirchner T. (1984). Modulation of sulfur dioxide-induced airways hyperresponsiveness in the
conscious dog. Int Arch Allergy Appl Immunol, 75, 188-190.
Li R; Meng Z. (2007). Effects of S02 derivatives on expressions of MUC5AC and IL-13 in human
bronchial epithelial cells. Archives of Toxicology 81, 867-874.
Li R; Meng Z; Xie J. (2007). Effects of sulfur dioxide on the expressions of MUC5AC and ICAM-1 in
airway of asthmatic rats. Regulatory Toxicology and Pharmacology, 48, 284-291.
Li YQ; Harrison RM. (1990). Comparison of indoor and outdoor concentrations of acid gases, ammonia
and their associated salts. Environ. Technol. 11: 315-326.
Liao D; Duan Y; Whitsel EA; Zheng ZJ; Heiss G; Chinchilli VM; Lin HM. (2004). Association of higher
levels of ambient criteria pollutants with impaired cardiac autonomic control: a population-based
study. Am J Epidemiol, 159, 768-777.
Liao D; Heiss G; Chinchilli VM; Duan Y; Folsom AR; Lin HM; Salomaa V. (2005). Association of
criteria pollutants with plasma hemostatic/inflammatory markers: a population-based study. J
Expo Anal Environ Epidemiol, 15, 319-328.
Liebhart J; Malolepszy J; Wojtyniak B; Pisiewicz K; Plusa T; Gladysz U; Chyrek-Borowska S; Siergiejko
Z; Rogalewska A; Szymanski W; Dziedziczko A; Banach-Wawrzynczak E; Tlappa J; Kurek M;
Malaczynska T; Grubska-Suchanska E; Lademan A; Szczeklik A; Sladek K; Bochenek G;
Duplaga M; Milanowski J; Trebas-Pietras E; Kuna P; Elgalal A; Kuprys I; Alkiewicz J;
Breborowicz A; Mlynarczyk W; Silny W; Kurzawa R; Sak I; Kruszewski J; Chmielewska D;
Droszcz W; Zaras E; Boznanski A; Jankowska R; Nittner-Marszalska M; Machaj G; Wrzyszcz
M; Balinska W; Rogala E; Rogala B; Gawlik R; Dobek R; Obojski A. (2007). Prevalence and risk
factors for asthma in Poland: Results from the PMSEAD study. Journal of Investigational
Allergology and Clinical Immunology, 17, 367-374.
Lin CA; Martins MA; Farhat SC; Pope CA, 3rd; Conceicao GM; Anastacio VM; Hatanaka M; Andrade
WC; Hamaue WR; Bohm GM; Saldiva PH. (1999). Air pollution and respiratory illness of
children in Sao Paulo, Brazil. Paediatr Perinat Epidemiol, 13, 475-488.
Lin CA; Amador Pereira LA; de Souza Conceicao GM; Kishi HS; Milani R, Jr.; Ferreira Braga AL;
Nascimento Saldiva PH. (2003a). Association between air pollution and ischemic cardiovascular
emergency room visits. Environ Res, 92, 57-63.
Lin CM; Li CY; Mao IF. (2004a). Increased risks of term low-birth-weight infants in a petrochemical
industrial city with high air pollution levels. Arch Environ Health, 59, 663-668.
Lin CM; Li CY; Yang GY; Mao IF. (2004b). Association between maternal exposure to elevated ambient
sulfur dioxide during pregnancy and term low birth weight. Environ Res, 96, 41-50.
R-33
-------
Lin M; Chen Y; Burnett RT; Villeneuve PJ; Krewski D. (2003b). Effect of short-term exposure to
gaseous pollution on asthma hospitalisation in children: a bi-directional case-crossover analysis. J
Epidemiol Community Health, 57, 50-55.
Lin M; Chen Y; Villeneuve PJ; Burnett RT; Lemyre L; Hertzman C; McGrail KM; Krewski D. (2004c).
Gaseous air pollutants and asthma hospitalization of children with low household income in
Vancouver, British Columbia, Canada. Am J Epidemiol, 159, 294-303.
Lin M; Stieb DM; Chen Y. (2005). Coarse particulate matter and hospitalization for respiratory infections
in children younger than 15 years in Toronto: A case-crossover analysis. Pediatrics, 116, e235-
240.
Lin S; Hwang SA; Pantea C; Kielb C; Fitzgerald E. (2004d). Childhood asthma hospitalizations and
ambient air sulfur dioxide concentrations in Bronx County, New York. Arch Environ Health, 59,
266-275.
Linn WS; Jones MP; Bailey RM; Kleinman MT; Spier CE; Fischer DA; Hackney JD. (1980). Respiratory
effects of mixed nitrogen dioxide and sulfur dioxide in human volunteers under simulated
ambient exposure conditions. Environ Res, 22, 431-438.
Linn WS; Shamoo DA; Spier CE; Valencia LM; Anzar UT; Venet TG; Hackney JD. (1983a). Respiratory
effects of 0. 75 ppm sulfur dioxide in exercising asthmatics: Influence of upper-respiratory
defenses. Environ Res, 30, 340-348.
Linn WS; Venet TG; Shamoo DA; Valencia LM; Anzar UT; Spier CE; Hackney JD. (1983b). Respiratory
effects of sulfur dioxide in heavily exercising asthmatics. A dose-response study. Am Rev Respir
Dis, 127, 278-283.
Linn WS; Avol EL; Shamoo DA; Venet TG; Anderson KR; Whynot JD; Hackney JD. (1984).
Asthmatics' responses to 6-hr sulfur dioxide exposures on two successive days. Arch Environ
Health, 39, 313-319.
Linn WS; Fischer DA; Shamoo DA; Spier CE; Valencia LM; Anzar UT; Hackney JD. (1985a).
Controlled exposures of volunteers with chronic obstructive pulmonary disease to sulfur dioxide.
Environ Res, 37, 445-451.
Linn WS; Shamoo DA; Anderson KR; Whynot JD; Avol EL; Hackney JD. (1985b). Effects of heat and
humidity on the responses of exercising asthmatics to sulfur dioxide exposure. Am Rev Respir
Dis, 131,221-225.
Linn WS; Avol EL; Peng RC; Shamoo DA; Hackney JD. (1987). Replicated dose-response study of
sulfur dioxide effects in normal, atopic, and asthmatic volunteers. Am Rev Respir Dis, 136, 1127-
1134.
Linn WS; Avol EL; Shamoo DA; Peng RC; Spier CE; Smith MN; Hackney JD. (1988). Effect of
metaproterenol sulfate on mild asthmatics' response to sulfur dioxide exposure and exercise.
Arch Environ Health, 43, 399-406.
Linn WS; Shamoo DA; Peng RC; Clark KW; Avol EL; Hackney JD. (1990). Responses to sulfur dioxide
and exercise by medication-dependent asthmatics: effect of varying medication levels. Arch
Environ Health, 45, 24-30.
R-34
-------
Linn WS; Gong H, Jr.; Shamoo DA; Anderson KR; Avol EL. (1997). Chamber exposures of children to
mixed ozone, sulfur dioxide, and sulfuric acid. Arch Environ Health, 52, 179-187.
Lipfert FW; Baty J; Miller J; Wyzga R. (2006a). PM2 5 constituents and related air quality variables as
predictors of survival in a cohort of U.S. military veterans. Inhal Toxicol 18, 645-657.
Lipfert FW; Morris SC; Wyzga RE. (2000a). Daily mortality in the Philadelphia metropolitan area and
size-classified particulate matter. J Air Waste Manag Assoc, 50, 1501-1513.
Lipfert FW; Perry HM, Jr.; Miller JP; Baty JD; Wyzga RE; Carmody SE. (2000b). The Washington
University-EPRI Veterans' Cohort Mortality Study: preliminary results. Inhal Toxicol, 12, 41-73.
Lipfert FW; Zhang J; Wyzga RE. (2000c). Infant mortality and air pollution: a comprehensive analysis of
U.S. data for 1990. J Air Waste Manag Assoc, 50, 1350-1366.
Lipfert FW; Perry HM, Jr.; Miller JP; Baty JD; Wyzga RE; Carmody SE. (2003). Air pollution, blood
pressure, and their long-term associations with mortality. Inhal Toxicol, 15, 493-512.
Lipfert FW; Wyzga RE; Baty JD; Miller JP. (2006b). Traffic density as a surrogate measure of
environmental exposures in studies of air pollution health effects: Long-term mortality in a cohort
of US veterans. Atmos Environ, 40, 154-169.
Lippmann M; Ito K; Nadas A; Burnett RT. (2000). Association of particulate matter components with
daily mortality and morbidity in urban populations. Res Rep Health Eff Inst, 5-72, discussion 73-
82.
Liu H; Jacob DJ; Bey I; Yantosca RM. (2001a). Constraints from 210Pb and 7Be on wet deposition and
transport in a global three-dimensional chemical tracer model driven by assimilated
meteorological fields. J Geophys Res, 106, 12,109-12,128.
Liu S; Krewski D; Shi Y; Chen Y; Burnett RT. (2003). Association between gaseous ambient air
pollutants and adverse pregnancy outcomes in Vancouver, Canada. Environ Health Perspect, 111,
1773-1778.
Liu S; Krewski D; Shi Y; Chen Y; Burnett RT. (2007). Association between maternal exposure to
ambient air pollutants during pregnancy and fetal growth restriction. J Expo Sci Environ
Epidemiol, 17, 426-432.
Liu X; Hegg DA; Stoelinga MT. (2001b). Numerical simulation of new particle formation the Northwest
Atlantic using the M5 mesoscale model coupled with sulfur chemistry. J Geophys Res, 106,
9697-9715.
Llorca J; Salas A; Prieto-Salceda D; Chinchon-Bengoechea V; Delgado-Rodriguez M. (2005). Nitrogen
dioxide increases cardiorespiratory admissions in Torrelavega (Spain). J Environ Health, 68, 30-
35.
Loomis D; Castillejos M; Gold DR; McDonnell W; Borja-Aburto VH. (1999). Air pollution and infant
mortality in Mexico City. Epidemiology, 10, 118-123.
Lovati MR; Manzoni C; Daldossi M; Spolti S; Sirtori CR. (1996). Effects of sub-chronic exposure to S02
on lipid and carbohydrate metabolism in rats. Arch Toxicol, 70, 164-173.
R-35
-------
Low RB; Bielory L; Qureshi AI; Dunn V; Stuhlmiller DF; Dickey DA. (2006). The relation of stroke
admissions to recent weather, airborne allergens, air pollution, seasons, upper respiratory
infections, and asthma incidence, September 11, 2001, and day of the week. Stroke, 37, 951-957.
Lu R; Turco RP; Jacobson MZ. (1997). An integrated air pollution modeling system for urban and
regional scales: 1. Structure and performance. J Geophys Res, 102, 6063-6079.
Luginaah IN; Fung KY; Gorey KM; Webster G; Wills C. (2005). Association of ambient air pollution
with respiratory hospitalization in a government-designated "area of concern": the case of
Windsor, Ontario. Environ Health Perspect, 113, 290-296.
Luke WT. (1997). Evaluation of a commercial pulsed fluorescence detector for the measurement of low-
level S02 concentrations during the Gas-Phase Sulfur Intercomparison Experiment. J Geophys
Res, 102, 16,255-16,265.
Lunn DJ; Thomas A; Best N; Spiegelhalter D. (2000). WinBUGS - A Bayesian modelling framework:
Concepts, structure, and extensibility. Stat Comput, 10, 325-337.
Lunn JE; Knowelden J; Roe JW. (1967). Patterns of respiratory illness in Sheffield infant schoolchildren..
Br J Prev Soc Med, 21, 7-16
Lunn JE; Knowelden J; Roe JW. (1970). Patterns of respiratory illness in Sheffield junior schoolchildren.
A follow-up study. Br J Prev Soc Med, 24, 223-228.
Luttmann-Gibson H; Suh HH; Coull BA; Dockery DW; Sarnat SE; Schwartz J; Stone PH; Gold DR.
(2006). Short-term effects of air pollution on heart rate variability in senior adults in Steubenville,
Ohio. J Occup Environ Med, 48, 780-788.
Mage D; Wilson W; Hasselblad V; Grant L. (1999). Assessment of human exposure to ambient
particulate matter. J Air Waste Manag Assoc, 49, 1280-1291.
Magnussen H; Jorres R; Wagner HM; von Nieding G. (1990). Relationship between the airway response
to inhaled sulfur dioxide, isocapnic hyperventilation, and histamine in asthmatic subjects. Int
Arch Occup Environ Health, 62, 485-491.
Mahrt L. (1998). Stratified atmospheric boundary layers and breakdown of models. Theor Comput Fluid
Dyn, 11,263-279.
Maisonet M; Bush TJ; Correa A; Jaakkola JJ. (2001). Relation between ambient air pollution and low
birth weight in the Northeastern United States. Environ Health Perspect, 109, 351-356.
Makri A; Stilianakis NI. (2008). Vulnerability to air pollution health effects. Int J Hyg Environ Health
211, 326-336. doi: 10.1016/j.ijheh.2007.06.005
Mannix RC; Phalen RF; Kenoyer JL; Crocker TT. (1982). Effect of sulfur dioxide-sulfate exposure on rat
respiratory tract clearance.Am Ind Hyg Assoc J, 43, 679-685.
Mar TF; Norris GA; Koenig JQ; Larson TV. (2000). Associations between air pollution and mortality in
Phoenix, 1995-1997. Environ Health Perspect, 108, 347-353.
R-36
-------
Mar TF; Norris GA; Larson TV; Wilson WE; Koenig JQ. (2003). Air pollution and cardiovascular
mortality in Phoenix, 1995-1997. In: HEI Special Report. Revised analyses of time-series studies
of air pollution and health, pp. 177-182. Boston, MA; Health Effects Institute.
Martins LC; Latorre Mdo R; Saldiva PH; Braga AL. (2002). Air pollution and emergency room visits due
to chronic lower respiratory diseases in the elderly: an ecological time-series study in Sao Paulo,
Brazil. J Occup Environ Med, 44, 622-627.
Martins MC; Fatigati FL; Vespoli TC; Martins LC; Pereira LA; Martins MA; Saldiva PH; Braga AL.
(2004). Influence of socioeconomic conditions on air pollution adverse health effects in elderly
people: an analysis of six regions in Sao Paulo, Brazil. J Epidemiol Community Health, 58, 41-
46.
Massman WJ; Pederson J; Delany A; Grantz D; Denhartog G; Neumann HH; Oncley SP; Pearson R;
Shaw RH. (1994). An evaluation of the regional acid deposition model surface module for ozone
uptake at 3 sites in the San-Joaquin valley of California. J Geophys Res, 99, 8281-8294.
Matsumi Y; Shigemori H; Takahashi K. (2005). Laser-induced fluorescence instrument for measuring
atmospheric SO 2. Atmos Environ, 39, 3177-3185.
Mautz WJ; Kleinman MT; Finlayson-Pitts B. (1988). Respiratory effects of acid containing
multicomponent pollutant atmospheres. Irvine, CA; Air Pollution Health Effects Laboratory;
Department of Community and Environmental Medicine; University of California; for California
Air Resources Board
McCurdy T. (1997). Modeling the dose profile in human exposure assessments: ozone as an example.
Reviews Toxicola, 1, 3-23.
McCurdy T. (2000). Conceptual basis for multi-route intake dose modeling using an energy expenditure
approach. J Expo Anal Environ Epidemiol, 10, 86-97.
McCurdy T; Glen G; Smith L; Lakkadi Y. (2000). The national exposure research laboratory's
consolidated human activity database. J Expo Anal Environ Epidemiol, 10, 566-578.
McDonnell WF; Abbey DE; Nishino N; Lebowitz MD. (1999). Long-term ambient ozone concentration
and the incidence of asthma in nonsmoking adults: the Ahsmog study. Environ Res, 80, 110-121.
McKenna DS; Konopka P; Grooss J-U; Gunther G; Muller R; Spang R; Offermann D; Orsolini Y. (2002).
A new chemical Lagrangian model of the stratosphere (CLaMS) 1. formulation of advection and
mixing. J Geophys Res, 107, 107(D16): 10.1029/2000JD000114.
Mebust MR; Eder BK; Binkowski FS; Roselle SJ. (2003). Models-3 community multiscale air quality
(CMAQ) model aerosol component. 2. Model evaluation. J Geophys Res, 108, D6,
10.1029/2001JD001410.
Melville GN. (1970). Changes in specific airway conductance in healthy volunteers following nasal and
oral inhalation of S02. West Indian Med J, 19, 231-235.
Meng Z; Zhang B; Ruan A; Sang N; Zhang J. (2002). Micronuclei induced by sulfur dioxide inhalation in
mouse bone-marrow cells in vivo. Inhal Toxicol, 14, 303-309.
R-37
-------
Meng Z; Geng H; Bai J; Yan G. (2003a). Blood pressure of rats lowered by sulfur dioxide and its
derivatives. Inhal Toxicol, 15, 951-959.
Meng Z; Qin G; Zhang B; Geng H; Bai Q; Bai W; Liu C. (2003b). Oxidative damage of sulfur dioxide
inhalation on lungs and hearts of mice. Environ Res, 93, 285-292.
Meng Z; Zhang B; Bai J; Geng H; Liu C. (2003c). Oxidative damage of sulfur dioxide inhalation on
stomachs and intestines of mice. Inhal Toxicol, 15, 397-410.
Meng Z; Bai W. (2004). Oxidation damage of sulfur dioxide on testicles of mice. Environ Res, 96, 298-
304.
Meng Z; Liu Y; Wu D. (2005a). Effect of Sulfur Dioxide Inhalation on Cytokine Levels in Lungs and
Serum of Mice. Inhal Toxicol, 17, 303-307.
Meng Z; Qin G; Zhang B. (2005b). DNA damage in mice treated with sulfur dioxide by inhalation.
Environmental and molecular mutagenesis, 46, 150-155.
Menzel DB; Keller DA; Leung K-H. (1986). Covalent reactions in the toxicity of S02 and sulfite.
Advances in Experimental Medicine and Biology, 197, 477-492.
Meranger JC; Brule D; Smith-Palmer T; Wentzell B; Donini JC. (1987). Indoor-outdoor levels of
nitrogen and sulphur species and their relation to air flow in Antigonish, Nova Scotia.
International Journal of Environmental Analytical Chemistry, 29, 61-72.
Metzger KB; Tolbert PE; Klein M; Peel JL; Flanders WD; Todd K; Mulholland JA; Ryan PB; Frumkin
H. (2004). Ambient air pollution and cardiovascular emergency department visits. Epidemiology,
15,46-56.
Metzger KB; Klein M; Flanders WD; Peel JL; Mulholland JA; Langberg JJ; Tolbert PE. (2007). Ambient
air pollution and cardiac arrhythmias in patients with implantable defibrillators. Epidemiology,
18, 585-592.
Michaud J-P; Grove JS; Krupitsky D. (2004). Emergency department visits and "vog"-related air quality
in Hilo, Hawai'i. Environ Res, 95, 11-19.
Michelozzi P; Forastiere F; Fusco D; Perucci CA; Ostro B; Ancona C; Pallotti G. (1998). Air pollution
and daily mortality in Rome, Italy. Occup Environ Med, 55, 605-610.
Miller KA; Siscovick DS; Sheppard L; Shepherd K; Sullivan JH; Anderson GL; Kaufman JD. (2007).
Long-term exposure to air pollution and incidence of cardiovascular events in women. N Engl J
Med, 356, 447-458.
Milner JT; Dimitroulopoulou C; ApSimon H. (2005). Indoor concentrations in buildings from sources
outdoors. In: Atmospheric Dispersion Modelling Liaison Committee. Annual Report 2004-2005 -
Annex B. Chilton, UK; The Atmospheric Dispersion Modelling Liaison Committee; Radiation
Protection Division; UK Health Protection Agency. Available:
http://www.admlc.org.uk/documents/ADMLC20042.pdf
Min JY; Min KB; Cho SI; Paek D. (2008). Combined effects of cigarette smoking and sulfur dioxide on
lung function in Koreans. J Toxicol Environ Health A, 71, 301-303.
R-38
-------
Mohorovic L. (2004). First two months of pregnancy - critical time for preterm delivery and low
birthweight caused by adverse effects of coal combustion toxics. Early Human Development
Elsevier Science Ltd, Oxford, UK: 2004 80: 2, 115-123.
Moolgavkar SH; Luebeck EG; Hall TA; Anderson EL. (1995). Air pollution and daily mortality in
Philadelphia. Epidemiology, 6, 476-484.
Moolgavkar SH; Luebeck EG; Anderson EL. (1997). Air pollution and hospital admissions for respiratory
causes in Minneapolis-St. Paul and Birmingham. Epidemiology, 8, 364-370.
Moolgavkar SH. (2000a). Air pollution and daily mortality in three U.S. counties. Environ Health
Perspect, 108, 777-784.
Moolgavkar SH. (2000b). Air pollution and hospital admissions for chronic obstructive pulmonary
disease in three metropolitan areas in the United States. Inhal Toxicol, 12, 75-90.
Moolgavkar SH. (2003a). Air pollution and daily mortality in two U.S. counties: season-specific analyses
and exposure-response relationships. Inhal Toxicol, 15, 877-907.
Moolgavkar SH. (2003b). Air pollution and daily deaths and hospital admissions in Los Angeles and
Cook counties. In: Revised Analyses of Time-Series Studies of Air Pollution and Health Special
Report. Boston: Health Effects Institute, 183-198.
Morello-Frosch R; Shenassa ED. (2006). The environmental "riskscape" and social inequality:
implications for explaining maternal and child health disparities. Environ Health Perspect, 114,
1150-1153.
Morris RD; Naumova EN; Munasinghe RL. (1995). Ambient air pollution and hospitalization for
congestive heart failure among elderly people in seven large US cities. Am J Public Health, 85,
1361-1365.
Morrow PE. (1986). Factors determining hygroscopic aerosol deposition in airways. Physiol Rev, 66,
330-376.
Mortimer KM; Tager IB; Dockery DW; Neas LM; Redline S. (2000). The effect of ozone on inner-city
children with asthma: identification of susceptible subgroups. Am J Respir Crit Care Med, 162,
1838-1845.
Mortimer KM; Neas LM; Dockery DW; Redline S; Tager IB. (2002). The effect of air pollution on inner-
city children with asthma. Eur Respir J, 19, 699-705.
Mueller PK; Collins JF. (1980). Development of a particulate sulfate analyzer. Report# P-1382F. Palo
Alto, CA: Electric Power Research Institute. Available from: NTIS, Springfield, VA; EPRI-EA-
1492.
Myers DJ; Bigby BG; Calvayrac P; Sheppard D; Boushey HA. (1986). Interaction of cromolyn and a
muscarinic antagonist in inhibiting bronchial reactivity to sulfur dioxide and to eucapnic
hyperpnea alone. Am Rev Respir Dis, 133, 1154-1158.
Myhre G; Berglen TF; Myhre CE; Isaksen ISA. (2004). The radiative effect of the anthropogenic
influence on the stratospheric sulfate aerosol layer. Tellus Ser B, 56, 294-299.
R-39
-------
Nadel JA; Salem H; Tamplin B; Tokiwa Y. (1965a). Mechanism of bronchoconstriction during inhalation
of sulfur dioxide; reflex involving vagus nerves. Arch Environ Health, 10, 175-178.
Nadel JA; Salem H; Tamplin B; Tokiwa Y. (1965b). Mechanism of bronchoconstriction during inhalation
of sulfur dioxide. J Appl Physiol, 20, 164-167.
Nadziejko C; Fang K; Narciso S; Zhong M; Su WC; Gordon T; Nadas A; Chen LC. (2004). Effect of
particulate and gaseous pollutants on spontaneous arrhythmias in aged rats. Inhal Toxicol, 16,
373-380.
Nafstad P; Haheim LL; Oftedal B; Gram F; Holme I; Hjermann I; Leren P. (2003). Lung cancer and air
pollution: a 27 year follow up of 16 209 Norwegian men. Thorax, 58, 1071-1076.
Nafstad P; Haheim LL; Wisloff T; Gram F; Oftedal B; Holme I; Hjermann I; Leren P. (2004). Urban air
pollution and mortality in a cohort of Norwegian men. Environ Health Perspect, 112, 610-615.
NAPAP. (1991). The experience and legacy of NAPAP. Report of the Oversight Review Board of the
National Acid Precipitation Assessment Program. Report # PB-92-128230/XAB. Washington,
DC; National Acid Precipitation Assessment Program, Council on Environmental Quality.
Neas LM; Dockery DW; Koutrakis P; Tollerud DJ; Speizer FE. (1995). The association of ambient air
pollution with twice daily peak expiratory flow rate measurements in children. Am J Epidemiol,
141, 111-122.
Neukirch F; Segala C; Le Moullec Y; Korobaeff M; Aubier M. (1998). Short-term effects of low-level
winter pollution on respiratory health of asthmatic adults. Arch Environ Health, 53, 320-328.
Newhouse CP; Levetin E. (2004). Correlation of environmental factors with asthma and rhinitis
symptoms in Tulsa, OK. Ann Allergy Asthma Immunol, 92, 356-366.
Nicks DK; Benner RL. (2001). Subminute measurements of S02 and low parts per trillion by volume
mixing ratios in the atmosphere. J Geophys Res [Atmos], 106, 27,769-727,776.
Nie A; Meng Z. (2005). Study of the interaction of sulfur dioxide derivative with cardiac sodium channel.
Biochim Biophys Acta, 1718, 67-73.
Nie A; Meng Z. (2006). Modulation of L-type calcium current in rat cardiac myocytes by sulfur dioxide
derivatives. Food Chem Toxicol, 44, 355-363.
Nodelman V; Ultman JS. (1999). Longitudinal distribution of chlorine absorption in human airways: a
comparison to ozone absorption. J Appl Physiol, 87, 2073-2080.
Norris G; YoungPong SN; Koenig JQ; Larson TV; Sheppard L; Stout JW. (1999). An association
between fine particles and asthma emergency department visits for children in Seattle. Environ
Health Perspect, 107, 489-493.
Nowak D; Jorres R; Berger J; Claussen M; Magnussen H. (1997). Airway responsiveness to sulfur
dioxide in an adult population sample. Am J Respir Crit Care Med, 156, 1151-1156.
R-40
-------
NRC. (1998). Research priorities for airborne particulate matter. I. Immediate priorities and a long-range
research portfolio. Committee on Research Priorities for Airborne Particulate Matter; Board on
Environmental Studies and Toxicology; Commission on Life Sciences; Commission on
Geosciences, Environment, and Resources; National Research Council (NRC); Washington, DC;
National Academy Press
NRC. (2004). Research priorities for airborne particulate matter. IV. Continuing research progress.
Committee on Research Priorities for Airborne Particulate Matter; Board on Environmental
Studies and Toxicology; Commission on Life Sciences; Commission on Geosciences,
Environment, and Resources; National Research Council (NRC); Washington, DC; National
Academy Press
NTP. (2005). Report on Carcinogens (Report #11). Research Triangle Park, NC: National Toxicology
Program (NTP); The National Institute of Environmental Health Sciences (NIEHS); National
Institutes of Health; Public Health Service; U.S. Department of Health and Human Services.
http://ntp.niehs.nih.gov/index.cfm?objectid=32BA9724-FlF6-975E-7FCE50709CB4C932
Nyberg F; Gustavsson P; Jarup L; Bellander T; Berglind N; Jakobsson R; Pershagen G. (2000). Urban air
pollution and lung cancer in Stockholm. Epidemiology, 11, 487-495.
NY DOH. (2006) A study of ambient air contaminants and asthma in New York City. ATSDR Final
Report #NTIS PB2006-113523.; Albany, NY; New York State Energy Research and
Development Authority; New York State Department of Health, for Atlanta, GA; Agency for
Toxic Substances and Disease Registry; U.S. Department of Health and Human Services.
O'Byrne PM. (2007). How much is too much? The treatment of mild asthma. Eur Respir J, 30, 403-406.
O'Dowd C; McFiggans G; Creasey DJ; Piijola L; Hoell C; Smith MH; Allan BJ; Plane JMC; Heard DE;
Lee JD; Pilling MJ; Kulmala M. (1999). On the photochemical production of new particles in the
coastal boundary layer. Geophys Res Lett, 26, 1707-1710.
O'Dowd CD; Jimenez JL; Bahreini R; Flagan RC; Seinfeld JH; Hamerl K; Pirjola L; Kulmala M;
Jennings SG; Hoffmann T. (2002). Marine aerosol formation from biogenic iodine emissions.
Nature, 417, 632-636.
O'Neill MS; Jerrett M; Kawachi I; Levy JI; Cohen AJ; Gouveia N; Wilkinson P; Fletcher T; Cifuentes L;
Schwartz J. (2003). Health, wealth, and air pollution: advancing theory and methods. Environ
Health Perspect, 111, 1861-1870.
Odman MT; Ingram CI (1996). Multiscale air quality simulation platform (MAQSIP): source code
documentation and validation. MCNC; technical report ENV-96TR002-vl.0. Research Triangle
Park, NC; Environmental Programs - MCNC-North Carolina Supercomputing Center. Available:
http://www.ce.gatech.edu/~todman/maqsip.pdf (9 September 2003).
Odman MT; Russel AG. (1999). Mass conservative coupling of non-hydrostatic meteorological models
with air quality models. In Grying SE, Batchvarova E (eds.), Air Pollution Modeling and its
Application XIII. New York, NY: Plenum Press.
Oftedal B; Nafstad P; Magnus P; Bjorkly S; Skrondal A. (2003). Traffic related air pollution and acute
hospital admission for respiratory diseases in Drammen, Norway 1995-2000. Eur J Epidemiol,
18, 671-675.
R-41
-------
Ohyama K-i; Ito T; Kanisawa M. (1999). The roles of diesel exhaust particle extracts and the promotive
effects ofN02 and/or S02 exposure on rat lung tumorigenesis. Cancer Lett, 139, 189-197.
Ostro B; Sanchez JM; Aranda C; Eskeland GS. (1996). Air pollution and mortality: results from a study
of Santiago, Chile. In: Papers from the ISEA-ISEE annual meeting; September 1994; Research
Triangle Park, NC. J. Exposure Anal. Environ. Epidemiol. 6: 97-114.
Ott WR. (1982). Concepts of human exposure to air pollution. Environ Int, 7, 179-196.
Padro J. (1996). Summary of ozone dry deposition velocity measurements and model estimates over
vineyard, cotton, grass and deciduous forest in summer. Atmos Environ, 30, 2363-2369.
Park H; Lee B; Ha EH; Lee JT; Kim H; Hong YC. (2002). Association of air pollution with school
absenteeism due to illness. Arch Pediatr Adolesc Med, 156, 1235-1239.
Park JK; Kim YK; Lee SR; Cho SH; Min KU; Kim YY. (2001a). Repeated exposure to low levels of
sulfur dioxide (S02) enhances the development of ovalbumin-induced asthmatic reactions in
guinea pigs. Ann Allergy Asthma Immunol, 86, 62-67.
Park JW; Lim YH; Kyung SY; An CH; Lee SP; Jeong SH; Ju YS. (2005a). Effects of ambient particulate
matter on peak expiratory flow rates and respiratory symptoms of asthmatics during Asian dust
periods in Korea. Respirology, 10, 470-476.
Park RJ; Stenchikov GL; Pickering; Dickerson RR; Allen DJ; Kondragunta S. (2001b). Regional air
pollution and its radiative forcing: studies with a single column chemical and radiation transport
model. J Geophys Res, 106, 28,751-28,770.
Park SK; O'Neill MS; Vokonas PS; Sparrow D; Schwartz J. (2005b). Effects of air pollution on heart rate
variability: the VA normative aging study. Environ Health Perspect, 113, 304-309.
Patterson E; Eatough DJ. (2000). Indoor/outdoor relationships for ambient PM2.5 and associated
pollutants: epidemiological implications in Lindon, Utah. J Air Waste Manag Assoc, 50, 103-110.
Peacock JL; Symonds P; Jackson P; Bremner SA; Scarlett JF; Strachan DP; Anderson HR. (2003). Acute
effects of winter air pollution on respiratory function in schoolchildren in southern England.
Occup Environ Med, 60, 82-89.
Peel JL; Tolbert PE; Klein M; Metzger KB; Flanders WD; Todd K; Mulholland JA; Ryan PB; Frumkin
H. (2005). Ambient air pollution and respiratory emergency department visits. Epidemiology, 16,
164-174.
Peel JL; Metzger KB; Klein M; Flanders WD; Mulholland JA; Tolbert PE. (2006). Ambient Air Pollution
and Cardiorespiratory Emergency Department Visits in Potentially Sensitive Groups.
Epidemiology, 17, S214-S215.
Peel JL; Metzger KB; Klein M; Flanders WD; Mulholland JA; Tolbert PE. (2007). Ambient air pollution
and cardiovascular emergency department visits in potentially sensitive groups. Am J Epidemiol,
165, 625-633.
Pekkanen J; Brunner EJ; Anderson HR; Tiittanen P; Atkinson RW. (2000). Daily concentrations of air
pollution and plasma fibrinogen in London. Occup Environ Med, 57, 818-822.
R-42
-------
Penard-Morand C; Charpin D; Raherison C; Kopferschmitt C; Caillaud D; Lavaud F; Annesi-Maesano I.
(2005). Long-term exposure to background air pollution related to respiratory and allergic health
in schoolchildren. Clin Exp Allergy, 35, 1279-1287.
Penard-Morand C; Schillinger C; Armengaud A; Debotte G; Chretien E; Pellier S; Annesi-Maesano I.
(2006). Assessment of schoolchildren's exposure to traffic-related air pollution in the French Six
Cities Study using a dispersion model. Atmos Environ 40, 2274-2287
Pereira LAA; Loomis D; Conceicao GMS; Braga ALF; Areas RM; Kishi HS; Singer JM; Bohm GM;
Saldiva PHN. (1998). Association between air pollution and intrauterine mortality in Sao Paulo,
Brazil. Environ Health Perspect, 106, 325-329.
Pereira PM; Saldiva PHN; Sakae RS; Bohm GM; Martins MA. (1995). Urban levels of air pollution
increase lung responsiveness in rats. Environ Res, 69, 96-101.
A; Goldstein IF; Beyer U; Franke K; Heinrich J; Dockery DW; Spengler JD; Wichmann HE.
(1996a). Acute health effects of exposure to high levels of air pollution in eastern Europe. Am J
Epidemiol, 144, 570-581.
A; Doring A; Wichmann HE; Koenig W. (1997). Increased plasma viscosity during an air pollution
episode: a link to mortality? Lancet, 349, 1582-1587.
A; Perz S; Doring A; Stieber J; Koenig W; Wichmann HE. (1999). Increases in heart rate during an
air pollution episode. Am J Epidemiol, 150, 1094-1098.
A; Liu E; Verrier RL; Schwartz J; Gold DR; Mittleman M; Baliff J; Oh JA; Allen G; Monahan K;
Dockery DW. (2000a). Air pollution and incidence of cardiac arrhythmia. Epidemiology, 11, 11-
17.
A; Skorkovsky J; Kotesovec F; Brynda J; Spix C; Wichmann HE; Heinrich J. (2000b).
Associations between mortality and air pollution in central Europe. Environ Health Perspect, 108,
283-287.
A; Dockery DW; Muller JE; Mittleman MA. (2001). Increased particulate air pollution and the
triggering of myocardial infarction. Circulation, 103, 2810-2815.
J; Hedley AJ; Wong CM; Lam TH; Ong SG; Liu J; Spiegelhalter DJ. (1996b). Effects of an
ambient air pollution intervention and environmental tobacco smoke on children's respiratory
health in Hong Kong. Int J Epidemiol, 25, 821-828.
Petroeschevsky A; Simpson RW; Thalib L; Rutherford S. (2001). Associations between outdoor air
pollution and hospital admissions in Brisbane, Australia. Arch Environ Health, 56, 37-52.
Petruzzi S; DeH'Omo G; Fiore M; Chiarotti F; Bignami G; Alleva E. (1996). Behavioural disturbances in
adult CD-I mice and absence of effects on their offspring upon S02 exposure. Arch Toxicol, 70,
757-766.
Phalen RF; Kleinman MT. (1987). 21-day exposure to mixed air pollutants: effects on lung airways and
macrophages. Irvine, CA; Air Pollution Health Effects Laboratory; Department of Community
and Environmental Medicine;University of California; for California Air Resources Board.
R-43
Peters
Peters
Peters
Peters
Peters
Peters
Peters
-------
Pickering KE; Thompson AM; Scala JR; Tao W-K; Simpson J; Garstang M. (1991). Photochemical
ozone production in tropical squall line convection during NASA Global Tropospheric
Experiment/Amazon Boundary Layer Experiment 2A. J Geophys Res, 96, 3099-3114.
Pickering KE; Thompson AM; Scala JR; Tao W-K; Dickerson RR; Simpson J. (1992). Free tropospheric
ozone production following entrainment of urban plumes into deep convection. J Geophys Res,
97, 17,985-18,000.
Pickering KE; Thompson AM; Tao W-K; Kucsera TL. (1993). Upper tropospheric ozone production
following mesoscale convection during STEP/EMEX. J Geophys Res, 98, 8737-8749.
Pickering KE; Thompson AM; Wang Y; Tao W-K; McNamara DP; KirchhoffWJH; Heikes BG; Sachse
GW; Bradshaw JD; Gregory GL; Blake DR. (1996). Convective transport of biomass burning
emissions over Brazil during TRACE A. J Geophys Res, 101, 23,993-24,012.
Pikhart H; Bobak M; Kriz B; Danova J; Celko MA; Prikazsky V; Pryl K; Briggs D; Elliott P. (2000).
Outdoor air concentrations of nitrogen dioxide and sulfur dioxide and prevalence of wheezing in
school children. Epidemiology, 11, 153-160.
Pikhart H; Bobak M; Gorynski P; Wojtyniak B; Danova J; Celko MA; Kriz B; Briggs D; Elliot P. (2001).
Outdoor sulphur dioxide and respiratory symptoms in Czech and Polish school children: a small-
area study (SAVIAH). Int Arch Occup Environ Health, 74, 574-578.
Pino P; Walter T; Oyarzun M; Villegas R; Romieu I. (2004). Fine particulate matter and wheezing
illnesses in the first year of life. Epidemiology, 15, 702-708.
Pinter A; Rudnai P; Sarkany E; Goczan M; Paldy A. (1996). Air pollution and childrens' respiratory
morbidity in the Tata area, Hungary. Central European Journal of Public Health, 4, 17-20.
Pinto JP; Turco RP; Toon OB. (1989). Self-limiting physical and chemical effects in volcanic eruption
clouds. J Geophys Res, 94, 11,165-11,174.
Piper JM; Mitchel EF, Jr.; Snowden M; Hall C; Adams M; Taylor P. (1993). Validation of 1989
Tennessee birth certificates using maternal and newborn hospital records. Am J Epidemiol, 137,
758-768.
Plunkett LM; Turnbull D; Rodricks JV. (1992). Differences between adults and children affecting
exposure assessment. In Guzelian, P S; Henry, D J; Olin, S S, eds Similarities and differences
between children and adults: implications for risk assessment Washington, DC: ILSI Press, 79-
96.
Poloniecki JD; Atkinson RW; de Leon AP; Anderson HR. (1997). Daily time series for cardiovascular
hospital admissions and previous day's air pollution in London, UK. Occup Environ Med, 54,
535-540.
Ponce de Leon A; Anderson HR; Bland JM; Strachan DP; Bower J. (1996). Effects of air pollution on
daily hospital admissions for respiratory disease in London between 1987-88 and 1991-92. The
APHEA project. Short term effects of air pollution on health: a European approach using
epidemiological time series data. J Epidemiol Community Health, 50, S63-S70.
R-44
-------
Ponce NA; Hoggatt KJ; Wilhelm M; Ritz B. (2005). Preterm Birth: The Interaction of Traffic-related Air
Pollution with Economic Hardship in Los Angeles Neighborhoods. Am J Epidemiol, 162, MO-
MS.
Ponka A. (1990). Absenteeism and respiratory disease among children and adults in Helsinki in relation
to low-level air pollution and temperature. Environ Res, 52, 34-46.
Ponka A. (1991). Asthma and low level air pollution in Helsinki. Arch Environ Occup Health, 46, 262-
270.
Ponka A; Virtanen M. (1994). Chronic bronchitis, emphysema, and low-level air pollution in Helsinki,
1987-1989. Environ Res, 65, 207-217.
Ponka A; Virtanen M. (1996). Asthma and ambient air pollution in Helsinki. J Epidemiol Community
Health, 50 Suppl 1, s59-62.
Ponka A; Savela M; Virtanen M. (1998). Mortality and air pollution in Helsinki. Arch Environ Health,
53,281-286.
Pool-Zobel BL; Schmezer P; Zeller WJ; Klein RG. (1990). In vitro and ex vivo effects of the air
pollutants S02 and NOx on benzo(a)pyrene activating enzymes of the rat liver. Exp Pathol, 39,
207-212.
Pool BL; Brendler S; Klein RG; Monarca S; Pasquini R; Schmezer P; Zeller WJ. (1988a). Effects of S02
or NOx on toxic and genotoxic properties of chemical carcinogens. II. Short term in vivo studies.
Carcinogenesis, 9, 1247-1252.
Pool BL; Janowsky I; Klein P; Klein RG; Schmezer P; Vogt-Leucht G; Zeller WJ. (1988b). Effects of
S02 or NOx on toxic and genotoxic properties of chemical carcinogens. I. In vitro studies.
Carcinogenesis, 9, 1237-1245.
CA, 3rd. (1989). Respiratory disease associated with community air pollution and a steel mill, Utah
Valley. Am J Public Health, 79, 623-628.
CA, 3rd; Schwartz J; Ransom MR. (1992). Daily mortality and PM10 pollution in Utah Valley.
Arch Environ Health, 47, 211-217.
CA, 3rd; Thun MJ; Namboodiri MM; Dockery DW; Evans JS; Speizer FE; Heath CW, Jr. (1995).
Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am J
Respir Crit Care Med, 151, 669-674.
CA, 3rd; Burnett RT; Thun MJ; Calle EE; Krewski D; Ito K; Thurston GD. (2002). Lung cancer,
cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA, 287,
1132-1141.
Possanzini M; De Santis F; Di Palo V. (1999). Measurements of nitric acid and ammonium salts in lower
Bavaria. Atmos Environ, 33, 3597-3602.
Prescott GJ; Cohen GR; Elton RA; Fowkes FG; Agius RM. (1998). Urban air pollution and
cardiopulmonary ill health: a 14.5 year time series study. Occup Environ Med, 55, 697-704.
Pope
Pope
Pope
Pope
R-45
-------
Price PS; Chaisson CF; Koontz M; Wilkes C; Ryan B; Macintosh D; Georgopoulos PG. (2003).
Construction of a comprehensive chemical exposure framework using person oriented modeling.
Annandale, VA; The Lifeline Group, for The Exposure Technical Implementation Panel,
American Chemistry Council. Available: http://www.thelifelinegroup.org/lifeline/docs.htm [6
March, 2007],
Pszenny AAP; Moldanova J; Keene WC; Sander R; Maben JR; Martinez M; Crutzen PJ; Perner D; Prinn
RG. (2004). Halogen cycling and aerosol pH in the Hawaiian marine boundary layer. Atmos
Chem Phys, 4, 147-168.
Purushothaman V; Georgopoulos PG. (1997). Computational tools to aid the estimation and visualization
of potential human exposure to ozone (Paper #68). In: Proceedings of the Air & Waste
Management Association Specialty Conference on Computing in Environmental Resource
Management, Research Triangle Park, NC. Pittsburgh, PA; Air & Waste Management
Association, pp. 17-28.
Purushothaman V; Georgopoulos PG. (1999a). Integrating photochemical modeling, geostatistical
techniques and geographical information systems for ozone exposure assessment. Report #ORC-
TR99-01. Piscataway, NJ; Ozone Research Center; Environmental and Occupational Health
Sciences Institute.
Purushothaman V; Georgopoulos PG. (1999b). Evaluation of regional emissions control strategies for
ozone utilizing population exposure metrics: a new GIS-based approach applied to the July 1995
episode over the OTAG domain. Report # ORC-TR99-02. Piscataway, NJ ; Ozone Research
Center; Environmental and Occupational Health Sciences Institute.
Qian Z; He Q; Lin HM; Kong L; Liao D; Yang N; Bentley CM; Xu S. (2007). Short-term effects of
gaseous pollutants on cause-specific mortality in Wuhan, China. J Air Waste Manag Assoc, 57,
785-793.
Qin G; Meng Z. (2005). Effect of sulfur dioxide inhalation on CYP1A1 and CYP1A2 in rat liver and
lung. Toxicol Lett, 160, 34-42.
Quist ME; Nasholm T; Lindeberg J; Johannisson C; Hogbom L; Hogberg P. (1999). Responses of a
nitrogen-saturated forest to a sharp decrease in nitrogen input. Journal of environmental quality,
28, 1970-1977.
Rabe KF; Adachi M; Lai CKW; Soriano JB; Vermeire PA; Weiss KB; Weiss ST. (2004). Worldwide
severity and control of asthma in children and adults: the global asthma insights and reality
surveys. The Journal of Allergy and Clinical Immunology, 114, 40-47.
Rahlenbeck SI; Kahl H. (1996). Air pollution and mortality in East Berlin during the winters of 1981 -
1989. International journal of epidemiology, 25, 1220-1226.
Ramadour M; Burel C; Lanteaume A; Vervloet D; Charpin D; Brisse F; Dutau H. (2000). Prevalence of
asthma and rhinitis in relation to long-term exposure to gaseous air pollutants. Allergy, 55, 1163-
1169.
Raub JA; Miller FJ; Graham JA; Gardner DE; O'Neil JJ. (1983). Pulmonary function in normal and
elastase-treated hamsters exposed to a complex mixture of olefin-ozone-sulfur dioxide reaction
products. Environ Res, 31, 302-310.
R-46
-------
Ravishankara AR. (1997). Heterogeneous and multiphase chemistry in the troposphere. Science, 276,
1058-1065.
Reithmeier C; Sausen R. (2002). ATTILA: atmospheric tracer transport in a Lagrangian model. Tellus,
54B, 278-299.
Rich DQ; Schwartz J; Mittleman MA; Link M; Luttmann-Gibson H; Catalano PJ; Speizer FE; Dockery
DW. (2005). Association of short-term ambient air pollution concentrations and ventricular
arrhythmias. Am J Epidemiol, 161, 1123-1132.
Rich DQ; Kim MH; Turner JR; Mittleman MA; Schwartz J; Catalano PJ; Dockery DW. (2006a).
Association of ventricular arrhythmias detected by implantable cardioverter defibrillator and
ambient air pollutants in the St Louis, Missouri metropolitan area. Occup Environ Med, 63, 591-
596.
Rich DQ; Mittleman MA; Link MS; Schwartz J; Luttmann-Gibson H; Catalano PJ; Speizer FE; Gold DR;
Dockery DW. (2006b). Increased risk of paroxysmal atrial fibrillation episodes associated with
acute increases in ambient air pollution. Environ Health Perspect, 114, 120-123.
Rich KE; Petkau J; Vedal S; Brauer M. (2004). A case-crossover analysis of particulate air pollution and
cardiac arrhythmia in patients with implantable cardioverter defibrillators. Inhal Toxicol, 16, 363-
372.
Riechelmann H; Maurer J; Kienast K; Hafner B; Mann WJ. (1995). Respiratory epithelium exposure to
sulfur dioxide—functional and ultrastructural alterations. Laryngoscope, 105, 295-299.
Riedel F; Kramer M; Scheibenbogen C; Rieger CH. (1988). Effects of S02 exposure on allergic
sensitization in the guinea pig. J Allergy Clin Immunol, 82, 527-534.
Robinson JM. (1989). On uncertainty in the computation of global emissions from biomass burning.
Climatic Change, 14, 243-261.
Roemer W; Hoek G; Brunekreef B. (1993). Effect of ambient winter air pollution on respiratory health of
children with chronic respiratory symptoms. Am Rev Respir Dis, 147, 118-124.
Roemer W; Hoek G; Brunekreef B; Haluszka J; Kalandidi A; Pekkanen J. (1998). Daily variations in air
pollution and respiratory health in a multicentre study: the PEACE project. Pollution Effects on
Asthmatic Children in Europe. Eur Respir J, 12, 1354-1361.
Roemer WH; van Wijnen JH. (2001). Daily mortality and air pollution along busy streets in Amsterdam,
1987-1998. Epidemiology, 12, 649-653.
Roger LJ; Kehrl HR; Hazucha M; Horstman DH. (1985). Bronchoconstriction in asthmatics exposed to
sulfur dioxide during repeated exercise. J Appl Physiol, 59, 784-791.
Romieu I; Meneses F; Ruiz S; Sienra JJ; Huerta J; White MC; Etzel RA. (1996). Effects of air pollution
on the respiratory health of asthmatic children living in Mexico City. Am J Respir Crit Care Med,
154, 300-307.
R-47
-------
Romieu I; Ramirez-Aguilar M; Moreno-Macias H; Barraza-Villarreal A; Miller P; Hernandez-Cadena L;
Carbajal-Arroyo LA; Hernandez-Avila M. (2004). Infant mortality and air pollution: modifying
effect by social class. J Occup Environ Med, 46, 1210-1216.
Rosenbaum A. (2005). The HAPEM5 user's guide: hazardous air pollutant exposure model, version 5.San
Francisco, CA; ICF Consulting ; Prepared for Office of Air Quality Planning and Standards;
Research Triangle Park, NC; ;U.S. Environmental Protection Agency: Available:
http://www.epa.gov/ttn/fera/hapem5/hapem5_guide.pdf [12 February, 2008],
Ross MA; Persky VW; Scheff PA; Chung J; Curtis L; Ramakrishnan V; Wadden RA; Hryhorczuk DO.
(2002). Effect of ozone and aeroallergens on the respiratory health of asthmatics. Arch Environ
Occup Health, 57, 568-578.
Rossi OV; Kinnula VL; Tienari J; Huhti E. (1993). Association of severe asthma attacks with weather,
pollen, and air pollutants. Thorax, 48, 244-248.
Rothman KJ; Greenland S. (1998). Modern epidemiology, Second edition. Philadelphia, PA: Lippincott-
Raven Publishers.
Routledge HC; Manney S; Harrison RM; Ayres JG; Townend JN. (2006). Effect of inhaled sulphur
dioxide and carbon particles on heart rate variability and markers of inflammation and
coagulation in human subjects. Heart, 92, 220-227.
Ruan A; Min H; Meng Z; Lu Z. (2003). Protective effects of seabuckthorn seed oil on mouse injury
induced by sulfur dioxide inhalation. Inhal Toxicol, 15, 1053-1058.
Rubinstein I; Bigby BG; Reiss TF; Boushey HAJ. (1990). Short-Term Exposure to 0.3 Ppm Nitrogen
Dioxide Does Not Potentiate Airway Responsiveness to Sulfur Dioxide in Asthmatic Subjects.
American Review of Respiratory Disease, 141, 381-385.
Ruidavets JB; Cassadou S; Cournot M; Bataille V; Meybeck M; Ferrieres J. (2005a). Increased resting
heart rate with pollutants in a population based study. J Epidemiol Community Health, 59, 685-
693.
Ruidavets JB; Cournot M; Cassadou S; Giroux M; Meybeck M; Ferrieres J. (2005b). Ozone air pollution
is associated with acute myocardial infarction. Circulation, 111, 563-569.
Russell A; Dennis R. (2000). NARSTO critical review of photochemical models and modeling. Atmos
Environ, 34, 2283-2324.
Rusznak C; Devalia JL; Davies RJ. (1996). Airway response of asthmatic subjects to inhaled allergen
after exposure to pollutants. Thorax, 51,1105-1108.
Ryerson TB; Buhr MP; Frost GJ; Goldan PD; Holloway JS; Hubler G; Jobson BT; Kuster WC; McKeen
SA; Parrish DD; Roberts JM; Sueper DT; Trainer M; Williams J; Fehsenfeld FC. (1998).
Emissions lifetimes and ozone formation in power plant plumes. J Geophys Res, 103, 22,569-
22,583.
Saez M; Tobias A; Munoz P; Campbell MJ. (1999). A GEE moving average analysis of the relationship
between air pollution and mortality for asthma in Barcelona, Spain. Stat Med, 18, 2077-2086.
R-48
-------
Saez M; Ballester F; Barcelo MA; Perez-Hoyos S; Bellido J; Tenias JM; Ocana R; Figueiras A; Arribas
F; Aragones N; Tobias A; Cirera L; Canada A. (2002). A combined analysis of the short-term
effects of photochemical air pollutants on mortality within the EMECAM project. Environ Health
Perspect, 110, 221-228.
Sagiv SK; Mendola P; Loomis D; Herring AH; Neas LM; Savitz DA; Poole C. (2005). A time-series
analysis of air pollution and preterm birth in Pennsylvania, 1997-2001. Environ Health Perspect,
113, 602-606.
Saiz-Lopez A; Plane JMC; Shillito JA. (2004). Bromine oxide in the mid-latitude marine boundary layer.
Geophys Res Lett, 31, L03111 doi 10.1029/2003GL018956.
Saldiva PH; Pope CA, 3rd; Schwartz J; Dockery DW; Lichtenfels AJ; Salge JM; Barone I; Bohm GM.
(1995). Air pollution and mortality in elderly people: a time-series study in Sao Paulo, Brazil.
Arch Environ Health, 50, 159-163.
Saldiva PHN; King M; Delmonte VLC; Macchione M; Parada MAC; Daliberto ML; Sakae RS; Criado
PMP; Silveira PLP; Zin WA; Bohm GM. (1992). Respiratory alterations due to urban air
pollution: an experimental study in rats. Environ Res, 57, 19-33.
Saldiva PHN; Lichtenfels AJFC; Paiva PSO; Barone IA; Martins MA; Massad E; Pereira JCR; Xavier
VP; Singer JM; Bohm GM. (1994). Association between air pollution and mortality due to
respiratory diseases in children in Sao Paulo, Brazil: a preliminary report. Environ Res, 65, 218-
225.
Samet JM; Dominici F; Zeger SL; Schwartz J; Dockery DW. (2000a). National morbidity, mortality, and
air pollution study. Part I: methods and methodologic issues. Report #94, Part 1. Cambridge, MA;
Health Effects Institute.
Samet JM; Zeger SL; Dominici F; Curriero F; Coursac I; Dockery DW; Schwartz J; Zanobetti A.
(2000b). The national morbidity, mortality, and air pollution study. Part II: morbidity, mortality,
and air pollution in the United States. Report #94, Part 2. Cambridge, MA; Health Effects
Institute.
Samoli E; Schwartz J; Wojtyniak B; Touloumi G; Spix C; Balducci F; Medina S; Rossi G; Sunyer J;
Bacharova L; Anderson HR; Katsouyanni K. (2001). Investigating regional differences in short-
term effects of air pollution on daily mortality in the APHEA project: a sensitivity analysis for
controlling long-term trends and seasonality. Environ Health Perspect, 109, 349-353.
Samoli E; Schwartz J; Analitis A; Petasakis Y; Wojtyniak B; Touloumi G; Spix C; Balducci F; Medina S;
Rossi G; Sunyer J; Anderson HR; Katsouyanni K. (2003). Sensitivity analyses of regional
differences in short-term effects of air pollution on daily mortality in APHEA cities. In: Revised
analyses of time-series studies of air pollution and health. HEI Special Report. Boston, MA:
Health Effects Institute; pp. 205-210. Available:
http://www.healtheffects.org/Pubs/TimeSeries.pdffl8 October, 2004],
Samet JM (1989) Issues in the Selection of Subjects for Clinical and Epidemiological Studies of Asthma.
In: Susceptibility to Inhaled Pollutants, Conference on the Susceptibility to Inhaled Pollutants
Williamsburg, Va. (Utell, MJ and Frank, R, eds). Philadelphia, PA ; American Society for
Testing and Materials pp 57-67
R-49
-------
Sarnat JA; Koutrakis P; Suh HH. (2000). Assessing the relationship between personal particulate and
gaseous exposures of senior citizens living in Baltimore, MD. J Air Waste Manag Assoc, 50,
1184-1198.
Sarnat JA; Schwartz J; Catalano PJ; Suh HH. (2001). Gaseous pollutants in particulate matter
epidemiology: confounders or surrogates? Environ Health Perspect, 109, 1053-1061.
Sarnat JA; Brown KW; Schwartz J; Coull BA; Koutrakis P. (2005). Ambient gas concentrations and
personal particulate matter exposures: implications for studying the health effects of particles.
Epidemiology, 16, 385-395.
Sarnat SE; Coull BA; Schwartz J; Gold DR; Suh HH. (2006a). Factors affecting the association between
ambient concentrations and personal exposures to particles and gases. Environ Health Perspect,
114, 649-654.
Sarnat SE; Suh HH; Coull BA; Schwartz J; Stone PH; Gold DR. (2006b). Ambient particulate air
pollution and cardiac arrhythmia in a panel of older adults in Steubenville, Ohio. Occup Environ
Med, 63, 700-706.
Scala JM; Garstang M; Tao W-K; Pickering KE; Thompson AM; Simpson J; Kirchhoff VWJH; Browell
EV; Sachse GW; L. TA; Gregory GL; Rasmussen RW; Khalil MAK. (1990). Cloud draft
structure and trace gas transport. J Geophys Res, 95, 17,015-17,030.
Scanlon PD; Seltzer J; Ingram RH, Jr.; Reid L; Drazen JM. (1987). Chronic exposure to sulfur dioxide.
Physiologic and histologic evaluation of dogs exposed to 50 or 15 ppm. Am Rev Respir Dis, 135,
831-839.
Schachter EN; Witek TJ, Jr.; Beck GJ; Hosein HB; Colice G; Leaderer BP; Cain W. (1984). Airway
effects of low concentrations of sulfur dioxide: dose-response characteristics. Arch Environ
Health, 39, 34-42.
Schildcrout JS; Sheppard L; Lumley T; Slaughter JC; Koenig JQ; Shapiro GG. (2006). Ambient air
pollution and asthma exacerbations in children: an eight-city analysis. Am J Epidemiol, 164, 505-
517.
Schouten JP; Vonk JM; de Graaf A. (1996). Short term effects of air pollution on emergency hospital
admissions for respiratory disease: results of the APHEA project in two major cities in The
Netherlands, 1977-89. In: The APHEA project. Short term effects of air pollution on health: a
European approach using epidemiological time series data. J Epidemiol Community Health, 50,
S22-S29.
Schwartz J. (1989). Lung function and chronic exposure to air pollution: a cross-sectional analysis of
NHANES II. Environ Res, 50, 309-321.
Schwartz J. (1991). Particulate air pollution and daily mortality in Detroit. Environ Res, 56, 204-213.
Schwartz J; Dockery DW. (1992a). Increased mortality in Philadelphia associated with daily air pollution
concentrations. Am Rev Respir Dis, 145, 600-604.
Schwartz J; Dockery DW. (1992b). Particulate air pollution and daily mortality in Steubenville, Ohio. Am
J Epidemiol, 135, 12-19; discussion 20-15.
R-50
-------
Schwartz J; Dockery DW; Neas LM; Wypij D; Ware JH; Spengler JD; Koutrakis P; Speizer FE; Ferris
BG, Jr. (1994). Acute effects of summer air pollution on respiratory symptom reporting in
children. Am J Respir Crit Care Med, 150, 1234-1242.
Schwartz J. (1995). Short term fluctuations in air pollution and hospital admissions of the elderly for
respiratory disease. Thorax, 50, 531.
Schwartz J; Morris R. (1995). Air pollution and hospital admissions for cardiovascular disease in Detroit,
Michigan. Am J Epidemiol, 142, 23-35.
Schwartz J; Spix C; Touloumi G; Bacharova L; Barumamdzadeh T; le Tertre A; Piekarksi T; Ponce de
Leon A; Ponka A; Rossi G; Saez M; Schouten JP. (1996). Methodological issues in studies of air
pollution and daily counts of deaths or hospital admissions. J Epidemiol Community Health, 50
Suppl 1, S3-11.
Schwartz J. (1997). Air pollution and hospital admissions for cardiovascular disease in Tucson.
Epidemiology, 8, 371-377.
Schwartz J. (2000). Daily deaths are associated with combustion particles rather than SO(2) in
Philadelphia. Occup Environ Med, 57, 692-697.
Schwartz J. (2001). Air pollution and blood markers of cardiovascular risk. Environ Health Perspect, 109
Suppl 3, 405-409.
Schwartz J. (2004). The effects of particulate air pollution on daily deaths: a multi-city case crossover
analysis. Occup Environ Med, 61, 956-961.
Schwartz J; Litonjua A; Suh H; Verrier M; Zanobetti A; Syring M; Nearing B; Verrier R; Stone P;
MacCallum G; Speizer FE; Gold DR. (2005). Traffic related pollution and heart rate variability in
a panel of elderly subjects. Thorax, 60, 455-461.
Seaman NL. (2000). Meteorological modeling for air quality assessments. Atmos Environ, 34, 2231-
2259.
Segala C; Fauroux B; Just J; Pascual L; Grimfeld A; Neukirch F. (1998). Short-term effect of winter air
pollution on respiratory health of asthmatic children in Paris. Eur Respir J, 11, 677-685.
Seinfeld JH; Pandis SN. (1998). Atmospheric Chemistry and Physics: From Air Pollution to Climate
Change. New York, NY; John Wiley.
Shami SG; Wolff RK; Hahn FF; Brooks AL; Griffith WC. (1985). Early cytokinetic and morphological
response of rat lungs to inhaled benzo(a)pyrene, gallium oxide, and S02. Environ Res, 37, 12-25.
Sheppard D; Saisho A; Nadel JA; Boushey HA. (1981). Exercise increases sulfur dioxide-induced
bronchoconstriction in asthmatic subjects. Am Rev Respir Dis, 123, 486-491.
Sheppard L; Levy D; Norris G; Larson TV; Koenig JQ. (1999). Effects of ambient air pollution on
nonelderly asthma hospital admissions in Seattle, Washington, 1987-1994. Epidemiology, 10, 23-
30.
R-51
-------
Sheppard L. (2003). Ambient air pollution and nonelderly asthma hospital admissions in Seattle,
Washington, 1987-1994. In Revised analyses of time-series studies of air pollution and health.
Special report, (pp. 227-230). Boston, MA: Health Effects Institute.
Sheppard L; Slaughter JC; Schildcrout J; Liu LJ; Lumley T. (2005). Exposure and measurement
contributions to estimates of acute air pollution effects. J Expo Anal Environ Epidemiol, 15, 366-
376.
Shi X. (1994). Generation of S03(-) and OH radicals in S03(2-) reactions with inorganic environmental
pollutants and its implications to S03(2-) toxicity. Journal of Inorg Biochemistry, 56, 155-165.
Shi X; Mao Y. (1994). 8-hydroxy-2'-deoxyguanosine formation and DNA damage induced by sulfur
trioxide anion radicals. Biochemical and Biophysical Research Communications, 205, 141-147.
Sillman S. (1995). The use ofNOy, H202 and HN03 as indicators for ozone-NOx-hydrocarbon sensitivity
in urban locations. J Geophys Res, 100, 14,175-14,188.
Sillman S. (2000). Ozone production efficiency and loss of NOx in power plant plumes: photochemical
model and interpretation of measurements in Tennessee. J Geophys Res, 105, 9189-9202.
Sim VM; Pattle RE. (1957). Effect of possible smog irritants on human subjects. J Am Med Assoc, 165,
1908-1913.
Simpson RW; Williams G; Petroeschevsky A; Morgan G; Rutherford S. (1997). Associations between
outdoor air pollution and daily mortality in Brisbane, Australia. Arch Environ Occup Health, 52,
442-454.
Sinclair AH; Tolsma D. (2004). Associations and lags between air pollution and acute respiratory visits in
an ambulatory care setting: 25-month results from the aerosol research and inhalation
epidemiological study. J Air Waste Manag Assoc, 54, 1212-1218.
Singh J. (1989). Neonatal development altered by maternal sulfur dioxide exposure. Neurotoxicology, 10,
523-527.
Smith AM; Keene WC; Maben JR; Pszenny AAP; Fischer E; Stohl A. (2007). Ammonia sources,
transport, transformation, and deposition in coastal New England during summer. J Geophys Res
112(D10S08): 10.1029/2006JD007574.
Smith E. (1994). Subject data supplied by the researchers for the recent controlled human studies
analyzed in the staff paper supplement and accompanying memorandum. Memorandum to
Docket No. A-84-25, Item IV-B-5.
Smith LG; Busch RH; Buschbom RL; Cannon WC; Loscutoff SM; Morris JE. (1989). Effects of sulfur
dioxide or ammonium sulfate exposure, alone or combined, for 4 or 8 months on normal and
elastase-impaired rats. Environ Res, 49, 60-78.
Snashall PD; Baldwin C. (1982). Mechanisms of sulphur dioxide induced bronchoconstriction in normal
and asthmatic man. British Medical Journal, 37, 118.
Snell RE; Luchsinger PC. (1969). Effects of sulfur dioxide on expiratory flow rates and total respiratory
resistance in normal human subjects. Arch Environ Health, 18, 693-698.
R-52
-------
Sole D; Camelo-Nunes IC; Wandalsen GF; Pastorino AC; Jacob CMA; Gonzalez C; Wandalsen NF;
Rosario Filho NA; Fischer GB; Naspitz CK. (2007). Prevalence of symptoms of asthma, rhinitis,
and atopic eczema in Brazilian adolescents related to exposure to gaseous air pollutants and
socioeconomic status. Journal of Investigational Allergology and Clinical Immunology, 17, 6-13.
Soyscth V; Kongerud J; Broen P; Lilleng P; Boe J. (1995a). Bronchial responsiveness, eosinophilia, and
short term exposure to air pollution. Arch Dis Child, 73, 418-422.
Soyseth V; Kongerud J; Haarr D; Strand O; Bolle R; Boe J. (1995b). Relation of exposure to airway
irritants in infancy to prevalence of bronchial hyper-responsiveness in schoolchildren. Lancet,
345,217-220.
Speizer FE; Frank NR. (1966). The uptake and release of S02 by the human nose. Arch Environ Health,
12, 725-728.
Spengler JD; Ferris BG; Dockery DW; Speizer FE. (1979). Sulfur dioxide and nitrogen dioxide levels
inside and outside homes and the implications on health effects research. Environ Sci Technol;
Vol/Issue: 13:10, Pages: 1276-1280.
Spix C; Wichmann HE. (1996). Daily mortality and air pollutants: findings from Koln, Germany. J
Epidemiol Community Health, 1, S52-S58.
Spix C; Anderson HR; Schwartz J; Vigotti MA; LeTertre A; Vonk JM; Touloumi G; Balducci F;
Piekarski T; Bacharova L; Tobias A; Ponka A; Katsouyanni K. (1998). Short-term effects of air
pollution on hospital admissions of respiratory diseases in Europe: a quantitative summary of
APHEA study results. Arch Environ Occup Health, 53, 54-64.
Stacy RW; Seal Jr E; House DE; Green J; Roger LJ; Raggio L. (1983). A survey of effects of gaseous and
aerosol pollutants on pulmonary function of normal males. Arch Environ Health, 38, 104-115.
Stara JF, Dungworth D, Orthoefer JG, Tyler WS,(eds) (1980) Long-term effects of air pollutants in canine
species. Washington, DC, Report No. EPA-600/8-80-014, US Environmental Protection Agency.
Stein AF; Lamb D; Draxler RR. (2000). Incorporation of detailed chemistry into a three-dimensional
Lagrangian-Eulerian hybrid model: application to regional tropospheric ozone. Atmos Environ,
34, 4361-4372.
Stempel DA; Boucher RC. (1981). Respiratory infection and airway reactivity. Med Clin North Am, 65,
1045-1053.
Stenchikov G; Dickerson R; Pickering K; Ellis WJ; Doddridge B; Kondragunta S; Poulida O; Scala J;
Tao W-K. (1996). Stratosphere-troposphere exchange in a midlatitude mesoscale convective
complex. 2. Numerical simulation. J Geophys Res, 101, 6837-6851.
Stevens RK; O'Keeffe AE; Ortman GC. (1969). Absolute calibration of a flame photometric detector to
volatile sulfur compounds at sub-part-per-million levels. Environ Sci Tech, 3, 652-655.
Stevens RK; Mulik JD; O'Keefe AE; Krost KJ. (1971). Gas chromatography of reactive sulfur gases in
air at the parts-per-billion level. Analytical chemistry, 43, 827-831.
R-53
-------
Stevenson DS; Dentener FJ; Schultz MG; Ellingsen K; van Noije TPC; Wild 0; Zeng G; Amann M;
Atherton CS; Bell N; Bergmann DJ; Bey I; Butler T; Cofala J; Collins WJ; Derwent RG; Doherty
RM; Drevet J; Eskes HJ; Fiore AM; Gauss M; Hauglustaine DA; Horowitz LW; Isaksen ISA;
Krol MC; Lamarque JF; Lawrence MG; Montanaro V; Mueller JF; Pitari G; Prather MJ; Pyle JA;
Rast S; Rodriguez JM; Savage NH; Shindell DT; Strahan SE; Sudo K; Szopa S. (2006).
Multimodel ensemble simulations of present-day and near-future tropospheric ozone. J Geophys
Res, 111, D08301.
DM; Burnett RT; Beveridge RC; Brook JR. (1996). Association between ozone and asthma
emergency department visits in Saint John, New Brunswick, Canada. Environ Health Perspect,
104, 1354-1360.
DM; Beveridge RC; Brook JR; Smith-Doiron M; Burnett RT; Dales RE; Beaulieu S; Judek S;
Mamedov A. (2000). Air pollution, aeroallergens and cardiorespiratory emergency department
visits in Saint John, Canada. J Expo Anal Environ Epidemiol, 10, 461-477.
DM; Judek S; Burnett RT. (2002). Meta-analysis of time-series studies of air pollution and
mortality: effects of gases and particles and the influence of cause of death, age, and season. J Air
Waste Manag Assoc, 52, 470-484.
DM; Judek S; Burnett RT. (2003). Meta-analysis of time-series studies of air pollution and
mortality: update in relation to the use of generalized additive models. J Air Waste Manag Assoc,
53,258-261.
Stock TH; Kotchmar DJ; Contant CF; Buffler PA; Holguin AH; Gehan BM; Noel LM. (1985). The
estimation of personal exposures to air pollutants for a community-based study of health effects
in asthmatics—design and results of air monitoring. Journal of the Air and Waste Management
Association, 35, 1266-1273.
Stockwell W; Middleton P; Chang J; Tang X. (1990). The second generation Regional Acid Deposition
Model chemical mechanism for regional air quality modeling. J Geophys Res, 95, 16343-16367.
Stockwell, WR; Kirchner, F; Kuhn, M; Seefeld S. (1997) A new mechanism for regional atmospheric
chemistry modeling. JGeophys Res- Atmospheres 102, 25,847 doi 97D00849.
Strandberg LG. (1964). S02 absorption in the respiratory tract: studies on the absorption in rabbit, its
dependence on concentration and breathing phase. Arch Environ Health, 9, 160-166.
Studnicka M; Hackl E; Pischinger J; Fangmeyer C; Haschke N; Kuhr J; Urbanek R; Neumann M;
Frischer T. (1997). Traffic-related N02 and the prevalence of asthma and respiratory symptoms
in seven year olds. Eur Respir J, 10, 2275-2278.
Suh HH; Spengler JD; Koutrakis P. (1992). Personal exposures to acid aerosols and ammonia. Environ
SciTech, 26, 2507-2517.
Sullivan J; Sheppard L; Schreuder A; IshikawaN; Siscovick D; Kaufman J. (2005). Relation between
short-term fine-particulate matter exposure and onset of myocardial infarction. Epidemiology, 16,
41-48.
Sunyer J; Anto JM; Murillo C; Saez M. (1991). Effects of urban air pollution on emergency room
admissions for chronic obstructive pulmonary disease. Am J Epidemiol, 134, 277-286.
Stieb
Stieb
Stieb
Stieb
R-54
-------
Sunyer J; Saez M; Murillo C; Castellsague J; Martinez F; Anto JM. (1993). Air pollution and emergency
room admissions for chronic obstructive pulmonary disease: a 5-year study. Am J Epidemiol,
137, 701-705.
Sunyer J; Castellsague J; Saez M; Tobias A; Anto JM. (1996). Air pollution and mortality in Barcelona. J
Epidemiol Community Health, 1, S76-S80.
Sunyer J; Spix C; Quenel P; Ponce-de-Leon A; Ponka A; Barumandzadeh T; Touloumi G; Bacharova L;
Wojtyniak B; Vonk J; Bisanti L; Schwartz J; Katsouyanni K. (1997). Urban air pollution and
emergency admissions for asthma in four European cities: the APHEA Project. Thorax, 52, 760-
765.
Sunyer J; Basagana X; Belmonte J; Anto JM. (2002). Effect of nitrogen dioxide and ozone on the risk of
dying in patients with severe asthma. Thorax, 57, 687-693.
Sunyer J; Atkinson R; Ballester F; Le Tertre A; Ayres JG; Forastiere F; Forsberg B; Vonk JM; Bisanti L;
Anderson RH; Schwartz J; Katsouyanni K. (2003a). Respiratory effects of sulphur dioxide: a
hierarchical multicity analysis in the APHEA 2 study. Occup Environ Med, 60, e2.
Sunyer J; Ballester F; Tertre AL; Atkinson R; Ayres JG; Forastiere F; Forsberg B; Vonk JM; Bisanti L;
Tenias JM; Medina S; Schwartz J; Katsouyanni K. (2003b). The association of daily sulfur
dioxide air pollution levels with hospital admissions for cardiovascular diseases in Europe (The
Aphea-II study). Eur Heart J, 24, 752-760.
Sutton MA; Dragosits U; Tang YS; Fowler D. (2000). Ammonia emissions from non-agricultural sources
in the UK. Atmos Environ, 34, 855-869.
Taggart SCO; Custovic A; Francis HC; Faragher EB; Yates CJ; Higgins BG; Woodcock A. (1996).
Asthmatic bronchial hyperresponsiveness varies with ambient levels of summertime air pollution.
EurRespir J, 9, 1146-1154.
Tan WC; Cripps E; Douglas N; Sudlow MF. (1982). Protective effect of drugs on bronchoconstriction
induced by sulphur dioxide. Thorax, 37, 671-675.
Tanaka H; Honma S; Nishi M; Igarashi T; Teramoto S; Nishio F; Abe S. (1998). Acid fog and hospital
visits for asthma: an epidemiological study. Eur Respir J, 11, 1301-1306.
Tanner RL; D'Ottavio T; Garber R; Newman L. (1980). Determination of ambient aerosol sulfur using a
continuous flame photometric detection system. I. Sampling system for aerosol sulfate and
sulfuric acid. Atmos Environ, 14, 121-127.
Tao W-K; Simpson J. (1993). The Goddard Cumulus Ensemble Model. Part I: model description. Terr
Atmos Oceanic Sci (TAO), 4, 35-71.
Tenias JM; Ballester F; Rivera ML. (1998). Association between hospital emergency visits for asthma
and air pollution in Valencia, Spain. Occup Environ Med, 55, 541-547.
Tenias JM; Ballester F; Perez-Hoyos S; Rivera ML. (2002). Air pollution and hospital emergency room
admissions for chronic obstructive pulmonary disease in Valencia, Spain. Arch Environ Health,
57,41-47.
R-55
-------
Thompson AJ; Shields MD; Patterson CC. (2001). Acute asthma exacerbations and air pollutants in
children living in Belfast, Northern Ireland. Arch Environ Health, 56, 234-241.
Thornton DC; Bandy AR. (1993). Sulfur dioxide and dimethyl sulfide in the central Pacific troposphere. J
Atmos Chem, 17, 1-13.
Thornton DCB, A. R. (1999). Sulfur dioxide distribution over the Pacific Ocean 1991-1996. J Geophys
Res [Atmos], 104, 5845-5854.
Thornton DCB, A. R.; Tu, F. H.; Blomquist, B. W.; Mitchell, G. M.; Nadler, W.; Lenschow, D. H.
(2002). Fast airborne sulfur dioxide measurements by atmospheric pressure ionization mass
spectrometry (APIMS). J Geophys Res [Atmos], 107. doi 10.1029/2002JD02289.
Timonen KL; Pekkanen J. (1997). Air pollution and respiratory health among children with asthmatic or
cough symptoms. American Journal of Respiratory & Critical Care Medicine, 156, 546-552.
Tobias A; Campbell MJ; Saez M. (1999). Modelling asthma epidemics on the relationship between air
pollution and asthma emergency visits in Barcelona, Spain. European journal of epidemiology,
15, 799-803.
Tolbert PE; Klein M; Peel JL; Sarnat SE; Sarnat JA. (2007). Multipollutant modeling issues in a study of
ambient air quality and emergency department visits in Atlanta. J Expo Sci Environ Epidemiol,
17, S29-S35.
Toumi R. (1994). BrO as a sink for dimethylsulfide in the marine atmosphere. Geophys Res Lett, 21, 117-
120.
Trenga CA; Koenig JQ; Williams PV. (1999). Sulphur dioxide sensitivity and plasma antioxidants in
adult subjects with asthma. Occup Environ Med, 56, 544-547.
Trenga CA; Koenig JQ; Williams PV. (2001). Dietary antioxidants and ozone-induced bronchial
hyperresponsiveness in adults with asthma. Arch Environ Health, 56, 242-249.
Triche EW; Belanger K; Bracken MB; Beckett WS; Holford TR; Gent JF; McSharry J-E; Leaderer BP.
(2005). Indoor heating sources and respiratory symptoms in nonsmoking women. Epidemiology,
16, 377-384.
Tsai SS; Goggins WB; Chiu HF; Yang CY. (2003a). Evidence for an association between air pollution
and daily stroke admissions in Kaohsiung, Taiwan. Stroke, 34, 2612-2616.
Tsai SS; Huang CH; Goggins WB; Wu TN; Yang CY. (2003b). Relationship between air pollution and
daily mortality in a tropical city: Kaohsiung, Taiwan. J Toxicol Environ Health A, 66, 1341-1349.
Tsai SS; Cheng MH; Chiu HF; Wu TN; Yang CY. (2006). Air pollution and hospital admissions for
asthma in a tropical city: Kaohsiung, Taiwan. Inhal Toxicol, 18, 549-554.
Tunnicliffe WS; Hilton MF; Harrison RM; Ayres JG. (2001). The effect of sulphur dioxide exposure on
indices of heart rate variability in normal and asthmatic adults. Eur Respir J, 17, 604-608.
R-56
-------
Tunnicliffe WS; Harrison RM; Kelly FJ; Dunster C; Ayres JG. (2003). The effect of sulphurous air
pollutant exposures on symptoms, lung function, exhaled nitric oxide, and nasal epithelial lining
fluid antioxidant concentrations in normal and asthmatic adults. Occup Environ Med, 60, el5.
Turco RP; Toon OB; Whitten RC; Hamill P; Keesee RG. (1983). The 1980 eruptions of Mount St.
Helens: physical and chemical processes in the stratospheric clouds. J Geophys Res, 88, 5299-
5319.
Turekian VC; Macko SA; Keene WC. (2001). Application of stable sulfur isotopes to differentiate
sources of size-resolved particulate sulfate in polluted marine air at Bermuda during spring.
Geophys Res Lett, 28, 1491-1494.
U.S. Code (2003). Clean Air Act, Section 109, national ambient air quality standards. U.S.C. 42: section
7409.
U.S. EPA. (1982). Air quality criteria for particulate matter and sulfur oxides. Report #EPA-600/8-82-
029aF-cF. Research Triangle Park, NC; Environmental Criteria and Assessment Office; Office of
Health and Environmental Assessment; U.S. Environmental Protection Agency.
U.S. EPA. (1986a). Review of the national ambient air quality standards for sulfur oxides: updated
assessment of scientific and technical information, addendum to the 1982 OAQPS staff paper.
Report # EPA/450/5-86-13. Research Triangle Park, NC; Strategies and Air Standards Division;
Office of Air Quality Planning and Standards; U.S. Environmental Protection Agency.
U.S. EPA. (1986b). Second addendum to air quality criteria for particulate matter and sulfur oxides
(1982): assessment of newly available health effects information. Report #EPA-600/8-86-020F
Research Triangle Park, NC; Environmental Criteria and Assessment Office; Office of Health
and Environmental Assessment; U.S. Envrionmental Protection Agency
U.S. EPA. (1992) Final guidelines for exposure assessment. Report #EPA/600/Z-92/001. Federal
Register 57(104):22888-22938. Washington, DC; Risk Assessment Forum; U.S.
Environmental Protection Agency.
U.S. EPA. (1994a) Methods for derivation of inhalation reference concentrations and application of
inhalation dosimetry. (No. EPA/600/8-90/066F). Research Triangle Park, NC; Office of Research
and Development; U.S. Environmental Protection Agency.
U.S. EPA. (1994b). Review of the national ambient air quality standards for sulfur oxides: assessment of
scientific and technical information, supplement to the 1986 OAQPS staff paper addendum.
Report # EPA/452/R-94/013. Research Triangle Park, NC; Air Quality Management Division;
Office of Air Quality Planning and Standards; U.S. Environmental Protection Agency
U.S. EPA. (1994c). Supplement to the second addendum (1986) to air quality criteria for particulate
matter and sulfur oxides (1982) (No. EPA-600/FP-93/002). Research Triangle Park, NC: Office
of Health and Environmental Assessment, Environmental Criteria and Assessment Office, U.S.
Environmental Protection Agency.
U.S. EPA. (1996d). Air quality criteria for particulate matter. Report # EPA/600/P-95/001aF-cF.
Research Triangle Park, NC: National Center for Environmental Assessment-RTP Office; Office
of Research and Development, U.S. Environmental Protection Agency.
R-57
-------
U.S. EPA. (1997a). Exposure factors handbook. Report # EPA/600/P-95/002Fa-c . Washington, DC;
National Center for Environmental Assessment; Office of Research and Development; U.S.
Environmental Protection Agency.
U.S. EPA. (1997b). National air pollutant emission trends 1990-1996. Report# EPA/454/R-97/011.
Reseach Triangle Park, NC; Office of Air Quality Planning and Standards; U.S. Environmental
Protection Agency
U.S. EPA. (2004). Air quality criteria for particulate matter. Report # EPA/600/P-99/002aF-bF. Research
Triangle Park, NC;: National Center for Environmental Assessment, Office of Research and
Development; U.S. Environmental Protection Agency.
U.S. EPA. (2005a). Guidelines for Carcinogen Risk Assessment.; Report EPA/630/P-03/001F.
Washington, DC; U.S. Environmental Protection Agency.
U.S. EPA. (2005b). Technical assistance document (TAD) for precursor gas measurements in the NCore
multi-pollutant monitoring network. Version 4. Report#EPA-454/R-05-003Research Triangle
Park, NC; Office of Air Quality Planning and Standards; U.S. Environmental Protection Agency.
U.S. EPA. (2006a). Acid rain program (ARP) 2005 progress report. Report # EPA-430-R-06-015.
Washington, DC: Office of Air and Radiation, Clean Air Markets Division; U.S. Environmental
Protection Agency, http://www.epa.gov/airmarkets/progress/docs/2005report.pdf.
U.S. EPA. (2006b). Air Quality Criteria for Ozone and Related Photochemical Oxidants. U.S.
Environmental Protection Agency, Washington, DC; EPA600/R-05/004aF.
U.S. EPA. (2006dc). Technology Transfer Network clearinghoU.S.e for inventories and emission factors.
Washington DC; U.S. Environmental Protection Agency.
http://www.epa.gov/ttn/chief/net/2002inventory.html
U.S. EPA. (2006d). Total risk integrated methodology (TRIM) air pollutants exposure model
documentation (TRIM.Expo/APEX, version 4). Volume II: technical support document. Research
Triangle Park, NC; U.S. Environmental Protection Agency.
U.S. EPA. (2007). S02 NAAQS Review Plan - Final. Research Triangle Park, NC; National Center for
Environmental Assessment; U.S. Environmental Protection Agency,
http ://nsdi .epa.gov/ttn/naaqs/standards/SO2/data/SO2_review_plan_draft_4-17-07 .pdf
U.S. Supreme Court. (2001). Whitman v. American Trucking Association. Volume 531, US 457, 99-1426
U.S. Surgeon General (1964). Smoking and Health. Report. Washington ,DC; US Department of Health,
Education and Welfare.
Van der Werf GR; Randerson IT; Collatz I; Giglio L. (2003). Carbon emissions from fires in tropical and
subtropical ecosystems. Glob Change Biol, 9, 547-562.
van der Zee SC; Hoek G; Boezen HM; Schouten IP; van Wijnen IH; Brunekreef B. (1999). Acute effects
of urban air pollution on respiratory health of children with and without chronic respiratory
symptoms. Occupational & Environmental Medicine, 56, 802-813.
R-58
-------
van der Zee SC; Hoek G; Boezen MH; Schouten JP; van Wijnen JH; Brunekreef B. (2000). Acute effects
of air pollution on respiratory health of 50-70 yr old adults. Eur Respir J, 15, 700-709.
van Noije TPC; Eskes HJ; Dentener FJ; Stevenson DS; Ellingsen K; Schultz MG; Wild O; Amann M;
Atherton CS; Bergmann DJ; Bey I; Boersma KF; Butler T; Cofala J; Drevet J; Fiore AM; Gauss
M; Hauglustaine DA; Horowitz LW; Isaksen ISA; Krol MC; Lamarque JF; Lawrence MG;
Martin RV; Montanaro V; MA%ller JF; Pitari G; Prather MJ; Pyle JA; Richter A; Rodriguez JM;
Savage NH; Strahan SE; Sudo K; Szopa S; van Roozendael M. (2006). Multi-model ensemble
simulations of tropospheric N02 compared with GOME retrievals for the year 2000. Atmos
Chem Phys, 6, 2943-2979.
Vedal S; Brauer M; White R; Petkau J. (2003). Air pollution and daily mortality in a city with low levels
of pollution. Environ Health Perspect, 111, 45-52.
Vedal S; Rich K; Brauer M; White R; Petkau J. (2004). Air pollution and cardiac arrhythmias in patients
with implantable cardioverter defibrillators. Inhal Toxicol, 16, 353-362.
Vedal SR, K.; Brauer, M.; White, R.; Petkau, J. (2004). Air pollution and cardiac arrhythmias in patients
with implantable cardiovascular defibrillators. Inhalation Toxicol, 16, 353-362.
Venners SA; Wang B; Xu Z; Schlatter Y; Wang L; Xu X. (2003). Particulate matter, sulfur dioxide, and
daily mortality in Chongqing, China. Environ Health Perspect, 111, 562-567.
Verhoeff AP; Hoek G; Schwartz J; van Wijnen JH. (1996). Air pollution and daily mortality in
Amsterdam. Epidemiology, 7, 225-230.
Vigotti MA; Rossi G; Bisanti L; Zanobetti A; Schwartz J. (1996). Short term effects of urban air pollution
on respiratory health in Milan, Italy, 1980-89. J Epidemiol Community Health, 50 Suppl 1, s71-
75.
Villeneuve PJ; Burnett RT; Shi Y; Krewski D; Goldberg MS; Hertzman C; Chen Y; Brook J. (2003). A
time-series study of air pollution, socioeconomic status, and mortality in Vancouver, Canada. J
Expo Anal Environ Epidemiol, 13, 427-435.
Villeneuve PJ; Chen L; Stieb D; Rowe B. (2006a). Associations between outdoor air pollution and
emergency department visits for stroke in Edmonton, Canada. European Journal of
Epidemiology, 21, 689-700.
Villeneuve PJ; Doiron MS; Stieb D; Dales R; Burnett RT; Dugandzic R. (2006b). Is outdoor air pollution
associated with physician visits for allergic rhinitis among the elderly in Toronto, Canada?
Allergy, 61, 750-758.
Vogt R; Crutzen PJ; Sander R. (1996). A mechanism for halogen release from sea-salt aerosol in the
remote marine boundary layer. Nature, 383, 327-330.
Von Glasow S, R.; Bott A; Crutzen PJ. (2002). Modeling halogen chemistry in the marine boundary
layer. 2. Interactions with sulfur and cloud-covered MBL. J Geophys Res, 107.
Von Mutius E; Sherrill DL; Fritzsch C; Martinez FD; Lebowitz MD. (1995). Air pollution and upper
respiratory symptoms in children from East Germany. Eur Respir J, 8, 723-728.
R-59
-------
Vutukuru S; Dabdub D. (2008). Modeling the effects of ship emissions on coastal air quality: A case
study of southern California. Atmos Environ, In Press, Corrected Proof.
Walters S; Griffiths RK; Ayres JG. (1994). Temporal association between hospital admissions for asthma
in Birmingham and ambient levels of sulphur dioxide and smoke. Thorax, 49, 133-140.
Wang AL; Blackford TL; Lee LY. (1996). Vagal bronchopulmonary C-fibers and acute ventilatory
response to inhaled irritants. Respir Physiol, 104, 231-239.
Wang C; Corbett JJ; Firestone J. (2007). Modeling Energy Use and Emissions from North American
Shipping: Application of the Ship Traffic, Energy, and Environment Model. Environ Sci Technol,
41, 3226-3232.
Wang LH; Milford JB; Carter WPL. (2000a). Reactivity estimates for aromatic compounds. Part 2.
uncertainty in incremental reactivities. Atmos Environ, 4349-4360.
Wang LH; Milford JB; Carter WPL. (2000b). Reactivity estimates for aromatic compounds. Part I:
uncertainty in chamber-derived parameters. Atmos Environ, 34, 4337-4348.
Wang TN; Ko YC; Chao YY; Huang CC; Lin RS. (1999). Association between indoor and outdoor air
pollution and adolescent asthma from 1995 to 1996 in Taiwan. Environ Res, 81, 239-247.
Wang X; Ding H; Ryan L; Xu X. (1997). Association between air pollution and low birth weight: a
community-based study. Environ Health Perspect, 105, 514-520.
Ward DJ; Roberts KT; Jones N; Harrison RM; Ayres JG; Hussain S; Walters S. (2002a). Effects of daily
variation in outdoor particulates and ambient acid species in normal and asthmatic children.
Thorax, 57, 489-502.
Ward DJ; Roberts KT; Jones N; Harrison RM; Ayres JG; Hussain S; Walters S. (2002b). Effects of daily
variation in outdoor particulates and ambient acid species in normal and asthmatic children
(correction). Thorax, 57, 752.
Ware JH; Ferris BG, Jr.; Dockery DW; Spengler JD; Stram DO; Speizer FE. (1986). Effects of ambient
sulfur oxides and suspended particles on respiratory health of preadolescent children. Am Rev
Respir Dis, 133, 834-842.
Warneck P. (1999). The relative importance of various pathways for the oxidation of sulfur dioxide and
nitrogen dioxide in sunlit continental fair weather clouds. Phys Chem Chem Phys, 1, 5471-5483.
Wellenius GA; Bateson TF; Mittleman MA; Schwartz J. (2005a). Particulate air pollution and the rate of
hospitalization for congestive heart failure among medicare beneficiaries in Pittsburgh,
Pennsylvania. Am J Epidemiol, 161, 1030-1036.
Wellenius GA; Schwartz J; Mittleman MA. (2005b). Air pollution and hospital admissions for ischemic
and hemorrhagic stroke among medicare beneficiaries. Stroke, 36, 2549-2553.
Wesely ML; Hicks BB. (2000). A review of the current status of knowledge on dry deposition. Atmos
Environ, 34, 2261-2282.
R-60
-------
Wheeler A; Zanobetti A; Gold DR; Schwartz J; Stone P; Suh HH. (2006). The relationship between
ambient air pollution and heart rate variability differs for individuals with heart and pulmonary
disease. Environ Health Perspect, 114, 560-566.
Whitfield RG; Richmond HM; Johnson TR. (1997). Overview of ozone human exposure and health risk
analyses used in the U.S. EPA's review of the ozone air quality standard. In: Proceedings of the
U.S.-Dutch International Symposium on Air Pollution in the 21st Century; April; Noordwijk, The
Netherlands. Argonne, IL: Argonne National Laboratory; ANL/DIS/CP-92660. Available:
http://www.osti.gov/dublincore/gpo/servlets/purl/9748-Nh3avx/webviewable/9748.pdf (22 July
2003).
WHO. (2005). Principles of characterizing and applying human exposure models. (IPCS harmonization
project document #3). Geneva, Switzerland: IPCS Project on the Harmonization of Approaches to
the Assessment of Risk from Exposure to Chemicals. WHO (World Health Organization
Available: http://whqlibdoc.who.int/publications/2005/9241563117_eng.pdff7 March, 2007],
Widdicombe J; Lee LY. (2001). Airway reflexes, autonomic function, and cardiovascular responses.
Environ Health Perspect, 109 Suppl 4, 579-584.
Widdicombe J. (2006). Reflexes from the lungs and airways: historical perspective. J Appl Physiol, 101,
628.
Widdicombe JG. (2003). Overview of neural pathways in allergy and asthma. Pulm Pharmacol Ther, 16,
23-30.
Wiley JA; Robinson JP; Cheng Y-T; Piazza T; Stork L; Pladsen K. (1991a). Study of children's activity
patterns: final report. Report # ARB-R-93/489. Sacramento, CA; California Air Resources Board.
Wiley JA; Robinson JP; Piazza T; Garrett K; Cirksena K; Cheng Y-T; Martin G. (1991b). Activity
patterns of California residents. Final report. Report #ARB/R93/487. Sacramento, CA; California
Air Resources Board.
Willis A; Jerrett M; Burnett RT; Krewski D. (2003). The association between sulfate air pollution and
mortality at the county scale: an exploration of the impact of scale on a long-term exposure study.
J Toxicol Environ Health A, 66, 1605-1624.
Wilson AM; Wake CP; Kelly T; Salloway JC. (2005). Air pollution, weather, and respiratory emergency
room visits in two northern New England cities: an ecological time-series study. Environ Res, 97,
312-321.
Wilson WE; Brauer M. (2006). Estimation of ambient and non-ambient components of particulate matter
exposure from a personal monitoring panel study. J Expo Sci Environ Epidemiol, 16, 264-274.
Winberry WT, Jr.; Ellestad T; Stevens B. (1999). Compendium method 10-4.2: determination of reactive
acidic and basic gases and strong acidity of atmospheric fine particles (< 2.5 mum). Report
#EPA/625/R-96/010a. Cinncinati, OH; Center for Environmental Research Information; Office of
Research and Development, U.S. Environmental Protection Agency. Available:
http://www.epa.gov/ttn/amtic/files/ambient/inorganic/mthd-4-2.pdf [2 May, 2007],
R-61
-------
Winterton DL; Kaufman J; Keener CV; Quigley S; Farin FM; Williams PV; Koenig JQ. (2001). Genetic
polymorphisms as biomarkers of sensitivity to inhaled sulfur dioxide in subjects with asthma.
Ann Allergy Asthma Immunol, 86, 232-238.
Wolff RK; Griffith WC; Henderson RF; Hahn FF; Harkema JR; Rebar AH; Eidson AF; McClellan RO.
(1989). Effects of repeated inhalation exposures to 1-nitropyrene, benzo[a]pyrene, Ga203
particles, and S02 alone and in combinations on particle clearance, bronchoalveolar lavage fluid
composition, and histopathology. J Toxicol Environ Health, 27, 123-138.
Wong CM; Atkinson RW; Anderson HR; Hedley AJ; Ma S; Chau PY-K; Lam T-H. (2002a). A tale of
two cities: effects of air pollution on hospital admissions in Hong Kong and London compared.
Environ Health Perspect, 110, 67-77.
Wong CM; Ma S; Hedley AJ; Lam T. (1999a). Does ozone have any effect on daily hospital admissions
for circulatory diseases? J Epidemiol Community Health 1999 53: 9, 580-581 5 ref.
Wong CM; Ma S; Hedley AJ; Lam TH. (2001a). Effect of air pollution on daily mortality in Hong Kong.
Environ Health Perspect, 109, 335-340.
Wong GW; Ko FW; Lau TS; Li ST; Hui D; Pang SW; Leung R; Fok TF; Lai CK. (2001b). Temporal
relationship between air pollution and hospital admissions for asthmatic children in Hong Kong.
Clin Exp Allergy, 31, 565-569.
Wong TW; Lau TS; Yu TS; Neller A; Wong SL; Tam W; Pang SW. (1999b). Air pollution and hospital
admissions for respiratory and cardiovascular diseases in Hong Kong. Occup Environ Med, 56,
679-683.
Wong TW; Tam WS; Yu TS; Wong AH. (2002a). Associations between daily mortalities from
respiratory and cardiovascular diseases and air pollution in Hong Kong, China. Occup Environ
Med, 59, 30-35.
Wong TW; Wun YT; Yu TS; Tam W; Wong CM; Wong AH. (2002b). Air pollution and general practice
consultations for respiratory illnesses. J Epidemiol Community Health, 56, 949-950.
Woodruff TJ; Parker JD; Kyle AD; Schoendorf KC. (2003). Disparities in exposure to air pollution
during pregnancy. Environ Health Perspect, 111, 942-946.
Wright AL; Hons FM; Lemon RG; McFarland ML; Nichols RL. (2007). Stratification of nutrients in soil
for different tillage regimes and cotton rotations. Soil & Tillage Research, 96, 19-27.
Wu D; Meng Z. (2003). Effect of sulfur dioxide inhalation on the glutathione redox system in mice and
protective role of sea buckthorn seed oil. Arch Environ Contain Toxicol, 45, 423-428.
Xu X; Ding H; Wang X. (1995). Acute effects of total suspended particles and sulfur dioxides on preterm
delivery: a community-based cohort study. Arch Environ Health, 50, 407-415.
Xu XP; Dockery DW; Wang LH. (1991). Effects of air pollution on adult pulmonary function. Arch
Environ Health, 46, 198-206.
R-62
-------
Yallop D; Duncan ER; Norris E; Fuller GW; Thomas N; Walters J; Dick MC; Height SE; Thein SL; Rees
DC. (2007). The associations between air quality and the number of hospital admissions for acute
pain and sickle-cell disease in an urban environment. Br J Haematol, 136, 844-848.
Yang CY; Tseng YT; Chang CC. (2003a). Effects of air pollution on birth weight among children born
between 1995 and 1997 in Kaohsiung, Taiwan. J Toxicol Environ Health A, 66, 807-816.
Yang CY; Chang CC; Chuang HY; Tsai SS; Wu TN; Ho CK. (2004a). Relationship between air pollution
and daily mortality in a subtropical city: Taipei, Taiwan. Environ Int, 30, 519-523.
Yang CY; Chen YS; Yang CH; Ho SC. (2004b). Relationship between ambient air pollution and hospital
admissions for cardiovascular diseases in Kaohsiung, Taiwan. J Toxicol Environ Health A, 67,
483-493.
Yang CY; Chen CJ. (2007). Air pollution and hospital admissions for chronic obstructive pulmonary
disease in a subtropical city: Taipei, Taiwan. J Toxicol Environ Health A, 70, 1214-1219.
Yang Q; Chen Y; Shi Y; Burnett RT; McGrail KM; Krewski D. (2003b). Association between ozone and
respiratory admissions among children and the elderly in Vancouver, Canada. Inhal Toxicol, 15,
1297-1308.
Yang Q; Chen Y; Krewski D; Burnett RT; Shi Y; McGrail KM. (2005). Effect of short-term exposure to
low levels of gaseous pollutants on chronic obstructive pulmonary disease hospitalizations.
Environ Res, 99, 99-105.
Yargicoglu P; Agar A; Gumuslu S; Bilmen S; Oguz Y. (1999). Age-related alterations in antioxidant
enzymes, lipid peroxide levels, and somatosensory-evoked potentials: effect of sulfur dioxide.
Arch Environ Contam Toxicol, 37, 554-560.
Yargicoglu P; Gumusluoriob; Agar A; Korgun DK; KucukatayVS. (2001). Effect of sulfur dioxide
inhalation on erythrocyte antioxidant status, food intake, and lipid peroxidation during aging.
Arch Envion Health 56, 53-57.
Yildirim Z; Kilic T; Koksal N; Kotuk M. (2005). Protective effect of ipratropium bromide on
bronchoconstriction induced by sulfur dioxide exposure during apricot sufurization processes that
causes asthma-like syndrome in agricultural environment. Pharmacol Res, 51, 479-482.
Zanis P; Trickl T; Stohl A; Wernli H; Cooper O; Zerefos C; Gaeggeler H; Schnabel C; Tobler L; Kubik
PW; Priller A; Scheel HE; KanterHJ; Cristofanelli P; Forster C; James P; Gerasopoulos E;
Delcloo A; Papayannis A; Claude H. (2003). Forecast, observation and modelling of a deep
stratospheric intrusion event over Europe. Atmos Chem Phys, 3, 763-777.
Zanobetti A; Schwartz J; Gold D. (2000). Are there sensitive subgroups for the effects of airborne
particles? Environ Health Perspect, 108, 841-845.
Zanobetti A; Schwartz J. (2001). Are diabetics more susceptible to the health effects of airborne particles?
Am J Respir Crit Care Med, 164, 831-833.
Zanobetti A; Schwartz J. (2002). Cardiovascular damage by airborne particles: are diabetics more
susceptible? Epidemiology, 13, 588-592.
R-63
-------
Zeger SL; Thomas D; Dominici F; Samet JM; Schwartz J; Dockery D; Cohen A. (2000). Exposure
measurement error in time-series studies of air pollution: concepts and consequences. Environ
Health Perspect, 108, 419-426.
Zeghnoun A; Czernichow P; Beaudeau P; Hautemaniere A; Froment L; Le Tertre A; Quenel P. (2001).
Short-term effects of air pollution on mortality in the cities of Rouen and Le Havre, France, 1990-
1995. Arch Environ Health, 56, 327-335.
Zhang JZ; Millero FJ. (1991). The rate of sulfite oxidation in seawater. Geochimica et Cosmochimica
Acta, 55, 677-685.
Zmirou D; Barumandzadeh T; Balducci F; Ritter P; Laham G; Ghilardi J-P. (1996). Short term effects of
air pollution on mortality in the city of Lyon, France, 1985-90. J Epidemiol Community Health,
1, S30-S35.
Zmirou D; Schwartz J; Saez M; Zanobetti A; Wojtyniak B; Touloumi G; Spix C; Ponce de Leon A; Le
Moullec Y; Bacharova L; Schouten J; Ponka A; Katsouyanni K. (1998). Time-series analysis of
air pollution and cause-specific mortality. Epidemiology, 9, 495-503.
R-64
------- |