vvEPA
United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
_EP_A/600/AP-95/001c
April 1995
External Review Draft
Air Quality
Criteria for
Particulate
Matter
Volume III of III
Review
Draft
(Do Not
Cite or
Quote)
Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
-------
DRAFT-DO NOT QUOTE OR CITE EPA/eoo/AP-95/ooic
April 1995
External Review Draft
Air Quality Criteria for
P articulate Matter
Volume III of
NOTICE
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Printed on Recycled Paper
-------
DISCLAIMER
This document is an external draft for review purposes only and does not constitute
U.S. Environmental Protection Agency policy. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
April 1995 III-ii DRAFT-DO NOT QUOTE OR CITE
-------
Air Quality Criteria for Particulate Matter
TABLE OF CONTENTS
Volume I
1. EXECUTIVE SUMMARY 1-1
2. INTRODUCTION 2-1
3. PHYSICS AND CHEMISTRY OF PARTICULATE MATTER 3-1
4. SAMPLING AND ANALYSIS OF PARTICULATE MATTER AND
ACID DEPOSITION : 4-1
5. SOURCES AND EMISSIONS OF SUSPENDED PARTICLES 5-1
6. ENVIRONMENTAL CONCENTRATIONS 6-1
Appendix 6A: Tables of Chemical Composition of PM 6A-1
7. EXPOSURE: AMBIENT AND INDOOR 7-1
Volume II
8. EFFECTS ON VISIBILITY AND CLIMATE 8-1
9. EFFECTS ON MATERIALS 9-1
10. DOSIMETRY OF INHALED PARTICLES IN THE
RESPIRATORY TRACT 10-1
11. TOXICOLOGY OF PARTICULATE MATTER CONSTITUENTS 11-1
Volume III
12. EPIDEMIOLOGY STUDIES OF HEALTH EFFECTS ASSOCIATED
WITH EXPOSURE TO AIRBORNE PARTICLES/ACID
AEROSOLS 12-1
Appendix 12A: Effects of Weather and Climate on Human Mortality and Their
Roles as Confounding Factors for Air Pollution 12A-1
13. INTEGRATIVE HEALTH SYNTHESIS 13-1
April 1995 III-iii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS
LIST OF TABLES ffl-ix
LIST OF FIGURES IH-xiii
AUTHORS, CONTRIBUTORS, AND REVIEWERS III-xvii
12. EPIDEMIOLOGY STUDIES OF HEALTH EFFECTS ASSOCIATED
WITH EXPOSURE TO AIRBORNE PARTICLES/ACID AEROSOLS 12-1
12.1 INTRODUCTION 12-1
12.1.1 Definition of Paniculate Matter and Measurement Methods .... 12-1
12.1.2 Epidemiologic Designs and Strategies 12-3
12.1.3 Guidelines for Assessment of Epidemiological Studies 12-5
12.2 METHODOLOGICAL CONSIDERATIONS 12-8
12.2.1 Exposure Measurement Error 12-9
12.2.2 Misclassification of Health Outcomes 12-11
12.2.3 Model Specification for Short-Term Exposures 12-11
12.2.4 Model Specification for Long-Term Exposures 12-19
12.2.5 Covariates and Confounders 12-25
12.2.6 Selection Bias 12-37
12.2.7 Publication Bias 12-38
12.2.8 Internal Consistency and Strength of Effects 12-38
12.2.9 Plausibility of Effects 12-40
12.2.10 Qualitative Versus Quantitative Studies 12-40
12.3 HUMAN HEALTH EFFECTS ASSOCIATED WITH SHORT-TERM
PM EXPOSURE 12-41
12.3.1 Mortality Effects Associated with Short-Term Particulate
Matter Exposures 12-44
12.3.1.1 Review of Short-Term Exposure Studies 12-46
12.3.1.2 Short-Term PM10 Exposure Associations with
Total Daily Mortality: Syntheses of Studies 12-70
12.3.1.3 Short-Term PM10 Exposure Associations with
Daily Mortality in Elderly Adults 12-75
12.3.1.4 Short-Term PM10 Exposure Associations with
Daily Mortality in Children 12-77
12.3.1.5 Short-Term PM10 Exposure Associations with
Daily Mortality in Other Susceptible Subgroups 12-78
12.3.2 Morbidity Effects of Short-Term PM Exposure 12-79
12.3.2.1 Hospitalization and Emergency Visit Studies 12-79
12.3.2.2 Respiratory Illness Studies 12-98
12.3.2.3 Pulmonary Function Studies 12-114
12.4 HEALTH EFFECTS FROM LONG-TERM EXPOSURE TO PM 12-129
12.4.1 Mortality Effects of Long-Term PM Exposures 12-129
12.4.1.1 Methodological Considerations 12-133
12.4.1.2 Population-Based Cross-Sectional Mortality
Studies 12-138
April 1995 HI-v DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
12.4.1.3 Prospective Mortality Studies 12-152
12.4.1.4 Summary of Long-Term Studies 12-168
12.4.2 Morbidity Effects of Long-Term PM Exposure 12-179
12.4.2.1 Respiratory Illness Studies 12-179
12.4.2.2 Pulmonary Function Studies 12-189
12.5 HUMAN HEALTH EFFECTS ASSOCIATED WITH ACID
AEROSOL EXPOSURE 12-194
12.5.1 Historical Evidence Evaluating the Relationship Between
Acid Aerosols and Health Effects 12-196
12.5.1.1 Meuse Valley 12-197
12.5.1.2 Donora 12-197
12.5.1.3 London Acid Aerosol Fogs 12-198
12.5.2 Quantitative Analysis of Earlier Acid Aerosol Studies 12-201
12.5.2.1 London Acute Mortality and Daily Acid Aerosol
Measurements 12-201
12.5.3 Studies Relating Acute Health Effects to Sulfates 12-205
12.5.3.1 Canadian Hospital Admissions Related to Sulfate
Acute Exposure Studies 12-206
12.5.3.2 Other Health Effects Related to Sulfate
Exposures 12-209
12.5.3.3 Studies Relating Acute Health Effects to Acidic
Aerosols 12-210
12.5.3.4 Acute Acidic Aerosol Exposure Studies of
Children 12-211
12.5.3.5 Acute Acid Aerosol Exposure Studies of Adults .. 12-218
12.5.4 Studies Relating Health Effects to Long-Term Exposure .... 12-228
12.5.4.1 Acid Mists Exposure in Japan 12-228
12.5.4.2 Studies Relating Chronic Health Effects to
Sulfate Exposures 12-229
12.5.4.3 Studies Relating Chronic Health Effects to Acid
Aerosols 12-237
12.5.4.4 Chronic Exposure Effects in Occupational
Studies 12-242
12.5.5 Summary of Studies on Acid Aerosols 12-244
12.6 DISCUSSION 12-246
12.6.1 Introduction and Basis for Study Evaluation 12-246
12.6.1.1 Differences Among Study Results 12-247
12.6.1.2 Importance of Comparisons Across Different
Cities 12-250
12.6.2 Sensitivity of PM Effects to Model Specification in
Individual Studies 12-251
12.6.2.1 Model Specification for Acute Mortality Studies .. 12-251
12.6.2.2 Model Specification for Morbidity Studies 12-279
12.6.2.3 Model Specification Issues: Conclusions 12-280
April 1995 HI-vi DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
12.6.3 Other Methodological Issues for Epidemiology Studies 12-280
12.6.3.1 Paniculate Matter Exposure Characterization .... 12-281
12.6.3.2 Exposure-Response Functions, Including
Thresholds 12-284
12.6.3.3 Adjustments for Seasonality, Time Lags, and
Correlation Structure 12-287
12.6.3.4 Adjustments for Meteorological Variables and
Other Confounders 12-290
12.6.3.5 Adjustments for Co-pollutants 12-295
12.6.3.6 Ecological Study Design 12-298
12.6.3.7 Measurement Error 12-299
12.6.4 Assessment Issues for Epidemiology Studies 12-300
12.6.4.1 Significance of Health Effects/Relevancy 12-300
12.6.4.2 Biological Mechanisms 12-302
12.6.4.3 Coherence 12-303
12.6.5 Meta-Analyses and Other Methods for Synthesis of Studies . . 12-305
12.6.5.1 Background 12-305
12.6.5.2 Meta-Analyses Using Studies Reviewed in This
Document 12-307
12.6.5.3 Discussion 12-312
12.7 SUMMARY AND CONCLUSIONS 12-314
REFERENCES 12-324
APPENDIX 12A: Effects of Weather and Climate on Human Mortality
and Their Roles as Confounding Factors for Air
Pollution 12A-1
13. INTEGRATIVE SYNTHESIS: KEY POINTS REGARDING PM EXPOSURE,
DOSIMETRY, AND HEALTH RISKS 13-1
13.1 INTRODUCTION 13-1
13.2 CHEMISTRY AND PHYSICS OF ATMOSPHERIC PARTICLES 13-2
13.2.1 Size Characteristics 13-2
13.2.2 Chemical Characterization of PM10 Particles 13-7
13.3 UNITED STATES PM CONCENTRATIONS AND EXPOSURE
CONSIDERATIONS 13-11
13.4 DOSIMETRY OF INHALED PARTICLES IN THE RESPIRATORY
TRACT 13-19
13.5 KEY HEALTH EFFECTS OF PARTICULATE MATTER 13-26
13.5.1 Short-Term PM Exposure Mortality Studies 13-26
13.5.1.1 PM10 Relative Risk Analyses 13-30
13.5.1.2 Fine Particles/Acid Aerosols Relative Risks 13-37
13.5.2 Long-Term PM10/PM2 5 Exposure Mortality Studies 13-39
13.5.2.1 Population-Based Cross-Sectional Mortality
Studies 13-39
April 1995 IH-vii DRAFT-DO NOT QUOTE OR CITE
-------
TABLE OF CONTENTS (cont'd)
Page
13.5.2.2 Prospective Mortality Studies 13-40
13.5.3 Morbidity Outcomes Associated with PM Exposure 13-47
13.5.3.1 Short-Term PM Exposure Hospital Admission
Studies 13-48
13.5.3.2 Short and Long-Term Exposure Respiratory Disease
Studies 13-48
13.5.3.3 Short and Long-Term Exposure Pulmonary Function
Studies 13-52
13.5.3.4 Comparison of the Effect of PM10 to PM2 5 on
Respiratory Disease and Pulmonary Function 13-58
13.5.4 Coherence of Epidemiologic Findings 13-61
13.6 HEALTH EFFECTS OF AMBIENT PM CONSTITUENTS 13-64
13.6.1 Mortality Effects of Acid Aerosols 13-65
13.6.2 Respiratory Illness Effects of Acid Aerosols 13-66
13.6.3 Pulmonary Function Effects of Acid Aerosols 13-68
13.7 BIOLOGICAL PLAUSIBILITY: POTENTIAL MECHANISMS OF
ACTION 13-70
13.7.1 Introduction 13-70
13.7.2 Characteristics of Observed Morbidity and Mortality 13-70
13.7.3 Influence of Particle Size, Chemical Composition, and
Respiratory Tract Deposition/Clearance 13-72
13.7.4 Potential Mechanisms of Causality Between Low Levels of
Paniculate Pollution and Health Effects 13-74
13.7.5 Biological Plausibility Conclusions 13-81
13.8 IDENTIFICATION OF POPULATION GROUPS POTENTIALLY
SUSCEPTIBLE TO HEALTH EFFECTS FROM PM EXPOSURE .... 13-82
13.9 IMPLICATIONS OF RELATIVE RISK ESTIMATES RISK GROUPS . . 13-86
REFERENCES 13-98
April 1995 Hl-viii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES
Number Page
12-1 Sample Size, Significance, and Other Characteristics of Recent
Studies on Daily PM/Mortality in U.S. Cities 12-17
12-2 Adjustments for Meteorological Factors in Some Recent Studies
Relating Mortality to Paniculate Matter 12-29
12-3 Summaries of Recently Published Epidemiological Studies Relating
Human Mortality to Ambient Levels of Particulate Matter 12-48
12-4 Summaries of Published PM10-Acute Mortality Effects Studies Based
on Various PM Measures 12-61
12-5 Comparison of Relative Risk Estimates for Total Mortality from
50 /ig/m3 Change in PM10, Using Studies Where PM10 Was Measured
or Was Calibrated for the Site 12-68
12-6 Basis for Evaluation of Time Series Studies on PM10-Mortality 12-69
12-7 Number and Rate of Patients Discharged from Short-Stay Hospitals,
by Age and First-Listed Diagnosis: United States, 1991 12-83
12-8 Number of First-Listed Diagnoses for Inpatients Discharged from
Short-Stay Non-Federal Hospitals, by ICD-9-CM Code, Age of Patient,
and Geographic Region of Hospital: United States, 1992-CON 12-84
12-9 Hospital Admissions Studies for Respiratory Disease 12-96
12-10 Hospital Admissions Studies for COPD 12-96A
12-11 Hospital Admissions Studies for Pneumonia 12-97
12-12 Hospital Admissions Studies for Heart Disease 12-97
12-13 Acute Respiratory Disease Studies 12-115
12-14 Acute Pulmonary Function Changes 12-130
12-15 Commonly-Based Cross-Sectional Studies (1960-1974 Mortality) 12-140
12-16 Commonly-Based Cross-Sectional Studies (1980 Mortality) 12-143
12-17 Relative Mortality Risks in Six Cities Adjusted Risks 12-154
April 1995 Hl-ix DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number Page
12-18 Estimated Relative Risks in Six U.S. Cities Associated with a
Range of Air Pollutants 12-161
12-19 Prospective Cohort Mortality Studies 12-167
12-20 Comparison of Log-Linear Regression Coefficients from Prospective
and "Ecologic" Analyses for U.S. Metropolitan Areas 12-170
12-21 Chronic Respiratory Disease Studies 12-190
12-22 Chronic Pulmonary Function Changes 12-195
12-23 Simultaneous Regressions of 1986 to 1988 Toronto Daily Summertime
Total Respiratory Admissions on Temperature and Various Pollution
Metrics 12-225
12-24 Comparison of Regressions of Daily Summertime Respiratory Admissions
on Pollution and Temperature in Toronto, Ontario, and Buffalo,
New York, 1988 Summer 12-226
12-25 Combined Estimates of Relative Risk of Increased Mortality from Acute
Exposure to Air Pollutants 12-312
12A-1 Summer Threshold Temperatures for Selected United State Cities 12A-7
12A-2 Estimates of Present-Day Heat-Related Mortality During an Average
Summer 12A-15
12A-3 Means and Standard Deviation for Summer Air Masses in Philadelphia . . 12A-18
12A-4 Daily Excessive Mortality (Summer Season) During Offensive Air
Masses 12A-19
13-1 Concentration Ranges of Various Elements Associated with Paniculate
Matter in the United States Atmosphere 13-10
13-2 Comparison of Fine-Versus Coarse-Mode Particles 13-12
13-3 Characterization of Urban PM10 Data From AIRS Network by Region
for the United States 13-15
13-4a PM10 Levels by Annual Average for Selected U.S. SMSAs for 1993 .... 13-17
April 1995 IH-x DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number Page
13-4b Selected U.S. PM10 Levels by Highest Second Maximum 24-Hour
Concentration for 1993 13-18
13-5 Comparison of Relative Risk Estimates for Total Mortality
from 50 jug/m3 Change in PM10, Using Studies Where PM10 Was
Measured or Was Calibrated for the Site 13-32
13-6 Mean Values of Variables Related to Daily Mortality in St. Louis
and Eastern Tennessee 13-38
13-7 Comparison of Univariate Regressions With Pollution on Previous
Day, Controlling for Weather and Season Variables 13-38
13-8 Estimated Relative Risks in Six U.S. Cities Associated with a Range
of Air Pollutants 13-43
13-9 Prospective Cohort Mortality Studies 13-46
13-10 Hospital Admissions Studies for Respiratory Disease 13-49
13-11 Hospital Admissions Studies for COPD 13-50
13-12 Hospital Admissions Studies for Pneumonia 13-51
13-13 Hospital Admissions Studies for Heart Disease 13-51
13-14 Acute Respiratory Disease Studies 13-53
13-15 Chronic Respiratory Disease Studies 13-57
13-16 Acute and Chronic Pulmonary Function Changes 13-59
13-17 Estimated Excess Mortality Per Day for Which an Increase
of 50 /xg/m3 PMio (24-h) Could be a Contributing Factor in
a Population of One Million 13-93
13-18 Estimated Numbers of Deaths Per Day in Cities of 10,000 to
10 Million for Which an Increase of 50 /ig/m3 PM10 Could be
a Contributing Factor 13-93
13-19 Association Between Cigarette Smoking Status and Excess Mortality
Risk From Air Pollution in the Six Cities Study 13-95
April 1995 IH-xi DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF TABLES (cont'd)
Number Page
13-20 Estimated Hospital Admissions Per Day in a Population of One Million
for Which and Increase of 50 /ig/m3 PM10 Could be a Contributing
Factor 13-97
13-21 Estimated Numbers of Hospital Admissions for Respiratory and
Cardiovascular Causes Per Day Due in Cities of 10,000 to 10 Million
for Which an Increase of 50 /*g/m3 PM10 (24-h) Could be a Contributing
Factor 13-97
April 1995 III-xii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES
Number Page
12-1 Examples of threshold function and non-zero background logistic
function 12-20
12-2 Mean daily mortality for PM quintiles, Philadelphia, PA 12-35
12-3 t-Ratios of PM coefficients versus sample size (days) from 11 recent
U.S. studies 12-40
12-4 Relative odds of incidence of lower respiratory symptoms smoothed
against 24-h mean PM10 on the previous day, controlling for temperature,
day of the week, and city 12-103
12-5 Relative odds of incidence of LRS smoothed against 24-h mean SO2
concentration on the previous day, controlling for temperature, city,
and day of the week 12-104
12-6 Relative odds of incidence of LRS smoothed against 24-h mean H+
concentration on the previous day, controlling for temperature, city,
and day of the week 12-104
12-7 Comparison of relative risks of exposure to 15 jtg/m3 of SO4,
as estimated by Dockery et al., 1993 (6-cities, prospective),
Pope et al., 1995 (151 cities, prospective), Ozkaynak and Thurston, 1987
(98 SMSAs, ecological), and Lipfert, 1993 (149 SMSAs, ecological) . . . 12-172
12-8 Comparison of relative risks of exposure to 25 fig/m3 of PM2 5,
as estimated by Dockery et al., 1993 (6-cities, prospective),
Pope et al., 1995 (50 cities, prospective), Ozkaynak and Thurston,
1987 (38 SMSAs, ecological), and Lipfert, 1993 (62 SMSAs,
ecological) 12-173
12-9 Comparison of relative risks of exposure to 100 /*g/m3 of TSP,
as estimated by Abbey et al., 1991 (California), and as estimated from
the data of Dockery et al., 1993, with and without a 10-year lag
(6-cities, prospective), and Lipfert, 1993 (149 SMSAs, ecological) .... 12-174
12-10 December 1962, London pollution episode 12-200
12-11 Time Series Plots of Daily Mortality, Pollution, and Temperature
in London, England, 1965 to 1972 12-204
12-12 Average number of adjusted respiratory admissions among all
168 hospitals by decile of the daily average sulfate level, lagged
1 day 12-210
April 1995 Hl-xiii DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
12-13 Association of moderate or severe cough with exposure-adjusted
hydrogen ion 12-220
12-14 Association of moderate or severe asthma rating with exposure-adjusted
hydrogen ion 12-221
12-15 Bronchitis in the last year, children 10 to 12 years of age in six U.S.
cities, by PM15 12-238
12-16 Bronchitis in the last year, children 10 to 12 years of age in six U.S.
cities, by hydrogen ion concentration 12-239
12-17 Relative risk of mortality for PM10 in Utah Valley, as a function of
several parametric and semiparametric models of time, temperature,
and dewpoint: (a) all causes, (b) respiratory causes, and
(c) cardiovascular causes 12-253
12-18 Relative risk of mortality for PM10 in Utah Valley, as a function of
several Poisson and Gaussian regression models of time, temperature,
and dewpoint: (a) all causes, (b) respiratory causes, and
(c) cardiovascular causes 12-255
12-19 Relative risk of mortality for PM10 in Utah Valley, as a function of
season: (a) all causes, (b) respiratory causes, and
(c) cardiovascular causes 12-256
12-20 Relative risk of mortality for PM10 in Utah Valley, as a function of
ozone indicator in the model: (a) all causes, (b) respiratory causes, and
(c) cardiovascular causes 12-257
12-21 Relative risk of mortality for PM10 in Utah Valley, as a function of
the moving average model: (a) all causes, (b) respiratory causes, and
(c) cardiovascular causes 12-259
12-22 Relative risk of total mortality for PM10 in Santiago, Chile, as a
function of (a) different models, (b) models for co-pollutants,
and (c) moving averages and lag times 12-260
12-23 Relative risk of total mortality for paniculate matter in St. Louis, as a
function of moving average and lag times: (a) PM10 and
(b) PM2 5 12-262
April 1995 IH-xiv DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
12-24 Relative risk of total mortality for participate matter in eastern Tennessee
as a function of moving average and lag times: (a) PM10 and
(b) PM2 5 12-263
12-25 Relative risk of total mortality for PM10 in Los Angeles, as a function of
(a) seasonal model and (b) models including co-pollutants 12-265
12-26 Relative risk of total mortality for PM10 in Chicago as a function of the
model for co-pollutants 12-267
12-27 Relative risk of total mortality for TSP in Steubenville: (a) different
models and (b) as a function of season 12-269
12-28 Relative risk of total mortality for TSP in Philadelphia 12-273
12-29 Relative risk of total mortality for TSP in Philadelphia, in the
(a) spring, (b) summer, (c) fall, and (d) winter 12-274
12-30 , Relative risk of mortality for TSP in Philadelphia, as a function of age,
averaging time, and temperature: (a) age < 65 and (b) age > 65 .... 12-278
12-31 Relative risk of mortality in five studies, where the paniculate matter
index is either total suspended particulates or PM10, in units of /*g/m3
grouped in quintiles 12-285
12-32 Smoothed nonparametric estimate of relative risk of mortality in three
studies, where the paniculate matter index is either total suspended
particulates or PM10, in units of /ig/m3 12-286
12-33 A conceptual model of sources and pathways for air pollution health
effects such as mortality, including a causal model of potential
confounding by co-pollutants 12-298
12-34 Summary of studies used in a combined EPA meta-analysis of PM10
effects on mortality with short averaging tunes (0-1 days) and
co-pollutants in model 12-309
12-35 Summary of studies used in a combined EPA meta-analysis of PM10
effects on mortality with longer averaging times (3-5 days) and no
co-pollutants in model 12-309
12-36 Summary of studies used hi a combined EPA meta-analysis of TSP
effects on mortality, with no co-pollutants in model 12-310
April 1995 IH_XV DRAFT-DO NOT QUOTE OR CITE
-------
LIST OF FIGURES (cont'd)
Number Page
12-37 Summary of studies used in a combined EPA meta-analysis of TSP
effects on mortality, with SO2 in the model 12-311
12-38 Summary of studies used in a combined meta-analysis of PM10
effects on mortality, with other pollutants in the model 12-311
12-39 Summary of PM10 effects on mortality 12-313
12A-1 Daily mortality: Allegheny County, PA, July, 1988 12A-6
12A-2 Relations between summer maximum temperature and daily mortality
in New York 12A-6
12A-3 Daily summer mortality for a New York heat wave, 1966 12A-9
13-1 Sampling fractions for an idealized ambient paniculate
mass distribution 13-3
13-2 Measured mass size distribution showing particles in nuclei and
accumulation modes of fine particles 13-5
13-3 Predicted tracheobronchial deposition fractions versus MM AD
of inhaled monodisperse and polydisperse aerosols 13-21
13-4 Predicted alveolar retained dose burden for 0.5 /*m MM AD
monodisperse aerosol and 2.55 pm MMAD polydisperse aerosol,
assuming a dissolution-absorption half-time of 10 days 13-23
13-5 Predicted alveolar retained dose burden for 0.5 fim MMAD
monodisperse aerosol and 2.55 /mi MMAD polydisperse
aerosol, assuming a dissolution-absorption half-time
of 1,000 days 13-24
April 1995 HI-xvi DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHAPTER 12. EPIDEMIOLOGY STUDIES OF HEALTH EFFECTS ASSOCIATED
WITH EXPOSURE TO AIRBORNE PARTICLES/ACID AEROSOLS
Principal Authors
Dr. Lester D. Grant-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Office, Research Triangle Park, NC 27711
Dr. Vic Hasselblad-29 Autumn Woods Drive, Durham, NC 27713
Dr. Kazuhiko Ito-New York University Medical Center, Institute of Environmental
Medicine, Long Meadow Road, Tuxedo, NY 10987
Dr. Laurence Kalkstein-University of Delaware, Center of Climatic Research, Newark,
DE 19716-2541
Dr. Dennis Kotchmar-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Fred Lipfert-23 Carll Court, Northport, NY 11768
Dr. Allan Marcus-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. George Thurston-New York University Medical Center, Institute of Environmental
Medicine, Long Meadow Road, Tuxedo, NY 10987
Contributors and Reviewers
Dr. Philip Bromberg-University of North Carolina, School of Medicine, Chapel Hill,
NC 27599-0126
Dr. Bert Brunekreef-The University of Wageningen, Department of Epidemiology and Public
Health, P.O. Box 238, 6700 A E Wageningen, The Netherlands
Dr. Richard Burnett-Health and Welfare Canada, 203 Environmental Health Center,
Tunney's Pasture, Ottwaw, Ontario, Canada K1A OL2
Dr. Robert Chapman-Health Effects Research Laboratory (MD-58), U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711
Dr. John Creason-Health Effects Research Laboratory (MD-58), U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711
April 1995 IH_xvii DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
Contributors and Reviewers (cont'd)
Dr. Douglas Dockery-Harvard School of Public Health, Environmental Epidemiology,
665 Huntington Avenue, Boston, MA 02115
Dr. Laurence Saul Kalkstein-University of Delaware, Center for Climatic Research,
Department of Geography, Newark, DE 19716-2541
Dr. Klea Katsouyanni-University of Athens, School of Medicine, Department of Hygiene and
Epidemiology, 1327 Athens, Greece
Dr. Patrick Kinney-New York University Medical Center, Institute of Environmental
Medicine, Long Meadow Road, Tuxedo, NY 10987
Dr. Aparna Koppikar, Human Health Assessment Group, Office of Health and
Environmental Assessment, U.S. Environmental Protection Agency, (8602), Waterside Mall,
401 M. St. S.E., Washington, DC 20460
Dr. Dennis Kotchmar-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Fred Lipfert-23 Carll Ct., Northport, NY 11768
Dr. Thomas Louis-University of Minnesota, School of Public Health, A-460 Mayo Building,
Box 197, 420 Delaware Street, S.E., Minneapolis, MN 55455
Dr. Joseph Lyon-University of Utah, Department of Family and Preventative Medicine,
50 North Medical Drive, Salt Lake City, UT 84132
Dr. David Mage-Atmospheric Research and Exposure Assessment Laboratory (MD-75),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Allan Marcus-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Suresh Moolgavkar-Fred Hutchinson Cancer Research Center, 1124 Columbia Street,
Seattle, WA 98104
Dr. William Nelson-Atmospheric Research and Exposure Assessment Laboratory (MD-56),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Bart Ostro-California Environmental Protection Agency, 2151 Berkeley Way, Annex 11,
Berkeley, CA 94704
April 1995 III-xviii DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
Contributors and Reviewers (cont'd)
Dr. C. Arden Pope, III-Brigham Young University, Department of Economics, Provo, Utah
84602
Dr. James Quackenboss-Environmental Monitoring Systems Laboratory, U.S. Environmental
Protection Agency, P.O. Box 93478, Las Vegas, NV 89193-3478
Dr. H. Daniel Roth-Roth Associates, 6115 Executive Boulevard, Rockville, MD 20852
Dr. Carl Shy-University of North Carolina, Department of Epidemiology, School of Public
Health, Campus Box 7400, Chapel Hill, NC 27599
Dr. Duncan Thomas-University of Southern California, Preventative Medicine Department,
1420 San Pablo Street, Los Angeles, CA 90033-9987
Dr. Dianne Wagener-National Center for Health Statistics, Mortality Statistics Branch,
Division of Vital Statistics, Center for Disease Control, 6526 Belcrest Road, Hyattsville,
MD 20782
Dr. William Wilson-Atmospheric Research and Exposure Assessment Laboratory (MD-75),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Mary C. White-Centers for Disease Control, National Center for Environmental Health,
4770 Buford Highway, NE, Atlanta, GA 30341-3724
Dr. Ronald Wyzga-3420 Hillview Avenue, Palo Alto, CA 94304
CHAPTER 13. INTEGRATIVE HEALTH SYNTHESIS
Principal Authors
Dr. Lester D. Grant-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Ms. Annie A. Jarabek-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Dennis Kotchmar-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Allan Marcus-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
April 1995 IH-xix DRAFT-DO NOT QUOTE OR CITE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (cont'd)
Principal Authors (cont'd)
Dr. James McGrath-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. William Wilson-Atmospheric Research and Exposure Assessment Labatory (MD-75),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Contributors and Reviewers
Dr. Judith A. Graham-Environmental Criteria and Assessment Office (MD-52),
U.S. Environmental Protection Agency, Research Triangle Park, NC 27711
Dr. Jeannette Wiltse, Office of Health and Environmental Assessment, Office of Research
and Development, U.S. Environmental Protection Agency (8601), Waterside Mall, 401 M.
St. S.E., Washington, DC 20460
April 1995 HI-xx DRAFT-DO NOT QUOTE OR CITE
-------
i 12. EPIDEMIOLOGY STUDIES OF HEALTH EFFECTS
2 ASSOCIATED WITH EXPOSURE TO AIRBORNE
3 PARTICLES/ACID AEROSOLS
4
5
6 12.1 INTRODUCTION
7 A rapidly growing body of epidemiologic data examines relationships between PM
8 concentrations and human health effects, ranging from respiratory function changes and
9 symptoms to exacerbation of respiratory disease and excess mortality associated with
10 premature death.
11 The purpose of this chapter is to review the epidemiological evidence relating health
12 effects to exposure to airborne particles. Much new information has appeared since EPA's
13 publication of the 1982 document on Air Quality Criteria for Paniculate Matter and Sulfur
14 Oxides (U.S. Environmental Protection Agency, 1982a), its second Addendum (U.S.
15 Environmental Protection Agency, 1986a), and a later Acid Aerosol Issue Paper (U.S.
16 Environmental Protection Agency, 1989). Information from these previous documents is
17 only concisely considered here to provide a background context for this chapter and to help
18 form the basis for evaluation of more recent publications.
19
20 12.1.1 Definition of Particulate Matter and Measurement Methods
21 As discussed in Chapter 3, "Particulate matter" is the generic term for a broad class of
22 chemically and physically diverse substances that exist as discrete particles (liquid droplets or
23 solids) over a wide range of sizes. Particles originate from a variety of stationary and
24 mobile sources and may be emitted directly or formed in the atmosphere by transformation
25 of gaseous emissions such as sulfur oxides (SOX), nitrogen oxides NOX), and volatile organic
26 compounds (VOCs). The chemical and physical properties of PM vary greatly with time,
27 region, meteorology, and source category, thus complicating the assessment of health and
28 welfare effects. Particles in ambient air usually occur in two somewhat overlapping bimodal
29 size distributions: (1) fine (diameter less than 2.5 jum) and (2) coarse (diameter larger than
30 2.5 ptm). The two size fractions tend to have different origins and composition, as discussed
31 in Chapter 3 along with other aspects concerning particle size and atmospheric chemistry.
April 1995 12-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 On July 1, 1987 (Federal Register, 1987), EPA published revisions to the PM NAAQS.
2 The principal revisions in 1987 included replacing TSP as the indicator for the ambient
3 standards with a new indicator that includes only particles with an aerodynamic diameter less
4 than or equal to a nominal 10 /xm (PM10).
5 The NAAQS revision to PM10 standards was based on several key conclusions as
6 summarized below:
7
8
9 (1) Health risks posed by inhaled particles are influenced by both the penetration and
10 deposition of particles in the various regions of the respiratory tract and the biological
11 responses to these deposited materials. Smaller particles penetrate furthest in the
12 respiratory tract. The largest particles are deposited predominantly in the extrathoracic
13 (head) region, with somewhat smaller particles depositing in the tracheobronchial
14 region; still smaller particles can reach the deepest portion of the lung, the pulmonary
15 region.
16
17 (2) The risks of adverse health effects associated with deposition of typical ambient fine
18 and coarse particles in the thoracic region (tracheobronchial and pulmonary deposition)
19 are markedly greater than those associated with deposition in the extrathoracic region.
20 Maximum particle penetration to the thoracic region occurs during oronasal or mouth
21 breathing.
22
23 (3) The size-specific indicator for primary standards should represent those particles small
24 enough to penetrate to the thoracic region. The risks of adverse health effects from
25 extrathoracic deposition of typical ambient PM are sufficiently low that particles
26 depositing only in that region can safely be excluded from the indicator.
27
28
29 A variety of PM sampling and measurement methodologies have been used in the
30 epidemiology studies discussed in this chapter. Some studies used earlier measures such as
31 British Smoke (BS), Coefficient of Haze (COHs) and Total Suspended Paniculate Matter
32 (TSP). The use and limitations on interpreting studies using such PM measures are discussed
33 in U.S. Environmental Protection Agency (1982, 1986). Additionally, current measures used
34 in more recent studies (e.g., PM2 5 and PM10), are defined and discussed earlier in this
35 document in Chapter 4 (Sampling and Analysis of Particulate Matter). Methodologies for
36 strong acid measurement are also discussed in U.S. Environmental Protection Agency (1989)
37 and in Chapter 4 of this document.
38
39
April 1995 12-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.1.2 Epidemiologic Designs and Strategies
2 The recent epidemiology studies to be discussed generally fall into four categories:
3 (1). short-terrn exposure studies related to acute effects, typically on a time scale of
4 one or a few days;
5
6 (2). prospective cohort studies, in which health outcomes for individuals recruited at
7 the same time are followed over a period of time, typically several years;
8
9 (3). cross-sectional epidemiology studies comparing at a single point in tune the health
10 effects of long term-exposures to air pollution of different populations, typically
11 assuming that exposure has occurred over a time interval of several years;
12
13 (4). metaanalyses and other syntheses of research studies.
14
15 Different studies have differing strengths and weaknesses. One limitation common to
16 all of the above different study designs is that only community-level air pollution information
17 is available, generally from one or a few air monitoring stations used to characterize PM and
18 other air pollution and weather exposures over a city or county. However, the acute studies
19 attempt to relate counts of the number of individuals with a specified health outcome to PM
20 exposures during the day when air pollution was measured in the region or possibly within a
21 few days after such exposure. The health endpoints reviewed here include death, hospital
22 admissions for respiratory or cardio-pulmonary causes, respiratory symptoms reported in a
23 diary by individuals on a selected panel of people who reside in the region, school absences,
24 and results of standard pulmonary function tests (PFT). Sometimes the health outcome data
25 are divided into demographic subgroups by age group, sex, or race. Some studies have also
26 divided the mortality data by primary and contributing causes of death on death certificates,
27 such as respiratory causes or cardio-vascular causes, compared with "control" causes that
28 were believed to have little relation to air pollution.
29 The strength of acute health effects/short-term exposure studies is that they allow
30 evaluations of a single region or community, comparing the response of a population of
31 individuals on one day with one set of pollution exposures to the response of the same
32 population (nearly) on another day with a different set of pollution and weather exposures.
33 In general, the daily health effects data are detrended so that only daily fluctuations in
34 outcome related to daily changes in exposure are evaluated. The detrending includes a
April 1995 12-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 variety of techniques to eliminate the effects of season and yearly changes in population
2 demographics, as well as control or adjustment for unpredictable variables that may affect
3 health outcome, including weather-related variables and shorter-term random events such as
4 influenza epidemics. The analyses of longitudinal data for London during 14 consecutive
5 winters was an important element of the 1986 Addendum and helped to establish a
6 relationship between excess mortality in London and high concentrations of airborne particles
7 measured as British Smoke (BS). Other indicators of airborne paniculate matter (PM) have
8 been used in many recent studies.
9 One limitation in the analysis of acute effects of recent PM exposure is the potential for
10 confounding of PM effects with those of other air pollutants or with meteorological variables.
11 The potential for confounding with other air pollutants may arise because processes that
12 produce PM may also produce the other pollutants. For example, incomplete combustion of
13 fossil fuels used in electrical power generation may produce sulfur dioxide (SO2) as well as
14 PM, so that emissions of both PM and SO2 emissions are high or low at the same time.
15 Moreover, SO2 may also form atmospheric sulfates, which constitute an important part of
16 fine particle mass in many eastern U.S. cities. Likewise, incomplete combustion of fossil
17 fuels in motor vehicles may directly generate PM and primary pollutants such as carbon
18 monoxide (CO) and NOX and indirectly contribute to secondary air pollutants such as ozone
19 (O3) and nitrates, with nitrates also being a PM component. Weather may be a contributing
20 factor to emissions, for example by increasing demand for electric power on very hot
21 summer days or very cold winter days, but meteorological conditions such as inversions are
22 also responsible for high concentrations of air pollutants. However, it is important to
23 remember that the potential for confounding of PM effects with weather or other air
24 pollutants does not necessarily mean that confounding actually biased the results in any given
25 study. Confounding must be evaluated on a case-by-case basis. Comparison of estimated
26 PM effects across different communities having different levels of a potentially confounding
27 factor may help to resolve questions about the role of any given potential confounder. This
28 is discussed in more detail in Section 12.2 below.
29 Prospective cohort studies follow individuals over an extended period of time. The
30 strength of such studies is that individual risk factors can be controlled or adjusted. Known
31 risk factors for mortality include age, sex, race, occupation, economic status, smoking status,
April 1995 12-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 use of alcoholic beverages, and body mass index. If the individuals selected are
2 representative of PM exposures across different communities, the effects of individual risk
3 factors can be separated from PM exposure effects. This epidemiologic design also allows
4 the evaluation of cumulative exposure to PM over the years, whereas the acute effects study
5 design only allows assessment of effects due to short-term exposure changes. One interesting
6 question is whether there are cumulative effects of chronic PM exposure greater than the sum
7 of daily acute effects, since chronic effects must include short-term effects. These strengths
8 of prospective cohort studies are greatly reduced if only occasional air pollution
9 measurements are available, so that only crude exposure comparisons across cities or regions
10 can be made.
11 Cross-sectional studies look only at highly aggregated community health outcomes, such
12 as mortality rates. With no individual-level exposure available, it is only possible to
13 compare different cities by statistical adjustment for demographic and climatological
14 differences and for average differences in levels of air pollutants or other community-wide
15 health risk factors. However, the data for such analyses may be obtained and analyzed
16 relatively easily, and such studies have served a useful historic role in hypothesis generation.
17 There is still much discussion about the appropriateness of using formal mathematical
18 methods known as "metaanalysis" in research syntheses (Shapiro, 1994). This approach used
19 in various other EPA documents, when applied properly, can provide useful guidance in
20 combining the results of diverse studies. Ultimately, synthesis of the results of the studies
21 reviewed here is clearly desirable, but must be guided by substantive knowledge about the
22 individual studies evaluated. For this reason, important methodological issues that affect
23 pertinent studies are discussed next, followed by evaluation of the studies themselves.
24
25 12.1.3 Guidelines for Assessment of Epidemiologic Studies
26 An important concept of epidemiologic information reviewed here concerns its
27 usefulness in demonstrating cause-effect relationships versus merely establishing associations
28 (which may be non-causal in nature) between PM exposures and various health effects. The
29 interpretation of epidemiologic data as an aid to inferring causal relationships between
30 presumed causal agents and associated effects has been previously discussed by several expert
31 committees or deliberative bodies faced with evaluation of controversial biomedical issues
April 1995 12-5 DRAFT-DO NOT QUOTE OR CITE
-------
1 (Surgeon General of the United States, 1964; U.S. Senate, 1968). Criteria selected by each
2 for determination of causality included: (1) strength of the association; (2) consistency of the
3 association, as evidenced by its repeated observation by different persons, in different places,
4 circumstances and time; (3) specificity of the association; (4) temporal relationship of the
5 association; (5) coherence of the association in being consistent with other known facts;
6 (6) existence of a biological gradient, or dose-response curve for the association; and
7 (7) biological plausibility of the association.
8 Hill (1965) further noted that strong support for likely causality suggested by an
9 epidemiologic association may be derived from experimental or semi-experimental evidence,
10 where manipulation of the presumed causative agent (its presence or absence, variability in
11 intensity, etc.) also affects the frequency or intensity of the associated effects. Importantly,
12 both Hill (1965) and the deliberative bodies or expert committees were careful to emphasize,
13 regardless of the specific set of criteria selected by each, that no one criterion is definitive by
14 itself nor is it necessary that all (except temporal relationship) be fulfilled in order to support
15 a determination of causality. Also, Hill (1965) and several of the expert groups noted that
16 statistical methods alone cannot establish proof of a causal relationship in an association nor
17 does lack of "statistical significance" of an association according to arbitrarily selected
18 probability criteria necessarily negate the possibility of a causal relationship. That is, as
19 stated by the U.S. Surgeon General's Advisory Committee on Smoking and Health (Surgeon
20 General of the United States, 1964): "The causal significance of an association is a matter of
21 judgment which goes beyond any statement of statistical probability." Apropos to this, Bates
22 (1992) has emphasized the importance of assessing the overall coherence of epidemiologic
23 findings of both morbidity and mortality effects at varying pollutant concentrations in making
24 judgements on likely causal relationships.
25 Taking into account the above considerations, the following guidelines can be stated by
26 which to judge the relative scientific quality of epidemiologic studies and their findings
27 reviewed here and to assist in their overall interpretation:
28 (1). Was the quality of the aerometric data used sufficient to allow for meaningful
29 characterization of geographic or temporal differences in study population
30 pollutant exposures in the range(s) of pollutant concentrations evaluated?
31
April 1995 12-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 (2). Were the study populations well defined and adequately selected so as to allow for
2 meaningful comparisons between study groups or meaningful temporal analyses of
3 health effects results?
4
5 (3). Were the health endpoint measurements meaningful and reliable, including clear
6 definition of diagnostic criteria utilized and consistency in obtaining dependent
7 variable measurements?
8
9 (4). Were the statistical analyses used appropriate and properly performed
10 and interpreted, including accurate data handling and transfer during
11 analyses?
12
13 (5). Were likely important confounding or co vary ing factors adequately controlled for
14 or taken into account in the study design and statistical analyses?
15
16 (6). Are the reported findings internally consistent, biologically plausible,
17 and coherent in terms of consistency with other known facts?
18
19 Few, if any, epidemiologic studies deal with all of the above points in a completely
20 ideal fashion; nevertheless, these guidelines provide benchmarks for judging the relative
21 quality of various studies and for selecting the best for use in criteria development. Detailed
22 critical analysis of all epidemiologic studies on PM health effects, especially in relation to all
23 of the above questions, is beyond the scope of this document. Of most importance for
24 present purposes are those studies which provide useful quantitative information on
25 exposure-effect or exposure-response relationships for health effects associated with ambient
26 air levels of PM currently likely to be encountered in the United States.
27 Accordingly, the following criteria were employed in selecting studies for detailed
28 discussion in the ensuing text:
29 (1). Concentrations of PM and/or other pollutants were reported, allowing for
30 evaluation of their separate or combined effects.
31
32 (2). Study results provide information on quantitative relationships between health
33 effects and ambient air PM levels of current concern (i.e., generally <250 /ig/m3
34 PM10).
35
36 (3). Key methodological issues were addressed, especially (a) in controlling for likely
37 important confounding factors and (b) in data collection, analysis, and interpretation
38 so as to minimize errors or potential biases.
39
40 (4). The study results have been reported in the open literature or are in
41 press, typically after having undergone peer review.
42
April 1995 12-7 DRAFT-DO NOT QUOTE OR CITE
-------
1 Extensive epidemiologic literature on the effects of occupational exposures to various
2 PM specific components is not reviewed here for several reasons:
3 (1). Such literature generally deals with effects of exposures to PM chemical species at
4 levels many times higher than those encountered in the ambient air by the general
5 population.
6
7 (2). Populations exposed occupationally mainly include healthy adults, self-selected to
8 some extent in terms of being better able to tolerate exposures to PM substances
9 than more susceptible workers seeking alternative employment or other groups
10 often at special risk among the general public (e.g., the old, the chronically ill,
11 young children, and asthmatics).
12
13 (3). Extrapolation of observed occupational exposure-health effects relationships (or
14 lack thereof) to the general public (especially population groups at special risk)
15 could, therefore, be potentially misleading in terms of demonstrating health effects
16 among healthy workers at higher exposure levels than would affect susceptible
17 groups in the general population.
18
19 The occupational literature does, however, demonstrate links between acute high level or
20 chronic lower level exposures to many different PM chemical species and a variety of health
21 effects, including: pulmonary function changes; respiratory tract diseases; morphological
22 damage to the respiratory system; and respiratory tract cancers. Some consideration of such
23 literature is provided in Chapter 11 on the toxicology of specific PM constituents as useful to
24 elucidate important points on observed exposure-effect relationships.
25
26
27 12.2 METHODOLOGICAL CONSIDERATIONS
28 Studies assessed in this chapter were evaluated for several factors of importance for
29 interpreting epidemiological studies. These include: (1) exposure measurement errors;
30 (2) misclassification of health outcomes; (3) model specification for acute studies; (4) model
31 specification for chronic studies; (5) covariates and confounders; (6) selection bias;
32 (7) publication bias; (8) internal consistency and strength of effects; and (9) plausibility of
33 observed effects.
34
April 1995 12-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.2.1 Exposure Measurement Error
2 Measurement error in exposure is potentially one of the most important methodological
3 problems in epidemiologic studies of PM. Ideally, personal monitors would be placed on all
4 subjects for the entire period of a study. Even then, some error associated with the
5 monitoring device itself would remain. Such intensive personal monitoring is generally not
6 feasible for large scale epidemiology studies of PM effects. Even personal monitoring,
7 because of the integrated multi-day sampling, does not adequately measure short-term peaks
8 nor long-term chronic exposures. Instead, PM exposure may also be estimated by source
9 description, personal monitors, in-home monitors, and fixed-site outdoor monitors.
10 The effect of exposure measurement error on epidemiologic estimation of exposure-
11 effect relationships has been studied by several authors: Shy et al. (1978), Gladen and Rogan
12 (1979), Clark (1982), Stefanski and Carroll (1985), Walker and Blettner (1985), Fuller
13 (1987), Lebret (1987), Schafer (1987), Whittemore and Keller (1988), Samet and Utell
14 (1990), Yoshimura (1990), Louis (1991), and Thomas et al. (1993). In general, exposure
15 measurement error independent of the health outcome results in estimated effects being
16 biased towards the null. For example, Whittemore and Keller (1988) found a 20%
17 misclassification rate of the exposure category to result in an underestimate of the logistic
18 regression coefficient by up to 50%. Also, Stefanski and Carroll (1985) found that even
19 without the independence of error related to outcome, the bias is towards the null in
20 situations where the probabilities of response are not extremely close to 0 or 1. Clark (1982)
21 reported measurement error to be biased towards the null in logistic regression when that
22 error was sampled from a normal or logistic regression, and also found bias towards the null
23 for certain multiple logistic models.
24 Several types of PM measurement errors can affect PM-mortality associations. One
25 type is analytical measurement error, including errors in weighing filters and in instrumental
26 response (e.g., light transmittance). The error, in this case, is the discrepancy between the
27 'true' concentration at the sampling point and the value obtained through certain analytical
28 techniques. Another type of error relates to exposure misclassification on an individual level
29 (e.g., activity patterns) and on a population level (e.g., a monitoring site's location relative
30 to population density center). Effects of individual level errors are harder to determine since
31 this requires data collection on human activity patterns. Population level exposure
April 1995 12-9 DRAFT-DO NOT QUOTE OR CITE
-------
1 misclassification errors may be explored using demographic data in conjunction with
2 information regarding the spatial variability of air pollution data.
3 The spatial variability in PM may affect PM mortality regression coefficients either
4 systematically or randomly (see appendix for an illustration). Systematic error may bias the
5 coefficient either upward or downward. Random errors bias the PM-mortality relationship
6 toward the null. The downward bias is, in general, a function of error-to-signal variance
7 ratio (Snedecor and Cochran, 1983). While extensive discussions exist on effects of
8 measurement»errors and possible methods of correction for more traditional epidemiologic
9 studies (Thomas et al., 1993), there has been virtually no systematic assessment of the nature
10 and impact of measurement errors in the time-series context (e.g., does averaging multiple-
11 day PM data reduce error?). Also, without knowledge about differential effects of errors for
12 each air pollution index, interpreting the relative significance of regression coefficients vis-a-
13 vis causality (e.g., those for PM vs. SO2) can be misleading.
14 A few examples in earlier literature, however, considered spatial variability errors for
15 PM and other pollutants (e.g., Clifton et al., 1959; Stalker et al., 1962). In Stalker et al's
16 study in Nashville, TN, a lower spatial precision was found for SO2 than TSP. More
17 recently, Kotchmar et al.'s (1987) study of five U.S. cities found fine particles (d < 2.5 /mi)
18 to be more uniformly spatially distributed than coarse particles (2.5
-------
1 having the least exposure error tends to dominate in such situations, even for the same
2 underlying relationships. In further simulation studies, Lipfert (1994b) showed that exposure
3 error also tends to obscure the true shape of the dose-response function.
4
5 12.2.2 Misclassification of Health Outcomes
6 Misclassification of the health outcome can occur whether the outcome is continuous,
7 such as a measure of pulmonary function, or dichotomous, such as the presence or absence
8 of respiratory symptoms. Lung function is typically measured with spirometry, a well-
9 standardized (Ferris, 1978) technique; measurement errors of the instruments collecting the
10 data have been carefully estimated, and random errors add to the error variance. On the
11 other hand, respiratory symptoms and disease are usually measured by questionnaire.
12 Responses to symptom questions are typically positively correlated and depend on the
13 interpretation of the respondent. A specific respiratory disease is likely to be reflected by
14 reporting of a constellation of symptoms; and, so, aggregate as well as single specific
15 symptom reports need to be considered. Obviously, questionnaire measurements that depend
16 on recent recall are better than those based on recall of events that occurred several years
17 ago. Questionnaires for cough and phlegm production, such as the British Medical Research
18 Council (BMRC) questionnaire (American Thoracic Society, 1969) and revisions of the
19 BMRC questionnaire (Ferris, 1978; Samet, 1978), have been standardized and validated for
20 English-speaking populations. These questionnaires and modified versions have also been
21 used to evaluate respiratory symptoms in non-English speaking populations, with varying
22 degrees of standardization or validation for specific populations studied. Misclassification of
23 primary cause of death on death certificates may also occur.
24
25 12.2.3 Model Specification for Short-Term Exposures
26 No single 'standard' statistical technique has been developed to examine short-term
27 associations of PM with mortality or morbidity, and a variety of approaches have been
28 employed by different researchers. This is partly due to lack of knowledge as to biological
29 mechanism(s) involved in the hypothesized link between population PM exposure and
30 response in the population (e.g., excess premature deaths). There are, however, a several
31 common issues that must be addressed in analyses of PM mortality/morbidity associations:
April 1995 12-n DRAFT-DO NOT QUOTE OR CITE
-------
1 • Treatment of long-term patterns and trends • Data Distribution
2 • Duration-dependence of exposure/response • Lag structure
3 • Functional form of exposure/response
4
5 No given statistical technique fully addresses all of these issues, and computational
6 complexity and limitations in the available data (e.g., only every-6th-day sampling frequency)
7 may not allow all investigators to employ all appropriate available statistical techniques.
8 Nevertheless, these issues are important not only to statistical validity, but also to possible
9 elucidation of biological mechanisms. This section considers such issues and some of the
10 techniques currently being used to model PM-mortality associations.
11
12 Treatment of long-term patterns and trends
13 Time-series observations of daily mortality are generally not random, but include 'long-
14 term' components (generally referring to peaks or cycles longer than two weeks or one
15 month). These long-term components include: sinusoidal annual cycles that are often regular
16 in their shapes (Rogot et al., 1976; Sakamoto-Momiyama, 1977); more irregular long-term
17 components, such as influenza epidemics (Glezen et al., 1982; Choi and Thacker, 1981;
18 Serfling, 1963), and; long-term trends that may be significant over many years.
19 Long-term components in mortality cause problems in the analyses of short-term
20 association of PM-mortality in two ways: (1) they can create a spurious (confounding)
21 mortality-PM association that can mask or bias short-term effects (Thurston and Kinney,
22 1995); and (2) they may induce significant autocorrelations in the residuals, which violate the
23 assumption of independent error in two of the most commonly used statistical methods, the
24 Ordinary Least Squares (OLS) and the Generalized Linear Models (GLMs), potentially
25 biasing significance tests (Cochrane and Orcutt, 1949; Liang and Zeger, 1993).
26 To address these problems, long-term components can be estimated by several methods
27 including: (1) weighted moving averages (e.g., Schimmel, 1978; Shumway et al., 1983;
28 Schwartz and Marcus, 1990); (2) trigonometric fit (e.g., Hexter and Goldsmith, 1971;
29 Schwartz, 1993); (3) Autoregressive Integrated Moving Average (ARIMA) fit (Box and
30 Jenkins, 1977), and; (4) Seasonal and Trend decomposition using Loess or STL (Cleveland
31 et al., 1990). Long-term components estimated by these methods may be included in the
April 1995 12-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 mortality regression or subtracted from each series to obtain deviation series, which in turn
2 may be used in subsequent regression analyses (see Appendix A). Distributional
3 characteristics of the original series, however, may be altered in such deviation series.
4 Removal of long-term components by filtering or by inclusion of long-term components
5 as covariates in the regression does not necessarily result in unautocorrelated regression
6 residuals. If the residuals are still autocorrelated, the error term would need to be further
7 modeled for such time dependent structures. Irregular systematic autocorrelations (e.g., day-
8 of-week effects) can be modeled by indicator dummy variables. When other than normal
9 distributions are assumed, an extension of the GLM procedure designed for correlated data
10 may also be used. This option is discussed below. While not controlling for these long-term
11 components can create bias and spurious associations, the majority of recent PM-mortality
12 studies have addressed this issue in various ways. However, an unintended consequence of
13 controlling for these long-term cycles (cycles > 1 month) is elimination of information
14 regarding possible delayed or prolonged effects.
15
16 Data Distribution
17 In theory, the underlying distribution of spontaneous death counts in a population in a
18 short time interval should follow the Poisson distribution, in which the only possible values
19 are whole numbers, the mean is proportional to the variance, and non-negative. A class of
20 statistical models, Generalized Linear Models (GLM), includes log-linear models suitable for
21 such count data (McCullagh and Nelder, 1989). The form of the relationship between the
22 expected dependent variable and covariates is: log(E(Yi))=Xib, where Y is death count on
23 day i, and Xj is a matrix of covariates on day i (in contrast to the ordinary linear model:
24 E(Yj)=Xjb). The maximum-likelihood estimates of the parameters are obtained by iterative
25 reweighted least squares with the adjusted dependent variable in the model. In the GLM log-
26 linear model, systematic effects are modeled as multiplicative, rather than additive, as in the
27 case of the classical linear model. This ensures that the counts are non-negative. In recent
28 U.S. studies of data from relatively small areas, these characteristics were apparent (e.g., in
29 Steubenville: mean=3.1, variance=3.1), and these Poisson regressions were employed.
30 Some studies (e.g., Schwartz, 1992) also included a parameter to allow for the greater-than-
31 predicted (by Poisson) variance of mortality ("over-dispersion"), which may result from
April 1995 12-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 model misspecifications (e.g., omitting some long-term components). In large cities, the data
2 distribution approximates normality, and use of OLS or Poisson model does make much
3 difference in the estimated coefficients (e.g., see Schwartz, 1993; Kinney et al., 1995).
4 While the GLM log-linear model can accommodate the Poisson distributed data found
5 in mortality series of small populations, GLM models share the same independent
6 observations assumption as the classical linear regression models. Thus, without
7 modification, they can not be adequately applied to correlated data, including serially
8 correlated time series data. Statistical problems with ignoring correlations in the data include
9 the inflation/deflation of variance of the estimated parameter and loss of efficiency, in the
10 sense that the uncertainty in the estimated parameter is greater than the uncertainty in the
11 best unbiased estimate (Liang and Zeger, 1993). Liang and Zeger extended GLM to the
12 analysis of longitudinal data (1986, 1986, 1988) by introducing a class of estimation
13 equations (i.e., Generalized Estimating Equations [GEEs]) that give consistent estimates of
14 regression parameters when there is dependence within observations. While the GEE has
15 been used in some recent PM-mortality studies (e.g., Schwartz , 1992; Pope et al., 1992),
16 issues exist regarding differences in specifications of covariance structures and sensitivity to
17 its application.
18
19 Duration Dependence of Exposure/Response
20 From the viewpoint of biological plausibility, the duration of exposure is likely to
21 influence the biological response. For example, a 24-h exposure to 200 /jg/m3 of PM10
22 would likely produce a different biological stress than a 96-h consecutive exposure to the
23 same level of PM. If this is the case, the PM-mortality regression coefficient or slope may be
24 dependent upon the duration of exposure. This potential phenomenon may be examined via
25 frequency (reciprocal of duration) domain time series analyses (see, for example, Bloomfield
26 (1976) or Shumway (1988)). In frequency domain analysis, the frequency dependence of the
27 'slope' can be assessed using gain spectra; the strength of association, or correlations for
28 each periodicity can be assessed using coherence spectra. Any inconsistency between the
29 'long-term' and the 'short-term' PM-mortality associations may be evaluated in these spectra,
30 and may suggest filtering of certain frequency regions. Most recent time-series studies of
31 PM-mortality associations have not explored these issues. One exception is an analysis of the
April 1995 12-14 DRAFT-DO NOT QUOTE OR CITE
-------
1 historical London data by Shumway et al. (1983), who reported that the strongest pollution-
2 mortality associations occur at cycles ranging between seven and 21 days (after filtering
3 mortality with a high-pass linear filter), implying that a pollution and temperature episode
4 must persist for effects to be seen. While such frequency domain analyses may give
5 comprehensive frequency break-downs of exposure-response relationships, they may still
6 miss certain complexities of exposure-response relationships, such as seasonal dependence,
7 even for a given segment of periodicity. Also, application of frequency domain techniques is
8 limited to data collected daily for a long period.
9
10 Lag structure
11 In evaluating epidemiological findings for consistency with causality, one of the criteria
12 suggested by Hill (1965) is a temporal relationship of the association: the exposure must
13 precede the biological response. Given sufficient number of observations, such temporal
14 relationship of exposure/response can be statistically examined through the cross-correlation
15 function (CCF) or the pair-wise correlation with lags. Any autocorrelations in the exposure
16 and/or response series may have to be removed prior to the CCF examination to obtain
17 'rational' CCF (Haugh and Box, 1977). A significant association found on the 'right' side
18 lag (exposure leading response) is less convincing if a similar extent of association is found
19 on the 'wrong' side lag (response leading exposure).
20 When PM has more than one significant lag (on the 'right' side), a distributed lag
21 model is suggested. The shape of distributed lag coefficients may also reflect duration
22 dependency of exposure/response relationship. In recent U.S. PM-mortality studies (see
23 Table 12-1), the length of the lagged PM variables used in the regression analyses range
24 from single day (current or previous day) to up to 4 previous days. The extent of description
25 regarding the lag length is not always comparable in these studies (some report CCF's,
26 others report the most significant lag and averaging length), and the choice of covariates'
27 lags likely influence significance of PM lags. Also, it should be noted that there is a
28 difference between the inclusion of multiple lagged PM variables and the inclusion of a
29 multi-day averaged PM variable (a form of constrained distributed lag model). Averaging
30 over multiple days may enhance relevant PM signals if a multi-day exposure is more
31 effective in causing health effects than a single day elevated PM level. On the other hand, if
April 1995 12-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 alternating high and low PM levels are more effective in producing health effects, then the
2 averaging would attenuate such signals. The existing literature seems to suggest the former.
3 What has been rarely described in detail is the lag structure among the covariates,
4 especially those between weather variables and pollutants. Elucidation of the lag structure
5 among covariates should explain the sequence of events occurring in the atmosphere (e.g.,
6 the passage of a weather front, a change in temperature, and a build-up of air pollution), and
7 may help understand the nature of 'confounding' among the weather and pollution
8 covariates. The diversity of PM and weather lag structures used in the published studies
9 complicates the intercomparison of results and the identification of an appropriate PM
10 exposure metric for use in setting standards (e.g., is a 24-h standard adequate?).
11
12 Functional form of exposure/response and threshold
13 Examination of possible functional forms of PM-mortality/morbidity associations is
14 complicated by several issues: (1) exposure duration dependence; (2) 'harvesting';
15 (3) sensitive sub-population; and (4) interaction with other covariates.
16 If the response is dependent on the duration, as well as the level, of PM exposure, then
17 separate exposure level/response curves for a range of duration may be needed. The slope
18 observed in a scatter plot of mortality vs. PM levels does not take into account differences in
19 exposure duration. Log-normally distributed PM levels suggest, in a time-series context, that
20 very high PM values do not persist except for a few unusual persisting pollution episodes.
21 Therefore, the shallower slope at high levels sometimes observed in the literature may simply
22 reflect the relatively ineffective impact of very short-term high level PM exposures.
23 'Harvesting' or more properly, 'mortality displacement', refers to a decrease in daily
24 mortality that follows the day(s) of elevated pollution levels. This phenomenon is postulated
25 because the people who die from the stress of air pollution are assumed to be those who
26 already suffer from pre-existing cardio-respiratory conditions and the degree of prematurity
27 of deaths may only be a few days. While there are some suggestions of this (e.g., Spix et
28 al., 1994), it was not apparent in past major episodes such as the 1952 London fog incident.
29 Also, statistical examination of this phenomenon is complicated by the necessary process of
30 time-series modeling (e.g., detrending and filtering).
April 1995 12-16 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-1. SAMPLE SIZE, SIGNIFICANCE, AND OTHER CHARACTERISTICS OF
RECENT STUDIES ON DAILY PM/MORTALITY IN U.S. CITIES
S
Ul
to
0
g
6
O
O
H
O
i
S
Q
H
W
Area
Birmingham, AL
Cincinnati, OH
Cook Co., IL
Detroit, MI
Kingston, TN
Los Angeles, CA
Santa Clara, CA
St. Louis, MO
Steubenville, OH
Philadelphia, PA
Utah Valley, UT
Period
1985-1988
1977-1982
1985-1990
1973-1982
1985-1986
1985-1990
1980-1986
1985-1986
1974-1984
1973-1980
1985-1989
Sample
Size
1,087
2,191
1,357
3,652
330
364
549
311
4,016
2,726
1,706
t-ratio
2.52
3.47
3.43
3.76
1.07
1.96
2.86
2.17
4.66
5.04
4.78
Population
884,000
873,224
5,300,000
1,200,000
640,887
8,300,000
1,400,000
2,356,460
163,099
1,688,710
260,000
Daily
Number
Deaths
17
21
117
53
16
153
18
56
3
48
3
PM measure
and mean3
PM10; 48
TSP; 76
PM10; 38
TSP; 87
PM10; 30
PM10; 58
COH; 67b
PM10; 28
TSP; 111
TSP; 77
PM10; 47
Lag and Average
same + 2 prev-day
same-day
same-day
prev. day
prev. day
same-day
same-day
prev. day
prev. day
same-l- prev-day
same +4prev-day
Reference
Schwartz (1993)
Schwartz (1994)
Ito et al. (1995)
Schwartz (1991)
Dockery et al. (1992)
Kinney et al. (1995)
Fairley (1990)
Dockery et al. (1992)
Schwartz and Dockery (1992b)
Schwartz and Dockery (1992a)
Pope et al. (1992)
a/ig/m3, unless otherwise noted
"12XCOH, unitless
Note: When multiple
models were
presented,
the model
with single pollutant (PM)
and weather, season
variables for the
entire year was chosen.
-------
1 The shape of the exposure/response relationship must be different for various segments
2 of the dying population. In fact, a certain fraction of non-accidental deaths is likely to be
3 from those not even subject to day-to-day changes in PM levels. The extent of prematurity
4 of deaths may be a few days for some, but could be many months for others.
5 Characterization of sensitive sub-populations is important in establishing meaningful
6 exposure/response curves.
7 Exposure/response curves may also be different depending upon the levels of other
8 factors (e.g., temperature, ozone). An addition of interactive terms in regression models is
9 not usually practiced in time-series studies, probably because of already highly collinear
10 variables. Some recent studies have examined interactions between weather variables and air
11 pollution variables, as noted in Section 12.2.5. Possible interactive effects may be examined
12 through more exploratory techniques, such as the Classification and Regression Trees, or
13 CART (Breiman et al., 1984). An exposure/response 'surface' of mortality on PM and other
14 factors may be estimated using these techniques.
15 In exploring the shape of possible exposure/response curves, rather than forcing a
16 certain functional form for the entire range of the data, smoothing, such as locally-weighted
17 regressions, or LOWESS (Cleveland, 1979), may be useful. Smoothing may be extended to
18 multivariate models using the Generalized Additive Model, or GAM (Hastie and Tibshirani,
19 1990), in which parametric and semi-parametric terms can also be included simultaneously.
20 These techniques have been used in some of the PM-mortality studies (e.g., Schwartz, 1992
21 and 1994). Some caution is required, however, in determining the functional form and any
22 'threshold' from these curves. PM data (TSP and PM10) for example, are often positively
23 skewed in their distributions and do not have very 'low' values (e.g., <20 /ig/rn3 PM10).
24 Also, the stability of the results depends on the neighborhood and the interval of PM scale,
25 or "span", used to compute each segment of the curve. Thus, it would be useful to report
26 the data density associated with these curves and information regarding the weights and the
27 span used in the construction of exposure/response curves.
28
29 Thresholds
30 The existence of thresholds can be argued both biologically and statistically. The
31 biological arguments have been given by several authors, including Stokinger (1972),
April 1995 12-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 Dinman (1972), and Waldron (1974). Methods for estimating threshold models have been
2 given by several authors including Quandt (1958), Hudson (1966), Hasselblad et al. (1976),
3 Crump (1984a,b), Crump and Howe (1985), Cox (1987), and Ulm (1991). However, the
4 concept of a threshold may be confused with the concept of a non-zero background.
5 A threshold model starts out completely flat, possibly above zero, and at some point begins
6 to curve upwards (see Figure 12-1). A non-zero background model begins above zero and
7 continually curves upward. However when fitting data, "... an additive background dose is
8 generally not distinguishable from a threshold" (Cox, 1987). Cox (1987) gives 10 real data
9 sets where thresholds have been estimated, and in every one of them it is possible to fit a
10 non-threshold model which fits nearly as well. Thus, for epidemiologic studies, the question
11 of thresholds may be difficult to resolve because of difficulties in estimation. When there is
12 substantial measurement error in the exposure variable or heterogeneity in threshold values in
13 a population, it may not be possible to identify a threshold using aggregate response data
14 such as mortality counts or hospital admissions.
15 However, many epidemiological studies reviewed in this section were structured to
16 develop linear or log-linear models with no such threshold, and in many cases, this
17 assumption has been supported by the data plots presented. However, it has also been shown
18 that it may be difficult to distinguish among alternative regression models with confidence,
19 presumably because the main outlying observations are controlled by factors not included in
20 the model. In such cases, linear and threshold models may have essentially equivalent
21 predictive power, and the most useful regression model will be one that admits the possibility
22 of a threshold and develops confidence limits for its value.
23
24 12.2.4 Model Specification for Long-Term Exposures
25 Cross-sectional studies are intended to reveal associations between air pollution and
26 health by virtue of similarities in spatial patterns, averaged over some suitably long term. In
27 order to be able to interpret any resulting statistical associations as potentially causal, all
28 other spatial factors that covary with air pollution or mortality rates must be accounted for or
29 "controlled." Rumford (1961) noted the necessity of considering the "total socioeconomic
30 complex" if valid estimates of the pollution effects are to be obtained. There may also be
April 1995 12-19 DRAFT-DO NOT QUOTE OR CITE
-------
1 spatially varying temporal factors, such as flu epidemics, heat waves, etc., that might
2 influence health differentially in certain locations.
0.8
0.6
0.4
0.2
Non-zero background
logistic function
Threshold function
2 3
Concentration
Figure 12-1. Examples of threshold function and non-zero background logistic function.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Chronic epidemiological studies of the health effects of community air pollution are
typically beset by two fundamental difficulties:
(1). The health endpoints of interest are also affected by factors other than air
pollution. Samet and Speizer (1993) noted that, "Few adverse effects of
environmental pollutants are specific, that is, uniquely associated with a single
agent." At current ambient conditions, in many instances, including mortality or
hospital utilization, air pollution plays a relatively minor role in the overall
etiology.
(2). As a result, it is necessary to study large populations in order to attain statistical
significance. This makes it difficult if not impossible to determine individual
exposures. Given that most people spend the majority of their tune indoors and
that routine air sampling is conducted outdoors, often at considerable distances
away, the exposure measure used in air pollution epidemiology is usually subject
to potential bias and to considerable uncertainty.
April 1995
12-20
DRAFT-DO NOT QUOTE OR CITE
-------
1 Uncertainties in exposure may be characterized on several different levels, listed in
2 approximate decreasing order of importance:
3
4 (1). the difficulty in partitioning observed health effects among collinear pollutants;
5
6 (2). effects of subjects' microenvironment, including outdoor air, workplace exposures,
7 indoor air in residences and other buildings, and air quality inside vehicles;
8
9 (3). selection of the appropriate exposure metric, with options ranging from hourly or
10 less to averages over multi-day episodes;
11
12 (4). within-city variations in outdoor air quality, with spatial variability depending
13 strongly (inversely) on averaging times and proximity to pollution sources; and
14
15 (5). instrumental and analytical errors, with the importance of such errors depending
16 on the study design (cross-sectional versus time-series) and whether they are
17 random or involve biases (e.g., filter artifacts as a source of bias in gravimetric
18 determination of suspended particle mass).
19
20 In general, the difference between measured outdoor concentrations and actual exposures
21 may be expected to vary from day-to-day and by pollutant species and particle size. Some
22 pollutants are distributed more uniformly throughout a community than others, and some
23 penetrate into indoor environments more readily than others. Thus the true exposure
24 collinearity among the components of urban air quality may be quite different from the
25 collinearity indicated by outdoor monitoring data.
26 It is common when analyzing a data set to discover that one or more key covariates for
27 the analysis were not measured. Schenker et al. (1983a) discuss socioeconomic status (SES),
28 passive smoking, and gender as important covariates in childhood respiratory disease studies.
29 Other covariates include age; temperature; and other pollutants, such as ozone or SO2. The
30 concern is that, had an omitted covariate been measured, then the estimate of the regression
31 coefficient for a dependent variable of interest would have been significantly different.
32 Although the problem is faced by many investigators, the literature on the field is relatively
33 sparse. For example, Kupper (1984) shows that high correlations between the variables just
34 described will result in "unreliable parameter estimates with large variances." Gail (1985)
35 considered the special case of omitting a balanced covariate from the analysis of a cohort
36 study and concluded that "In principle, the bias may be either toward or away from zero,
37 though in typical examples—the bias is toward zero. In applications with additive or
April 1995 12-21 DRAFT-DO NOT QUOTE OR CITE
-------
1 multiplicative regression, there is no bias." Neither paper provided information on how to
2 attempt to correct for the bias or to investigate the possible bias in a given situation.
3 Most studies of respiratory disease and PM exposures discussed here measured
4 important covariates such as age, socioeconomic level of the parents, gender, and parental
5 smoking habits. The estimated effect (regression coefficient of disease on PM exposure) will
6 be overestimated if a missing covariate is positively (or negatively) correlated with both
7 exposure and outcome. The estimated effect will be underestimated if positively correlated
8 with either exposure or outcome but negatively correlated with the other. For example,
9 Ware et al. (1984), found that parents with some college education were more likely to
10 report respiratory symptoms but less likely to use a gas stove, leading to an underestimate of
11 the health effect of exposure to gas stove emissions if education were left out of the analysis.
12
13 The Ecological Fallacy
14 Epidemiological studies of air pollution health effects are necessarily observational, that
15 is, involving naturally occurring rather than manipulated environmental conditions
16 (Kleinbaum et al., 1982). Since the characterization of individual environmental exposures is
17 clearly impractical in the large populations that are required, such a study is likely to be
18 "ecological" as well as observational, i.e., involving the study of group attributes rather than
19 those of individuals (Piantidosi et al., 1988). According to Kleinbaum et al. (1982), the
20 primary feature of an ecological study is the lack of knowledge of the joint distribution of the
21 study factor (i.e., exposure to air pollution) and the disease within each group. The primary
22 weakness in ecological regression relates to the lack of specificity of the affected individuals
23 and the exposed individuals, because groups are used in the regression analysis. This
24 problem is most critical when the pollutant is very localized (such as emissions from a toxic
25 waste dump) or when the disease is relatively rare (such as leukemia). In such cases, an
26 ecological analysis may identify the incorrect agent.
27 However, the problem diminishes for broadly distributed regional pollutants, such as
28 fine particles or sulfates, and for mortality from all causes or from very common causes
29 (such as heart disease). In such instances, it is reasonable to assume that all the cases were
30 exposed at least to some degree, although the particulars of their exposures will vary
April 1995 12-22 DRAFT-DO NOT QUOTE OR CITE
-------
1 because of indoor-outdoor pollutant relationships and other local factors. In these cases, the
2 DRFs may be incorrect in shape or magnitude.
3 Greenland and Morganstern (1989) present examples in which ecological analyses
4 clearly derive incorrect results. The concept of "effect modification" is one example in
5 which the effect (i.e., slope of the dose-response function) depends on the proportion of
6 exposed subjects in each of the regions used for ecological regression. (In the time-series
7 analog, the slope may vary from day to day.) In this case, the esophageal cancer rate for
8 non-smokers was constant across all regions, but rates for smokers increased as the
9 proportion of smokers in the region decreased. Such a situation might occur due to selective
10 survival: a low proportion of smokers could result from smokers' higher rates of cancer and
11 heart disease and subsequent demise from these conditions. In this example, the numbers of
12 smokers' cancer deaths was constant across regions even though the absolute numbers of
13 smokers varied by a factor of 1.67. It is hard to conceive a physiological rationale for such
14 a situation, since mortality rates are much too low to deplete the population, even for
15 smokers. Greenland and Morgenstern's other two examples also had special situations
16 involving two competing cancer risks, smoking being the "confounding" variable. In both
17 examples, the regional rates for the two risk factors were strongly negatively correlated (r =
18 -1 and -0.7). One of the most important checks to be made in a multifactor ecological
19 analysis is for the existence of such confounding situations. If the desired risk factor is
20 strongly correlated (+ or -) with other variables, disentanglement will be difficult if not
21 impossible in all cases, ecological or not. These examples also involve use of a linear
22 regression model in situations where the effects are established as multiplicative. The
23 confounding factor (smoking) increased the risk by factors of five and ten, respectively. In
24 such extreme situations, it is not surprising to find anomalous results. In Greenland and
25 Morgenstern's examples, the difficulties arise from assumed properties of the regions, not
26 from differences between group and individual behavior. One might thus argue that the
27 chances of such situations seriously affecting an analysis diminish as the size of the groups
28 decreases and their numbers and diversity increase (Robinson, 1950). However, Piantidosi et
29 al. (1988) compared regression coefficients for a group of physiological and dietary
30 variables, contrasting the "true" values obtained from individuals with those obtained from
April 1995 12-23 DRAFT-DO NOT QUOTE OR CITE
-------
1 various aggregations. In four out of the thirteen cases, the true values lay outside the
2 confidence limits for the aggregated values, with errors in both directions.
3 For cross-sectional analyses, spatial variation constitutes the within-group variance at
4 issue with respect to the ecological fallacy. Here, the ecological hypothesis is that each of
5 the cities or locations studied has the same within-city spatial distribution of air quality
6 (assuming that adequate monitoring networks are not always available) and the same within-
7 city distributions of potential confounding variables such as age, race, poverty
8 neighborhoods, etc. This is not likely to be true in general, but it can be seen that these
9 considerations favor the use of the smallest practical units for geographic analysis. As larger
10 geographic units are used for analysis, for example, Standard Metropolitan Statistical Areas
11 (SMSAs) or larger, the representativeness of air monitoring is likely to diminish. In some of
12 the studies reviewed below, only one monitoring station was available in a given
13 metropolitan area.
14
15 "OvercontroL " This term refers to indiscriminate use of large numbers of nonpollutant
16 variables in a cross-sectional regression. In U.S. Environmental Protection Agency (1986),
17 the example cited (Selvin et al., 1984) involved use of 17 socioeconomic and four weather
18 variables in a cross-sectional analysis of 3082 counties and 410 county groups. Some of the
19 variables in this study that might be considered questionable included divorce rate, elevation,
20 and land area, among others. In this case, there were adequate degrees of freedom to
21 accommodate these many independent variables; but the question then becomes does
22 inclusion of a particular variable act to obscure a "true" pollution relationship? Although
23 some answers to this question may well be idiosyncratic, the questionable variable must be
24 correlated with mortality and with air pollution for such obscuration to occur. However, in
25 the typical interrelated data sets found in national cross-sectional studies, the relationship
26 with the dependent variable must be evaluated dynamically, after control for other important
27 variables like age, race, and income.
28 A good rule of thumb to preclude the "overcontrol" criticism may be to only consider
29 those independent variables for which a relationship with the health endpoint can be shown
30 exogenously, preferably with individual level data. Since it is generally not possible to make
31 such a showing for air pollution variables at current ambient levels, the burden of proof will
April 1995 12-24 DRAFT-DO NOT QUOTE OR CITE
-------
1 rest on showing that the pollutant still has an effect even when all known non-pollution
2 factors are included in a joint regression. In general, this may require the detailed evaluation
3 of a range of models, rather than the selection of an a priori "best" model. Certainly, given
4 the likelihood of some confounding in any event, maximizing the effects of the pollution
5 variables should not be the rationale for selecting a particular regression model.
6
7 12.2.5 Covariates and Confounders
8 Confounding
9 Stellman (1987) stated that confounding is the "cause of great angst among
10 epidemiologists." The term refers to the incorrect assignment of an effect to a given agent
11 when in fact a third variable (the confounder) is responsible. Such a situation requires that
12 the confounder have an effect on the outcome variable and be correlated with the first agent.
13 In other words, a confounder must have the property of different distributions for exposed
14 and nonexposed subjects (Miettinen and Cook, 1981). Smoking may be a good example,
15 since it increases exposure to fine particles and adversely affects health. Heat waves may be
16 another, since urban air pollution often increases with temperature as do mortality and
17 hospitalization rates. In ecological case-control studies of environmental factors, in which an
18 exposed city is compared to an unexposed city, confounding is likely because there are many
19 other ways in which two population groups may differ. As numbers of locations or time
20 periods increase and regression methods become appropriate, opportunities for serious
21 confounding are diminished. It is possible to devise examples which feature serious
22 confounding (i.e., Greenland and Morganstern, 1989); but according to Stellman (1987,
23 p. 165); "Rarely, however, does confounding itself, especially from unidentified sources, live
24 up to its reputation by introducing seriously spurious associations."
25
26 Confounders and choice of covariates
27 To pose a problem in epidemiologic analyses, Confounders must: (1) be a risk factor
28 for the outcome; (2) be associated with the exposure variable; and (3) not be an intermediate
29 step in the causal path between the exposure and the outcome (Rothman, 1986). The risk
*»
30 factor need not be causal in this case. Thus, many weather variables, as well as co-
31 pollutants, qualify as potential Confounders for PM-mortality or morbidity associations.
April 1995 12-25 DRAFT-DO NOT QUOTE OR CITE
-------
1 The causality of various weather and pollution variables may or may not be clear,
2 however. For example, an extreme (hot or cold) temperature is known to cause excess
3 mortality, and laboratory human and animal studies support biologically plausible
4 mechanisms for such observations. Thus, simultaneous inclusion of temperature and pollution
5 variables in a time-series regression seems reasonable, although there also is a chance that
6 this may result in under-estimation of pollution coefficients because temperature is also
7 correlated with meteorological conditions that cause air pollution build-up. Wind speed is
8 clearly a good predictor of air pollution build-up, but is not directly causally related to health
9 outcomes. Therefore, inclusion of wind speed in mortality/morbidity regression is not
10 recommended for air pollution epidemiology. Barometric pressure is an example of a
11 variable whose effect (within the range of day-to-day variation) on physiological functions is
12 less clear (Tromp, 1980). It is associated with certain physiological changes such as blood
13 pressure, but this may be due to its association with temperature change, which is also
14 related to change in blood pressure. Barometric pressure is also correlated with air pollution
15 levels. Thus, while there is a need to address potential confounders, care must be taken that
16 the regression model selected is not over-specified.
17 Common air pollution variables, such as SO2, O3, NO2, and CO are all known to cause
18 health effects, independent of PM, at some level. On a day-to-day basis, the concentrations
19 of these air pollutants, as well as PM, are all correlated to varying degrees, due to the
20 meteorological conditions that control dispersion of these pollutants. Care must therefore be
21 taken when including these correlated pollution variables in a health effect regression, as
22 their coefficients may be unstable. Furthermore, the significance of coefficients for each
23 variable may be influenced by their measurement errors, rather than their causal strengths.
24 Thus, without external information regarding differential error and some description of
25 collinearity among the covariates, interpretation of these multiple regressions with collinear
26 variables can be misleading. Under circumstances where various collinear variables are
27 present, and each one of the pollutants is suspected of causality to differing degrees, a single
28 pollutant model may result in over-estimation of the coefficient for that pollutant, while a
29 multiple pollutant model may result in under-estimation of each pollutant's coefficient.
30 Thus, ideally both results should be considered. Separation of possible effects from these
31 various correlated pollutants may be difficult from a single study, but may be possible by
April 1995 12-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 evaluating the consistency of coefficients across studies in which the levels and the extent of
2 collinearity of co-pollutants vary. To facilitate such collective understanding (or even meta-
3 analysis), it is crucial for each study to include systematic description of collinearity among
4 the covariates (e.g., correlation of the estimated parameters), levels of each pollutant, and
5 discussion of biological plausibility for each variable at the observed levels.
6 While the parsimony of a model is generally desirable, blind reliance on the automatic
7 variable selection schemes based on the F-statistic, such as stepwise regressions, or the use
8 of other criteria, based on residual error and number of parameters (Akaike, 1973; Schwarz,
9 1978) is not appropriate for epidemiologic purposes, as the objective is not to develop a
10 parsimonious model, but to assess the impact of pollution while adequately 'controlling' for
11 other covariates.
12
13 Season specific exposure/response
14 Removal or control of long-term seasonal cycles alone does not eliminate the possible
15 differences in short-term PM exposure/response relationships across different seasons. This is
16 likely because: (1) exposure patterns may be different among seasons due to the difference
17 in human activity patterns (e.g., percent of time spent outdoors); (2) differences exist in
18 indoor/outdoor air exchange patterns (e.g., open windows, air conditioning); (3) there can be
19 a lack of PM signals in certain seasons and/or a lack of certain chemical components (e.g.,
20 sulfates exist more in the summer) in PM that may have health effects; and (4) the influence
21 of other factors, such as temperature and co-pollutants, likely varies over time. Not only the
22 strength of associations, but also the lag structure of PM-mortality associations, may
23 therefore vary across seasons.
24 The seasonal specificity in PM-mortality associations may be explored by examining
25 cross-correlation functions for each season, provided sufficient sample size is available. If
26 such non-uniformity of PM-mortality relationships exists across the seasons, the year-round
27 relationship may underestimate or mask within-season associations. Thus, it may be naive to
28 apply one model specification without considering possible changes in lags. Season specific
29 analyses may be more appropriate in certain cases. However, when the total data base is not
30 large, the small sample size in such season specific analysis may compromise statistical
31 power to detect 'small' PM-mortality/morbidity associations.
April 1995 12-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 Weather and Climate: Factors in Studies Relating PM to Mortality
2 The confounding impact of weather on acute mortality can be significant, and
3 determination of the differential and/or synergistic impacts of weather and PM on death rates
4 is necessary in any PM-mortality analysis. There is much variation among studies as to how
5 weather should be considered or controlled. Also, weather and PM concentrations are
6 themselves related, and this collinearity problem can render any effort to adjust for weather
7 rather difficult. The role of weather, e.g., day-to-day meteorology (especially extreme
8 events) is very important when analyzing PM effects on acute mortality.
9 Some authors have conducted weather/pollution/mortality evaluations in Steubenville,
10 OH; Philadelphia, PA; London, England; Birmingham, AL; and Utah County, UT as well as
11 other locales (Table 12-2). In all of these studies reported significant associations are
12 reported between human mortality and PM; and, in some cases, the relationship extends to
13 levels well below the current PM NAAQS. Several also allude to significant
14 weather-mortality relationships. For Steubenville, a positive non-linear relationship among
15 temperature, dew point, and mortality was detected. When dummy variables were used to
16 denote hot days, humid days, and hot/humid days, the hot/humid days were a significant
17 regression models however, neither temperature nor dew point proved to be significant
18 mortality predictor. When seasonal variations were controlled for in their Poissonpredictors
19 of mortality (Schwartz and Dockery, 1992b). In a study of British Smoke in London,
20 Schwartz and Marcus (1990) controlled for temperature and humidity and improved the
21 goodness of fit of the model significantly compared to the model with no meteorological
22 variables. The reduction of otherwise unexplained variability in the daily mortality series
23 had little effect on the regression coefficient for British Smoke, but increased its statistical
24 significance.
25 More recent analyses (Kalkstein et al., in press) suggest that controls for weather may
26 have not been fully adequate in the studies cited above. Many PM-mortality studies used
27 temperatures, categories by temperature percentiles, squared temperature and dewpoint
28 values, moving averages of temperature, and mean temperatures for groupings of days (see
29 Table 12-2), which may not provide adequate bases to detect true weather-mortality
30 relationships. It is also probably not feasible to assume that cities within a wide range of
31 climates demonstrate similar responses of daily mortality to daily changes in
April 1995 12-28 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-2. ADJUSTMENTS FOR METEOROLOGICAL FACTORS IN SOME
RECENT STUDIES RELATING MORTALITY TO PARTICULATE MATTER
Location
Studied and Authors
London
Schwartz and
Marcus (1990)
Philadelphia
Schwartz and
Dockery (1992a)
Steubenville
Schwartz and
Dockery (1992b)
Utah
Pope, Schwartz and
Ransom (1992)
Erfurt.
East Germany
Spixetal. (1993)
Birmingham
Schwartz (1993)
Steubenville
Moolgavkar et al. (1995)
Philadelphia
Wyzga and Lipfert (1994)
Pollution
Data and Treatment
British Smoke measurements
from 7 stations;
logarithmic and square root
transformations;
TSP samples collected
routinely at two monitors;
supplemented by sampling
every sixth day at several
sites; daily means and lags
used
TSP from one monitor;
ranked by levels and sorted
into quartiles;
PM10 level from one site;
up to 7-day lagged moving
averages; divided into
quintiles used as dummy
variables;
Suspended participates;
0-3 day lags; logarithmic
transformation;
PM10 level averaged from
all (1-2) city monitors;
divided into quartiles used
as dummy variables;
TSP from one monitor and
SO2 (two series)
24 h averages of Oj and
TSP from several city
monitors; lags of 0 to 4
days tested
Mortality
Data and Treatment
Daily total death counts,
including respiratory
and cardiovascular
causes; sensitivity to
filtering
Daily total, elderly,
<65, pulmonary
disease, pneumonia,
cardiovascular and
cancer mortality;
Poisson regression
using GEE;
Daily total mortality;
Poisson and weather
model regressions;
Daily total, non-
accidental respiratory,
cardiovascular, and all
other causes of non-
accidental mortality;
Poisson regression;
Daily total mortality;
Poisson regression;
autocorrelation
adjustment for
"harvesting"
Daily total mortality;
Poisson regression
using GEE; dummy
variables for year and
day of week;
biannual cycle filters;
Daily total non-
accidental mortality;
Poisson regression; with
and without GEE; full
year and season; serial
correlation unimportant.
Daily non-accidental
deaths (non-elderly and
elderly); linear filtering;
variable for time over
entire period in stepwise
regression and forced
OLS.
Weather Data
and Treatment
Temperature and RH;
grouped plots of
temperature and humidity
versus mortality;
autoregressive model.
Mean 24 h temp and DP
including squared
transformation;
indicator variables for
season, hot, cold, humid,
and hot humid days, and
year.
Mean 24 h temp and DP;
year as random effect;
indicator variables for
hot, humid, and hot
humid days;
Temp and RH; dummy
variables used for 10 °F
ranges, previous day's
temp, 5-day temp moving
average, and humidity;
linear time trend, random
year effect;
Daily mean temp, RH,
precipitation; indicator
variables used for very
cold days and hot days
for different thresholds,
various lags;
Mean 24 h temp and DP;
dummy variables same as
Utah study, plus cold
days; 3-day moving lags;
Mean 24 h temp and DP;
indicator variable for hot
and humid days;
temperature quintiles;
Daily maximum temp.;
daily change in
barometric pressure;
dummy variables for
winter and seasonality
Pollution/
Weather Impact
British Smoke is
significant predictor of
mortality; temp/humidity
control increased
significance, as did
autocorrelation adjustment
Significant TSP association
mortality, strongest among
elderly and respiratory
patients; hot days, mean
DP, other weather factors
also associated
Nonlinear association
between TSP and daily
mortality; hot humid days
associated with daily total
mortality;
Relative risk of death
increased monotonically
with the mean PM10 level
for each quintile; also
observed when weather
controlled;
Effects of air pollution
smaller than influenza and
weather effect; significant
SO2
Significant association
between PM10 and daily
mortality; extremely hot
weather also associated
with excess mortality;
relation to temperature.
TSP influence on mortality
greatly reduced when SO2
included in analysis;
choice of SO2 series and
season had large impact on
mortality results.
Strong relationship
between temp, and
mortality; seasonal
adjustments very
important; TSP-temp.
interaction; most mortality
with TSP on hot days.
1 weather or in air pollution, and there are possibly some regional similarities in response
2 which have not been adequately explored. In a reanalysis of Philadelphia PM-mortality
April 1995
12-29
DRAFT-DO NOT QUOTE OR CITE
-------
1 relationships, Schwartz (1994) took a more flexible approach to examine the possibility of
2 confounding weather impacts. The reanalysis utilized Hastie and Tibshirani's (1990)
3 "Generalized Additive Model" to detect and control for nonlinearities in the dependence of
4 daily mortality on weather; nevertheless, and yielded findings similar to the original
5 Philadelphia study. On the other hand, Moolgavkar (1995) asserts that the role of weather
6 was improperly evaluated within the Steubenville study, and suggests that weather-related
7 variables be evaluated more comprehensively in future PM-mortality analyses.
8 To further interpret possible weather impacts, Schwartz (1994) stated that the similar
9 responses to air pollution in the mild weather of Philadelphia (mean daily temperature of
10 57 °F) and the cold weather of London suggest that weather does not seriously confound the
11 PM-mortality relationship. Schwartz (1994) further noted that similarity in temperature and
12 humidity on high and low air pollution days (when different mortality responses are seen),
13 "... also would seem to eliminate weather as a potential confounder." However, it is
14 possible that these studies do not remove the total confounding influence of weather,
15 especially because of their dependence on mean temperatures and other meteorological
16 surrogates which may not fully reflect weather variation associated with biological responses.
17 Other studies have attempted to assess the differential impact of PM and weather on
18 acute mortality. For example, Ostro (1993) summarized studies which show strong
19 associations between exposures to PM10 and total daily mortality for many urban areas in the
20 U.S., Europe, and Canada and noted remarkably consistent results across regions.
21 Ito et al. (1993) showed that daily mortality in London was significantly associated with
22 aerosol acidity levels and British Smoke. The amount of variation explained by weather was
23 smaller than that explained by air pollution. Ito's work confirms results obtained by others
24 on London's mortality/PM/weather relationship (Schwartz and Marcus, 1990; Thurston
25 et al., 1989; Mazumdar et al., 1982). However, it should be noted that London's marine
26 climate is rather benign when compared to many large American cities, as thermal extremes
27 are unusual, so that the mortality variations attributable to variations in weather during
28 London winters may be relatively smaller than in most U.S. cities studied.
29 Some studies for cities exhibiting higher climate variation yielded somewhat different
30 results. Wyzga (1978) used COH as a surrogate measure of PM concentration, and
31 determined that high COH values are associated with increased mortality in Philadelphia.
April 1995 12-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 However, he recognized the potential impact of extreme weather as well, and noted that heat
2 waves may also be responsible for large numbers of extra deaths. In a recent study in which
3 weather was treated in a more sophisticated manner (Wyzga and Lipfert, 1995), the impact
4 of O3 concentrations and weather on acute mortality were evaluated and results were
5 compared to TSP. The authors concluded that a determination of O3 and TSP impacts is
6 most difficult because of the influence of confounders, particularly weather. Also, use of
7 different explanatory models yields disparate results, with pollution impacts ranging, "...from
8 essentially no effect to response similar to that associated with a 10 °F increase in ambient
9 temperature" (Wyzga and Lipfert, 1995). This evaluation appeared to find a synergistic
10 relationship between weather and pollution, as days with maximum temperatures exceeding
11 85 °F contributed most to the TSP-mortality associations.
12 Several other studies have uncovered synergistic relationships. Ramlow and Kuller
13 (1990) found that daily mortality was most closely associated with the daily average
14 temperature of the previous day rather than any pollution measure in Allegheny County, PA
15 where the mortality effect varied with temperature. In a study which attempts to determine
16 synergistic relationships between weather and pollution on mortality in Los Angeles,
17 Shumway et al. (1988) determined that mortality is, "...an additive nonlinear function of
18 temperature and pollution, whereas there may be significant interactions present, especially
19 when low or high temperatures are combined with high pollution levels." The authors found
20 that model-predicted average mortality values increased at both temperature extremes when
21 particulate levels were held constant. Two evaluations in the Netherlands found also
22 temperature extremes in summer and winter to be significantly correlated with mortality
23 variation. Kunst et al. (1993) and Mackenbach et al. (1993) determined that the relationship
24 between temperature and mortality is linear, producing a U-shaped temperature curve, with
25 minimum mortality rates observed between 10 to 15 °C. Kunst found that summer acute
26 mortality was not influenced by air pollution variations.
27 Although weather seems to induce mortality increases when temperatures are either
28 very warm or very cold, the impact of weather as a confounder or as a modifier of the effect
29 of air pollution varies seasonally. For example, the impact of weather on acute mortality in
30 winter is much more difficult to evaluate, and thermal relationships are decidedly weaker
31 (WHO/WMO/UNEP, in press). Thus, although the temperature/mortality relationship might
April 1995 12-31 DRAFT-DO NOT QUOTE OR CITE
-------
1 follow the U-shaped curve discussed in Kunst et al. (1993), the variation of mortality around
2 the cold portion of the curve is much larger, and the magnitude of mortality change for a
3 given change at low temperatures is smaller. This implies that adjustment procedures for
4 weather as a confounder should not be identical for summer and winter.
5 Another means of evaluating weather as a differential seasonal confounder within
6 PM-mortality studies is to include variables to adjust for seasonal cycles. A recent Chicago
7 and Los Angeles (Ito et al., in press) study, where seasonal cycles were determined using
8 sine/cosine curves, found only 10 to 20% of the PM variance attributable to seasonal cycles
9 in Los Angeles and an even lower percentage of explained variance in Chicago.
10 In general, either air pollution or weather (especially temperature) can appear as an
11 additive main effect (possibly nonlinear) in a mortality time series model. Which is more
12 important depends on both the range of variation in pollution and weather, and on the
13 strength of their association with mortality in any study.
14 Many studies used mean meteorological data (e.g., mean daily temperature or relative
15 humidity) or running means, which may underestimate the impact of meteorological extremes
16 upon human mortality. Thus some mortality attributed to weather variation may have been
17 appropriated to PM due to inadequate control for meteorology within the models used. Some
18 studies have effectively evaluated maximum or minimum daily temperature, but virtually
19 none have determined how weather situations (comprised of combinations of several
20 variables) contribute to heightened mortality. This implicitly assumes that humans respond to
21 specific individual meteorological variables such as temperature, rather than to the interaction
22 of numerous meteorological variables which comprise a weather situation (e.g., combined
23 temperature and relative humidity stress).
24
25 Application of Synoptic Weather Patterns for Adjusting Daily Mortality Series
26 The use of arbitrary temperature increments, various weather "dummy variables", and
27 moving averages may have the effect of dampening the true meteorological confounding
28 impact within PM-mortality studies (Kalkstein, 1991). These approaches ignore the fact that
29 the meteorology of a locale is defined by discrete, identifiable weather situations, and
30 humans, animals, and pollution concentration respond to these situations. One means of
31 identifying the meteorological character of these situations (which vary through space) is to
April 1995 12-32 DRAFT-DO NOT QUOTE OR CITE
-------
1 graph the frequency of daily temperature-dewpoint temperature combinations for a particular
2 location. Such a plot clearly demonstrates that certain temperature-dewpoint modes exist,
3 containing a high proportion of days. These modes represent "air masses", which frequently
4 invade a region. Many fewer days reside between the modes (these represent "transition
5 situations" between air masses). Thus, days can be clustered within air mass categories,
6 which represent meteorologically realistic combinations of weather elements occurring at a
7 locale. A more comprehensive picture of weather can be determined if air mass designators
8 are inputted within PM-mortality analyses, rather than arbitrary thermal divisions, which are
9 not based on meteorological reality. Since the air masses tend to persist for several days for
10 any city, they also provide a natural method for defining weather episodes and duration of
11 exposure during these episodes. This "synoptic climatological approach" is gaining
12 increasing favor among climatologists, health scientists, and ecologists, but has been
13 distinctly under-utilized to date within available PM-mortality analyses.
14 The synoptic climatological classification method proposed by Kalkstein is easily
15 implemented. The amount of subjective meteorological and bioclimatological judgement used
16 in defining distinct weather regimes can vary, with some approaches nearly completely
17 automated (Davis and Kalkstein, 1990; Eder et al., 1994). One approach requires use of
18 24 daily meteorological measurements: temperature, dewpoint, surface barometric pressure,
19 East-West and North-South components of windspeed, and cloud cover observed at 6-h
20 intervals. Principal component scores (linear combinations of variables measured at each
21 time point) are then grouped using several cluster formation strategies. This generally results
22 in 9 to 11 distinct weather regimes, some with only a few days being clearly transitional
23 between two primary regimes. It is generally possible to assign meaningful meteorological
24 characterizations to the major groupings.
25 A "synoptic climatological approach" was employed to model climate as a holistic unit
26 composed of multiple meteorological elements in a study of the impact of summer weather
27 and pollution on St. Louis mortality (Kalkstein, 1991). The methodology provides a more
28 complete depiction of the weather than any one-variable surrogate (e.g., temperature), as it
29 assumes that people respond to combined characteristics of the entire umbrella of air (or "air
30 mass") that surrounds them (Kalkstein, 1993). Each day is designated by air mass type, and
31 those air masses associated with heightened mortality are deemed "offensive". In the St.
April 1995 12_33 DRAFT-DO NOT QUOTE OR CITE
-------
1 Louis study, an offensive air mass was determined which possessed mean daily mortality
2 values of 33% above baseline levels (Kalkstein, 1991). Those days within the offensive air
3 mass with the highest mortality totals were the hottest and/or part of a multiple-day
4 sequence. However, PM values within all offensive category days averaged close to the
5 summer mean for St. Louis and were found to have little effect on mortality.
6 A more complete study using the synoptic methodology for four U.S. cities
7 (Philadelphia, Birmingham, Cleveland, Seattle) supported the St. Louis findings (Kalkstein
8 et al., in press; Kalkstein and Smoyer, 1993). In addition, an attempt was made to
9 determine if PM contributed to heightened mortality in a differential manner during offensive
10 and non-offensive air mass situations in Philadelphia. Days were divided into two groups of
11 PM concentration quintiles: one for the offensive air mass and the second for all days within
12 non-offensive air masses. Mean daily mortality was then calculated for each quintile
13 (Figure 12-2). The results indicated that PM had little impact on mortality when the
14 offensive air mass was present. In fact, mean mortality was lower in the highest PM
15 concentration quintile than in the lowest PM quintile during offensive air mass days. It is
16 noteworthy that the mean daily mortality for all PM quintiles within the offensive air mass
17 was well above the baseline and, in some cases, approached or exceeded 10 additional deaths
18 per day. Kalkstein suggested that the magnitude of the mortality totals on offensive air mass
19 days was due almost exclusively to the associated stressful weather and that PM
20 concentration played almost no role. However, for days not within the offensive air mass, a
21 small systematic increase in mortality (which was not statistically significant) was noted as
22 PM concentration increased. Nevertheless, mean daily mortality was below the baseline even
23 within the highest PM concentration quintile when the weather was not offensive. Kalkstein
24 concluded that PM has small impact on mortality in these cities during nonstressful weather,
25 and when the weather is stressful, the "add-on" effect of high PM concentration is
26 undetectable.
27
April 1995 12-34 DRAFT-DO NOT QUOTE OR CITE
-------
oo
OS
u> to i—
to
o
o
1
o
C
O
n
vj i
»• §
n>
o
=S K-
C si 5'
" « (TO
3 1
I i
er
OQ
* ~
rc
offe
O
o
&
§•'
ff
ing
8* g
CD
5:
fD
TO
TO
ft
CA
z
- ff
O 03
O ^
8"
*r)
I S
ff B-
s* a
* «
CO
»-• >—; a
3 or 3
ft s
& ^
«2 o
CA 2-
I 8"
ff 8-
ff
o
I
e
8"
a
! 1
co ,5-
a «
CA £
O O.
35--
ft
fD CA
en
I! ffi
§•• i
CA n;
nr <
1 "
§
-h
(T(
I
rt
"i
fa
a
O
i
B.
the pro
eans
o
o
H
3*
I
rt
1
CA
CA
5'
3
i i
^T fc5
PJ Jjl
tn f^
O
K> _
8
*•
o
>-t>
CO
ia TO
n 3
B i
a
8
(3 rt «
5- o °
8-
81
f?5
3
B.
CO
oa
o
Si
I
5
a
S
1
a
•a
Mean Daily Mortality (variation from baseline)
i
a
82.
z
•g
I
•a
-------
1 Thus, disparate results from several studies suggest that relative roles of meteorology
2 and PM concentration in contributing to human mortality are not readily differentiated.
3 However, there is growing evidence that weather is, at minimum, a significant confounder
4 that must be considered in attempting to estimate PM/mortality relationships. See Appendix
5 12A for more discussion of specific weather impacts on mortality.
6
7 Confounding by Epidemics
8 Concern exists that the increased incidence of illness or mortality associated with
9 changes in air pollution during the winter season may not indicate a causal relationship
10 because of confounding influences of contagious illnesses epidemics. Infectious respiratory
11 illness (e.g., the "flu") strongly influences mortality. An underlying or contributing cause
12 for changes in air pollution or in contagious illness may be weather changes. Confounding
13 due to epidemics may be adjusted statistically to some extent by use of filtering, but this is at
14 best suitable for time series with a normally distributed response, and filtering of time series
15 may perform better when there is some recurrent medium-to-long wave pattern to outbreaks
16 of the disease in a given population. Without some recurrence pattern, filtering may only
17 eliminate evidence of longer-term persistence of health effects related to air pollution. Close
18 inspection of the time course of infectious respiratory illness outbreaks in populations reveals
19 that outbreaks do not appear on a regular schedule from year to year (Henderson et al.,
20 1979; Murphy et al., 1981; Chapman et al., 1981; Denny et al., 1983). In any given year, a
21 number of important respiratory pathogens may not appear at all in a given population.
22 Thus, fixed-cycle curve-smoothing techniques may not accurately describe the time course of
23 respiratory illness outbreaks in populations. Several investigators have subsequently used
24 long-term nonparametric methods such as lowess smoothers or generalized additive models
25 (GAM) to adjust mortality series for aperiodic fluctuations that may include time-extended
26 outbreaks of respiratory disease (Pope, 1994; Schwartz, 1994, 1995).
27 It is sometimes possible to evaluate the effect of epidemics on health outcome time
28 series by comparison with adjacent communities. Pope (1991c) evaluated the possible effect
29 on hospital admissions of contagious illnesses such as influenza (which is known to cause a
30 substantial number of deaths in the elderly) and respiratory syncytial virus (RSV, which
31 affects a substantial number of children and is often mistakenly diagnosed as influenza).
April 1995 12-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 There was particular interest in the possibility that infectious diseases occurred more often
2 during the winters when the Utah Valley steel mill was open, and less often during the
3 winter when the steel mill was closed, purely by chance. Pope writes that "The few
4 diagnoses where the agent of disease was specified limited opportunities to directly observe
5 epidemics of any specific infectious agent. Bronchitis and asthma admissions for preschool-
6 age children were more than twice as high in Utah Valley during periods when the mill was
7 operating than when it was closed. The potential of highly localized epidemics of contagious
8 respiratory disease that were correlated coincidentally with the operation of the steel mill
9 cannot be completely ruled out. If the association were strictly spurious, however, the same
10 correlation would probably be observed in neighboring communities unaffected by the mill's
11 pollution. Such correlations were not observed."
12 The ability to directly observe diagnosed cases of influenza or RSV would allow a
13 direct adjustment of health outcome time series for community-wide incidence of occurrence
14 of the disease, which could in turn be exacerbated by co-occurring air pollution. Some
15 progress in obtaining data on outbreaks of influenza-like illnesses (ILI) may be possible using
16 recently established data bases, such as the CDC volunteer physician surveillance network.
17 Evidence exists that these 140 family physicians make good sentinels for epidemics of ILI
18 (Buffington et al., 1993). Data are provided to CDC on a weekly basis, which seems
19 appropriate to the level of filtering that may be needed to adjust daily tune series of health
20 outcomes and air pollution for co-occurring respiratory diseases.
21
22 12.2.6 Selection Bias
23 The possibility of selection bias is a concern of every assessment. Selection bias would
24 require selection of participants based on exposure and also health outcome. Because most
25 epidemiologic studies of these exposures are population based, there is little possibility of
26 selection bias affecting health end-points. Nevertheless, the possible loss of subjects by non-
27 participation and attrition differentially associated with both exposure and health studies must
28 be considered.
29
30
31
April 1995 12-37 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.2.7 Publication Bias
2 Publication bias, and the related "file drawer problem", refers to increased likelihood of
3 publication of studies that have positive results, leading to a bias in the literature reviewed
4 towards positive results (Rosenthal, 1979). The "file drawer problem" is used to describe
5 negative studies not submitted or published. However, large scale epidemiological studies
6 are published regardless of their results. In fact, the study design and methods usually
7 appear as papers prior to publication of results. Analyses of preexisting data sets such as
8 mortality and hospital admissions present a different problem. Here the results can be
9 calculated using fewer resources, and the likelihood of unpublished negative results increases.
10 There is no adequate way to ascertain or correct for this bias; it does suggest, however, that
11 more weight be attached to the large scale epidemiological studies.
12
13 12.2.8 Internal Consistency and Strength of Effects
14 Internal consistency is always a check on the validity of a study, but often the authors
15 do not report sufficient detail by which to check for such consistency. For known risk
16 factors for respiratory effects, a study should provide evidence of expected associations.
17 Furthermore, certain patterns of age or gender relationships to observed health outcomes
18 should be expected. For example, studies of lung function should show a decrease as a
19 function of increasing age if the subjects are old enough and the follow-up time is long
20 enough. Consistency between studies provides a further indication of the overall strength of
21 the total data base.
22
23 Sample size and power of reported PM-mortality associations
24 Since the size of the 'relative risks' and the extent of associations found in recent
25 observational studies of PM-mortality are not 'large', such associations are unlikely to be
26 shown in a 'small' sample size (i.e., a limited number of days). This can be particularly
27 problematic if one plans to analyze the data using PM data collected at the current U.S.
28 sampling frequency (i.e., every-6th-day). It should be noted that a majority of the existing
29 studies that reported significant PM-mortality associations used PM data that were collected
30 daily. A determination of the sample size required to find the observed association in a given
31 community is not simple, because power may be dependent on not only sample size, but also
April 1995 12-38 DRAFT-DO NOT QUOTE OR CITE
-------
1 on: (1) the population size of the community (to produce certain number of deaths per day);
2 (2) the levels of PM; (3) the proportion of susceptible populations (e.g., age/race/gender
3 distribution); (4) the location and number of PM sampling sites to estimate representative PM
4 exposure of the population, and; (5) the model specification. Also, determining the expected
5 'effect' size from the published studies alone may be misleading because of potential
6 'publication bias' towards significant effects. With this caveat in mind, one can illustrate the
7 effects of sample size and the above mentioned factors on the significance of PM/total daily
8 mortality associations, by examining the t-ratios of the PM coefficients reported by recent
9 U.S. PM-mortality studies (Table 12-1). When both multi-pollutant models and single
10 pollutant models were presented, a single pollutant model was selected here. All the models
11 included weather variables. When both Poisson models (log-linear GLM) and OLS models
12 were presented (Schwartz, 1993; Kinney et al., 1995), both gave essentially identical t-ratios,
13 and therefore the results for Poisson models are shown. Despite the magnitude of differences
14 in various studies' population/mean deaths, the key predictor of t-ratio appears to be the
15 number of study days (sample size).
16 In a simple linear regression, the t-ratio for the null hypothesis of a regression
17 coefficient being zero is a function of square-root of sample size, with its slope being
18 r/O-r2)0-5, where r is the underlying size of the correlation between the dependent variable
19 (e.g., mortality) and the explanatory variable (e.g., PM). The plot (Figure 12-3) of these t-
20 ratios versus square-root of sample days from Table 12-1 in fact shows the t-ratio's strong
21 linear dependency on the square-root of sample size. The magnitude of PM-mortality
22 associations seen in these studies, as reflected in the slope (r=0.083, if the slope is equated
23 to r/Q-r2)0-5, requires about n=600 days for the association to be significant at 0.05 (two-
24 tailed), or n=400 for one-tailed test at 0.05 level. The required sample size observations to
25 detect this size of r with 80% power is about 800 days. Therefore, findings of statistical non-
26 significance of PM effect may reflect inadequate power to detect an effect of this magnitude
27 if sample size is limited.
28
29
April 1995 12-39 DRAFT-DO NOT QUOTE OR CITE
-------
5-
4
o
"? 31
^
2-
1-
Philadelphia
t
Utah*
• Steub.
Cook
•
Detroit
Santa Clara
Cincinnati
St. Louis/ Birmingham
* Kingston
2505001,000 2,000
sample size (days)
(square-root)
—I—
4,000
Figure 12-3. t-ratios of PM coefficients versus sample size (days) from 11 recent U.S.
studies.
1 12.2.9 Plausibility of Effects
2 Health outcomes measured should be ones for which there are plausible bases to suspect
3 that they could be affected by PM exposure. Health outcome measures of greatest interest in
4 PM epidemiologic studies include: lung function measurements, respiratory symptoms and
5 illness occurrence, hospital admissions, and mortality.
6
1 12.2.10 Qualitative Versus Quantitative Studies
2 Studies which do not measure PM exposure but only infer it from some other related
3 indicator (e.g., coal consumption, level of mining activity, etc.) are by definition qualitative.
4 Other studies that measure PM exposures in broad categories (e.g., low, medium, high) and
5 do not relate the exposure to the health effects in an exposure-response fashion are also
6 qualitative. A quantitative study is one in which PM exposures are estimated with reasonable
April 1995
12-40
DRAFT-DO NOT QUOTE OR CITE
-------
1 accuracy as a continuous variable, important potential covariates are adequately controlled,
2 and the response variable reliably varies as a function of increased PM exposure.
3
4 12.3 HUMAN HEALTH EFFECTS ASSOCIATED WITH SHORT-TERM
5 PM EXPOSURE
6 Some of the earliest indications that short-term ambient air paniculate matter or acid
7 aerosols exposure may be associated with human health effects were derived from the
8 investigation of historically well-known, major air pollution episode events. These include
9 the Meuse Valley (Belgium), Donora, PA (USA), and London (UK) episodes.
10 Firket (1931) described December 1930 fog in the Meuse Valley and morbidity and
11 mortality related to them. More than 60 persons died from this fog and several hundred
12 suffered respiratory problems, with a large number becoming complicated with
13 cardiovascular insufficiency. The mortality rate during the fog was more than 10 times
14 higher than normal. Those persons especially affected were the elderly, those suffering from
15 asthma, heart patients and other debilitated individuals. Most children were not allowed
16 outside during the fog and few attended school. Unfortunately, no actual measurements of
17 pollutants in ambient air during the episode are available by which to establish clearly their
18 relative roles in producing the observed health effects.
19 Schrenk et al. (1949) reported on the atmospheric pollutants and health effects
20 associated with the Donora smog episode of October 1948. A total of 5,910 persons (or
21 42.7%) of the Donora population experienced some effect. The air pollutant-ladened fog
22 lasted from the 28th to the 30th of October, and during a 2-week period 20 deaths occurred,
23 18 of them being attributed to the fog. An extensive investigation by the U.S. Public Health
24 Service concluded that the health effects observed were mainly due to an irritation of the
25 respiratory tract. Mild upper respiratory tract symptoms were evenly distributed through all
26 age groups and, on average, were of less than four days duration. Cough was the most
27 predominant symptom; it occurred in one-third of the population and was evenly distributed
28 through all age groups. Dyspnea (difficulty in breathing) was the most frequent symptom in
29 the more severely affected, being reported by 12% of the population, with a steep rise as age
30 progressed to 55 years; above this age, more than half of the persons affected complained of
31 dyspnea. While no single substance could be clearly identified as being responsible for the
April 1995 12-41 DRAFT-DO NOT QUOTE OR CITE
-------
1 October 1948 episode, the observed health effects syndrome seemed most likely to have been
2 produced by two or more of the contaminants, i.e., SO2 and its oxidation products together
3 with PM, as among the more significant highly elevated contaminants present.
4 Based on the Meuse Valley mortality rate, Firket (1931) estimated that 3,179 sudden
5 deaths would likely occur if a pollutant fog similar to the Meuse Valley one occurred in
6 London. An estimated 4,000 deaths did later indeed occur during the London Fog of 1952,
7 as noted by Martin (1964). During the fog of 1952, evidence of bronchial irritation,
8 dyspnea, bronchospasm and, in some cases, cyanosis is clear from hospital records and from
9 the reports of general practitioners. There was a considerable increase in sudden deaths from
10 respiratory and cardiovascular conditions. The nature of these sudden deaths remains a
11 matter for speculation since no specific cause was found at autopsy. Evidence of irritation of
12 the respiratory tract was, however, frequently found and it is not unreasonable to suppose
13 that acute anoxia due either to bronchospasm or exudate in the respiratory tract was an
14 important factor. Also, the United Kingdom Ministry of Health (1954) reported that in the
15 presence of moisture, aided perhaps by the surface activity of minute solid particles in fog,
16 some sulfur dioxide is oxidized to trioxide. It is possible that sulfur trioxide, dissolved as
17 sulfuric acid in fog droplets, appreciably augmented the harmful effects of PM and/or other
18 pollutants.
19 The occurrence of the above episodes and resulting marked increases in mortality and
20 morbidity associated with acute exposures to very high concentrations of air pollutants
21 (notably including PM and SO2 in the mix):
22
23 (1) left little doubt about causality in regard to induction of serious health effects by
24 very high concentrations of air pollutants;
25
26 (2) stimulated the establishment of air monitoring networks in major urban areas and
27 control measures to reduce air pollution; and
28
29 (3) stimulated research to identify key causative agents contributing to urban air
30 pollution effects and to characterize associated exposure-response relationships.
31
32 Besides evaluating mortality associated with major episodes, the 1982 criteria
33 document (U.S. Environmental Protection Agency, 1982) also focused on epidemiology
34 studies of more moderate day-to-day variations in mortality within large cities in relation to
April 1995 12-42 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM pollution. Evaluating risks of mortality at lower exposure levels, the 1982 criteria
2 document concluded that studies conducted in London by Martin and Bradley (1960) and
3 Martin (1964) yielded useful, credible bases by which to derive conclusions concerning
4 quantitative exposure-effect relationships. The 1986 addendum to the 1982 criteria document
5 (U.S. Environmental Protection Agency, 1986) also considered several additional acute
6 exposure mortality studies of London, England during the 1958-1959 through 1971-1972
7 winter periods conducted by Mazumdar et al. (1982), Ostro (1984), Shumway et al. (1983),
8 and by EPA (later published in Schwartz and Marcus, 1990). After reviewing these various
9 data analyses, and taking into account the previously reviewed London results and the above
10 noted epidemiological considerations, the following conclusions were made (U.S.
11 Environmental Protection Agency, 1986):
12
13 (1). Markedly increased mortality occurred, mainly among the elderly and chronically
14 ill, in association with BS and SO2 concentrations above 1,000 /ig/m3, especially
15 during episodes when such pollutant elevations occurred for several consecutive
16 days;
17
18 (2). During such episodes, coincident high humidity or fog was also likely important,
19 possibly by providing conditions leading to formation of H2SO4 or other acidic
20 aerosols;
21
22 (3). Increased risk of mortality is associated with exposure to BS and SOj levels in the
23 range of 500 to 1,000 /ig/m3, for SO2 most clearly at concentrations in excess of
24 «700/ig/m3; and
25
26 (4). Convincing evidence indicates that relatively small, but statistically significant,
27 increases in the risk of mortality exist at BS (but not SO2) levels below 500
28 Mg/m3, with no indications of any specific threshold level having been
29 demonstrated at lower concentrations of BS (e.g., at < 150 /ig/m3). However,
30 precise quantitative specification of the lower PM levels associated with mortality
31 is not possible, nor can one rule out potential contributions of other possible
32 confounding variables at these low PM levels.
33
34 The extensive epidemiological research that ensued has greatly advanced our knowledge
35 regarding the above issues, especially roles played by PM and SO2 in mortality and
36 morbidity associated with both episodic and non-episodic (lower level) exposures to these
37 and/or other co-occurring pollutants. Key studies and findings from such research on
38 mortality associated with short-term exposures to paniculate matter are evaluated in the
April 1995 12.43 DRAFT-DO NOT QUOTE OR CITE
-------
1 following section. Additional discussion on validity of model specification is in
2 Section 12.6.2.
3
4 12.3.1 Mortality Effects Associated With Short-Term Participate Matter
5 Exposures
6 Mortality is a synonym for death rate, the ratio of the total number of deaths in a
7 specified area to the population, generally figured in terms of a number of deaths per i.e.,
8 100,000 of population. For a mortality rate, the numerator is the number of persons dying
9 during a stated period; the denominator is the total population within which the deaths
10 occurred (MacMahon and Pugh, 1970). Thus, the mortality rate expresses the frequency
11 with which members of a general population die. The difference between an increase in the
12 death rate and the cessation of life for an individual is important. Mortality is a population
13 statistic, death relates to an individual. Factors effecting changes in population mortality in
14 relation to population exposure to PM may exert themselves in both similar and dissimilar
15 ways as contrasted to factors associated with the cessation of life for an individual and
16 personal PM exposure. Also, premature death and the related aspect of length of time of
17 potential life lost are important aspects of the end point of interest. The biological
18 plausibilities of mortality and population PM exposure may be different and involve different
19 factors than death and exposure for an individual.
20 The National Center for Health Statistics (NCHS) mortality statistics used in most U.S.-
21 mortality studies published were compiled in accordance with the World Health Organization
22 (WHO) regulations, which specify that member nations classify causes of death by the
23 current Manual of the International Statistical Classification of Diseases, Injuries, and Causes
24 of Death (World Health Organization, 1977). Causes of death for 1979-91 were classified
25 according to the manual. For earlier years, causes of death were classified according to the
26 revisions then in use—1968 through 1978, Eighth Revision; 1958 through 1967, Seventh
27 Revision; and 1949 through 1957, Sixth Revision. Changes in classification of causes of
28 death due to these revisions may result in discontinuities in cause-of-death trends.
29 National Center for Health Statistics tabulations of cause-of-death statistics are based
30 solely on the underlying cause of death. The underlying cause is defined by WHO as the
31 disease or injury that initiated the sequence of events leading directly to death, or as the
April 1995 12-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 circumstances of the accident or violence that produced the fatal injury. It is selected from
2 the conditions entered by the physician in the cause-of-death section of the death certificate.
3 When more than one cause or condition is entered by the physician, the underlying cause is
4 determined by the sequence of conditions on the certificate, provisions of the ICD, and
5 associated selection rules.
6 One index of the quality of reporting causes of death is the proportion of death
7 certificates coded to the Ninth Revision, Chapter XVI, Symptoms, signs, and ill-defined
8 conditions (ICD-9 Nos. 780-799). Although deaths occur for which the underlying causes
9 are impossible to determine, this proportion indicates the care and consideration given to the
10 certification by the medical certifier. This proportion also may be used as a rough measure
11 of the specificity of the medical diagnoses made by the certifier in various areas. In 1991,
12 1.12% of all reported deaths in the United States were assigned to Symptoms, signs, and ill-
13 defined conditions, the same as 1990. However, trends in the percent of deaths assigned to
14 this category vary by age. Although the percent of deaths from this cause for all ages
15 combined generally has remained stable since 1980, decreases have occurred for the age
16 group 55 to 64 years since 1983; and for 10-year age groups from 15 to 44 years since 1988.
17 Between 1990 and 1991, the percent increased for all age groups, except for those 15 to 44
18 and 55 to 64 years.
19 Mortality statistics are based on information code by the States and provided to NCHS
20 through the Vital Statistics Cooperative Program and from copies of the original certificates
21 received by the NCHS from the State Registration Offices. The National Center for Health
22 Statistics (1993) reported that in 1991, in the United States, the death rate was 860.3 deaths
23 per 100,000 population. In 1991 a total of 2,169,518 deaths occurred in the United States.
24 The first leading causes of death — diseases of the heart; malignant neoplasms; and
25 cerebrovascular diseases accounted for 64% of deaths. Chronic obstructive pulmonary
26 disease and allied conditions surpassed accidents in 1991 as the fourth leading cause.
27 In 1991, life expectancy at birth reached a record high at 75.5 years. For those
28 between 65 and 70 years of age, the average number of years of life remaining is 17.4 years.
29 Women currently are expected to outlive men by an average of 6.9 years and white persons
30 are expected to outlive black persons by an average 7.0 years. In 1991, the age-adjusted
31 death rate for males of all races was 1.7 times that for females. In 1991, the age-adjusted
April 1995 12-45 DRAFT-DO NOT QUOTE OR CITE
-------
1 death rate for the black population was 1.6 times that for the white population. The annual
2 asthma death rate was constantly higher for blacks than for whites during the period 1980
3 through 1990; for blacks, the rate increased 52% (from 2.5 to 3.8 per 100,000), compared
4 with a 45% increase (from 1.1 to 1.6 per 100,000) for whites. (United States Center for
5 Disease Control, 1994). National Centers for Health Statistics (1994) reported that for
6 January 1985 through December 1992 trends in mortality from diseases of the heart
7 (including coronary heart disease) showed that for the modeled period, provisional death
8 rates decreased. Mortality also showed a seasonal pattern with death rates higher in winter.
9
10 12.3.1.1 Review of Short-Term Exposure Studies
11 The decade or so since the previous criteria document addendum was released (U.S.
12 Environmental Protection Agency, 1986a) has been an active period for the reporting of time
13 series analyses of associations between human mortality and acute exposures to PM. In the
14 beginning of this period, various PM measures of only indirect applicability to the standard
15 setting process (e.g., TSP, BS, KM, or COH) were usually employed. However, in the last
16 few years the analyses have more often employed PM10 as a measure of PM. This is
17 because sufficient routine PM10 ambient measurement data have finally begun to be available
18 for such statistical analyses to be conducted in a wide variety of locales. The focus of this
19 section is on detailed assessments of those studies conducted since the PM criteria document
20 addendum (U.S. Environmental Protection Agency, 1986a). Of special interest are studies
21 that have employed PMjQ in their analyses of the human mortality effects of acute exposures
22 to PM; although studies employing other indices of PM exposures are be summarized in
23 tables and discussed in the text, as appropriate.
24 As shown in Table 12-3 a variety of PM metrics have been employed in time-series
25 studies relating PM to acute mortality. These have included gravimetric measures, such as
26 total suspended paniculate matter (TSP) and PM10, the former of which includes a significant
27 portion of nonrespirable particles. In addition, many studies have employed data from
28 various samplers that yield BS or KM optical measurements of particle reflectance of light,
29 or coefficient of haze (COH) optical measurements of particle transmission of light. All of
30 these latter metrics are most directly related to ambient elemental carbon concentration (e.g.,
31 see Bailey and Clayton, 1982, and Cass et al., 1984), but only indirectly related to particle
April 1995 12-46 DRAFT-DO NOT QUOTE OR CITE
-------
1 mass, as the relationship with mass will vary as sampled particle size, shape, color, and
2 surface characteristics vary over time and between sites. Hence, unless side-by-side
3 calibrations of these optical measurements are made against direct mass measurements
4 obtained by collocated gravimetric monitoring instruments, such optical measurements cannot
5 be readily converted to quantitative estimates of ambient PM mass concentrations or
6 associated PM-mortality relationships. Thus, given the diversity of nonequivalent PM
7 metrics employed across many of the reviewed epidemiology studies, attempting quantitative
8 intercomparisons between results of all of the various reviewed epidemiologic studies
9 necessarily introduce additional uncertainties, although attempts have been made using
10 assumed conversion factors (Schwartz, 1991; Ostro, 1993, 1993; Dockery and Pope, 1994;
11 and Pope etal., 1995).
12 The two studies using KM as the PM metric employed very different approaches to the
13 same data set from Los Angeles during 1970 to 1979. The study by Shumway et al. (1988)
14 evaluated long-wave associations and found significant KM-mortality associations, but this
15 analysis failed to account for seasonal effects. The KM study by Kinney and Ozkaynak
16 (1991) more appropriately studied the short-wave associations of multiple pollutants, finding
17 KM to be significantly associated with total mortality, but collinearities among KM, NO2
18 and CO made it impossible to separately estimate a PM association.
19 Similar to the KM studies, BS studies are quite varied in approach. Thurston et al.
20 (1989) applied a high pass filter (similar to that employed by Kinney and Ozkaynak, 1991) to
21 the 1963 to 1972 London, England wintertime mortality-pollution data set, whereas Ito et al.
22 (1993) analyzed a subset of the same data using prewhitening and autoregressive techniques.
23 In both, BS, SO2, and H2SO4 were all found to be significantly associated with mortality, but
24 the effects of each were not separable due to the high collinearity among these pollution
25 metrics. Katsouyanni et al. (1990) similarly found an association between total mortality and
26 all pollutants measured in Athens, Greece, during 1975 to 1987, though they reported that
27 BS was the most strongly associated of the pollutants. A separate randomized block analysis
28 of SO2 and by-cause mortality during this same period (Katsouyanni, 1990) found significant
29 SO2 effects, but also noted that SO2 and BS were correlated at r = 0.73, suggesting that BS
30 may also have been associated. A subsequent analysis by Katsouyanni et al. (1993) of
31 summer heat wave periods found a significant temperature-SO2 interaction term, and the
32 suggestion of an interaction (p < 0.2) for BS.
April 1995 12-47 DRAFT-DO NOT QUOTE OR CITE
-------
f
TABLE 12-3. SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
KM (mean = 25;
SD = 11)
Total, respiratory, and cardiovascular mortality in
Los Angeles County during 1970 to 1979 related to
O3, CO, SO2, NO2, HC, daily max. temperature,
relative humidity, and KM (a paniculate matter
metric of optical reflectance by particles, related to
the ambient carbon concentration). Low pass filter
used to eliminate short-wave, so that only long-
wave associations are studied.
Frequency domain analyses indicated significant
short- and long-wave associations with KM. The
filtered (i.e., long-wave) data analysis indicated that
air pollution (including KM) was significantly
associated with seasonal variations in LA mortality.
Shumway et al.
(1988)
Ni
-U
oo
>*x
?
O
O
2
O
H
O
I
TSP (OECD Method)
(Lyons, France: 3 year
mean = 87 /tg/m3)
(Marseilles, France
3 y mean = 126 /tg/m3)
BS
DO
(mean = 90.1 /*g/m3)
(24-h avg. daily max. =
709 /ig/m3)
Daily total, respiratory, and cardiac mortality for
persons >65 years of age tested for associations
with SO2 and TSP during 1974 to 1976 in Lyons
and Marseilles, France. Temperature also
considered in analyses.
Daily total mortality analyzed for associations with
BS, SO2, and H2SO4 in London, England, during
1963 to 1972 winters. Mean daily temperature and
relative humidity also considered.
No significant mortality associations found with TSP,
while SO2 was reportedly associated with total
elderly deaths in both cities. Seasonality addressed
by analyzing deviations from 3-year average of
31-day running means of variables. However, lags
of temperature not considered and probable seasonal
differences in winter/summer temperature-mortality
relationship not addressed.
PM, SO2, and H2SO4 all indicated as having
significant associations with mortality (0, 1 day lag).
Temperature also correlated (negatively) with
mortality, but with a 2-day lag. Seasonality
addressed by studying only winters and by applying a
high-pass filter to the series and analyzing residuals.
Derriennic et al.
(1989)
Thurston et al.
(1989)
O
3
w
-------
O
sl
TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
COH (monthly mean
range = 9 to 12)
Daily total, respiratory, cancer, and circulatory
associations with daily COH in Santa Clara
County, CA, during 1980 to 1982 and 1984 to
1986 winters. Daily mean temperature and
relative humidity at 4 PM also considered.
An association found between COH and increased
mortality, even after making adjustments for
temperature, relative humidity, year, and
seasonally.
Fairley (1990)
BS
Daily total mortality in Athens, Greece, and
surrounding boroughs during 1975 to 1987
related to BS, SO2, N02, O3, and CO2 using
multiple regression.
During winter months 1983 to 1987, the daily
number of deaths was positively and statistically
significantly associated with all pollutants, but the
association was strongest with BS.
Katsouyanni et al.
(1990a)
to
O
o
z
o
H
O
d
o
H
W
BS (annual mean range
= 51.6to73.3ftg/rn3)
(maximum daily value
= 790 /*g/m3)
For the period 1975 to 1982 in Athens, Greece,
199 days with high SO2 (> 150 /xg/m3) were
each matched on temperature, year, season, day
of week, and holidays with two low SO2 days.
Mortality by-cause comparisons made between
groups by analysis of variance by randomized
blocks. BS correlated with SO2 at r = 0.73, but
not directly employed in the analysis.
Mortality was generally higher on high SO2 days,
with the difference being most pronounced for
respiratory conditions. BS levels for each group
not provided, and BS-SO2 confounding not
addressed, limiting interpretability of results.
Katsouyanni et al.
(1990b)
-------
£ TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
5: RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
KM
(mean = 25;
SD = 11)
Shumway et al. (1988) 1970 to 1979 Los Angeles
mortality dataset analyzed using a high-pass filter
to allow investigation of short-wave (acute)
associations with environmental variables (by
removing seasonality effects). Environmental
variables considered in regression analyses
included temperature, relative humidity,
extinction coefficient, carbonaceous paniculate
matter (KM), SO2, NO2, CO, and O3.
Analyses demonstrated significant associations
between short-term variations in total mortality and
pollution, after controlling for temperature. Day-
of-week effects found not to affect the
relationships. The results demonstrated significant
mortality associations with O3 lagged 1 day, and
with temperature, NO2, CO, and KM. The latter
three pollutants were highly correlated with each
other, making it impossible to separately estimate
paniculate matter associations with mortality.
Kinney and Ozkaynak
(1991)
to
U)
o
TSP
(mean = 87 /ug/m3)
(24-h avg. range:
46 to 137 fig/m3,
5th to 95th percentiles)
Total deaths in Detroit, MI, 1973 to 1982
analyzed using Poisson methods. Environmental
variables considered included TSP, SO2, O3,
temperature, and dew point. Seasonality
controlled via multiple dummy weather and time
variables.
Significant associations reported between TSP and
mortality in autoregressive Poisson models (RR of
100 jug/m3 TSP = 1.06). However, most TSP
data estimated from visibility, which is best
correlated with the fine aerosol (and especially
sulfate) portion of the TSP. Thus, results suggest
a fine particle association.
Schwartz (1991)
TSP
(mean = 77 j^g/m3)
(max. = 380 /ig/m3)
(5th to 95th percentiles =
37 to 132 ng/m3)
Total and cause specific daily mortality in
Philadelphia, PA during 1973 to 1980 related to
daily TSP and SO2 (n =2,700 days). No other
pollutants considered in the analysis. Poisson
regression models, using GEE methods, included
controls for year, season, temperature, and
humidity. Autocorrelation addressed via
autoregressive terms in model.
Strongest associations found with pollution on the
same and prior days. Total mortality (mean =
48/day) estimated to increase 7% (95% C.I. =
4 to 10%) for a 100 /ig/m3 increase in TSP.
Cause-specific effects of TSP were larger (as %).
SO2 associations were non-significant in
simultaneous models with TSP, but correlations of
their coefficients not reported
Schwartz and Dockery
(1992a)
-------
2.
TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
H
6
o
iO
d
§
o
*>
o
TSP
(mean = 69 /*g/m3)
(5th to 95th percentiles
32 to 120 /*g/m3)
TSP
(mean = 111 jig/m3)
(24-h avg. range:
36 to 209 jug/m3,
10th to 90th percentiles)
TSP
(mean =113 /ig/m3)
(10th to 90th percentiles
38 to 212
Age and cause-specific daily mortality in
Philadelphia, PA during 1973 and 1990 related
to daily TSP, SO2, and O3. Other environmental
variables included were: temperature,
barometric pressure, humidity, and precipitation.
Various models employed, including poisson and
autoregessive. Prefiltering methods also applied
to remove long-waves in data.
Daily total mortality in Steubenville, OH,
between 1974 to 1984 related to TSP, SO2,
temperature, and dew point. Poisson regression
employed, because of very low death counts/day
(mean = 3.1). Regressions controlled for
season by including dummy variables for winter
and spring, and autoregressive methods also used
to address any remaining autocorrelation.
Daily mortality in Steubenville, OH during 1974
to 1984 related to TSP, SO2, temperature, and
dew point (to allow comparisons of results with
Schwartz and Dockery, 1992b). Poisson method
employed. Analyses done overall and by-season.
TSP effect found only hi whiter season. TSP
never significant in by-cause analyses of those
< 15 or >65 years of age. TSP effects
weakened by the addition of other pollutants
(TSP-SO2 r = 0.57). However, the inclusion of
barometric pressure and precipitation in these
models may have acted as surrogates for PM,
potentially confounding results. Correlations
between TSP and these variables not presented.
In regressions controlling for season and
weather, previous day's TSP was a significant
predictor of daily mortality. SO2 was less
significant in regressions, becoming
nonsignificant when entered simultaneous with
TSP. Auto-regressive models gave similar
results.
In single pollutant models, the TSP coefficient
was the same as Schwartz and Dockery (1992b),
but TSP effects were found to be attenuated by
SO2 inclusion in the model. SO2 was also
attenuated by the addition of TSP. It is
concluded that TSP and SO2 effects cannot be
separated in this dataset. Intercorrelations
among these variables not presented.
Li and Roth (1995)
Schwartz and Dockery
(1992b)
Moolgavkar et al.
(1995)
-------
£• TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
U RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
BS
(mean =
90.1 /xg/m3)
(range = 0 to
350 /tg/m3)
Further analysis of London, England data (1965 to 1972)
examined by Thurston et al. (1989). Spectral and advanced
time series methods applied, including prewhitening and
auto- regressive (AR) moving average (MA) methods.
Environmental variables considered included BS, SO2,
H2SO4, temperature, and relative humidity.
Estimated pollution mean effect of 2 to 7 % of all London
winter deaths (mean = 281/day). However, the various
pollutants' effects were not separated. Independent model
test on the 1962 episode confirmed the appropriateness of
such methods. Long-wave addressed by considering
winters only and by prewhitening the data.
Ito et al.
(1993)
BS Daily total mortality in Athens, Greece, during July, 1987 Mean daily temperature above 30 °C found to be
(range = 50 to (when a major "heat wave occurred) compared to the deaths significantly associated with mortality. The main effects
250 fig/m3) in July during the previous 6 years. Environmental of all air pollutants nonsignificant, but the interaction
variables considered included: temperature, discomfort between high air pollution and temperature were
index (DI), BS, SO2, and BS. Confounding effects of day- significant for SO2 and suggestive (p < 0.20) for ozone
of-week, month, and long-term trends addressed via dummy and BS.
variables in OLS regression models.
Katsouyanni
et al. (1993)
H
6
o
2
o
H
/O
G
O
H
W
n
HH
H
W
Suspended Particles
(SP) (range = 10 to
650 /ig/m3)
BS (mean =
83 fig/w?) (range =
18 to 358 ftg/m3)
Daily total mortality in Erfurt, East Germany, during 1980
to 1989 (median = 6/day) related to S02, SP, T, RH, and
precipitation. SP measurements made only 1988 to 1989.
Autoregresssive Poisson models employed (due to low
deaths/day) also included indicator variables for extreme
temperatures and adjustments for trend, season, and
influenza epidemics.
Daily total mortality in Athens, Greece, during 1984-1988
(mean = 38/day) related to BS, SO2, CO, T, and RH.
Autoregressive OLS models employed also included
indicator variables for season, day of week, and year.
Both SO2 and SP were found to be significantly associated Spix et al.
with increased mortality. In a simultaneous regression, SP (1993)
remained significant while SO2 did not. Correlations of
these coefficients not provided, however. Pollution effect
size similar to that for meteorology.
BS, SO2, and CO all found to be individually significantly Touloumi et
associated with increased mortality. In simultaneous
regressions, the size of all coefficients declined, with SO2
still significant and BS approaching significance. CO was
no longer significant, but was highly correlated with BS (r
= 0.74) in the data.
al. (1994)
-------
TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
TSP
(mean = 52 /xg/m3;
SD = 19.6 /ig/m3)
Daily total and cause-specific mortality in Cincinnati,
OH, (mean total = 21/day) during 1977 to 1982
related to TSP, temperature, and dew point. Poisson
model employed with dummy variables for each month
and for eight (unspecified) categories of temperature
and dew point. Linear and quadratic time trend terms
also included. Spline and nonparametric models also
applied. Autocorrelation not directly addressed.
TSP was significantly associated with increased risk of total
mortality. The relative risk was higher for the elderly and
for those dying of pneumonia and cardiovascular disease.
However, the analysis failed to consider other pollutants,
and there remains the potential for within-month, long-wave
confoundings.
Schwartz
(1994a)
0
O
2
O
H
O
W
O
90
n
»— i
H
W
TSP
(mean =
375 /*g/m3)
(maximum =
1,003 /ig/m3)
PM10
(mean = 47 /ig/m3)
(24 h max. =
365 ftg/m3)
(5 day max. =
297 /.g/m3)
Daily deaths during 1989 in two residential areas in
Beijing, China, (mean total deaths = 21.6/day) related
to TSP and SO2 using Poisson methods. Controlling
indicator variables for quintiles of temperature and
humidity, as well as for Sunday also included.
Long-wave confounding and autocorrelation not
directly addressed. However, season-specific results
presented.
Total, respiratory, and cardiovascular mortality in
Utah County, UT, during 1985 to 1989 related to
5-day moving average PM10, temperature, and
humidity. Time trend and a random year terms also
included in autoregressive Poisson models employed.
Seasonally not directly addressed in this basic model,
but the addition of four seasonal dummy variables
changed results little.
Significant mortality associations found for In (SO2) and In Xu et al.
(TSP). Associations were strongest for chronic respiratory (1994)
diseases. In simultaneous regressions, S02 was significant,
but not TSP. However, the two pollutants were highly
correlated with each other (r = 0.6), as well as with
temperature. In season-specific analyses, both pollutants
were significant in summer, but only SO2 in winter.
A significant positive association between total non- Pope et al.
accidental mortality and PM10 was observed, the strongest (1992)
association being with the 5-day moving average of PM10.
The association was largest for respiratory disease, the next
largest for cardiovascular, and the lowest for all other.
Association noted below 150 /ig/m3 PM10. The possible
influence of other pollutants discussed, but not directly
addressed.
-------
TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
^g PM Measure
Ul (Concentrations)
Study Description
Results and Comments
Reference
PM10
St. Louis, MO:
(mean = 28 /*g/m3)
(24 hr max. = 97
Kingston/Harriman, TN
(mean = 30 /ig/m3)
(24 h max. = 67 pig/m3)
Total mortality in St. Louis, MO, and
Kingston/Harriman, TN (and surrounding
counties), during September 1985 to August 1986
related to PM10, PM25, SO2, NO2, O3, H + ,
temperature, dew point, and season using auto-
regressive Poisson models.
Statistically significant daily mortality associations
found with PM10 and PM2 5 in St. Louis, but not
with other pollutants. In Kingston/Harriman, PM10
and PM2 5 approached significance, while other
pollutants did not. Seasonality was reduced by
season indicator, variables, but within season long
wave cycles not directly addressed.
Dockery et al.
(1992)
ts>
PM
10
(mean = 48 /*g/m3)
(24 h max. = 163 /ig/m3)
Total daily mortality in Birmingham, AL, from Significant associations found between total mortality Schwartz (1993)
August 1985 to December 1988 related to PM10, and prior day's PM10. Various models gave similar
temperature, and dew point. Poisson models results, as did eliminating all days with PM10 > 150
employed addressed seasonal long wave influences /«g/m3. However, the possible role of other
by the inclusion of 24 sine and cosine terms having pollutants not evaluated.
periods ranging from 1 mo to 2 years.
Autoregressive linear models also applied.
PM
10
O
O
Z
s
O
G
s
w
§
o
(mean = 40
Total, cardiovascular, cancer, and respiratory
mortality in Toronto, Canada, during 1972 to 1990
(24 h max. = 96 pig/m3) related to PM10, TSP, SO4, CO, O3> temperature,
and relative humidity. Nineteen-day moving
average filtered data used in OLS regressions.
Sixty-three hundred and three PM10 values
estimated based on TSP, SO4, COH, visibility
(Bext) and temperature data, using model developed
from 200 PM10 sampling days during the period.
Significant associations found between all pollutants
considered and mortality, after controlling for
weather and long wave influences. However, it was
not possible to separate the PM10 association from
other paniculate measures considered. Simultaneous
PM and ozone regressions gave significant
coefficients for each, but intercorrelations among the
pollutants not presented.
Ozkaynak et al.
(1994)
-------
TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
PM10
(mean = 58 /ig/m3)
(24 h max. =
177 /ig/m3)
Total mortality in Los Angeles, CA, during 1985 to 1990
related to PM10, O3, CO, temperature, and relative
humidity. Poisson models employed addressed seasonal
long-wave influences by including multiple sine and cosine
terms ranging from 1 mo to 2 years in periodicity. OLS
and long linear models also tested. Winter and summer
analyzed separately also.
Association between PM10 and mortality found to be
only mildly sensitive to modeling method. CO also
individually significant. The addition of either CO or
O3 lowered the significance of PM10 in model
somewhat, but the PM10 coefficient was not as
affected, indicating minimal effects on the PM10
association by other pollutants in this case.
Kinney et
al. (1995)
PM
10
S)
o
5
I
O
O
H
O
O
(mean = 38
(24 h max.
128 /tg/m3)
Total mortality in Los Angeles, CA and Chicago, IL
during 1985 through 1990 related to PM10, O3, and
temperature. Analysis focused on importance of monitor
choice to modeling results. Poisson models used
addressed seasonal long wave influences by including
multiple sine/cosine terms ranging from 1 mo to 2 years
in periodicity.
Average of multiple sites' PM10 found to be
significantly associated with mortality in each city after
controlling for season, temperature and ozone. Other
pollutants and relative humidity not yet considered.
Individual sites' PM10 varied from non-significant to
strongly significant. Also, dividing the data by season
diminished the significance of the multi-site average
PM10 in mortality regressions. Both site selection and
sample size concluded to influence results.
Ito et al.
(1995)
n
H-*
H
W
-------
TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
to
Tj
H
6
o
z
s
o
PM10
(mean = 115 /xg/m3)
(24 h max.
367 / 65 years old)
(mean = 63/day) in Sao Paulo, Brazil during May 1990
through April 1991 related to two day avg. of PM10, SO2,
NOX, O3, and CO, and to temperature, humidity, and day
of week. Season addressed by including seasonal and
monthly dummy variables. Temperature addressed using
three discrete dummy variables.
Significant association found between PM10 and daily Ostro et al.
mortality, even after addressing potential confounders (1995a)
(e.g., weather), other pollutants, lag structure, and
outliers. Strongest associations found for respiratory
deaths. SO2 and NO2 also significantly associated
individually, but only PM10 remained significant
when all were added simultaneously to the
regression. Correlations of the coefficients not
reported.
Significant association found between respiratory Saldiva et al.
deaths and NOX, but no other pollutants. No such (1994)
association found for non-respiratory deaths.
However, auto-correlation not addressed. Also,
inter-correlations of the pollutant coefficients not
reported (but NOX - PM10 correlation = 0.68)
Significant associations found between total elderly Saldiva et al.
deaths and all pollutants considered. In a (1994)
simultaneous regression, PM10 was the only pollutant
which remained significant. The PM10 coefficient
actually increased in this regression, suggesting
interpollutant interactions. Correlations of the
pollutant coefficients not provided.
i
o
-------
TABLE 12-3 (cont'd). SUMMARIES OF RECENTLY PUBLISHED EPIDEMIOLOGICAL STUDIES
RELATING HUMAN MORTALITY TO AMBIENT LEVELS OF PARTICULATE MATTER
PM Measure
(Concentrations)
Study Description
Results and Comments
Reference
PM10 (Cook County
median = 37 /ig/m3;
max = 365 ftg/m3)
(Salt Lake County
median = 35 /ig/m3;
max = 487
Total, respiratory, circulatory, and cancer mortality in
Cook County, 1985 to 1990. Elderly, total by race and
sex also evaluated. Poisson regression with seasonal
adjustments, meteorological variables, and pollen tested.
In Salt Lake Count, total and elderly mortality. One daily
station in Cook County and two daily monitoring stations
in Salt Lake County, plus multiple -day stations.
Average and single site PM10 were significant
predictions of PM10 in Cook County for total,
elderly, cancer, and elderly white mortality, marginal
for respiratory, circulatory, and elderly black.
Significant Fall and Spring mortality in Cook
County, not Summer or Winter. No significant
effects in Salt Lake County. No copollutants.
Styer et al.
(1995)
PMi0 (variable by
month and year)
Reanalysis of Utah County mortality, 1985 to 1992,
broken down by year, season cause and place of death.
PM10 was entered as a dichotomous variable, less or
greater than 50 /tg/m3. No adjustment for copollutants or
for weather in Poisson regression, except for daily
minimum temperature. Poisson regression, not GEE.
Variations in RR did not appear to be associated with Lyon et al.
high or low PM10 days. High RR for cancer deaths, (1995)
age < 60, at home. Highest RR in spring.
Increased RR for sudden infant death syndrome.
Patterns appear noncausal.
-------
A PM study using BS as the paniculate matter index which has carefully addressed the
potential confounding effects of other pollutants and temperature was recently conducted by
Touloumi et al. (1994) for daily all-cause mortality in Athens, Greece during 1984 through
1988. In this study, BS (mean = 83 /ig/m3), SO2 (mean = 45 Aig/m3), CO (mean =
6 jug/m3), temperature, and relative humidity were all modeled separately and
simultaneously, giving a range of estimates for PM effects, depending on the model
specification. The five years of data employed provided ample numbers of records for the
analysis (e.g., n = 1684 for BS). Temperature associations were simply but effectively
modeled. The authors examined the bivariate temperature-mortality plot and noted that a
mortality minimum occurred at approximately 23 °C daily mean temperature. They then
defined two temperature variables: one as the daily mean temperature deviation below 23 °C;
the other as the daily deviation above 23 °C, whichever was relevant, thereby allowing a
separate modeling of the cold and hot weather effects on mortality. It was found that the
square of each of these (lagged one day) gave the best fit of the mortality, and these terms
were used in subsequent pollutant models. Multiple monitoring stations were averaged
(e.g., 5 for smoke) after filling in missing observations from available data on the same day
at other sites, providing spatially representative exposure estimates. Ordinary least squares
modeling was applied, which is acceptable in this case, given the relatively large number of
mortality counts/day (mean = 38 deaths/day, SE = 12) in this metropolitan area. In
mortality regressions on pollutants, day-of-week, season, hot (> 23 °C) and cold (<23 °C)
temperature deviations squared, and relative humidity terms were also included. Although
the use of only a dummy term for each season could not have fully addressed the
within-season long wave mortality trends shown in time series plots, autoregressive modeling
did address any resulting residual autocorrelation. Also, the use of a single annual sine
curve with a periodicity of 1 year (phase not reported) gave similar results.
Separating the effects of the various air pollutants was attempted in this analysis of
Athens mortality, but proved challenging. The log of pollutant concentrations were entered
into the basic model both individually and simultaneously. All pollutants considered were
individually significant at the p = 0.0001 level. When copollutants were entered
simultaneously, SO2 was the least affected, both in terms of coefficient size and statistical
significance. The BS coefficient dropped in size by 50% when entered with SO2 in the
April 1995 12-58 DRAFT-DO NOT QUOTE OR CITE
-------
model, and its statistical significance weakened, as expected when correlated variables are
entered together, but remained significant (p = 0.02, jtwo-tailed test). However, SO2
declined by less than 30 percent and remained significant at the p = 0.002 level in
simultaneous regressions. The CO coefficient decreased in size by 75% and became clearly
non-significant when entered with either SO2 or BS. The authors noted, however, that these
pollutants are highly intercorrelated over time (e.g., for CO and BS, r = 0.79). Thus, while
the most consistent mortality association, both in terms of size and significance of its
coefficient, appears to be with SO2 in this city, the colinearities among these primary,
combustion-related, air pollutants prevents a quantitative apportionment of effects to
individual pollutants. The authors acknowledged this fact, concluding that relatively
low-level air pollution has a small but real effect on mortality. Using BS alone as the index
of ambient air pollution, the authors report that a 10% decrease in BS was associated with a
0.75% decrease in total mortality. Using an on-site calibration with PM10 (PM10 = 8.70 +
0.832 x BS) developed for this city (Katsouyanni, 1995) yields a mean PM10 of 77.7 jig/m3,
and a relative risk (RR) of 1.07 for a 100 /*g/m3 increase in PM10 (i.e., to 203 jig/m3 BS).
However, when the BS coefficient from the simultaneous regression with other pollutants is
employed, the estimated RR of that 100 /xg/m3 increase in PM10 drops to 1.03. Thus, the
estimate of the total mortality RR of a one day 100 jjg/m3 increase in PM10 implied by this
work ranges from 1.07 to 1.03, depending on whether the PM metric is entered into the
regression singly or in combination with other pollutants, respectively.
Total suspended paniculate matter studies have also yielded mixed results as to the
relative role of PM, versus other pollutants, in mortality. For example, Schwartz (1991)
examined total mortality in Detroit during 1973 to 1982, finding TSP to be more strongly
associated with mortality than SO2 However, the correlation between SO2 and TSP was not
reported, other pollutants likely to have been present (e.g., CO and NO2) were not
considered in the analysis, and most of the TSP values were estimated from visibility
records, which are most strongly correlated with fine particles (e.g., see Ozkaynak et al.,
1986). Schwartz and Dockery's analysis of 1974 to 1984 mortality in Steubenville, OH
(1992), similarly concluded that TSP was more significant than SO2 but failed to consider
other pollutants and did not report the correlation between SO2 and TSP in this valley locale.
A Schwartz (1994a) analysis of Cincinnati, OH, mortality during 1977 to 1982 also found a
April 1995 12-59 DRAFT-DO NOT QUOTE OR CITE
-------
TSP-mortality association, but did not consider other pollutants. Derriennic et al. (1989)
examined mortality among the elderly in two French cities during 1974 to 1976 and found
mortality associations with S02, but not with TSP (though the model specification for
temperature did not address possible lag structure or season). Spix et al. (1993) found
significant suspended particle (SP) and SO2 associations with mortality in Erfurt, East
Germany, during 1980 to 1989, with SP remaining significant in simultaneous regressions,
despite very high SO2 levels. Xu et al. (1994) found significant SO2 and TSP associations
(other pollutants not considered) in Bejing, China, but found that SO2 (not TSP) remained
significant in simultaneous regressions.
Overall, qualitatively examining the KM, BS, and TSP time-series studies summarized
in Table 12-4 reveals that these various PM metrics are often associated with mortality. The
strength and interpretation of that association can vary depending on the number of other
pollutants included and on the way they are considered in the analysis. In the above
discussed cases where more pollutants were considered, other pollutants were often found to
also be associated with mortality, sometimes less strongly and sometimes more strongly than
for the PM metric. Moreover, in the cases where the correlations among the significant
pollutants were reported, it was consistently found that the PM metric was correlated with
these other pollutants. Also, the relationships of all these various PM metrics with PM10
mass are not generalizable, and likely vary between seasons and from place to place. Thus,
although these various analyses are strongly supportive of an ambient air pollution effect on
mortality throughout the world, and are generally consistent with the hypothesis of a PM
effect on mortality, they are of more limited usefulness in trying to quantitatively assess PM
mortality associations (e.g., as a relative risk per jug/m3).
Table 12-4 includes summaries of numerous recently reported PM10-mortality studies.
Among the first was a study of total, respiratory, and cardiovascular mortality in Utah
County, UT during 1985-1989 (Pope et al., 1992). In this study, the various daily counts of
mortality were regressed on the 5-day moving average PM10, as well as on temperature,
humidity, a time-trend term, and random year terms. While only one site was used to
represent the whole county's PM10 level, comparisons with two other PM10 sites indicated
spatial consistency (correlation between sites > 0.95). Autoregressive Poisson methods were
used because of the low total mortality counts (mean = 2.7/day) in this relatively small
April 1995 12-60 DRAFT-DO NOT QUOTE OR CITE
-------
> TABLE 12-4. SUMMARIES OF PUBLISHED PM10-ACUTE MORTALITY
S: EFFECTS STUDIES BASED ON VARIOUS PM MEASURES
§
u.
Health Outcome Synthesis Study
Total Mortality Ostro (1993)
Dockery and Pope
(1994)
10
OS
t-1 Respiratory Mortality Dockery and Pope
(1994)
O
> Cardiovascular Mortality Dockery and Pope
^ (1994)
6
0
O
H
0
0
H
W
Location
London UK
Steubenville OH
Philadelphia PA
Santa Clara CA
St. Louis MO
Kingston TN
Birmingham AL
Utah Valley UT
Philadelphia PA
Detroit MI
Steubenville OH
Santa Clara CA
Birmingham AL
Utah Valley UT
Philadelphia PA
Santa Clara CA
Birmingham AL
Utah Valley UT
Philadelphia PA
Santa Clara CA
Original PM
Measurement
BS
TSP
TSP
COH
PM10
PM10
PM10 (3d)
PM10 (5d)
TSP (2d)
TSP
TSP
COH
PM10 (3d)
PM10 (5d)
TSP (2d)
COH
PM10 (3d)
PM10 (5d)
TSP (2d)
COH
Mean
Equivalent PM10
80
61
42
37
28
30
48
47
40
48
61
35
48
47
40
35
48
47
40
35
Percent Change
Per 10 /ig/m3
PM10 Equivalent
0.3
0.6
1.2
1.1
1.5
1.6
1.0
1.5
1.2
1.0
0.7
0.8
1.5
3.7
3.3
3.5
1.6
1.8
1.7
0.8
95 Percent
Confidence Interval
(0.29, 0.31)
(0.44, 0.84)
(0.96, 1.44)
(0.73, 1.51)
(0.1, 2.9)
(-1.3,4.6)
(0.2, 1.5)
(0.9, 2.1)
(0.7, 1.7)
(0.5, 1.6)
(0.4, 1.0)
(0.2, 1.5)
(-5.8, 9.4)
(0.7, 6.7)
(0.1, 6.6)
(1.5, 5.6)
(-1.5, 3.7)
(0.4, 3.3)
(1.0, 2.4)
(0.1, 1.6)
n
si
-------
1 population (260,000). Using this model, a significant positive association was found between
2 total non-accidental mortality and PM10, and the authors concluded that a 100 /*g/m3 increase
3 in the 5-day average PM10 concentration was associated with a 16% increase in mortality.
4 Analyses presented indicate that the use of concurrent day PM10, rather than a 5-day
5 average, would have resulted in an effect estimate roughly half that reported in terms of the
6 5-day average PM10 (in deaths per /ig/m3). A "control" disease category (i.e., one unlikely
7 to be affected by air pollution) was not considered, per se. However, deaths due to causes
8 other than respiratory or cardiovascular were considered, and found not to be associated with
9 PM. Respiratory deaths were more strongly associated with PM10 than any other cause.
10 Both of these results support the biological plausibility of a PM-mortality association.
11 Furthermore, the PM10-mortality association was noted by the authors to be present well
12 below the existing 24-h average PM10 standard of 150 /ig/m3. The authors dismiss other air
13 pollutants as having negligible influence by comparing them to their respective present air
14 quality standards without directly modeling the possibility that other (correlated) air
15 pollutants might also influence mortality. On the other hand, Pope (1995) reported that
16 PM10, and SO2 were only weakly correlated (r = 0.19), acid aerosol (H+) levels were below
17 8 nmoles/m3, and the introduction of 03 into the model actually strengthened the PM10
18 association. Thus, other pollutants appear unlikely to be a major confounder in this analysis,
19 and this study supports the hypothesis of a PM10 air pollution effect on human mortality.
20 Dockery et al. (1992) investigated the relationship between multiple air pollutants and
21 total daily mortality during the one year period between September 1985 and August 1986 in
22 two communities: St. Louis, MO; and Kingston/Harriman, TN and surrounding counties.
23 In the latter locale, the major population center considered is Knoxville, TN, some 50 Km
24 from the air pollution monitoring site employed. In each study area, total daily mortality
25 was related to PM10, PM2 5, S02, N02, O3, H+, temperature, dew point, and season using
26 autoregressive Poisson models. In St. Louis, after controlling for weather and season,
27 statistically significant associations were found with both prior day's PM10 and PM2.5, but
28 not with any lags of the other pollutants considered. In the Kingston/Harriman vicinity,
29 PM10 and PM2 5 approached significance in the mortality regression, while the other
30 pollutants did not. In both cities, very similar PM10 coefficients are reported, implying a 16
31 to 17 percent increase in total mortality per 100/ng/m3 of PM10. While autocorrelation was
April 1995 12-62 DRAFT-DO NOT QUOTE OR CITE
-------
1 accounted for, seasonality was only addressed by season indicator (dummy) variables, which
2 could not address any within-season long wave influences. However, in both places, only
3 one daily monitoring station was employed to represent community exposure levels, and no
4 information regarding the representativeness of these sites are provided (e.g., correlations
5 with other sites' data). Also, using mortality data for Knoxville, TN (50 km from the
6 Kingston/Harriman, TN monitoring site at which PM concentrations were measured) in the
7 PM-mortality analysis raises questions about the representativeness of the exposure estimates
8 in that city. Furthermore, the number of days for which pollution data are available for
9 time-series analyses is limited in this data set, especially for H+ (e.g., only 220 days had H+
10 values at the St. Louis site). In the words of the authors: "Because of the short monitoring
11 period for daily particulate air pollution, the power of this study to detect associations was
12 limited." However, despite this limitation, consistent PM10 coefficients were found in each
13 of these two cities.
14 A longer record was examined for a total mortality-PM10 relationship in Birmingham,
15 AL during August, 1985 through December, 1988 (Schwartz, 1993). In this work, Poisson
16 modeling was used to address small count effects (mean mortality =17.1 deaths/day),
17 season was addressed by the inclusion of 24 sine and cosine terms having periods ranging
18 from 1 to 24 mo, and weather was modeled using various specifications of temperature and
19 relative humidity. Autocorrelation was addressed using autoregressive parameters, as
20 required, and day of week dummy variables were also included. In these analyses,
21 significant associations were found between total daily mortality and the average of the three
22 prior day's PM10 concentration. It was noted that averaging fewer days weakened the PM10-
23 mortality association, which is consistent with expectation that multiple day pollution
24 episodes are of the greatest health concern. The analysis did not look at any other pollutants,
25 making it impossible to directly assess whether the association noted is due to PM10 alone, or
26 also in part to some other collinear pollutant (e.g., SO2) not considered in the analysis.
27 However, a variety of modeling approaches gave similar results, as did eliminating all days
28 with PM10 > 150 ptg/m3, indicating that the PM-mortality associations noted are not
29 dependent on model choice, or limited to elevated pollution days only.
30 Ozkaynak et al (1994) related total daily mortality in Toronto, Ontario during 1972-
31 1990 to daily PM10, TSP, SO4, CO, O3, temperature, and relative humidity. A 19-day
April 1995 12_63 DRAFT-DO NOT QUOTE OR CITE
-------
1 moving average equivalent high-pass filter was used to prefilter out long-wave cycles in the
2 data and to reduce autocorrelation. OLS regression was employed, as the distribution of
3 mortality data tend toward the normal in larger cities such as Toronto (mean deaths =
4 40/day) once seasonal cycles are removed. In this dataset, 6,303 PM10 daily values were
5 estimated based on TSP, SO4, COH, visibility (i.e., Relative Humidity corrected Bext, the
6 extinction coefficient derived from airport visibility observations), and temperature data,
7 using a model developed from 200 actual PM10 sampling days during the study period. This
8 limits the usefulness of the results for distinguishing PM10 in the analyses, as it is derived
9 from other particulate matter metrics and from variables which may themselves be causally
10 related to mortality (e.g., temperature). For example, estimated PM10 is correlated at r =
11 0.95 with TSP, and r = 0.27 with temperature. In the analyses, all pollutants considered
12 were significantly associated with daily mortality. The simultaneous regression of total
13 mortality on both O3 and PM10 yielded significant coefficients for each. The PM10 mean
14 effect (at 41 /Ag/m3) was reported to be 2.3% of total mortality. However, the authors found
15 that it was not possible to separate the PM10-mortality association from that for the other
16 particulate matter metrics considered.
17 Kinney et al. (1995) investigated total daily mortality in Los Angeles, CA during
18 1985-1990 (mean = 153 deaths/day), relating it to PM10, O3, CO, temperature, and relative
19 humidity in order to assess the sensitivity of the PM-mortality association to model type and
20 model specification. Pollution data were averages of all sites available (e.g., 4 for PM10 and
21 8 for O3 and CO) after first filling in missing days at each site based on available data from
22 other sites (thereby addressing error from day to day variation in site availability). Although
23 the data were collected over 6 years, the PM10 sampling was conducted only every sixth day,
24 so only 364 days could be included in the analyses, limiting the power of the analysis to
25 detect associations. Poisson models were employed which addressed seasonal long wave
26 influences by including sine and cosine terms ranging in periodicity from 1 to 24 mo in
27 periodicity. OLS and log-linear models also considered. Weather was modeled initially by
28 including only same-day maximum temperature and relative humidity in regressions, but
29 sensitivity analyses also considered dummy variables for extreme temperature and up to 3-d
30 lags of all weather variables. Winter and summer were also modeled separately. In these
31 various analyses, PM10 was generally found to be significantly associated with mortality after
April 1995 12-64 DRAFT-DO NOT QUOTE OR CITE
-------
1 controlling for weather and season, with a relative risk (RR) estimate of approximately 1.05
2 (CI = 1.0 to 1.10) reported for a 100 ptg/m3 increase in PM10. Durbin - Watson (DW)
3 statistics indicated only modest autocorrelation in these models (1.8
-------
1 expected due to reduced sample size. However, the PM10 coefficient was not as affected by
2 season-specific analyses, indicating consistent associations throughout the year. Overall,
3 multi-site averaging and larger sample sizes were shown to strengthen the PM10-mortality
4 association, but the results (and the fact that a very basic model specification was employed)
5 leaves open the possibility that other co-pollutants or more elaborate weather specifications
6 could account for part of the Chicago PM10-mortality association.
7 Styer et al. (1995) considered total, respiratory, circulatory, and cancer deaths in Cook
8 County, IL (Chicago) and elderly total deaths in Salt Lake County, UT. The mean number
9 of total, respiratory circulatory, and cancer deaths in Cook County were 117 for all
10 nonaccidental causes, 83 of them elderly (age 65 and over), 10 from respiratory causes (ICD
11 9 codes 11, 35, 472 to 519, 710.0, 710.2, 710.4), 56 from circulatory causes (ICD 9 codes
12 390 to 459), 28 from cancer (ICD 9 codes 140 to 209) per day. They also broke down total
13 mortality by race and by sex. There were two daily PM10 stations in Salt Lake County, one
14 daily station and up to 12 measurements per day from other monitoring stations in Cook
15 County. Models were fitted using Poisson regression, with adjustments for mean daily
16 temperature, specific humidity, and average daily pressure, but with no other air pollutant in
17 the model. Pollen counts and other meteorological variables were evaluated but were not
18 found to significantly improve one fitted model; semi-parametric and parametric models for
19 PM10 were tested, with lags up to 5 days. Seasonal adjustments were significant.
20 The overall PM10 effect was found to be statistically significant overall in Cook
21 County. Spring and Autumn showed significant PM10 effects, whereas Winter and Summer
22 did not. Elderly mortality had twice the excess risk of total mortality. Respiratory deaths in
23 Cook County had nearly three times the response to PM10 as total mortality, but was only
24 marginally significant. The cancer death effect of PM10 was about twice as high as for all
25 cause mortality and statistically significant, unlike findings by Schwartz and Dockery (1992a)
26 for Philadelphia using TSP. Styer et al. (1995) also found no effect of PM10 on elderly
27 mortality in Salt Lake County. The best PM10 predictor for most of the Cook County
28 analyses performed by Styer et al. (1995) was a 3-day moving average (lags 0, 1, 2). While
29 often PM10 lags were evaluated, no other pollutants were tested. Relative risks are reported
30 in Tables 12-3, 12-5, and 12-6. The total mortality RR for 50 jug/m3 in Cook County can be
31 estimated as 1.04 (95% confidence interval 1.00 to 1.08) and is consistent with other studies.
April 1995 12-66 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ostro et al. (1995a) considered total, respiratory, and cardiovascular daily deaths
2 (mean = 55, 8, and 18 deaths/day, respectively) in Santiago, Chile during 1989-1991,
3 examining their relationship to ambient PM10, O3, SO2, and NO2, and to daily minimum and
4 maximum temperature and humidity. To improve exposure estimate representativeness,
5 multiple sites' daily data were averaged for each pollutant (e.g., 4 sites for PM10), though
6 the maximum from all 4 PM10 sites for each day was also considered in some analyses.
7 In this work, most regressions employed the log of PM10, as it showed the highest
8 associations with total mortality in exploratory analyses. OLS regression was employed for
9 most total mortality regressions because a test of normality was not rejected for the total
10 mortality data, though Poisson regressions were employed for cause-specific analyses, in
11 view of their lower daily counts. Also, sensitivity analyses were conducted for various model
12 types: the total mortality RR of the mean PM10 concentration (115 ^g/m3) ranged from 1.04
13 to 1.09 (1.12 with a 3-day average mean PM10 employed). Seasonal influences were
14 alternatively addressed by various methods, including seasonal stratification, the inclusion of
15 sine/cosine trigonometric terms for 2.4, 3, 4, 6, and 12 mo periodicities, prefiltering, and
16 the use of various non-parametric fits of temperature: the PM10 RR estimate ranged from
17 1.04 to 1.11, with the lowest mean PM10 risk provided by the OLS model with
18 5 trigonometric terms included (RR = 1.04). Investigations of mortality by-cause and age
19 indicated that the strongest PM10 associations were for respiratory-specific deaths (RR =
20 1.15) and for the elderly (RR = 1.11). Other pollutants were also considered separately and
21 simultaneously with PM10 in a total mortality regression which also contained 36 dummy
22 variables (one for each month of the study). In this model, the individually significant
23 pollutants were: log(PM10) (RR at mean = 1.05 ; CI = 1.01 to 1.08); SO2 (RR at mean =
24 1.01; CI = 1.00 to 1.03), and; NO2 (RR at mean = 1.02 ; CI = 1.01 to 1.04). Thus, all
25 three pollutants had similar levels of significance in this model. Neither the log of S02 nor
26 the log of NO2, were investigated, however. In multi-pollutant regressions, only log(PM10)
27 stayed significant. The intercorrelations of the various pollutants' coefficients in this model
28 were not reported, but they were likely high, given that the pollutants themselves were highly
29 intercorrelated over time, e.g., r(PM10-NO2) = 0.73. Overall, these results suggest that, of
30 the pollutants considered, PM10 is the air pollutant most strongly associated with mortality in
April 1995 12-67 DRAFT-DO NOT QUOTE OR CITE
-------
I
h-
o\
oo
O
§>
H
6
O
2
o
o
G
O
TABLE 12-5. COMPARISON OF RELATIVE RISK (RR) ESTIMATES FOR TOTAL MORTALITY FROM 50 jtg/m3
CHANGE IN PM10, USING STUDIES WHERE PM10 WAS MEASURED OR WAS CALIBRATED FOR THE SITE
Study
Utah Valley, UT
St. Louis, MO
Kingston, TN
Birmingham, AL
Athens, Greece
Toronto, ON Canada
Los Angeles, CA
Chicago, IL
Santiago, Chile
Chicago, IL
Reference
Pope et al.(1992)
Dockery et al. (1992)
Dockery et al. (1992)
Schwartz (1993)
Touloumi et al. (1994)
Ozkaynak et al. (1994)
Kinney et al. (1995)
Ito et al. (1995)
Ostro et al. (1995a)
Styer et al. (1995)
PM10
Mean
47
28
30
48
78
40
58
38
115
37
(/ig/m3)
Maximum
297
97
67
163
306
96
177
128
367
365
Other Pollutants
In Model
None
None, winter
None, summer
Max O3, summer
Avg O3, summer
None
03
None
03
None
None
S02, CO
None
None
O3, CO
O3, CO
None
None
None, Poisson
SO2, Poisson
NO2, Poisson
O3, Poisson
None
Lag Times, d
< 4d
< 4d
< 4d
< 4d
< 4d
< 3d
< 3d
< 3d
<; 3d
< 3d
1 d
1 d
Od
1 d
1 d
< 3d
1 d
< 4d
1 d
1 d
1 d
1 d
3d
RRper
50 /*g/m3
1.08
1.085
1.11
1.19
1.14
1.08
1.06
1.085
1.09
1.05
1.034
1.015
1.025
1.025
1.017
1.025
1.04
1.07
1.0223
1.026a
1.043a
1.026a
1.04
95 Percent
Confidence Interval
(1.05, 1.11)
(1.03, 1.35)
(0.92, 1.35)
(0.96, 1.47)
(0.92, 1.41)
(1.005, 1.15)
(0.98, 1.15)
(0.94, 1.25)
(0.94, 1.26)
(1.01, 1.10)
(1.025, 1.044)
(1.00, 1.03)
(1.015, 1.034)
(1.00, 1.055)
(0.99, 1.036)
(1.005, 1.05)
(1.035, 1.06)
(1.04, 1.10)
(1.003, 1.042)
(1.005, 1.047)
(1.020, 1.066)
(1.005, 1.047)
(1.00, 1.08)
"Calculated on basis of 50 /*g/m3 increase, from 50 to
n
-------
I
I— '
to
o\
O
H
6
o
1
o
1
o
o
TABLE 12-6
Study
Utah Valley, UT
St. Louis, MO
Kingston, TN
Birmingham, AL
Athens, Greece
Toronto, ON Canada
Los Angeles, CA
Chicago, IL
Santiago, Chile
Chicago, IL
. BASIS FOR EVALUATION OF TIME SERIES STUDIES ON PM10-MORTALITY CITED IN
TABLE 12-5.
Reference
Pope et al. (1992)
Dockery et al. (1992)
Dockery et al. (1992)
Schwartz (1993)
Touloumi et al. (1994)
Ozkaynak et al.
(1994)
Kinney et al. (1995)
Ito et al. (1995)
Ostro et al. (1995a)
Styer et al. (1995)
Period
1985-1989
1985-1986
1985-1986
1985-1988
1984-1988
1972-1990
1985-1990
1985-1990
1989-1991
1985-1990
Other Pollutants In
Model Lag Addressed
Pollutants Temp
None 0-4 d < 1 d
PM25, S04, H+, < 3 d < 1 d
SO2, NO2, O3
PM25, SO4, H+, < 3 d < 1 d
S02,N025, 03
None 0-3 d < 3 d
SO2, CO Id Id
TSP, PM2 5, S04, 0 d 0 d
O3, COH, NO2, SO2
O3, CO Id < 3 d
O3, CO < 3 d < 3 d
O3, SO2, NO2 < 4 d Id
None < 5 d < 2 d
Multiple
Methods
Yes
No
No
Yes
Yes
No
Yes
No
Yes
Yes
Correl. of
B's Given
No 1,436
No 311
No 330
No 1,087
No 1,684
No 6,506
Yes 364
Yes 1,357
No 779
No 1,357
-------
1 this setting. Moreover, sensitivity analysis suggest that the elderly having respiratory
2 diseases are most susceptible to adverse effects from ambient air pollution excursions.
3 A reanalysis of deaths in Utah Valley, UT, from 1985 to 1992 was carried out by Lyon
4 et al. (1995). The data were extensively categorized by year, season, cause, age, and place
5 of death. Based on quintile plots, the authors concluded that excess mortality increased
6 steeply at about 50 /zg/m3 and consequently used only a dichotomous indicator of PM10
7 greater than 50 pig/m3, rather than any linear or nonlinear function of PM10. No other
8 pollutants were used, and the only meteorological variable used in the model was minimum
9 daily temperature. Relative risk (called rate ratio) was calculated from a Poisson regression
10 model without tune series structure adjustment by GEE. However, a linear time trend was
11 used to adjust for decreasing mortality rates over the years. The authors found an apparently
12 random pattern of increased RR, by year, season, age, cause, and place of death. Among
13 their results, they noted the following: strongest effect in spring, not winter; largest
14 contribution to excess mortality from age 75 and over dying in hospital; largest RR for ages
15 15 to 59 dying at home from cancer; increased RR for sudden infant death syndrome.
16 The choice of a 5-day mean PM10 as the exposure metric was based on an earlier study
17 (Pope et al., 1992). However, dichotomizing the PM10 metric at 50 /ig/m3 may have cost a
18 great deal of useful information, possibly including a substantial exposure measurement error
19 or misclassification problem. Since this PM10 metric cannot be scaled to RR increments over
20 other ranges of values, we were not able to include this study in the subsequent tables of this
21 section. However, the authors estimate an excess mortality of 4% for PM10 above
22 50 /ig/m3, roughly consistent with other studies.
23
24 12.3.1.2 Short-Term PM10 Exposure Associations With Total Daily Mortality:
25 Syntheses of Studies
26 Most of the studies summarized in Table 12-3 and discussed in more detail above
27 considered daily mortality in the entire population (i.e., all ages) and due to all causes,
28 though some also considered sub-populations. Considering all these studies in one overall
29 assessment of PM effects on mortality is not a straightforward task, given the variety of
30 models and model specifications employed but, as noted above, this has been attempted
31 previously. Table 12-4 presents a synthesis of two recently published summaries of the PM
32 literature which attempted to convert all results to a PM10-equivalence basis and provide
April 1995 12-70 DRAFT-DO NOT QUOTE OR CITE
-------
1 quantitative intercomparisons (Ostro, 1993; Dockery and Pope, 1994). As also noted above,
2 other such summaries have been conducted using TSP as the reference PM metric (Schwartz,
3 1991; Schwartz, 1994b), but many of the same studies were considered in those two
4 PM10-equivalent summaries, so the TSP-equivalent results are not tabulated here.
5 The results presented in Table 12-4 suggest about a 1 percent change in acute total
6 mortality for a 10 jug/m3 change in PM10, but the estimates range from 0.3 to 1.6% (i.e., a
7 factor of 5). While most of the 95% confidence intervals (CI's) of these estimates overlap,
8 CI's of the highest and lowest estimates do not overlap, indicating significant differences
9 between these estimates. Note that the effects indicated here for a 10 /xg/m3 change cannot
10 be consistently converted to other PM increments (e.g., 50 or 100 jwg/m3 PM10), as
11 differences in model specification (e.g., linear versus log models) will cause them to differ in
12 their conversions to other particle concentration levels. The reasons for the approximately
13 five-fold effect estimate difference noted among studies are not obvious from the information
14 provided by these references, but one factor appears to be the PM exposure averaging time,
15 as estimates using multiple day PM10 averages are all 1 % or higher. This is not unexpected,
16 given that any lagged effects from prior days of PM10 exposure will be added to the effects
17 estimate when a multi-day average is employed, increasing the estimated effect on a per
18 Mg/m3 basis. It is also important to note that other air pollutants have generally not yet been
19 addressed in reaching the coefficients reported here. Also, coefficient variation is to be
20 expected, given that the composition (and, therefore, toxicity) of the PM, as well as the
21 population make-up, in each city can be expected to differ. Moreover, the conversions from
22 other PM metrics to PM10 must necessarily introduce additional uncertainty. This is made
23 apparent here when comparing the estimates for Santa Clara, CA from these two sources,
24 each having its own, somewhat different, estimate of the equivalent PM10 and of the PM10
25 effect. Although not all of these results may therefore be the most appropriate available for
26 quantifying a PM10 effect, they do consistently indicate an association between acute PM
27 exposure and increased daily mortality. Moreover, the by-cause results also reported in these
28 summaries indicate that PM effect estimates are greater for respiratory causes, lending
29 support to the biological plausibility of the PM associations being causal.
30 In an attempt to better quantify the daily PM10-total acute mortality association
31 indicated by the above discussions, Table 12-5 presents a summary of the total mortality
April 1995 12-71 DRAFT-DO NOT QUOTE OR CITE
-------
1 relative risks (RR) of a 50 /*g/m3 increase in PM10 estimated from nine studies reviewed in
2 the above section which employed PM10 data in their analysis of total mortality data (or
3 which had on-site PM10 reference data to convert other PM metrics with more certainty).
4 This selection of studies does not in any way mean to dismiss the other studies discussed
5 above as less important: these studies are selected for this analysis solely because they can
6 most readily be intercompared and referred to the present PM10 standard. The RR's
7 calculated were based upon a 50 /ig/m3 increase above the mean PM10, which is the order of
8 magnitude of the difference between the maximum and mean in these cities and roughly
9 approximates the estimated effects of a typical day experiencing an exceedance of the present
10 PM10 standard, relative to the average case. This is important to note, because in non-linear
11 models such as were often employed in the studies in Table 12-4, the RR estimate associated
12 with a given /ig/m3 PM10 increase will vary depending upon the baseline concentration to
13 which it is added.
14 From the results presented in Table 12-5, it is apparent that these studies generally have
15 yielded at least marginally significant PM10 coefficients, but that the resultant excess risk
16 estimates vary by a factor of five across these studies (from 1.5% to 8.5% per 50 /xg/m3).
17 The mean and maximum PM10 concentration data are noted for each study. If the PM10
18 coefficient decreased as the mean level of PM10 decreased, then confounding would be
19 suggested. However, the data presented indicate that the variability in coefficients is not a
20 function of PM10 level, as sites with high or low PM10 concentrations can report either high
21 or low RR's. In Table 12-6, an attempt is made to concisely summarize the statistical
22 methodology characteristics of each study, in order to determine if any of these factors are
23 important to the variability observed from study to study in the PM10 RR estimate. Of all
24 factors examined in this table, the one consistently present with higher PM10 RR's is when
25 other pollutants have not been simultaneously considered in the model. Indeed, those studies
26 which considered PM10 both alone and with other pollutants in the model yielded consistently
27 smaller (and usually more marginally significant) PM10 relative risks when the other
28 pollutants were simultaneously considered. This influence ranges from roughly a 20 to 50
29 percent reduction in the excess risk associated with 100 /Ltg/m3 in PM10 (e.g., in Athens,
30 Greece, the PM10 RR declines from 1.07 to 1.03 when other pollutants are considered).
31 However, such a reduction is to be expected when co-linear variables are added, and the
April 1995 12-72 DRAFT-DO NOT QUOTE OR CITE
-------
1 "true" PM10 RR is likely to lie between the single pollutant and multi-pollutant model
2 estimates.
3 Another factor which clearly affected the PM10 RR from some of the studies listed in
4 Table 12-5 was the PM10 averaging period. Both of the studies which utilized multi-day
5 averages of PM10 in their regressions (i.e., Utah Valley, UT and Birmingham, AL) are
6 among the higher RR estimate studies. As discussed above, this would be expected, but the
7 increase indicated for these studies is not as large as might be expected. Indeed, in
8 sub-analyses included by Pope et al. (1992), the PM10 mortality risk is indicated to be
9 roughly doubled by using a five day average versus a single day concentration, while
10 sub-analyses presented by Ostro et al. (1995) for Santiago also indicate approximately a
11 doubling in the PM10 RR when a 3 day average is considered (i.e., from RR = 1.04 for a
12 single day PM10 value to RR = 1.07 for a 3d average PM10 value). This may be due to the
13 fact that, since autocorrelation exists in the PM10 concentrations from day to day, the single
14 day concentration is "picking up" some of the effect of multi-day pollution episodes, even
15 though they are not explicitly modeled. Also, most studies show a maximum same-day or
16 one day lag PM-mortality association, with the PM10 "effect" dropping off on subsequent
17 days. These results suggest that a multi-day average PM10 concentration may provide a
18 more relevant index to gauge the effects of acute PM10 exposures than the present single-day
19 based PM10 standard. However, the present standard is in terms of a one-day concentration,
20 so these estimates, while no more valid than those for a multi-day average, are more directly
21 relevant to the present regulatory process, and will be examined further.
22 It appears from Table 12-5 that the total acute mortality relative risk estimate associated
23 with a 50 /wg/m3 increase in the one-day 24-h average PM10 can range from 1.015 to 1.085,
24 depending upon the site (i.e., the PMj0 and population composition) and also upon whether
25 PM10 is modeled as the sole index of air pollution or not. Relative Risk estimates with PM10
26 as the only pollutant index in the model range from RR = 1.025 to 1.085, while the PM10
27 RR with multiple pollutants in the model range from 1.015 to 1.025. The former range
28 might be viewed as approximating an upper bound of the best estimate, as any mortality
29 effects of co-varying pollutants are likely to be "picked up" by the PM10 index, while the
30 multiple pollutant model range might be viewed as approximating a lower bound of the best
31 estimate, as the inclusion of highly correlated covariates are likely to weaken the PM10
April 1995 12-73 DRAFT-DO NOT QUOTE OR CITE
-------
1 estimate, even if they are not themselves causal. Both estimates should be considered in
2 assessing the potential effects of PM10. Overall, consistently positive PM-mortality
3 associations are seen throughout these analyses, despite the use of a variety of modeling
4 approaches, and after controlling for major confounders such as season, weather, and
5 co-pollutants, with the 24-h average 50 /zg/m3 PM10 total mortality effect estimate apparently
6 being in approximately the RR = 1.025 to 1.05 range.
7
April 1995 12-74 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.3.1.3 Short-Term PM10 Exposure Associations with Daily Mortality in Elderly
2 Adults
3 Of the studies in Table 12-3 and/or discussed above, only a few directly examined the
4 elderly as a potentially sensitive sub-population. Certainly, since the highest mortality rates
5 are among the elderly, this is a population which surely dominated the total mortality
6 analyses discussed above, and it is therefore logical to assume that the bulk of the total
7 mortality effects suggested by these studies are among the elderly. Also, as noted earlier,
8 during the historic London, 1952 pollution episode the greatest increase in mortality rate was
9 among older citizens and those with respiratory diseases. However, the analysis of deaths in
10 the elderly population in France by Derriennic et al. (1989) discussed previously found no
11 associations with TSP, while SO2 was associated with total elderly deaths in both cities
12 studied. No PM10 or fine particle metric was considered, however. Similarly, Li and Roth
13 (1995) did not find a significant association between TSP and daily deaths in the elderly.
14 However, an analysis by Schwartz (1994c) of mortality in Philadelphia, PA during 1973
15 through 1980 comparing mortality during the 5% highest versus the 5% lowest TSP days
16 found the greatest increase in risk of death in those aged 65 to 74 and those >74 year of age
17 (mortality risk ratios = 1.09 and 1.12, respectively, between high and low TSP days). Also,
18 in their time series analyses of Philadelphia daily mortality during this period, Schwartz and
19 Dockery (1992a) found a significantly higher TSP-mortality coefficient (fi = 0.000910 ±
20 0.000161) for persons older than 65 years of age than for the younger population (6 =
21 0.000271 ± 0.000206). These coefficients indicated an effect size for the elderly roughly
22 three times that for the younger population (10% versus 3%, respectively, for a 100 jig/m3
23 TSP increase).
24 In addition, two recent PM10 analyses directly considered the question of
25 PM10-mortality associations among the elderly population (> 65 years of age), and these
26 provide further relevant insights into this question. The first of these two analyses was
27 conducted by Saldiva and Bohm (1994) during May 1990 through April 1991 in Sao Paulo.
28 Environmental variables considered included PM10, SO2, NOX, O3, CO, temperature, and
29 humidity. PM10 was not measured gravimetrically, but was instead evaluated via beta gauge
30 instruments calibrated to mass. Pollutants were considered in the analysis in the form of
31 2-day moving averages of concentration (i.e., averages of the same-day and the prior day's
32 concentration). Monitoring data from multiple sites were averaged for each pollutant (e.g., 8
April 1995 12-75 DRAFT-DO NOT QUOTE OR CITE
-------
1 sites for PM10). Multiple regression models estimated the association between daily
2 mortality and air pollution controlling for month of year, temperature, relative humidity, and
3 day of week. Because of the large number of daily deaths (mean = 63/day), Gaussian
4 regression models were appropriately used for the basic analysis. Poisson models using the
5 generalized estimating equation of Liang and Zeger (1986) were also applied for comparison.
6 Temperature effects were crudely accounted for through the use of three dummy variables (T
7 < 8 °C; 8 °C < T < 12 °C; 13 °C < T < 18 °C) in the basic = model. Regression
8 results indicated that, when studied separately, PMi0, SO2, NOX, and CO were all
9 significantly associated with mortality. In a simultaneous regression of mortality on all
10 pollutants, however, PM10 was the only pollutant that remained significant. In fact, the
11 PM10 coefficient actually increased, suggesting confounding among these correlated
12 pollutants. Thus, as noted by the authors, "the close interdependence exhibited by the
13 concentrations of measured pollutants suggests that one has to be cautious when ascribing to
14 a single pollutant the responsibility of causing an adverse health effect". Nevertheless,
15 multiple regression models, including those considering all pollutants simultaneously,
16 consistently attributed the association found with mortality among the elderly to PM10. The
17 reported PM10 relative risk (RR = 1.13 for a 100 jug/m3 increase) is higher than noted above
18 for total mortality studies addressing multiple pollutants (100 /ig/rn3 RR « 1.03 to 1.05),
19 supporting past observations that the elderly represent a population especially sensitive to
20 health effects of air pollution.
21 A second recent study directly examining PM10-mortality associations in the elderly was
22 that by Ostro et al. (1995) in Santiago, Chile. For the overall population, the PM10 100
23 Mi/m3 RR estimate was 1.08, but for the population aged 65 and greater, it rose to an
24 estimate of RR = 1.11 in the same model specification. Thus, these directly comparable
25 estimates (i.e., using the same model specification and population) suggest that the elderly
26 experience roughly a 40 percent higher excess risk from exposure to PM air pollution than
27 the total population.
28 Overall, considering the historical pollution episode evidence and the results of recent
29 PM10-mortality analyses considering elderly populations, it seems evident that elderly adults
30 represent a population especially at risk to the mortality implications of acute exposure to air
31 pollution, including PM.
April 1995 12-76 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.3.1.4 Short-Term PM10 Exposure Associations with Daily Mortality in Children
2 As with analyses of PM-mortality associations for the elderly, few studies have directly
3 examined PM-mortality associations in children. While the previously discussed London Fog
4 episode yielded the greatest increased risk in the older population (e.g., the episode mortality
5 risk versus the week before the episode increased by a factor of 2.74 for persons >45 years
6 old), the second highest increase in risk was in the neo-natal group (ratio = 1.93 for children
7 < 1 year) (United Kingdom Ministry of Health, 1954). More recently, as described above,
8 Schwartz (1994c) examined increased risk of death in Philadelphia, PA for relatively high
9 versus low TSP days during 1973 to 1980 by age, concluding that no pattern of increased
10 risk emerged until age 35 and above (e.g., the high/low TSP mortality ratio for < 1 year of
11 age was 1.01). The author noted increased risk of death on high PM days for children 5 to
12 14 years old, which he suggested may be due to their greater time spent outdoors than other
13 ages, though he notes that the small numbers of deaths in this age group suggest caution in
14 such interpretations.
15 A recent analysis of PM10 pollution and mortality in Sao Paulo, Brazil provides further
16 insight into the potential mortality effects of PM10 on children. Saldiva et al. (1994) studied
17 respiratory mortality among children < 5 yrs old in Sao Paulo during May 1990 to April
18 1991. The environmental variables considered included PM10, SO2, NOX, O3, CO,
19 temperature, and humidity. PM10 was not measured gravimetrically, but was evaluated via
20 beta gauge instruments calibrated to mass. Pollutants were considered in the analysis in the
21 form of 3-day moving averages of concentration (i.e., averages of the same-day and the two
22 prior day's concentrations). Monitoring data from multiple sites were averaged for each
23 pollutant (e.g., 8 sites for PM10). Prior to the analysis, mortality counts were adjusted using
24 a square root transformation to address their non-normal distribution, which results in part
25 from low daily counts (mean = 3.0 deaths/day). Season was addressed by including both
26 seasonal and monthly dummy variables in all regressions. Weather was only crudely
27 addressed, in that only two dummy variables for extreme temperature and two for extreme
28 relative humidity were considered. Day-of-week effects were addressed by the inclusion of
29 six dummy variables, but none were significant. Autocorrelation was not directly addressed
30 in the analyses. Despite the limited data set size, a significant mortality association was
31 found with NOX, but not with any other pollutant. No such association was found for
April 1995 12-77 DRAFT-DO NOT QUOTE OR CITE
-------
1 non-respiratory mortality, which is supportive of the interpretation of the air
2 pollution-respiratory mortality association as causal. In the multiple pollutant model, the
3 PM10 coefficient actually becomes negative (though non-significant), which is likely due to
4 its high intercorrelation with NOX over time (r = 0.68). The high interdependence between
5 NOX and most of the other pollutants caused the authors to note that "interplay among
6 pollutants causing respiratory damage is very difficult to exclude". Thus, while there
7 appeared to be an air pollution association with mortality in children in this case, this study
8 found the strongest association with NOX, though the high intercorrelation among pollutants
9 makes it difficult to designate the effects noted to any one pollutant in this case.
10 Overall, there is an indication among these various analyses that children may be more
11 susceptible to the mortality effects of air pollution exposure than the population in general,
12 but it is difficult, given the limited and somewhat conflicting available results to ascribe any
13 such association to PM pollution in particular.
14
15 12.3.1.5 Short-Term PM10 Exposure Associations with Daily Mortality in Other
16 Susceptible Subgroups
17 Throughout the results and discussions presented above regarding the effects of acute
18 PM exposure on human mortality, a consistent trend was for the effect estimates to be higher
19 for the respiratory mortality category. As discussed above, this lends support to the
20 biological plausibility of a PM air pollution effect, as the breathing of toxic particles would
21 be expected to most directly affect the respiratory tract, and these results are consistent with
22 this expectation. For example, the respiratory mortality relative risk estimates presented in
23 Table 12-4 are all higher than the risks for the population as a whole. Of particular interest
24 is to compare the relative risk values for each study, which yield the most direct and
25 appropriate comparisons. In the case of the Santa Clara study (Fairley, 1990), the
26 respiratory mortality RR of PM was 4.3 times as large as for deaths as a whole (i.e.,
27 3.5/0.8, in Table 12-4). In the case of the Philadelphia, PA study (Schwartz and Dockery,
28 1992a), the respiratory mortality RR of PM was 2.7 times as large as for death as a whole
29 (i.e., 3.3/1.2, in Table 12-4). In the case of the Utah Valley study (Pope et al., 1992), the
30 respiratory mortality RR of PM10 was 2.5 times as large as for deaths as a whole (i.e.,
31 3.7/1.5, in Table 12-4). In the case of the Birmingham, AL study (Schwartz, 1994), the
32 respiratory mortality RR of PM10 was 1.5 times as large as for deaths as a whole (i.e.,
April 1995 12-78 DRAFT-DO NOT QUOTE OR CITE
-------
1 1.5/1.0, in Table 12-4). More recently, the Santiago, Chile PM10 study by Ostro et al.
2 (1995a), reported that the respiratory mortality RR of PM10 was 1.8 times as large as for
3 deaths as a whole (i.e., 1.15/1.08 RR for a 100 /xg/m3). Thus, in these studies, the PM RR
4 for respiratory diseases is indicated to range from 50 to over 400% higher for respiratory
5 disease categories than for all causes of death, indicating that increases in respiratory deaths
6 are a major contributor to the overall PM-mortality associations noted previously.
7 Moreover, since evidence suggests that an acute pollution episode is most likely be inducing
8 its primary effects by stressing already compromised individuals (rather than, for example,
9 inducing chronic respiratory disease from a single air pollution exposure episode), the above
10 results indicate that persons with pre-existing respiratory disease represent a population
11 especially at risk to the mortality implications of acute exposures to air pollution, including
12 PM.
13 In overall summary, the time-series mortality studies reviewed in this and past PM
14 criteria documents provide strong evidence that ambient air pollution can cause increases in
15 daily human mortality. Recent studies provide confirmation that such effects occur at routine
16 ambient levels, and suggest that such effects extend below the present U.S. air quality
17 standards. Furthermore, these new PM studies are consistent with the hypothesis that PM is
18 a causal agent in the mortality impacts of air pollution. Overall, the PM10 relative risk
19 estimates derived from the most recent PM10 total mortality studies suggest a 24-h average
20 50 ng/m3 PM10 acute exposure effect of the order of RR = 1.025 to 1.05 in the general
21 population, with higher relative risks indicated for the elderly sub-population and for those
22 with pre-existing respiratory conditions, both of which represent sub-populations especially at
23 risk to the mortality implications of acute exposures to air pollution, including PM.
24
25 12.3.2 Morbidity Effects of Short-Term PM Exposure
26 12.3.2.1 Hospitalization and Emergency Visit Studies
27 Introduction to Hospitalization Studies
28 Hospitalization for a respiratory illness diagnosis can provide a measure of the
29 respiratory morbidity status of a community during a specified time frame. Such respiratory
30 diagnosis include hospitalization for asthma, pneumonia and influenza. Various factors affect
31 the epidemiology of admissions for these diagnosis. Thomson and Philion (1991) suggest
April 1995 12-79 DRAFT-DO NOT QUOTE OR CITE
-------
1 some factors shown to be independently associated with respiratory hospitalization include
2 poor socioeconomic level, lack of breast feeding, type of heating, and exposure to
3 second-hand tobacco smoke.
4 In the last decade, large increases have occurred in asthma hospitalization rates in the
5 pediatric population. While this pattern has been seen in all age, race and gender groups the
6 most severely affected group were urban blacks (Gerstman et al., 1993). This increase was
7 largest among 0 to 4 years old with blacks having approximately 1.8 times the increase of
8 whites (Gergen and Weiss, 1990). During this time total hospitalization decreased while
9 admissions for lower respiratory tract disease also had a slight decrease (Gergen and Weiss,
10 1990).
11 There are differences in the frequency of admission for asthma by age and gender
12 (Skobeloff et al., 1992). Asthma morbidity is known to exhibit seasonal periodicity. For
13 persons ages 5 through 34 years hospitalization peaked in September through November
14 whereas mortality trends peaked in June through August. For individuals 65-years-old or
15 older, both asthma hospitalization and mortality demonstrated increases during December
16 through February (Weiss, 1990). Crane et al. (1992) states that the most valid and reliable
17 marker of asthma readmission is the number of hospitalization admissions for asthma in the
18 previous 12 mo. In New York City, Carr et al. (1992) found large geographic variations for
19 asthma hospitalization with the highest rate concentrated in the city's poorest neighborhoods.
20 The patients are heavily dependent on hospital outpatient departments and emergency rooms
21 for their ambulatory care. Differences in medical practice styles, reflecting the exercise of
22 physicians discretion in the way illnesses are treated, are important determinants of temporal
23 variation and geographic variation in hospital utilization for many medical conditions.
24 Storr and Lenney (1989) observed a long term variation in children's hospitalization for
25 asthma and school holidays. The admission rate fell during holidays and there were two or
26 more peaks during terms. The pattern is consistent with a largely viral etiology for asthmatic
27 attacks throughout the year. They postulated that school holidays disrupt the spread of viral
28 infectious in a community, with synchronization of subsequent attacks. Travel during
29 holidays may facilitate acquisition of new viral strains by the community.
April 1995 12-80 DRAFT-DO NOT QUOTE OR CITE
-------
1 Based on a total of 450,000 hospitalizations for asthma and an estimated U.S.
2 population of 10,000,000 asthmatics, the incidence of hospitalization for all asthmatic
3 subjects is about 45 per 1,000 asthmatics (National Institutes of Health, 1991).
4 Beard et al. (1992) evaluated interobserver variability during data collection for a
5 population based study of asthma using medical record information. The results suggested
6 that data collection was carried out reliably in this study. Osborne et al. (1992) evaluated the
7 diagnosis of asthma in 320 inpatient and outpatient records bearing the diagnosis of asthma
8 for the period 1970 through 1973 and 1980 through 1983 in a health maintenance
9 organization (HMO). The majority of charts examined exhibited a clinical picture consistent
10 with asthma. The increases in "definite asthma" among outpatients from the 1970s to the
11 1980s reflected increasing chart documentation among physicians. Jollis et al. (1993) study
12 of hospital insurance claim information to include medicare indicated that insurance claim
13 data lack important diagnostic and pragmatic information when compared with concurrently
14 collected clinical data in the study of ischemic heart disease as an example.
15 Wennberg et al. (1984) found that hospital admissions for the following diagnosis
16 related groups showed a very high variation by hospital market area: pediatric pneumonia,
17 pediatric bronchitis and asthma, chronic obstructive lung disease, and adult bronchitis and
18 asthma. Richardson et al. (1991) found that adjusted admission rates for respiratory distress
19 (COPD, asthma, bronchitis, and pneumonia) varied up to 3.09-fold between the highest and
20 lowest hospital market areas in 1986 for the state of Ohio. The reasons for differences
21 between hospital market areas are found in the incidence of illness, variability of local
22 resources, access to care, practice styles of area physicians, numbers of physicians and
23 pulmonologists, inconsistencies in diagnoses, conflicting treatment methodologies, lack of
24 consensus of care, quality of outpatient care, and varying criteria for admission among
25 principal variables. For example, Wennberg et al. (1984) documented great geographic
26 variability in hospital admission rates for adult community acquired pneumonia. This
27 variation suggests that physicians do not use consistent criteria for hospitalization. Specific
28 indications for admission do exist such as the Appropriateness Evaluation Protocol (AEP).
29 Substitution of outpatient for inpatient care is a major strategy promoted to reduce health
30 care cost and as such the majority of patients with community acquired pneumonia are
31 treated as outpatients.
April 1995 12-81 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fedson et al. (1992) state that vaccination practices may play a role in hospitalization
2 rates for influenza and associated respiratory disorders. Despite public health
3 recommendations for influenza vaccination for elderly persons, the vaccine has not been
4 widely used, in the United States only 32% of elderly persons may be vaccinated each year.
5 During the influenza outbreak period most persons with respiratory conditions requiring
6 hospitalization (92%) resided in the community rather than in personal care homes. Also
7 while previous epidemiologic studies arbitrarily defined outbreak periods as the first three
8 months of the year this study indicated that hospitalization discharges for influenza occurred
9 during the period December 1 through February 28.
10 The number and rate of patients discharged by age and first-listed diagnosis in the
11 United States in 1991 are shown in Table 12-7 for all conditions, respiratory disease, heart
12 and circulatory diseases and neoplasma. The number and rate for pneumonia of the
13 respiratory diseases listed are highest for all ages primarily due to the high number and rate
14 for 65 years and over. Disease of the respiratory system approximately represent 10% of all
15 conditions. Specific diseases of the respiratory system are shown in Table 12-8 for 1992 in
16 which five groupings predominate. Pneumonia organism unspecified is the largest group.
17
18 Hospital Admission Studies
19 This section discusses studies of hospital admissions, outpatient visits and emergency
20 room visits, both within the United States and from countries with different medical care
21 systems which may have different medical care practices. Most hospitalization studies
1 consider at least two different classes of admissions. Thus, the studies were not grouped by
2 class, but the tabled results are presented by class.
3 Bates and Sizto (1983, 1986, 1987) reported results of a study relating hospital
4 admissions in southern Ontario to air pollution levels. Data for 1974, 1976, 1977, and 1978
5 were discussed in the 1983 paper. The 1985 analyses evaluated data up to 1982 and showed:
6 (1) no relationship between respiratory admissions and S02 or COHs in the winter; (2) a
7 complex relationship between asthma admissions and temperature in the winter; and (3) a
8 consistent relationship between respiratory admissions (both asthma and nonasthma) in
9 summer and sulfates and ozone, but not to summer COH levels. However, Bates and Sizto
10 note that the data analyses are now complicated by long-term trends in respiratory disease
April 1995 12-82 DRAFT-DO NOT QUOTE OR CITE
-------
oo
Ui
TABLE 12-7. NUMBER AND RATE OF PATIENTS DISCHARGED FROM SHORT-STAY HOSPITALS, BY AGE
AND FIRST-LISTED DIAGNOSIS: UNITED STATES, 1991a
First-listed diagnosis
ICD-9-CD
code
All ages
Under
15 years
15-44
years
45-64
years
Number of patients discharged in
All conditions
Diseases of the respiratory
system
Acute respiratory
infections
Pneumonia
Asthma
Diseases of the circulatory
system including heart disease
Neoplasms
460-519
460-466
480-486
493
390-459
140-239
31,098
3,052
518
1,088
490
5,338
2,001
2,498
736
220
214
187
28
52
11,620
500
68
133
128
396
363
6,173
530
75
152
85
1,509
626
65 years
and over
thousands
10,806
1,286
156
589
90
3,405
960
All ages
Rate of
1,241.1
121.8
20.7
43.4
19.6
213.1
79.9
Under 15
years
15-44
years
patients discharged per
453.2
133.6
39.8
38.9
33.9
5.1
9.5
993.4
42.7
5.8
11.4
10.9
33.9
31.0
45-64
years
65 years
and over
10,000 population
1,321.6
113.4
16.0
32.5
18.2
323.1
133.9
3,403.1
405.2
49.2
185.5
28.5
1,072.4
302.3
^ aDischarges from non-Federal hospitals. Excludes newborn infants. Diagnostic groupings and code number inclusions are based on the International Classification
of Diseases, 9th Revision, Clinical Modification (ICD-9-CD).
Adapted from National Center for Health Statistics (1993).
O
O
25
O
d
o
O
-------
VO
TABLE 12-8. NUMBER OF FIRST-LISTED DIAGNOSES FOR INPATIENTS DISCHARGED FROM SHORT-STAY
NON-FEDERAL HOSPITALS, BY ICD-9-CM CODE, AGE OF PATIENT, AND GEOGRAPHIC REGION OF
HOSPITAL: UNITED STATES, 1992-CON
First-listed diagnosis ICD-9-CM code
h— *
1
2
0
^
6
0
1
0
o
a
o
n
s
Diseases of the respiratory system
Acute Bronchitis
Viral Pneumonia
Pnuemoccal pneumonia
Other bacterial pneumonia
Pneumonia, other specified organisms
Broncho pneumonia, organism unspecified
Pneumonia, organism unspecified
Influenza
Bronchitis, unspecified
Chronic Bronchitis
Emphysema
Asthma
460-519
466
480
481
482
483
485
486
487
490
491
492
493
Total
2,923
251
39
53
202
20
45
700
13
23
201
29
463
Under
15 years
735
149
27
6
11
*
16
145
*
9
*
193
15-44
years
460
21
*
10
26
*
*
87
*
*
7
*
117
Age
45-64
years
Number
501
23
*
10
31
*
*
108
*
6
52
8
78
Region
65 years
and over
Northeast
of first-listed diagnoses in
1,227
58
*
27
134
8
22
360
6
*
141
18
76
635
49
*
10
34
*
9
134
*
*
45
6
116
Midwest
thousands
704 1
64
12
14
47
5
9
177
*
*
40
8
113
South
,139
104
12
17
87
8
22
287
*
11
86
12
152
West
445
33
10
12
35
*
5
101
*
*
30
*
83
*Figure does not meet standard of reliability or precision.
Adapted from National Center for Health Statistics (1994).
-------
1 admissions unlikely related to air pollution, but they nevertheless hypothesize that observed
2 effects may be due to a mixture of oxidant and reducing pollutants which produce intensely
3 irritating gases or aerosols in the summer but not in the winter. In a more recent paper,
4 Bates and Sitzo (1987) extend the time period through 1983 and include additional air
5 sampling data not available previously. The monitoring was from 17 air sampling stations
6 and included O3, sulfate fraction, SO2, NO2, and COH. Stepwise multiple regressions
7 confirmed the earlier findings that there was a consistent summer relationship between
8 sulfates and O3 with hospital admissions. The analyses did not adjust for time trends, trends
9 within the summer season, or serial correlation.
10 Lipfert and Hammerstrom (1992) conducted a 6-year study of hospital admissions in
11 southern Ontario for 1979 to 1985. Daily hospital admissions were obtained from the
12 Ontario Ministry of Health, the same data base used by Bates and Sitzo (1983, 1986, 1987).
13 The primary focus of the study was on respiratory illness in one of the following ICD codes:
14 466 acute bronchitis, 480 to 482 or 485 pneumonia, 490 to 492 chronic bronchitis,
15 emphysema, or 493 asthma. Three regions were defined with slightly different air pollution
16 exposures, based on data from the Ontario Ministry of the Environment for SO2, NO2, O3,
17 sulfate fraction, COH, and TSP. Some stations monitored every three or six days and
18 averages were taken by region for those monitors present. A Box-Jenkins ARIMA multiple
19 regression model was used to analyze the data. Bivariate correlations were calculated
20 between the pollutants and respiratory illness. Stepwise multiple regressions did not include
21 TSP as a significant factor, but O3 was significant for January and February and SO2 was
22 significant for some regions in July and August.
23 Burnett et al. (1994) studied hospital admissions in southern Ontario, using a broader
24 area than that used by Bates and Sitzo (1983, 1986, 1987). The respiratory admissions were
25 for 1983 to 1988 and were restricted to the ICD9 codes of 466, 480 to 486, 490 to 494, and
26 496. The non respiratory control admissions included the codes of 280 to 281.9, 345 to 347,
27 350 to 356, 358 to 359.5, 530 to 534, 540 to 543, 560 to 569, 571, 572, 574 to 578, 594,
28 and 600. Twenty-two monitoring stations were used to estimate daily O3 and sulfate fraction
29 data; meteorological data came from 10 different stations. The daily fluctuations in
30 admissions were related to the pollution and meteorological data after subtracting a 19 term
31 linear trend as discussed by Shumway et al. (1983). The rates were analyzed using a random
April 1995 12-85 DRAFT-DO NOT QUOTE OR CITE
-------
1 effects model, where hospitals were assumed to be random. The estimates were obtained
2 using the generalized estimating equations (GEE) of Liang and Zeger (1986). In general,
3 O3, sulfate fraction, and temperature were all predictors of hospital admissions; but O3
4 tended to be more significant than did sulfate fraction. The models predicted about a 3 %
5 increase in respiratory hospital admissions for about a 14 /ig/m3 concentration of sulfate
6 fraction.
7 Thurston et al. (1994) studied hospital admissions in the Toronto metropolitan area.
8 during the months of July and August of 1986, 1987, 1988 and restricted to the following
9 causes: total respiratory (ICD9 codes 466, 480, 481, 482, 485, 490 to 493), asthma (493),
10 and non respiratory control (365, 430, 431, 432, 434, 435, 531, 543, 553.3, 537, 540, 541,
11 542, 543, 590). There were no stated restrictions on age. Pollution data consisted of acidity
12 (H+) and sulfate data measured at three sites during the three summer seasons. In addition,
13 O3, NO2, and SO2 and daily 24-h PM2 5 and PM10 were measured at several other stations.
14 Meteorological measurements were available from two of the monitoring sites. Ordinary
15 least squares analyses were calculated after the environmental variables were detrended. The
16 data for the three summers were combined. In general, O3 was the strongest predictor of
17 hospital admissions above the strong effect of temperature. There was some suggestion of an
18 effect from PM10, especially for total respiratory admissions. Non-linear temperature terms
19 were not fitted.
20 Sunyer et al. (1991, 1993) studied daily emergency room admissions for COPD in
21 adults in Barcelona, Spain. The original study included admissions for the years 1985 and
22 1986. A specially trained physician collected data from clinical records from the four largest
23 hospitals in Barcelona. A panel of chest physicians defined expressions used to determine
24 the diagnosis of COPD. Seventeen manual samplers and two automatic samplers took 24-h
25 measurements of SO2, black smoke, CO and 03. Neither SO2 nor black smoke exceeded the
26 European Community standards. A Box-Jenkins ARIMA (auto-regressive integrated moving
27 average) tune series model was used to analyze the results. COPD was found to be related
28 to SO2, black smoke, and CO. The relationship with black smoke was especially pronounced
29 for temperatures greater than 11.7° C. In the later paper, Sunyer et al. (1993) included the
30 larger time period of 1985 to 1989. The study was restricted to individuals in the four
31 largest hospitals at least 14 years of age. Fifteen manual samplers provided SO2 and black
April 1995 12-86 DRAFT-DO NOT QUOTE OR CITE
-------
1 smoke measurements. Ridge regression (a modification of standard multiple linear
2 regression) was used to analyze the daily admissions, but the analyses were done separately
3 by season. Ridge regression is a conservative method of handling collinear variables, but it
4 does not take into account the effects of non-normality of counts. Lag variables to adjust for
5 the autocorrelation were selected according to the methodology of Box and Jenkins (1979).
6 Significant changes in admissions were found for both SO2 and black smoke for the winter
7 season, but only SO2 was significant in the summer.
8 Hospital admissions for all hospitals in the Birmingham, AL, SMSA were studied by
9 Schwartz (1994e). The admissions were restricted to pneumonia (ICD-9 codes 480 to 487)
10 and chronic obstructive pulmonary disease (COPD) (ICD-9 codes 490 to 496) from
11 January 1, 1986 to December 31, 1989. Only persons age 65 were included in the analysis.
12 Daily pollution estimates of PM10 and O3 were computed by averaging all Birmingham
13 stations reporting on a given day. The author used three different models for the analysis
14 including (1) Fourier series adjustments for season with linear and quadratic terms for
15 temperature, dew point, and time trend, (2) a similar model with cubic splines used instead
16 of Fourier series, and (3) a nonparametric approach. Serial correlation was adjusted for
17 using the generalized estimating equations of Liang and Zeger (1986). The various models
18 gave reasonably similar results. The relative risk of pneumonia was found to be about 1.16
19 (1.05 to 1.28) corresponding to an increase of 100 jig/m3 of PM10. The relative risk of
20 COPD was found to be about 1.24 (1.05 to 1.45) for an increase of 100 /xg/m3 of PM10.
21 Associations with O3 were found to be slightly weaker.
22 Schwartz (1994d) also studied hospital admissions for the elderly in Detroit, restricted
23 to pneumonia (ICD-9 codes 480 to 486) and chronic obstructive pulmonary disease (COPD)
24 (ICD-9 codes 490 to 496) from January 1, 1986 to December 31, 1989. Only persons age
25 65 or older were included in the analysis. Separate counts were constructed for asthma (493)
26 and all other COPD (491 to 492 and 494 to 496). Daily pollution estimates of PM10 and O3
27 were computed by averaging all Detroit metropolitan area stations reporting on a given day.
28 The author used three different approaches to the analysis, including a nonparametric
29 approach. Serial correlation was adjusted for using autoregressive terms which were
30 estimated using the generalized estimating equations of Liang and Zeger (1986). The various
31 models gave reasonably similar results. The multiple logistic regression coefficient for
April 1995 12-87 DRAFT-DO NOT QUOTE OR CITE
-------
1 pneumonia was 0.00115 (0.00039) which corresponds to an odds ratio of about 1.06 (1.02 to
2 1.10) for an increase of 50 /ng/m3 of PM10. The multiple logistic regression coefficient for
3 COPD was 0.00202 (0.00059) which corresponds to an odds ratio of about 1.11 (1.04 to
4 1.17) for an increase of 100 /ig/m3 of PM10. Associations with O3 were also found, but the
5 dose response relationship was not as consistent.
6 Schwartz (in press) studied hospital admissions for all hospitals in Spokane, WA. The
7 admissions were restricted to respiratory disease (ICD-9 codes 460 to 519) from January 1,
8 1988 to December 31, 1990. Only individuals > 65 yrs were included in the analysis.
9 Daily pollution estimates of PM10 and O3 were computed by averaging all Spokane stations
10 reporting on a given day. PM10 values averaged 46 j^g/m3 with 10 and 90 percentile values
11 of 16 and 83 Mg/m3. The author used three different models for the analysis, including (1)
12 Fourier series adjustments for season with linear and quadratic terms'for temperature, dew
13 point, and tune trend, (2) a similar model with cubic splines used instead of Fourier series,
14 and (3) a nonparametric approach. Serial correlation was adjusted for using the generalized
15 estimating equations of Liang and Zeger (1986). The various models gave reasonably similar
16 results. The relative risk of respiratory disease was about 1.08 (1.04 to 1.14) corresponding
17 to an increase of 50 pig/m3 of PM10. Associations were also found with O3, giving a relative
18 risk of 1.24 (1.00 to 1.54) for an increase of 50 ^tg/m3. Inclusion of both pollutants in the
19 model had little effect on either estimate.
20 Ponka and Virtanen (1994) studied hospital admissions for exacerbations of chronic
21 bronchitis (ICD-9 code 491) and emphysema (ICD-9 code 492) in Helsinki, Finland during
22 1987 to 1989. Individuals with the diagnosis of asthma (ICD9 code 493) were excluded.
23 Sulfur dioxide was measured hourly at four stations, NO2 at two stations, and O3 at one
24 station; TSP was measured every other day at four stations and every third day at two
25 stations. Meteorological information was available from a single station but the location was
26 not specified. Daily admissions were analyzed using Poisson regression as described by
27 McCullagh and Nelder (1989). The model included variables for season, day of week, year,
28 and influenza epidemics. The authors report that the day of week variables effectively
29 reduced the autocorrelation, and so autocorrelation terms were not included due to their
30 difficulty of interpretation. For persons < 65 years old, the only effects seen were with SO2
31 on the same day or three days previous. For individuals older than age 64, the only effect
April 1995 12-88 DRAFT-DO NOT QUOTE OR CITE
-------
1 seen was for NO2 six days previous. Although these results are difficult to interpret, the
2 study did not find any results suggesting a PM effect.
3 Ponka (1991) also studied hospital admissions for asthma (ICD9 code 493) in Helsinki
4 during 1987 to 1989. Persons with the diagnosis of bronchiolitis were excluded. Sulfur
5 dioxide was measured hourly at four stations, NO2 at two and O3 at one; TSP was measured
6 every other day at four stations and every third day at two. Meteorological information was
7 available from a single station. The analysis was done using simple and partial age specific
8 correlations of asthma admissions with mean daily concentrations of SO2, NO2, NO, CO,
9 TSP, O3, temperature, wind speed and humidity. No adjustment was made for season or
10 serial correlation. TSP was found to be significantly correlated with hospital admissions, but
11 was less correlated than some of the other pollutants.
12 White et al. (1994) studied asthma outpatient clinic visits of children at Grady
13 Memorial Hospital in Atlanta. The encounter forms for each child between June 1, 1990 and
14 August 31, 1990 were abstracted, excluding visits when pneumonia or bronchiolitis was
15 mentioned. Hourly O3 measurements were available from two stations in the area. PM10
16 data were available from the middle of July, but data before that time had to be estimated
17 using visibility data from Hartsfield International Airport. Clinic visits were increased when
18 O3 exceeded 0.11 ppm. Using a Poisson regression model, the estimated increase, as
19 measured by a rate ratio, was 1.02 (CI = 0.96, 1.13) for a 10 ^g/m3 increase in PM10.
20 Tseng et al. (1992) studied quarterly hospital discharges for asthma (ICD-9 Code 493)
21 from the computerized hospital inpatient data base of the Medical and Health Department of
22 Hong Kong. The study ran from the second quarter of 1983 to the last quarter of 1989. The
23 discharges were split into four groups: under age 1, age 1 to 4, age 5 to 14, and adult.
24 Quarterly averages of SO2, NO2, O3, TSP and RSP values were obtained from the
25 environmental protection unit of the Hong Kong Government. Multiple regression analyses
26 were performed on the hospitalization rates using the four different age groups as the
27 dependent variables and the pollution values as the independent values. Season and year
28 were used as covariates, but no meteorological variables were included in the analyses. The
29 significant correlations were between TSP and hospitalization rates for children aged 1 to
30 4 and children aged 5 to 14. The correlations for RSP tended to be similar, but smaller in
31 magnitude.
April 1995 12-89 DRAFT-DO NOT QUOTE OR CITE
-------
1 Queiros et al. (1990) studied asthmatic admissions and Emergency Room visits to the
2 Pediatric Department of the hospital de S. Joao during the period from 1983 to 1987. The
3 hospital serves the Oporto area of Portugal. Air pollution was estimated from measurements
4 of SO2 and black smoke (BS) taken daily at four stations. The admissions were adjusted so
5 that the values represented deviations from the average for a particular month or year. No
6 correlation was found between daily, monthly, or quarterly mean admissions or visits and BS
7 levels but SO2 levels were correlated with monthly mean admissions. The authors concluded
8 that there was no evidence for pollution effects on admissions or visits.
9 During January 1985, large parts of Europe from western Germany to Great Britain
10 experienced a pollution event that was traced to emission sources in Central Europe. This
11 event was tracked by monitoring stations in several countries as it moved from east to west,
12 and then finally dissipated over the North Sea. Very high levels of PM, SO2, and NOX were
13 reported. Wichmann et al. (1989) studied mortality, hospital admissions, ambulance
14 transports and outpatient visits for respiratory and cardiovascular disease in West Germany
15 during the 1985 event. During that time, daily suspended paniculate matter reached 600
16 /ig/m, SO2 reached 830 /-tg/m3, and NO2 reached 410 jig/m3. Total mortality rose
17 immediately with the increase in pollution (January 16, 1985), and reached a maximum on
18 January 18. The increase in mortality was about 8 percent. Similarly, increases in hospital
19 admissions (15 percent), outpatient visits (12 percent), and ambulance transports (28 percent)
20 were seen. Wichmann et al. (1988a,b) reported on other events in 1986 and 1987 which
21 related lung function changes to SO2 levels but did not report PM data.
22 Walters et al. (1994) studied hospital admissions in Birmingham, England. The
23 admissions were restricted to asthma or acute respiratory disease (ICD9 codes of 466, 480 to
24 486, and 490 to 496) for the period of April 1988 to March 1990. No age restrictions were
25 indicated. Seven monitoring stations were used to estimate British standard black smoke and
26 sulfur dioxide. Meteorological information came from the University of Birmingham
27 Department of Geography. The data were divided into four seasons for analysis to control
28 for seasonal variation in all variables. Stepwise multiple regression models were fitted to the
29 hospital admissions data using pollution and meteorological variables as independent
30 variables. Marginally significant regression coefficients were found for both pollutants for
31 both endpoints, especially in the winter season. Additional analyses were run using 2-day
April 1995 12-90 DRAFT-DO NOT QUOTE OR CITE
-------
1 lags of the pollution variables, and some of these were marginally significant. This study
2 adds little to the effect of particulate matter on respiratory hospital admissions because of the
3 difficulties in comparing black smoke to particulate fractions.
4 In another study, (Schwartz et al., 1993), emergency room visits for 8 hospitals in the
5 greater Seattle area were abstracted for the period September 1, 1989 to September 30, 1990.
6 Asthma was defined as a diagnosis of ICD9 Codes 493, 493.01, 493.10, 493.90 and 493.91.
7 Sulfur dioxide was measured at an industrial site, PM10 was available from a residential area
8 north of town, and O3 was measured at a site 20 km east of town. Poisson regression as
9 described by McCullagh and Nelder (1983) was used to estimate the effect of pollution on
10 asthma visits with adjustments for serial correlation using the method of Zeger and Liang
11 (1986). Logistic regression coefficient estimated from the Poisson regression gave a values
12 of .0036 (.0012) for PM10. The pollution monitors were located far from the study
13 population, but the analyses of partial data suggested that the station produced estimates that
14 were highly correlated with the local data.
15 Urgent hospital admissions for respiratory illnesses in Montreal, Canada were collected
16 from 14 hospitals from 1984 to 1988, and were split into asthma and non-asthma admissions
17 Delfino et al. (1994). The definitions were similar to those used by Bates and Sitzo (1987).
18 City-wide averages of O3, PM10, and sulfate fraction were calculated from seven selected
19 monitoring stations. PM10 was measured every sixth day, and values for the other five days
20 were estimated. A high-pass filter was used to eliminate yearly seasonal trends (see
21 Shumway et al., 1983). Weather variables included temperature and humidity. Regression
22 analyses with and without autoregressive terms found few significant relationships between
23 the health endpoints and the various pollutants.
24 Duclos et al. (1990) studied hospital admissions for respiratory and non-respiratory
25 conditions during several forest fires in northern California. The fires commenced on
26 August 30, 1987, and TSP levels increase to about 300 ptg/m3 from a background level
27 generally below 100 /ig/m3. The analysis consisted of comparing observed versus expected
28 rates without adjustment for serial correlation or other factors. Although there was a
29 significant increase in visits for respiratory conditions, the same pattern appeared for visits
30 for injuries.
April 1995 12-91 DRAFT-DO NOT QUOTE OR CITE
-------
1 Pope (1991) studied hospital admissions in the Salt Lake Basin during the period
2 surrounding the shut-down or strike of the steel mill. According to Pope (1991), PM10
3 pollution in the Utah Valley came from many sources, but the primary source was a
4 45-year-old integrated steel mill with coke ovens, blast furnaces, open hearth furnaces, and a
5 sintering plant. When in operation, the mill emitted 82 to 92% of the valley's industrial
6 PM10 pollution and 50 to 70% of the total Utah Valley PM10 emissions. The steel mill shut
7 down from August 1, 1986 to September 1, 1987. Winter PM10 levels were approximately
8 twice as high when the mill was open compared to when it was closed. Three mountain
9 areas of central and north central Utah were monitored for admissions to three local
10 hospitals. Daily admissions for asthma, bronchitis, and pneumonia were recorded. PM10,
11 SO2, and NO2 levels were monitored at a site 5 km northeast of the steel mill. Admissions
12 for bronchitis and asthma were higher during periods of operation of the steel mill when
13 compared to other areas of Utah. Logistic regressions were generally not significant, but
14 respiratory hospital admissions were associated with monthly mean PM10 levels.
15 Hefflin et al. (1994) compared the number of emergency room visits in southeast
16 Washington state for twelve respiratory disorders for each day of 1991 with daily PM10
17 levels. During two dust storms on October 16 and 21, 1991 PM10 reached 1,689 and
18 1,035 /-ig/m3, respectively. Other pollutants were not measured. Particulates in rural eastern
19 Washington, which are volcanic in origin, fall mostly in the PM10 fraction and belong to the
20 plagioclase (glass) mineral class of aluminum silicates and other oxides. The authors used a
21 Poisson regression model to predict daily emergency room visits as a function of season,
22 relative humidity, and one and 2-day lags of PM10 pollution. Variances were estimated using
23 the generalized estimating equations with an exchangeable correlation structure as described
24 by Liang and Zeger (1986). Daily emergency room totals for each disorder, except
25 respiratory allergy, had a statistically significant inverse correlation with mean daily
26 temperature. The maximum observed/ expected ratio for respiratory disorders from the dust
27 storms on October 16 and 21 was 1.2. The author considered this relatively low ratio for
28 such high pollution days as indicating that the high PM10 levels probably had a minimal
29 public health impact. A statistically significant relationship between a year of daily PM10
30 levels for emergency room visits for bronchitis and sinusitis was found, although the
April 1995 12-92 DRAFT-DO NOT QUOTE OR CITE
-------
1 estimated regression coefficient indicated a small effect. Ten other disorders, including
2 asthma, pneumonic influenza, and COPD did not show this relationship.
3 Thurston et al. (1992) studied hospital admissions for respiratory disease among all
4 ages in Buffalo, Albany, and New York City during July and August, 1988-1989. Three
5 monitoring stations (one per city) measured sulfate, H + , and ozone. A linear regression
6 analysis on filtered data showed relative risk of 1.05 (1.01, 1.10) for sulfate. No results
7 were given for measures of PM.
8 Schwartz (1994f) studied hospital admissions for elderly patients in Minneapolis during
9 1986 to 1989. Exposure measurements were obtained from 6 monitoring stations which
10 measured PM10 and O3. The mean PM10 value was 36 /zg/m3, the 10th percentile was 18
11 and the 90th was 58. The mean O3 value was 26 ppb, the 10th percentile was 11 and the
12 90th was 41. An autoregressive Poisson model with 8 categories of temperature and dew
13 point, month, year, linear and quadratic time trend was used to analyze the data. The
14 estimated odds ratio for a 50 pig/m3 increase in PM10 was 1.25 (1.10, 1.44) for COPD
15 (ICD9 490 to 496) and 1.08 (1.01, 1.15) for pneumonia (ICD9-480 to 487).
16 Schwartz (in press) studied respiratory hospital admissions (ICD9-460-519) for elderly
17 patients in New Haven and Tacoma during 1988 to 1990. For New Haven, the PM10 data
18 was averaged from all monitoring stations giving data. The mean PM10 was 41, the 10th
19 percentile 19 and the 90th percentile 67 /xg/m3. The mean O3 percentile 16 and the 90th
20 percentile 45 jig/m3. The mean SO2 was 30 ppb, the 10th percentile 9 and the 90th
21 percentile 61. A Poisson log-linear regression model using with a 19 day moving average
22 filter was used to analyze the data. Temperature and dew point were adjusted for in the
23 moving average. The odds ratio for respiratory hospital admissions for a 50 /ig/m3 increase
24 in PM10 was 1.06 (1.00, 1.13). Using a two day lag SO2 term in the model, the odds ratio
25 was 1.07 (1.01, 1.14). The same analysis was run for the Tacoma data. The odds ratio for
26 respiratory hospital admissions for a 50 ^g/m3 increase in PM10 was 1.10 (1.03, 1.17).
27 Using a two day lag SO2 term in the model, the odds ratio was 1.11 (1.02, 1.20).
28 Schwartz and Morris (in press) studied ischemic heart disease hospital admissions
29 (ICD9 410 to 414, 427 and 428) for the elderly in Detroit from 1986 to 1989. There were
30 from 2 to 1 1 PM10 monitoring stations operating during the study period, and data were
31 available for 82% of possible days The mean PM10 was 48 j^g/m3, the 10th percentile 22
April 1995 12-93 DRAFT-DO NOT QUOTE OR CITE
-------
1 and the 90th percentile 82. The mean S02 was 25 ppb, the 10th percentile 11 and the 90th
2 percentile 44. A Poisson auto-regressive model using GEE was used to analyze the data with
3 dummy variables for temperature, month, and linear and quadratic time trend. The odds
4 ratio for hospital admissions for ischemic heart disease for a 50 /ig/m3 increase in PM10 was
5 1.06 (1.02, 1.10). Using O3, CO and S02 in the model resulted in an odds ratio of 1.06
6 (1.02, 1.10).
7 Cardiac and respiratory hospital admissions in 168 acute care hospitals in Ontario, CA
8 for 1983 to 1988 calendar years were studied by Burnett et al. (in press). The cardiac
9 admissions were defined as ICD9 codes 410, 413, 427, and 428, and the respiratory
10 admissions as codes 466, 480 to 486, 490 to 494 and 496. No other age restrictions were
1 1 given. Twenty-two monitoring stations were used to estimate daily O3 zone and sulfate
12 fraction data. Meteorological information came from 10 different stations. The daily
13 fluctuations in admissions were related to the pollution and meteorological data after
14 subtracting a 19 term linear trend as discussed by Shumway et al. (1983). The rates were
15 analyzed using a random effects model, where hospitals were assumed to be random. The
16 estimates were obtained using the generalized estimating equations (GEE) of Liang and Zeger
17 (1986). The sulfate fraction, O3, and temperature were all predictors of hospital admissions,
18 with O3 more significant than the sulfate fraction. The models tended to predict about a 3 to
19 4% increase in respiratory admissions and about a 2 to 3% increase in cardiac admissions
20 with about a 13 jttg/m3 increase in the concentration of sulfate fraction.
21
22 Hospital Admission Studies Summary
23 Hospitalization data can provide a measure of the morbidity status of a community
24 during a specified time frame. Hospitalization data specific for respiratory illness diagnosis
25 or more specifically for COPD and pneumonia give a measure of the respiratory status.
26 Such studies provide an outcome measure that relates to mortality studies for total and
27 specified respiratory measures. Tables 12-9 through 12-12 summarize these studies. These
28 studies associate hospitalization data with PM measures. Many of the same factors and
29 concerns related to the mortality studies are at issue for these studies also.
30 Both COPD and pneumonia hospitalization studies show moderated but statistically
31 significant relative risks in the range of 1.06 to 1.25 resulting from an increase of 50
April 1995 12-94 DRAFT-DO NOT QUOTE OR CITE
-------
1 in PM10 or its equivalent. There is a possible suggestion of a relationship to heart disease,
2 but the evidence is very inconclusive. The admission studies of respiratory disease show a
3 similar effect. The hospitalization studies in general use very similar analysis methodologies
4 and the majority of the papers are written by a single author. Overall, these studies are
5 indicative of health outcome related to PM. They are also supportive of the mortality
6 studies, especially with the more specific diagnosis relationships.
7 Schwartz (in press) reviewed the hospital admission and mortality studies of paniculate
8 matter and ozone. The hospitalization results were based on the studies of Thurston et al.,
9 (1992), Schwartz (1994e), Burnett et al. (1994), Schwartz et al. (1995), Schwartz (1994f),
10 Sunyer et al. (1993), Schwartz (1994d), and Burnett et al. (1995). Summary tables for all
11 respiratory admissions showed relative risks ranging from 1.10 to 1.20 per 100 /*g/m3 PM10
12 (or equivalently, 1.05 to 1.10 per 50 /xg/m3 PM10). Summary tables for COPD admissions
13 showed relative risks ranging from 1.15 to 1.57 per 100 jttg/m3 PM10 (or equivalently, 1.07
14 to 1.25 per 50 jig/m3 PM10). Schwartz (in press) argues that because there is no significant
15 heterogeneity in the relative risks across studies that
16 "This suggests that confounding by other pollutants or weather is not the source of
17 these associations, since the coincident weather patterns and levels of other
18 pollutants varied greatly across the studies. In particular, studies in the western
19 United States (Spokane, Tacoma) had very low levels of sulfur dioxide, and much
20 less humidity than [sic] in the eastern United States locations."
21 However, tests for homogeneity are known to have very little power against specific
22 alternatives, and so this conclusion may not be appropriate (Hunter and Schmidt, 1989).
23 The hospitalization studies usually compared daily fluctuations in admissions about a
24 long term (e.g., 19 day) moving average. These fluctuations were regressed on PM
25 estimates for the time period immediately preceding or concurrent with the admissions.
26 Some authors considered lags up to 5 days, but the best predictor usually was the most recent
27 exposure. Some morbidity outcomes associated with hospitalization may be appropriately
28 associated with concurrent admission, while others may require several days of progression
29 to end in an admission. Exposure-response lag periods are not yet well examined.
April 1995 12-95 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-9. HOSPITAL ADMISSIONS STUDIES FOR RESPIRATORY DISEASE
Ul
to
Study
Burnett et al. (1994)
All ages in Ontario,
Canada, 1983-1988
Thurston et al. (1994)
All ages in Ontario,
Canada, July and August,
1986-1988
Thurston et al. (1992)
All ages in Buffalo,
Albany, New York City,
July and August, 1988-1989
Schwartz (in press)
Elderly in New Haven,
1988-1990
Schwartz (in press)
Elderly in Tacoma, 1988-
1990
PM Type &
No. Sites
9 monitoring
stations
measuring
sulfate
3 monitoring
stations
measuring
sulfate, TSP,
and PMjo
3 monitoring
stations (one
per city)
measuring
sulfate, H+
PM10
monitoring
stations
averaged, no.
of stations not
given
PM10
monitoring
stations
averaged, no.
of stations not
given
PM Mean Ave. Count
& Range per Day
sulfate means 108
ranged from 3.1
to 8.2 /tg/m3
mean sulfate 14.4
ranged 38 to 124
(nmole/m3), PM10
30 to 39 /ig/m3,
TSP 62 to 87
/tg/rn3
(values not given) Buffalo, 24
Albany, 12,
New York,
137
mean =41, 8.1
10% tile = 19,
90% tile = 67
mean = 37, 4.2
10% tile = 14,
90% tile = 67
Model Type
&Lag
Structure
Lin. regress.
on filtered
data, 1-d lag
best
Linear
regression on
filtered data,
0-d lag best
Linear
regression on
filtered data
Poisson log-
linear
regression,
19 day mov.
ave. filter,
0-d lag best
Poisson log-
lin. regress.
19 day mov.
ave. filter,
0-d lag best
Other pollutants
measured
Ozone
Ozone, H + ,
SO2, N02
Ozone, H +
Weather &
Other Factors
Temperature
Temperature
Temperature
Result*
Pollutants (Confidence
in model Interval)
none 1 .03
(1.02,
none PM10
1.09
(0.96,
ozone PM10
1.01
(0.87,
1.04)
1.22)
1.15)
ozone (not given for
PM measures)
Ozone (ppb):
mean = 29;
10% tile = 16;
90% tile = 45;
SO2 (ppb): mean
= 30; 10% tile
= 9; 90% tile
= 61
Ozone (ppb):
mean = 25;
10% tile = 13;
90% tile = 36;
SO2 (ppb): mean
= 17; 10% tile
= 6; 90% tile
= 28
Temperature
and dew point
adjusted for in
the moving
average
Temperature
and dew point
adjusted for in
the moving
average
none 1 .06
(1.00,
SO2(2day 1.07
lag) (1.01,
none 1.10
(1.03,
SO2(2day 1.11
lag) (1.02,
1.13)
1.14)
1.17)
1.20)
* Relative risk calculated from parameters given by author assuming a 50 /xg/m3 increase in PM10 on 100 /ig/m3 increase in TSP.
-------
TABLE 12-10. HOSPITAL ADMISSIONS STUDIES FOR COPD
3.
Study
Sunyer et al. (1993)
Adults in Barcelona,
1985-1989
Schwartz (1994f)
Elderly in Minneapolis,
1986-1989
PM Type &
No. Sites
15 monitoring
stations
measuring
black smoke
6 monitoring
stations
measuring
PM10
PMMean
& Range
whiter 33% tile =
49, 67% tile =
77, summer 33%
tile = 36, 67%
tile = 55
mean = 36, 10%
2.2 tile = 18,
90% tile = 58
Ave. Model Type
Count & Lag
per Day Structure
12 Autoregressive
linear regression
analysis, 0-d lag
best
2.2 Autoregressive
Poisson model,
1-d lag best
Other
pollutants
measured
Sulfur dioxide,
whiter 33% tile =
49 /ig/m3, 67% tile
= 77, summer
33% tile = 36,
67% tile = 55
Ozone, mean = 26
ppb, 10% tile =
11,90% tile = 41
Weather &
Other
Factors
min temp,
dummies for
day of week
and year
8 categories of
temp. & dew
pt., month,
year, lin. &
quad, time
trend
Result*
Pollutants (Confidence
hi model Interval)
none whiter: 1.15
(1.09, 1.21)
summer: 1.05
(0.98, 1.12)
SO2 whiter: 1.05
(1.01, 1.09)
summer: 1.01
(0.97, 1.05)
none 1.25
(1.10, 1.44)
ON
* Relative risk calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100 /tg/m3 increase in TSP.
\^
?
o
o
^
9
W
O
&
n
H
m
-------
TABLE 12-11. HOSPITAL ADMISSIONS STUDIES FOR PNEUMONIA
NJ
H
6
o
1
o
H
O
W
g
O
Study
Schwartz (1994f)
Elderly in Minneapolis,
1986-1989
Schwartz (1994e)
Elderly in Birmingham,
1986-1989
PM Type &
No. Sites
6 monitoring
stations
PMMean
& Range
mean = 36,
10%tile = 18,
measuring PM10 90% tile = 58
1 to 3
monitoring
stations
mean = 45,
10% tile = 19.
90% tile = 77
Ave. Model Type
Count & Lag Other pollutants
per Day Structure measured
6.0 Autoregressive Ozone: mean 26 ppb;
Poisson mod., 10% tile 11; 90%
1-d lag best tile 41
5.9 Autoregressive Ozone: mean 25 ppb;
Poisson mod., 10% tile 14; 90%
0-d lag best tile 37
measuring PM10
Schwartz (1994d)
Elderly in Detroit
1986-1989
2 to 11 PM10
mon. stations,
data for 82% of
possible days
mean = 48,
10% tile = 22,
90% tile = 82
15.7 Poisson auto- Ozone: mean 21 ppb;
regress, mod. 10% tile 7; 90%
using GEE, tile 36
0-d lag best
Weather &
Other Pollutants
Factors in model
8 categories of none
temp. & dew pt.,
month, year, lin.
& quad, time
trend
7 cat. of temp. & none
dew pt., month,
year, lin. &
quad, time trend
Dummy variables ozone
for temp, month,
lin. & quad, time
trend
Result*
(Confidence
Interval)
1.08
(1.01,1.15)
1.09
(1.03, 1.15)
1.06
(1.02, 1.10)
TABLE 12-12. HOSPITAL ADMISSIONS STUDIES FOR HEART DISEASE
Study
Schwartz and Morris
(in press)
Elderly in Detroit
1986-1989
Ischemic Heart Disease
Burnett et al. (in press)
All ages in Ontario,
Canada, 1983-1988
Cardiac disease
admission
PM Type &
No. Sites
2 to 11 PM10
monitoring
stations, data
available for
82% of possible
days
22 sulfate
monitoring
stations
PM Mean
& Range
mean = 48,
10% tile = 22,
90% tile = 82
station means
ranged from 3.0
to 7.7 in the
summer and 2.0
and 4.7 in the
winter
Ave. Model Type
Count & Lag Other pollutants
per Day Structure measured
44.1 Poisson auto- SO2, mean = 25
regressive ppb, 10% tile =
Weather &
Other Pollutants
Factors in model
Dummy vars. none
for temp, month,
Result*
(Confidence
Interval)
1.06
(1.02, 1.10)
model using 11, 90% tile = 44 lin. & quad.
GEE, 0-d lag CO, mean 2.4
best ppm, 10% tile 1.
90% tile = 3.8
time trend ozone,
2, CO, SO2
14.4 Linear Ozone averaged 36 Temperature none
regression on a ppb
19 day linear
filter, 1-d lag
best
included in
separate analyses
by summer and ozone
winter
1.06
(1.02, 1.10)
1.04
(1.03, 1.06)
1.04
(1.03, 1.05)
2 * Relative risk calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100 pig/m3 increase in TSP.
w
-------
1 12.3.2.2 Respiratory Illness Studies
2 Respiratory illness and the factors determining its occurrence and severity are important
3 public health concerns. This section discusses epidemiologic findings relating estimates of
4 PM exposure to respiratory illness. This effect is of public health importance because of the
5 widespread potential for exposure to PM and because the occurrence of respiratory illness is
6 common (Samet et al., 1983; Samet and Utell, 1990). Of added importance is the fact that
7 recurrent childhood respiratory illness may be a risk factor for later susceptibility to lung
8 damage (Glezen, 1989; Samet et al., 1983; Gold et al., 1989).
9 The PM studies generally used several different standard respiratory questionnaires that
10 evaluated respiratory health by asking questions about each child's and adult's respiratory
11 disease and symptom experience daily, weekly or over a longer recall period. The reported
12 symptoms and diseases characterize respiratory morbidity in the cohorts studied. A brief
13 discussion of aspects of epidemiology of respiratory morbidity provides a background for
14 studies examining PM exposure in relation to respiratory health. Respiratory morbidity
15 typically includes specific diseases such as asthma and bronchitis, and broader syndromes
16 such as upper and lower respiratory illnesses.
17 Asthma is characterized by reversible airway obstruction, airway inflammation, and
18 increased airway responsiveness to stimuli (National Institutes of Health, 1991). The
19 prevalence of physicians diagnosed asthma among children under age 18 is 6.3/100 (U.S.
20 Department of Health and Human Services, 1994). From 1982 through 1992, the prevalence
21 of asthma among persons aged 5 to 34 years (for whom the diagnosis is likely most accurate)
22 increased 42%, from 3.4 (401 deaths) to 4.9 (569 deaths) (United States Center for Disease
23 Control, 1995). Overall, an estimated 4.9% of the total U.S. population or over 12 million
24 people, have asthma (U.S. Department of Health and Human Services, 1994). Asthma
25 patients develop clinical symptoms such as wheezing and dyspnea after exposure to allergens,
26 environmental irritants, viral infections, cold air, or exercise. Exacerbations of asthma are
27 acute or subacute episodes of progressively worsening shortness of breath, cough, wheezing,
28 chest tightness, or some combination of these symptoms associated with decreased levels of
29 various measures of forced expiratory volume. Although viral respiratory tract infections are
30 common asthma triggers, especially in young children (National Institutes of Health, 1991),
31 symptoms such as wheezing may occur without an infectious cause.
April 1995 12_98 DRAFT-DO NOT QUOTE OR CITE
-------
1 Chronic bronchitis in adults is defined as a clinical disorder characterized by excessive
2 mucous secretion in the bronchial tube with an associated chronic productive cough on most
3 days for a minimum of 3 months of the year for not less than 2 successive years (American
4 Thoracic Society, 1962). The diagnosis can only be made after excluding other disorders
5 with similar symptoms. In contrast, Morgan and Taussig (1984) state that a clear definition
6 and etiology of chronic bronchitis in childhood have not yet been described. They
7 characterize chronic bronchitis in children as a symptom complex consisting of a chronic or
8 recurrent "wet" cough, increased phlegm production, and wheezing that may be associated
9 with evidence of airway inflammation. Symptoms and findings observed in children with
10 physician-diagnosed chronic bronchitis commonly include recurrent respiratory infections and
11 wheezing, with chronic phlegm production and chronic cough being less prevalent (Burrows
12 and Lebowitz, 1975). Respiratory syncytial virus (RSV) and parainfluenza virus are isolated
13 in cases of bronchitis (Chanock and Parrott, 1965), but symptoms of bronchitis may occur
14 without an infectious cause.
15 Viral respiratory illnesses can be subdivided by predominant anatomic site of
16 involvement in the respiratory tract: rhinitis (the common cold), pharyngitis, laryngitis,
17 laryngotracheo bronchitis (croup), tracheobronchitis, bronchiolitis, and pneumonia. In many
18 instances, signs and symptoms referable to more than one site (e.g., pharyngitis, laryngitis,
19 and rhinitis) may occur at the same time in the same patient.
20 The common cold is the leading acute upper respiratory illness in the United States.
21 Rhinoviruses lead the list as the most common group of viruses that cause colds in adults and
22 children. Other common viruses include coronaviruses, parainfluenza virus, respiratory
23 syncytial virus, and influenza virus. The number of colds acquired per year decreases with
24 age. Infants and preschool children have the highest incidence (4 to 8 colds per year), and
25 adults generally have two to five colds per year. Rhinovirus accounts for about 30% of
26 colds with a defined etiologic cause and has a characteristic seasonal pattern with the highest
27 prevalence of rhinovirus colds in early fall and a second smaller peak in the spring. Many
28 factors, including age of the patient, season of year, geographic location, and type of
29 population are important determinants of etiology. In most adults and older children signs
30 and symptoms of the common cold are marked by a dry, scratchy or sore throat. Usual
31 progression leads to a watery nasal discharge associated with the feeling of irritated nasal
April 1995 12-99 DRAFT-DO NOT QUOTE OR CITE
-------
1 mucosa and sneezing. Typically, symptoms and responses on respiratory questionnaire for
2 upper respiratory illness include throat irritation, acute cough, cough with phlegm, wheeze,
3 runny nose, breathing difficulty, fever, and earache.
4 Lower respiratory illnesses are generally classified into one of four clinical syndromes:
5 croup (laryngotracheobronchitis), tracheobronchitis, bronchiolitis, and pneumonia
6 (Glezen and Denny, 1973; Wright et al., 1989; McConnochie et al., 1988). In a study in
7 Tucson, the most common diagnosis during the first year of life was bronchiolitis, which
8 accounts for 60% of all lower respiratory illness (Wright et al., 1989). The most common
9 signs and symptoms associated with lower respiratory illnesses were wet cough (85%),
10 wheeze (77%), tachypnea (48%), fever (54%), and croupy cough (38%) as reported by
11 Wright et al. (1989). A few infectious agents are presumed to cause the majority of
12 childhood lower respiratory illness. Bacteria are not thought to be common causes of lower
13 respiratory illness in nonhospitalized infants in the United States (Wright et al., 1989).
14 Seventy-five percent of the isolated microbes were one of four types: RSV, parainfluenza
15 virus types 1 and 3, and Mycoplasma pneumoniae (Glezen and Denny, 1973; McConnochie
16 et al., 1988). Respiratory syncytial virus is particularly likely to cause lower respiratory
17 illness during the first two years of life. More than half of all illnesses diagnosed as
18 bronchiolitis, for which an agent was identified, were positive for RSV (Wright et al., 1989).
19 Wright et al. (1989) noted that studies that rely on parental reports of symptoms may
20 underestimate illness. Asking parents about illnesses at the end of the first year of life
21 revealed that one-third of them failed to report illnesses diagnosed by pediatricians.
22 Various studies of lower respiratory illness have reported rates based on visits to
23 physicians ranging from about 20 to 30 illnesses/100 children in the first year of life (Glezen
24 and Denny, 1973; Wright et al., 1989; Denny and Clyde, 1986; McConnochie et al., 1988).
25 Glezen and Denny (1973) reported that the rate for lower respiratory illnesses ranged from
26 24/100 person-years in infants under one year of age and decreased steadily each year
27 through the preschool years, tending to level off in school children (age 12 to 14 years) to
28 about 7.5 illnesses/100 person-years. Several factors affect the rate of lower respiratory
29 illness in children, including age, immunologic status, prior viral infections, siblings of early
30 school age, level of health, SES (Chanock et al., 1989), day care attendance, home dampness
31 and humidity, environmental tobacco smoke, NO2, PM, and other pollutants. Rates also
April 1995 12-100 DRAFT-DO NOT QUOTE OR CITE
-------
1 depend on method of illness ascertainment. Studies in the United States (Wright et al., 1989;
2 Denny and Clyde, 1986; McConnochie et al., 1988) indicated that the overall pattern and
3 incidence of lower respiratory illness is consistent in different geographic regions during the
4 two decades covered by the studies, suggesting that diagnosis and infectious agents have
5 changed little in that time period. Lower respiratory illness remains one of the major causes
6 of childhood morbidity in the United States (McConnochie et al., 1988).
7 Over the past 4 decades, a large body of epidemiologic evidence has accumulated that
8 indicates that respiratory illness events in childhood (mostly viral) are important determinants
9 (risk factors) for the future risk of chronic respiratory symptoms and disease in adult life
10 (Samet et al., 1983; Denny and Clyde, 1986; Britten et al., 1987; Glezen, 1989; Gold et al.,
11 1989). Based on such data, it seems likely that any factor such as PM that could be
12 responsible for increasing the risk of childhood respiratory illness and symptoms would be of
13 considerable public health importance not only with regard to immediate morbidity, but also
14 in relation to its contribution to long-term morbidity from chronic respiratory disease.
15
16 Studies of Respiratory Illness in Children
17 Schwartz et al. (1994) analyzed respiratory symptoms in children from the Harvard Six
18 Cities Studies. The cities included Watertown, MA; St. Louis, MO; Portage, WI; Kingston-
19 Harriman, TN; Steubinville, OH; and Topeka, KS. Daily diaries of respiratory symptoms
20 were collected from the parents of 1844 school children for one year starting in September,
21 1984. A centrally located residential monitor measured SO2, NO2, and O3 on a continuous
22 basis, PM2 5 and PM10 were collected by a dichotomous sampler and aerosol acidity was
23 measured daily. A multiple logistic regression model was used to analyze the data, adjusting
24 for serial correlation by autoregressive terms estimated using the generalized estimating
25 equations of Liang and Zeger (1986). The only weather variable included in the model was
26 temperature, using both linear and quadratic terms.
27 In order to avoid the seasonal component of respiratory illness, the analysis was
28 restricted to the months of April through August. During this period the PM2 5 values had a
29 median value of 18 /xg/m3 with 10th and 90th percentile values of 7.2 and 37.0 ^ig/m3. The
30 PM10 values had a median value of 30 /zg/m3 with 10th and 90th percentile values of 13 and
31 53 fj,g/m3. Sulfate fractions were estimated from the PM2 5 filters. Cough was significantly
April 1995 12-101 DRAFT-DO NOT QUOTE OR CITE
-------
1 related to all pollutants except acidity. The strongest relationships for cough were found
2 with PM10 and O3, and these effects appeared to be independent of each other. An increase
3 of 30 fig/m3 in PM10 was associated with an odds ratio for cough of 1.28 (1.07 to 1.54).
4 Fitting a Generalized Additive Model (non-parametric) showed that cough incidence
5 increased monotonically with PM10 concentration, and there was no evidence of non-
6 linearity. Lower respiratory symptoms (LRS) were also related to all pollutants except
7 acidity. Strongest relationships were found with PM10 and sulfate fraction, and these effects
8 appeared to be independent of each other. An increase of 30 ^ig/rn3 in PM10 was associated
9 with an odds ratio for lower respiratory symptoms of 1.53 (1.20 to 1.95). There was no
10 evidence of non-linearity, as shown in Figure 12-4. Comparable analyses for SO2 and H+
11 are shown in Figures 12-5 and 12-6. Note that these curves show an inconsistent relationship
12 at lower exposure estimates. Although these non-parametric models do not provide
13 confidence intervals, it is clear that the relationship between cough and PM10 is stronger
14 than for either SO2 or H+.
15 Pope et al. (1991) studied respiratory symptoms in asthmatic school children in the
16 Utah Valley. Participants were selected from samples of 4th and 5th grade elementary
17 students in 3 schools in the immediate vicinity of PM10 monitors in Orem and Lindon, Utah
18 and were restricted to those who responded positively to one of: (a) ever wheezed without a
19 cold; (b) wheezed for 3 days out of a week for a month or longer; (c) had a doctor say the
20 "child has asthma". This resulted in 34 subjects who were included in the final analyses.
21 PM10 monitors operated by the Utah State Department of Health collected 24 hr PM10
22 samples from midnight to midnight (range 11 to 195 jug/m3). There was limited monitoring
23 of SO2, NO2, and O3. Lower respiratory disease was defined as the presence of at least one
24 of: trouble breathing, dry cough, or wheezing. A fixed effects logistic regression analysis
25 was calculated using each person as his own control and low temperature as a covariate.
26 Estimated odds ratios for upper respiratory disease per PM10 increase of 50 /ig/m3 was 1.20
27 (1.03, 1.39); for lower respiratory illness, it was 1.28 (1.06, 1.56).
28 Pope et al. (1991) also studied asthmatics aged 8 to 72 in the Utah Valley, selected
29 from those referred by local physicians. This resulted in 21 subjects who were included in
30 the final analysis. PM10 monitors operated by the Utah State Department of Health collected
31 24 hr samples from midnight to midnight; and values ranged from 11 to 195 /ig/m3.
April 1995 12-102 DRAFT-DO NOT QUOTE OR CITE
-------
1.8-
eo
5
•5
M
T3
T3
O
-------
1.6-
W 1.4
5
•5
1.2-
1.0-
10 20 30
S02 (ppb)
40
Figure 12-5. Relative odds of incidence of LRS smoothed against 24-h mean SO2
concentration on the previous day, controlling for temperature, city,
and day of the week.
Source: Schwartz et al. (1994)
2.0-
co
5
•5
«
S
1.0-
50 100 150 200
Hydrogen Ion (nm/m3)
250 300
Figure 12-6. Relative odds of incidence of LRS smoothed against 24-h mean H+
concentration on the previous day, controlling for temperature, city,
and day of the week.
Source: Schwartz et al. (1994)
April 1995
12-104 DRAFT-DO NOT QUOTE OR CITE
-------
1 In a follow up study, Pope and Dockery (1992) enrolled non-asthmatic symptomatic and
2 asymptomatic children in the Utah Valley, selected from samples of 4th and 5th grade
3 elementary students in three schools in the immediate vicinity of PM10 monitors in Orem and
4 Lindon Utah. A questionnaire identified 129 children who were mildly symptomatic and 60
5 were selected. An additional 60 with no symptoms were recruited. PM10 monitors operated
6 by the Utah State Department of Health collected 24 hr samples from midnight to midnight;
7 PM10 values ranged from 11 to 195 /ig/m3. Limited SO2, NO2, and O3 monitoring was
8 conducted. Low temperature was used to adjust for weather, but no adjustment was made
9 for humidity. Upper respiratory symptoms had a logistic regression coefficient of .00519
10 (.00203) and lower respiratory symptoms had a coefficient of .00658 (.00205) in the
11 symptomatic sample using a 5-day moving average of PM10. These correspond to odds
12 ratios of 1.30 and 1.39 respectively for an increase of 50 jig/m3 in PM10. No consistent
13 effects were seen in the asymptomatic sample, although all effects tended to increase with
14 PM. Only minimum temperature was used to adjust for weather.
15 Ostro et al. (1995b) studied a panel of 83 African-American asthmatic children aged 7
16 to 12 recruited from four allergy and pediatric clinics in central Los Angeles and two asthma
17 camps in the summer of 1992. The analysis focused on the daily reporting of respiratory
18 symptoms including shortness of breath, cough, and wheeze. Daily air monitoring at three
19 fixed sites included O3, PM10, NO2, and SO2. PM10 levels ranged from 20 to 101 /zg/m3
20 and O3 from 10 to 160 ppb. Daily temperature, humidity, rainfall, pollens and molds were
21 also used as covariates. A logistic regression allowing for repeated measures with variances
22 estimated by generalized estimating equations was used to estimate effects of the pollutants
23 and covariates. Both PM10 and O3 were associated with increased shortness of breath, and
24 the authors could not separate the effect of the two pollutants. The odds ratio for an increase
25 of 56 jug/m3 PM10 was 1.58 (1.05, 2.38). No effects were seen with cough or wheeze.
26 Schwartz et al. (1991) analyzed acute respiratory illness in children in five German
27 communities. Children's hospitals, pediatric departments and pediatricians were asked to fill
28 out a short questionnaire for each visit for croup or obstructive bronchitis over a 2-year
29 period. A diagnosis of croup was defined as acute stenotic subglottic laryngotracheitis. Not
30 all doctors reported for the full 2 years—a loss of about 50%. Thus, participation was about
31 50%. Areas chosen to represent a wide range of air pollution exposure included: Duisburg
April 1995 12-105 DRAFT-DO NOT QUOTE OR CITE
-------
1 and Kohn in the highly industrialized areas of Northrhine-Westfalia and Stuttgart, and
2 Tubingen/Reutlingen and Freundenstadt in South Germany. Two to four monitors were
3 located in each study areas and 24 hr measurements were taken of TSP, SO2, and NO2. TSP
4 was measured by low volume sampler, NO2 by chemiluminescence, and SO2 by the
5 conductometric method, and were expressed in /xg/m3. Poisson regression analysis as
6 described by McCullagh and Nelder (1983) was used to estimate the effect of pollution on
7 croup and obstructive bronchitis, with adjustments for serial correlation using the method of
8 Zeger and Liang (1986). The model included terms for season (annual and biannual sine and
9 cosine terms), weather (temperature and relative humidity), and drop-outs. Logistic
10 regression coefficients estimated from the Poisson regressions gave values of 0.1244
11 admissions/log(TSP) (.0309), 0.4161 (.156) for NO2, and 0.0831 (.0352) for SO2. The log
12 TSP coefficient was not significant with either NO2 or SO2 were included in the model.
13 Hoek and Brunekreef (1993) and Hoek (1992) studied a general population sample of
14 112 children aged 7 to 12 who lived in a non-urban area proximal to Wageningen, ND.
15 Acute respiratory symptoms of the children were recorded in a diary by their parents,
16 including throat irritation, cough, cough with phlegm, wheeze, runny nose, and a variety of
17 other symptoms. PM10 was measured daily (3PM to 3PM) with an instrument similar to the
18 Sierra Anderson 241 dichotomous sampler. SO2 was measured using fluorescence, and NO2
19 was measured using chemiluminescence. Logistic regression analyses including first order
20 autoregressive terms were used to analyze the data and included ambient temperature and day
21 of study as covariates. The PM10 coefficient for any upper respiratory illness was 0.0026
22 (0.0013). This corresponds to an odds ratio of 1.14 (95% confidence interval of (1.00 to
23 1.30) for an increase of 50 /ig/m3 PM10. Most other coefficients were not significant.
24 Braun-Fahrlander et al. (1992) studied daily respiratory disease symptoms in preschool
25 children in 4 areas of Switzerland. A sample of 840 children was chosen from Basel and
26 Zurich. One-twelfth of the sample was recruited each month from November 1985 to
27 November 1986. A physician conducted a standardized questionnaire with the parents.
28 Parents recorded daily symptoms including cough without runny nose, breathing difficulty,
29 and fever with earache and sore throat. TSP was measured daily (method not given) and
30 NO2 by Palmes tubes both outside the apartment and inside the room where the child stayed
31 most frequently. Children lived within 6 km of an outdoor monitor which measured TSP,
April 1995 12-106 DRAFT-DO NOT QUOTE OR CITE
-------
1 NO2, SO2, and O3. Multiple logistic regression analysis was used to explain differences in
2 upper respiratory symptom incidence. Analysis terms included temperature, season, city,
3 and a risk strata based on a cross-sectional analysis. Variances were adjusted using the
4 method of Liang and Zeger (1986). The TSP coefficient for upper respiratory symptoms was
5 0.00454 (0.00174), corresponding to an odds ratio of 1.57 per TSP increase of 100 /ig/rn3.
6 Neither NO2, SO2, or O3 were significant.
7 Hoek and Brunekreef (1994) studied respiratory function and pulmonary symptoms in
8 more than 1000 children in 4 towns in the Netherlands. Children aged 7 to 11 in Deurne,
9 Enkhuizen, Venlo, and Nijmegen were studied during one of three winters (1987/88,
10 1988/89, 1989/90). During the study, respiratory symptoms data were collected daily by
11 diary. PM10 was measured daily (3PM to 3PM) with an instrument similar to the Sierra
12 Anderson 241 dichotomous sampler, SO2 by fluorescence, and NO2 by chemiluminescence.
13 Separate logistic regressions were performed for 9 locations (six groups of subjects of
14 Deurne and one in each other town) using a first order autoregressive model. The
15 coefficients were combined using the inverse variance weighting method. The odds ratio for
16 the incidence of cough associated with 100 /zg/m3 PM10 increase was 1.10 (0.67,1.79).
17 PM10 odds ratios for upper and lower respiratory illness were also not statistically
18 significant. Nor was the incidence of acute respiratory symptoms significantly related to
19 PM10, SO2, NO2, or sulfate.
20 In a study by Roemer et al. (1993) of children with chronic respiratory symptoms,
21 parents of children in grades 3 to 8 in two small nonindustrial towns in the Netherlands were
22 given questionnaires about respiratory symptoms. Seventy-four of the 313 children with
23 positive responses (cough or shortness of breath) were included in the study. PM10 was
24 measured daily using the Sierra Anderson 24 dichotomous sampler. SO2, NO2, and black
25 smoke were also measured. Several symptoms including asthma attack, wheeze, and cough
26 were marginally associated with PM10. The logistic regression coefficient for wheeze was
27 .00224 (.00115) per unit increase in the same day's PM10 level. The coefficient for
28 broncho-dilator use was .00210 (.00085). SO2 and black smoke were also marginally related
29 to several of the symptoms. Seasonal variation did not appear to be considered in the model.
30 Hoek and Brunekreef (1995) studied respiratory symptoms in 300 children aged 7 to
31 11 years in Duerne and Enkhuizen, The Netherlands. The study was designed as an ozone
April 1995 12-107 DRAFT-DO NOT QUOTE OR CITE
-------
1 study, but SO2, NO2, and PM10 were also measured (PM10 ranged 13 to 124 /ig/m3; O3
2 ranged 22 to 107 ppb). A symptom diary similar to that used in the Harvard Six Cities
3 Study was used to obtain daily information on cough, phlegm, wheeze, runny nose, and other
4 respiratory symptoms. A multiple logistic model with first order autoregressive residuals
5 was used. Additional analyses using ARIMA models to allow for autocorrelation confirmed
6 results of the logistic analyses. Nearly all logistic regression coefficients were non-
7 significant and negative. The analyses of cough in Deurne gave an estimated odds ratio of
8 0.93 for a 50 /xg/m3 increase in PM10 on the same day. Analyses of other endpoints, lag
9 times, and pollutants gave similar results.
10 Relationships between air pollution indices for 84 standard metropolitan statistical areas
11 (SMSA's) mostly of 100,000 to 600,000 people in size and indices of acute morbidity effects
12 were studied by Ostro (1983), Hausman et al. (1984), and Ostro (1987), using data derived
13 from the National Center for Health Statistics (NCHS) Health Interview Survey (HIS) of
14 50,000 households comprising about 120,000 people. Ostro (1983) used HIS data to assess
15 the prevalence of illness and illness-related restrictions in activity in the United States. Data
16 on either restricted activity days (RADs) or work loss days (WLDs) were aggregated over a
17 year, and correlated with annual TSP levels, controlling for temperature, wind, precipitation,
18 population density, and smoking. Using the 1976 survey, a significant relationship between
19 TSP and both outcomes was found, with RAD's being more significant. Sulfate fractions
20 were not significantly related to either outcome. Ozone was not measured. The explained
21 variation was much higher for RADs than for WLDs. The average of air pollution monitors
22 for each city was used, rather than aerometric data aggregated for smaller geographic units in
23 relationship to individuals residing nearby for whom HIS data were included in the analysis.
24 Hausman et al. (1984) analyzed the same data, but used Poisson regression analysis using a
25 fixed effects model that compared deviations from the city mean levels of illness and
26 short-term pollution as the exposure variable. Significant associations between 2-week
27 average TSP levels and RADs or WLDs were found. The magnitude of the within city
28 effects was similar to the magnitude of the between city effects seen earlier. Demographic
29 factors were controlled for on an individual basis, along with climatic conditions.
30 Ostro (1987) applied the Hausman et al. (1984) techniques to analyze HIS results from
31 1976 to 1981 in relation to estimates of fine particle (FP) mass. That is, for adults aged
April 1995 12-108 DRAFT-DO NOT QUOTE OR CITE
-------
1 18 to 65, days of work loss (WLDs), restricted activity days (RADs) and respiratory-related
2 restricted activity days (RRADs) measured for a 2-week period before the day of the survey
3 were used as measures of morbidity and analyzed in relation to estimated concurrent 2-week
4 averages of FP or lagged in relation to estimated 2-week FP averages from two to four
5 weeks earlier. The FP estimates were produced from the empirically derived regression
6 equations of Trijonis. These equations incorporated screened airport data and 2-week
7 average TSP readings at population-oriented monitors, using data taken from the metropolitan
8 area of residence. Various potentially confounding factors (such as age, race, education,
9 income, existence of a chronic health condition, and average 2-week minimum temperature)
10 were controlled for in the analyses. The morbidity measures (WLDs, RADs, RRADs), for
11 workers only or for all adults in general, were consistently found to be significantly
12 (p <0.01 or <0.05) related to lagged FP estimates (for air quality 2 to 4 weeks prior to the
13 health interview data period), when analyzed for each of the individual years from 1976 to
14 1981. However, less consistent associations were found between the health endpoints and
15 more concurrent FP estimates.
16 Ostro and Rothschild (1989) studied acute respiratory morbidity based on an analysis of
17 1976 to 1981 HIS data. Ozone measurements were taken from EPA's SAROAD monitoring
18 network, and FP measurements were estimated from airport visibility data. The endpoints of
19 the analysis included minor restrictions in activity and work loss. Using a multiple
20 regression analysis, both endpoints showed a relationship to FP.
21
22 School Absences Studies
23 Most school absences are due to acute conditions (Klerman, 1988). Respiratory
24 conditions are the most frequent cause, particularly influenza and the common childhood
25 infectious diseases. School absences are also caused by injuries, digestive system conditions
26 and ear infections. Kornguth (1990) notes the following characteristic of school absent
27 children: (1) as mothers level of education or family income increased the likelihood of their
28 children being absent decreased; and (2) days absent due to illness are related to source of
29 medical care and to type of health insurance coverage. Children with a wide range of
30 chronic illnesses miss more school than their healthy peers. There is only tentative evidence
31 that school absent rates of individual children vary directly with the severity of their health
April 1995 12-109 DRAFT-DO NOT QUOTE OR CITE
-------
1 problem (Weitzman, 1986). Parcel et al. (1979) found that children with asthma have a
2 significantly higher absentee rate than do nonasthmatic children. Children who smoke and
3 whose parents smoke are more likely to be absent from school for minor ailments (Charlton
4 and Blair, 1989). Whether this increased likelihood of absence is due to genuine health
5 problems or to a generally negative attitude to school in children who take up smoking to
6 boost their self-esteem, is unclear.
7 Most excessive school absence is probably the result of factors outside the health care
8 sphere (Klerman, 1988). Chaotic family environments, lack of achievement motivation,
9 understaffed and uninviting schools, and other societal problems, are undoubtedly the major
10 reason for absenteeism. Excessive school absence is a profound educational and social
11 problem in the United States (Weitzman et al., 1986). Despite the fact that the majority of
12 school absences are reported as being health related, data suggest that demographic and
13 educational characteristics of students have a much greater influence on absence behavior
14 than do health-related factors. Since school absence rates reflect both health and non-health
15 related factors, it is important that investigators recognize the nonspecific nature of the
16 measure and account for non-health related influences appropriately (Weitzman, 1986). Such
17 non-health related potential problems with the data include the following: data are difficult
18 to collect, individual data as compared to aggregate data; different coding in schools for
19 tardy or leaves school early for sickness; and, records may not be computerized at school
20 making retrospective studies more difficult (Weitzman, 1986).
21 Ransom and Pope (1992) studied elementary school absences in connection with the
22 steel strike in the Utah Valley. Data for school absences from 1985 to 1991 were obtained
23 from two sources: (1) district-wide attendance averages by grade level from the Provo
24 School District, and (2) daily absenteeism records from the Northridge Elementary School in
25 Orem. The Northridge School was much closer to the steel mill than were the schools in the
26 Provo School District. Daily PM10 measurements were made at three sites (Linden, Provo,
27 and Orem), but only the Linden site collected daily measurements for the entire time period
28 of the study. Some SO2 and O3 measurements were available, but these values tended to be
29 well below the National Ambient Air Quality standards. Meteorological information was
30 available from the Brigham Young University weather station. Regression analyses were
31 conducted, taking into account several covariates including month of study, snowfall,
April 1995 12-no DRAFT-DO NOT QUOTE OR CITE
-------
1 Christmas and Thanksgiving holidays, and low temperature. The best PM10 predictor was a
2 4-week moving average. A highly significant increase of about 2% in the absence rates
3 (absolute increase) for an increase of 100 /*g/m3 increase in the 4-week average PM10 was
4 found both sets of data, and the coefficient was similar even when a dummy variable was
5 added for the strike. No adjustments were made for periods of increased influenza cases.
6
7 Studies of Respiratory Illness in Adults
8 Lawther et al. (1970) reported on studies carried out from 1954 to 1968 mainly in
9 London, using a diary technique for self-assessment of day-to-day changes in symptoms
10 among bronchitic patients. A daily illness score was calculated from the diary data and
11 related to BS and SO2 levels and weather variables. Pollution data for most of the London
12 studies were mean values from the group of sites used in the mortality/morbidity studies of
13 Martin (1964). In early years of the studies, when pollution levels were generally high, well
14 defined peaks in illness score were seen when concentrations of either BS or SO2 exceeded
15 1,000 /ig/m3. With later reductions in pollution, the changes in condition became less
16 frequent and of smaller size. From the series of studies as a whole, up to 1968, it was
17 concluded that the minimum pollution levels associated with significant changes in the
18 condition of patients was a 24-h mean BS level of -250 ^g/m3 together with a 24-h mean
19 SO2 concentration of -500 jug/m3 (0.18 ppm). A later study reported by Waller (1971)
20 showed that, with much reduced average levels of pollution, there was an almost complete
21 disappearance of days with smoke levels exceeding 250 /ng/m3 and SO2 levels over
22 500 jig/m3 (0.18 ppm). As earlier, some correlation remained between changes in the
23 conditions of the patients and daily concentrations of smoke and SO2, but the changes were
24 small at these levels and it was difficult to discriminate between pollution effects and those of
25 adverse weather. The analysis of the Lawther et al. (1970) study was made prior to the
26 availability of current statistical methods such as poisson regression using generalized
27 estimating equations. The large differences seen by Lawther et al. (1970) at high levels
28 would undoubtedly remain significant regardless of the analysis technique. Marginally
29 significant results must be viewed with caution.
30 Dusseldorf et al. (1994) studied respiratory symptoms in 32 adults living near a large
31 steel plant in Wijk aan Zee, The Netherlands. During the study period PM10 levels ranged
April 1995 12-111 DRAFT-DO NOT QUOTE OR CITE
-------
1 from 36 to 137 |ug/m3. Diary information on acute respiratory symptoms, medication use,
2 and presence of fever was collected. Peak flow measurements were also taken. The study
3 was conducted from 11 October 1993 to 22 December 1993, and the average number of days
4 per subject was 66. A logistic regression model was used and to control autocorrelation, a
5 linear time series model was also fitted. Both models gave similar results and so the logistic
6 regression coefficients converted to odds ratios for 100 ^g/m3 were reported. These were
7 converted to odds ratios for 50 /ig/m3. The odds ratio for cough on PM10 (lag zero) was
8 1.31 (0.9, 1.76). The other endpoints of phlegm, shortness of breath and wheeze showed
9 lesser effects. Using PM10 lagged one, two, and three days showed little effect.
10 Lebowitz et al. (1982) studied 117 families in Tucson, AZ selected from a stratified
11 sample of families in geographical clusters from a representative community population
12 included in an ongoing epidemiologic study. Both asthmatic and non-asthmatic families were
13 evaluated over a 2-year period using daily diaries. The health data obtained were related to
14 various indices of environmental factors derived from simultaneous micro-indoor and outdoor
15 monitoring in a representative sample of houses for air pollutants, pollen, fungi, algae and
16 climate. Monitoring of air pollutants and pollen was carried out simultaneously. Two-month
17 averages of indoor TSP ranged from 2.1 to 169.6 /ig/m3. Cyclone measurements of
18 respirable particulate (RSP) ranged from below minimum detectable limits up to 28.8 jig/m3;
19 CO and NO2 measurements were also taken, but no S02 monitoring was reported. TSP and
20 pollen were reported to be related to symptoms in both asthmatics and non-asthmatics, but
21 the authors reported that the statistical analyses used were all qualitative (because of low
22 sample size) and statistical significance was not computed.
23 Whittemore and Korn (1980) studied asthmatics in seven communities in the Los
24 Angeles area. Panelists were located by consulting local physicians and were followed for
25 34 weeks from May 7 to December 30 in the years 1972 to 1974 from the communities of
26 Santa Monica, Anaheim, Glendora, Thousand Oaks, Garden Grove, and Covina. Dairies
27 were filled out weekly by the participants who gave daily information on symptoms.
28 Monitoring stations were placed in each community near an elementary school. TSP, RSP,
29 suspended sulfates, suspended nitrates, SO2, and photochemical oxidants were measured.
30 NO2 was also measured but the data were determined to be unreliable. Because of the
31 colinearities and measurement errors, only TSP and photochemical oxidants were actually
April 1995 12-112 DRAFT-DO NOT QUOTE OR CITE
-------
1 included in the analyses. A logistic model was used for each individual that included the
2 presence of an attack on the previous day, meteorology, day of study, day of week, and
3 pollutants. Regression coefficients were combined using both a fixed and random effects
4 model. Both photochemical oxidants and TSP were found to be significantly related to
5 symptoms, even when the other pollutant was included in the model. The coefficient for
6 TSP for both models was .00079 (standard error not given). This corresponds to an odds
7 ratio of 1.08 for a 100 /ig/m3 increase in TSP.
8 Ostro et al. (1991) studied adult asthmatics recruited from clinic patients in Denver.
9 Diagnosis of asthma was based on physical exam confirmed by lung function tests. The
10 panel of 207 recorded daily symptoms and medication use from November 1987 to February
11 1988. Ambient air pollutants measured were sulfates, nitrates, PM2 5, nitric acid, H+, and
12 SO2 at a downtown Denver monitor two miles from the clinic. Logistic regression analysis
13 was used with adjustment for autocorrelation by creating an instrumental variable; the final
14 regression used Proc Autoreg in SAS. The coefficient for log(PM2 5) was .0006 (.0053) for
15 asthma and .0012 (.0043) for cough. H+ was the only pollutant near statistical significance,
16 having an estimate coefficient of .0031 (.0042) for asthma and .0076 (.0038) for cough. The
17 coefficients cannot be compared directly with other studies because of the log transformation,
18 and attempts to convert them based on mean values give unreasonable answers.
19 Ostro et al. (1993) studied respiratory symptoms in non-smoking adults aged 18 or
20 more in Southern California from September 1978 to March 1979. The analysis was
21 restricted to those 321 subjects who completed diaries for the entire 181-day period. The
22 health endpoints included upper respiratory illness, lower respiratory illness, and eye
23 irritation. Air pollution data for the Glendora, Covina, and Azusa areas were obtained from
24 the Los Angeles County Air Pollution Control District Station in Azusa and included O3,
25 NO2, SO2, and sulfate fraction of PM. Temperature, rain, and humidity were used as
26 meteorological covariates. A multiple logistic regression analysis was run using the three
27 health endpoints. Ozone, sulfate fraction, and gas stove use were associated with significant
28 odds ratios for lower respiratory tract illness. The odds ratio for gas stove use, 1.23, was
29 well within the range reported in a meta-analysis of studies of nitrogen oxides by Hasselblad
30 et al. (1992), but COH was not significantly related to lower respiratory illness. Only ozone
April 1995 12-113 DRAFT-DO NOT QUOTE OR CITE
-------
1 was related to upper respiratory illness or eye irritation. The author did not report that
2 adjustments were made for serial correlation of the health outcomes.
3
4 Acute Respiratory Illness Studies Summary
5 This category include several different endpoints, but the majority of authors reported
6 results on at least two of: (1) upper respiratory illness, (2) lower respiratory illness, or (3)
7 cough (See Table 12-13). The results for upper respiratory illness are very inconsistent: two
8 studies estimate a relative risk near 1.00 whereas four others obtain estimates between 1.14
9 and 1.55. These relative risks are all estimated for an increase of 50 /xg/m3 in PM10 or its
10 equivalent The relative risks for lower respiratory illness are spread between 1.01 and 2.03,
11 but all are positive. The relative risks for cough include two below 1.0 and go as high as
12 1.51. All of these are suggestive of an effect. Whereas the hospital admission studies were
13 all done in a similar manner and resulted in very similar results, these studies are done with
14 very different designs and give very inconsistent results.
15
1 12.3.2.3 Pulmonary Function Studies
2 Pulmonary function studies are part of any comprehensive investigation of possible
3 effects of any air pollutant. Measurements can be made in the field, they are noninvasive,
4 and their reproductibility has been well documented. Guidelines for reference values and
5 interpretative strategies of lung function tests have been prepared (American Thoracic
6 Society, 1991). Various factors are important determinants of lung function measures.
7 Lung function in childhood is primarily related to general stature (as measured by
8 height) and as children enter late childhood, by age. The growth patterns differ between
9 males and females. Compared to girls, boys show larger size-adjusted (usually height or
10 height2) average values for various measures of lung function (Wang et al., 1993a,b).
11 Moreover, growth of measures derived from forced expiratory maneuvers (e.g., forced vital
12 capacity-FVC and forced expiratory volume one-second-FEVj) continues for a longer period
13 of time in males, beyond the time when height growth is complete (Wang et al., 1993a,b).
14 Lung function begins to decline with age in the 3rd to 4th decades (Tager et al., 1988) and
15 continues to do so monotonically as people age. Cigarette smoking, the presence of chronic
April 1995 12-114 DRAFT-DO NOT QUOTE OR CITE
-------
*
>•*.
1— »
S Study
Schwartz et al. (1994)
300 elementary school
children in Six-Cities in
U.S., 1984-1988
TABLE 12-13
PM Type &
No. Sites
PM10
monitoring in
each city
PM Mean
& Range
median 30
jtg/m3 10th
percentile =
13, 90th
percentile =
53
. ACUTE RESPIRATORY
Ave.
Rate
per Day
3.1
Model Type
& Lag Structure
Autoregressive
logistic
regression using
GEE
DISEASE STUDIES
Other
pollutants
measured
Ozone, NO2,
SO2
Weather &
Other
Factors
Temperature
Other
pollutants
in model
none
SO2
ozone
Result*
(Confidence
Interval)
1.51 (1.12, 2
1.39(0.98, 1
1.49(1.10, 2
.05)
.96)
.01)
O
o
o
H
O
w
o
o
HH
H
ffl
Pope et al. (1991),
students in the Utah
Valley, winter 1989-1990
PM1?
monitoring
stations at 3
sites
Pope et al. (1991), PM10
asthmatic children in the monitoring
Utah Valley, winter 1989- stations at 3
1990 sites
Pope and Dockery (1992), PMi0
symptomatic children in monitoring
the Utah Valley, winter stations at 2
1990-1991 sites
mean = 46
jtg/m3,
range = 11 to
195
(not given)
mean = 46
/ig/m3,
range = 11 to
195
(not given)
mean = 76 (not given)
range = 7 to
251
Fixed effects Limited Variables for none
logistic monitoring of temperature
regression NO2, SO2, and and time trend
ozone. Values
were well
below the
standard
Fixed effects Limited Variables for none
logistic monitoring of low
regression NO2, SO2, and temperature
ozone. Values and time trend
were well
below the
standard
Autoregressive none
logistic
regression using
GEE
Variable for
low
temperature
none
Relative risk calculated from parameters given by author assuming a 50 /xg/m3 increase in PM10 on 100 /xg/m3 increase in TSP.
Upper resp.
1.20(1.03, 1.39)
Lower resp.
1.28(1.06, 1.56)
Upper resp.
0.99(0.81, 1.22)
Lower resp.
1.01 (0.81, 1.27)
Upper resp.
1.20(1.03, 1.39)
Lower resp.
1.27(1.08, 1.49)
Cough
1.29(1.12, 1.48)
-------
TABLE 12-13 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
a
VO
VO
t— »
to
1
1— '
ON
G
>
H
G
0
^
O
H
x^
s
W
Study
Pope and Dockery (1992),
asymptomatic children in
the Utah Valley, whiter
1990-1991
Hoek and Brunekreef
(1993), respiratory disease
in school children aged 7
to 12 hi Wageningen,
Netherlands, whiter
1990-1991
Schwartz et al. (1991)
Study of acute respiratory
illness hi children in
5 German communities,
1983-1985
PM Type &
No. Sites
PM10
monitoring
stations at
2 sites
Two to 4
monitoring
stations
measured
PM10
Two to 4
monitoring
stations in
each area
measured
TSP
* Relative risk calculated from parameter
Ave.
PM Mean Rate
& Range per Day
mean = (not given)
76 ng/m ,
range = 7 to
251
max. = (not given)
110/ig/m3
medians ranged 0.5 to 2.9
from 17 to
56 /*g/m3,
10% tiles from
5 to 34, 90%
tiles from 41 to
118
s given by author assuming a 50
Model Type
&Lag
Structure
Autoregressive
logistic
regression
using GEE
Autoregressive
logistic
regression
using GEE
Autoregressive
Poisson
regression
using GEE
/ig/m3 increase in
Other
pollutants
measured
none
Max SO2 =
105 jtg/m3,
max NO2 =
127 /ig/m3
median SO2
levels ranged
from 9 to
48 /*g/m3,
median NO2
levels ranged
from 14 to
5 /ig/m3
PM10 on 100 n
Weather & Other
Other pollutants
Factors in model
Variable for none
low temperature
Variable for none
ambient
temperature and
day of study
Most none (TSP
significant was not
terms of day significant
of week, time when NO2
trend, and added to model)
weather
(terms not
listed)
g/m3 increase in TSP.
Result*
(Confidence
Interval)
Upper resp.
0.99
(0.78, 1.26)
Lower resp.
1.13
(0.91, 1.39)
Cough
1.18
(1.00, 1.40)
Upper resp.
1.14
(1.00, 1.29)
Lower resp.
1.06
(0.86, 1.32)
Cough
0.98
(0.86, 1.11)
1.26
(1.12, 1.42)
n
H
m
-------
TABLE 12-13 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
VO
IJJ
^1
O
£>
3
6
o
o
H
0
c!
3
W
Study
Schwartz et al. (1994)
Study of respiratory
symptoms in 6 U.S.
cities, 1984-1988
Braun-Fahrlander et al.
(1992)
Study of preschool
children in four areas of
Switzerland
Roemer et al. (1993)
Study of children with
chronic respiratory
symptoms in Wageningen,
The Netherlands
Dusseldorfetal. (1994)
Study of adults near a
steel mill in The
Netherlands
PM Type &
No. Sites
Daily
monitoring
of PM10,
PM2 5 at
each city
Daily
monitoring of
TSP
Daily
monitoring of
PM10
Daily
monitoring of
PM10, iron,
sodium,
silicon, and
manganese
Ave.
PM Mean Rate
& Range per Day
median PM,0 (not given)
= 30 /ig/m ,
10% tile = 13,
90% tile = 53
median PM2 5
= 18 /*g/m ,
10% tile = 7,
90% tile = 37
(not given) 4.4
6 days above .094
110/ig/m3 incidence
rate
mean PM10 (not given)
= 54 ^g/m3,
range =
4 to 137)
Model Type
&Lag
Structure
Autoregressive
logistic
regression
using GEE
Logistic
regression
Autoregressive
logistic
regression
Logistic
regression
Other
pollutants
measured
SO2, median
= 4 ppb, 10%
tile = 1, 90%
tile = 18
NO2, median
= 13 ppb,
10% tile 5,
90% tile = 24,
ozone
SO2, NO2, and
ozone levels
not given
SO2 and NO2
means not
given
Geometric
mean iron
= 501 ng/m3,
manganese
= 17 ng/m3,
silicon =
208 ng/m3
Weather &
Other
Factors
temperature,
day of week,
city or
residence
city, risk
strata, season,
temperature
(not given)
(not given)
Other
pollutants
in model
all two
pollutant
models were
fitted with
minimal
effect on PM
none
none
none
Result*
(Confidence
Interval)
Cough
(PM10 lag 1)
1.51
(1.12,2.05)
Upper resp.
(PM10 lag 2)
1.39
(0.97, 2.01)
Lower resp.
(PM10 lag 1)
2.03
(1.36,3.04)
Upper resp.
1.55
(1.10,2.24)
Cough
(not given,
probably less
than one)
Cough
1.14
(0.98, 1.33)
n
i—i
H
W
* Relative risk calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100 |*g/m3 increase in TSP.
-------
TABLE 12-13 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
2.
£
K>
00
Study
Ostro et al. (1991)
Study of adult asthmatics
in Denver, Colorado
November 1987 to
February 1988
Ostro et al. (1993)
Study of non-smoking
adults in Southern
California
PM Type &
No. Sites
Two monitors
provided daily
measurements of
PM2.5
Apparently one
site (Azusa).
PM measurements
included sulfate
fraction and
COHS
PMMean
& Range
22 /ig/m3,
range = 0.5
to 73
mean sulfate
= 8 /ig/m3,
range = 2 to
37 mean
COHS = 12
per 100 ft,
range = 4 to
Ave.
Rate
per Day
15 (out of 108)
4.2/person for
lower,
10.2/person,
upper
26
Model Type
&Lag
Structure
Autoregressive
logistic
regression
Logistic
regression
Other
pollutants
measured
nitric acid,
sulfates,
nitrates, SO2,
and hydrogen
ion
ozone, mean
= 7 pphm,
range = 1 to
Weather & Other
Other pollutants
Factors in model
day of survey, none
day of week,
gas stove,
minimum
temperature
temperature, none
rain humidity
28
Result*
(Confidence
Interval)
Cough
1.09
(0.57,2.10)
Sulfates:
Upper resp.
0.91
(0.73, 1.15)
Lower resp.
1.48
(1.14, 1.91)
Relative risk calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100 /xg/m3 increase in TSP.
0
O
1
O
a
1
O
W
n
i—i
H
m
-------
1 obstructive lung disease and, in some cases, asthma are some factors related to more rapid
2 declines in lung function in adults (Tager et al., 1988; Vedal et al., 1986).
3 Factors in the environment undoubtedly influence the natural history of the growth and
4 decline of lung function. Three such factors, viral respiratory illness, active smoking and
5 passive exposure to tobacco smoke products, are briefly summarized here. As in older
6 children and adults clinically inapparent alteration in lower airway function occur during URI
7 in infants (Martinez et al., 1990). Both differences in the caliber or length of the airway and
8 differences in the elasticity of the lungs and chest wall may exist between infants who
9 subsequently have wheezing with a viral lower respiratory tract illness and those who do not
10 have wheezing with a similar illness. Thus the initial airway caliber, length, or both (and
11 perhaps the structure of the lung parenchyma) may predispose infants to wheezing in
12 association with common viral respiratory infection (Martinez et al, 1988; Tager et al., 1993;
13 Martinez et al., 1991; Martinez et al., 1995).
14 Active smoking is the major risk factor for chronic airflow limitation. As a group,
15 cigarette smokers have more rapid reductions in lung function with age relative to non-
16 smokers. In approximately 15 to 20% of long-term regular smokers, this increased loss of
17 lung function leads to the development of symptomatic chronic obstructive lung disease.
18 Smoking cessation can be associated recovery of a very small amount of function and a
19 lessening of the rate of decline of function (Dockery et al, 1983). However, such cessation
20 amongst persons with far advanced chronic obstructive lung disease has little effect on the
21 overall course of the disease.
22 Passive exposure to products of tobacco smoke generated by parental smoking
23 consistently has been associated with alterations in lung function in infants and children.
24 Maternal smoking, in particular, has demonstrated an exposure-response association with
25 reduced lung function. The extensive body of evidence in support this association has been
26 reviewed by the U.S. Environmental Protection Agency (1992). The issue of passive
27 exposure to tobacco smoke has particular conceptual relevance to the issue of the health
28 effect of ambient PM, since tobacco smoke is a major PM source in indoor environments.
29
April 1995 U-119 DRAFT-DO NOT QUOTE OR CITE
-------
1 Studies of Pulmonary Function in Children
2 Dockery et al. (1982) studied changes in lung function in school age children as the
3 result of air pollution episodes in Steubenville, OH — one of the cities in Harvard Six City
4 studies. Steubenville was known to have large changes in SO2 and TSP exposures, and these
5 occurred in the fall of 1978, the fall of 1979, the spring of 1980, and the fall of 1980.
6 During each period, lung function measurements (FEV0 75 and FVC) were taken prior to the
7 episode and within a week after the episode. Linear regression was used to estimate the
8 effect of pollution on each child separately. The slopes were summarized by period and
9 combined in a total summary. The pooled slopes were significantly different from zero for
10 both TSP and SO2 for both FEV0 75 and FVC. The median slope for FEV0 75 with TSP was
11 -0.018 ml per pig/m3 and for FVC it was -0.081 ml per /zg/m3.
12 During November, 1984, Dassen et al. (1986) obtained baseline pulmonary function
13 data for a sample of more than 600 children aged 6-11 in the Netherlands. Then, a subset of
14 the same children (N = 62) was retested again in January, 1985, during an air pollution
15 episode when 24-h mean values for TSP (hi-vol samples), RSP (respirable suspended
16 paniculate, cyclone sampler), and S02 (acidimetric technique) measured via a 6-station
17 network all reached the range of 200 to 250 /ig/m3. Lung function values of 62 children
18 were taken at the end of the episode. Growth adjusted FVC values decreased by an average
19 of 62 ml (11), FEVj 0 values decreased by 50 ml (10), and PEFR values decreased by
20 219 ml/sec (62). These decreases were all statistically significant. Several lung function
21 parameters showed statistically significant average declines of 3 to 5% at second (episode)
22 testing in comparison with each child's own earlier baseline values, including decrements
23 seen on the second day of the episode in both FVC and FEV levels, as well as in measures
24 reflecting small airway functioning (i.e., maximum mid-expiratory flow and maximum flow
25 at 50% vital capacity). Declines from their original baseline values for these parameters
26 were still observed 16 days after the episode upon retesting of another subset of the children,
27 but no differences were found between baseline and retest values for a third subset of
28 children reevaluated 25 days after the episode. The 24-h mean TSP, RSP, and SO2 levels
29 measured in the 100 to 150 /ig/m3 range just prior to the last lung function tests may not
30 have been sufficient to cause observable pulmonary function effects in children.
April 1995 12-120 DRAFT-DO NOT QUOTE OR CITE
-------
1 Quackenboss et al. (1991) reported results of a lung function study of asthmatic
2 children aged 6 to 15 years in Tuscon, AZ. The data were collected over two week periods
3 from May 1986 to November 1988. Peak flow rates (PEFR) were measured with mini-
4 Wright peak flow meters with three tests during each of the four times per day (morning,
5 noon, evening, bed). Activity patterns were recorded in diaries, as well as symptoms and
6 medication use. PM2 5, PM10, and NO2 measurements were made both inside and outside
7 the home during the two week period for 50% of the homes. PM2 5 levels were elevated in
8 homes with environmental tobacco smoke. Exposures for the remaining homes were
9 estimated statistically. A random effects linear model was used to estimate the effect of
10 pollutants and other covariates on PEFR. The NO2 levels had the greatest effect on PEFR
11 rates, but the indoor PM2 5 levels were associated with a 15 ml/s decrease in morning PEFR
12 (within day change) per unit increase of PM2 5 in jug/m3. The relationships were unaffected
13 by the inclusion of weather variables such as temperature, wind speed, and dew point.
14 Pope et al. (1991) studied pulmonary function (PEFR) in asthmatic school children in
15 the Utah Valley. The group of participants was selected from samples of 4th and 5th grade
16 elementary students in 3 schools in the immediate vicinity of PM10 monitors in Orem and
17 Lindon Utah. Participants were restricted to those who responded positively to one of: ever
18 wheezed without a cold, wheezed for 3 days out of a week for a month or longer or had a
19 doctor say the "child has asthma". This resulted in 34 subjects being included in the final
20 analyses. PM10 monitors were operated by the Utah State Department of Health. 24 hr
21 samples were collected from midnight to midnight, and PM10 values ranged from 11 to 195
22 /ig/m3. There was limited monitoring of SO2, NO2, and ozone. PEFR values were averaged
23 across participants, and the deviations were analyzed using single period and polynomial-
24 distributed lag models. The estimated coefficient for PM10 was -0.0110 1/min (0.0082).
25 This coefficient corresponds to a 9.2 ml decrease in PEFR for a 50 /xg/m3 increase in PM10.
26 This effect was not statistically significant, but using a five day moving average of PM10 did
27 result in a significant regression coefficient. The relationship was not affected by the
28 inclusion of low temperature as a covariate.
29 Pope et al. (1991) also studied pulmonary function (PEFR) in asthmatics aged 8 to 72
30 in the Utah Valley, selected from those referred by local physicians. This resulted in 21
31 subjects being included in the final analysis. PM10 monitors operated by the Utah State
April 1995 12-121 DRAFT-DO NOT QUOTE OR CITE
-------
1 Department of Health collected 24 hr samples from midnight to midnight (PM10 range 11 to
2 195 /ig/m3). There was limited monitoring of SO2, NO2, and ozone. PEFR values were
3 averaged across participants, and the deviations were analyzed using single period and
4 polynomial-distributed lag models. The estimated coefficient for PM10 was -0.0175 1/min
5 (0.0092), corresponding to a 14.6 ml decrease in PEFR for a PM10 50 pig/m3 increase. This
6 effect was not statistically significant, but using a five day moving average of PM10 did
7 result in a significant regression coefficient. The relationship was not affected by the
8 inclusion of low temperature as a covariate.
9 Pope and Dockery (1992) studied non-asthmatic symptomatic and asymptomatic Utah
10 Valley children selected from samples of 4th and 5th grade elementary students in three
11 schools near PM10 monitors in Orem and Lindon, Utah. A questionnaire identified 129
12 children who were mildly symptomatic and 60 were selected; and 60 more with no symptoms
13 were selected. Utah State Department of Health PM10 monitors collected 24 hr samples
14 from midnight to midnight; PM10 ranged from 11 to 195 /zg/m3. For purposes of analyses,
15 five day moving averages of PM10 were used for exposure estimates. Limited monitoring of
16 SO2, NO2, and O3 was conducted. Mean deviations of PEF were computed for each
17 individual. Weighted least squares regression found a -0.00060 (0.00020) decrease in PEF
18 per unit increase in PM10 in symptomatic children and a -0.00042 (0.00017) change in PEF
19 per unit increase in PM10 in asymptomatic children. No relationship between low
20 temperature and PEF was found.
21 Koenig et al. (1993) studied two groups of elementary school children, one during the
22 school year 1988 to 1989 and another during the school year 1989 to 1990. The subjects in
23 the first study included 326 children, 24 of whom were asthmatics. During the second year,
24 only 20 asthmatics were studied (14 of which were in the original study). The FVC and
25 FEVj o were measured for each child in September, December, February, and May of each
26 year. Fine particulates were considered to be the primary pollutant of interest and were
27 measured by light scattering using a nephelometer. 12 hour averages (7:00 PM to 7:00 AM)
28 were used as the exposure measure. Additional information on PM2 5 was collected and
29 shown to be linearly related to light scattering (r2 = 0.945). A mixed model was used to
30 analyze the data. The model included random effects terms for the individuals and fixed
31 effects terms for height, temperature, and light scattering. No relationship was found
April 1995 12-122 DRAFT-DO NOT QUOTE OR CITE
-------
1 between light scattering and lung function in the larger sample, but a significant relationship
2 was found in the asthmatics. When converted to PM2 5 units, the decrease in FEV1>0 was -
3 0.0017 (0.0006) liters/(jug/m3). Effects of other pollutants were not considered.
4 Silverman et al. (1992) studied 36 asthmatic children over a 10-day period in the
5 summer and a 10-day period in the winter in Toronto, Canada. The first study (17 subjects)
6 and the second study (19 subjects) were selected from a pool of 800 asthmatic children from
7 the Gage Research Institute in Toronto. Patients were selected if they had a diagnosis of
8 asthma and if they experienced wheezing at least a few times a week. Lung function
9 measurements were obtained at the start and end of each day. Subjects carried a portable
10 monitor which measured SO2, NO2, and particulate matter. The first study measured
11 particles less then 25 microns, the second less than 10 microns. The regression coefficient
12 of FEVj 0 on PM was -.00078 liters/(/ig/m3) for the summer and .00018 liters/0*g/m3) for
13 the winter for Study 1, and -.00165 and .00283 for the summer and winter in Study 2. No
14 standard errors were given. The SAS analysis procedure was not specified, and there was no
15 mention of a repeated measures design. Results from analyses using SO2 and NO2 as
16 exposure variables were not reported.
17 Hoek and Brunekreef (1993) studied pulmonary function in 112 children aged 7 to 12
18 who lived in a non-urban area near Wageningen, ND. Spirometry was performed every
19 three weeks for a total of six times. One more measurement was made during an air
20 pollution episode. PM10 was measured daily (3PM to 3PM) with an instrument similar to the
21 Sierra Anderson 241 dichotomous sampler. SO2 was measured by fluorescence and NO2 by
22 chemiluminescence. Linear regression analysis using the SAS procedure AUTOREG yielded
23 an estimated coefficient for FEV! with PM10 of -.00055 liters/dug/m3) (.00010) and for PEF
24 of -.00082 (liters/s)/(/xg/m3) (.00050). Lagged PM10 values gave similar coefficients. SO2
25 and black smoke coefficients were similar in magnitude. Thus both FEVj and PEF showed
26 decreases related to pollution measures, but it was not possible to assign the effects to a
27 particular pollutant.
28 Hoek and Brunekreef (1994) studied pulmonary function and respiratory symptoms in
29 children aged 7 to 11 in the towns of Deurne, Enkhuizen, Venlo, and Nijmegen, The
30 Netherlands. Each child was studied six to ten times during one of three winters (1987/88,
31 1988/89, 1989/90). Measurements of FEV were obtained along with information on
April 1995 12-123 DRAFT-DO NOT QUOTE OR CITE
-------
1 respiratory symptoms. PM10 was measured daily (3 pm to 3 pm), as were SOj and NO2.
2 Linear regression analysis using the SAS procedure MODEL with the %AR macro was used.
3 The estimated coefficient for FEV! with PM10 was -.00010 liters/Og/m3) (.00006) and the
4 estimated coefficient for PEF was -.00082 (liters/s)/Og/m3) (.00029). Lagged PM10 values
5 gave similar coefficients. PM10 and NO2 coefficients remained significant after adjusting for
6 ambient temperature, but pollutants such as SO2, HONO, sulfate and nitrate did not. Other
7 adjustments for factors such as relative humidity, self-reported colds, and learning effects did
8 not affect magnitudes of estimated coefficients.
9 Brunekreef et al. (1991) further analyzed data from Dockery et al. (1982) on pulmonary
10 function in children in Steubenville, OH as part of the Harvard Six-Cities Study discussed
11 earlier. Linear decreases in forced vital capacity (FVC) with increasing TSP concentrations
12 were found, and slopes were determined for linear relationships fitting the data for four
13 different observation periods (fall, 1978; fall, 1979; spring, 1980; fall, 1980). The slope of
14 FVC vs. TSP was calculated for 335 children with three or more observations during any of
15 the four study periods, with 194 having been tested during more than one study period.
16 Individual regression coefficients for each child using pollution as the independent variable
17 were calculated. The distribution of coefficients was then trimmed to eliminate outliers.
18 Slopes for TSP using one and five day averages were significantly lower than zero for both
19 FVC and FEV0 75. No overall dose-response relationship was estimated.
20 Dusseldorf et al. (1994) studied pulmonary function in 32 adults living near a large
21 steel plant in Wijk aan Zee, The Netherlands. During the study period PM10 levels ranged
22 from 36 to 137 /xg/m3. Peak flow measurements (PEFR) were measured twice daily using a
23 Mini Wright peak flow meter. Diary information on acute respiratory symptoms, medication
24 use, and presence of fever was also collected. The study was conducted from 11 October
25 1993 to 22 December 1993, and the average number of days per subject was 66. Multiple
26 linear regression analysis with adjustment for first order autocorrelation. The regression
27 coefficient for evening PEFR on PM10 (lag zero) was -.054 1/min/jug/m3 (0.022), and the
28 regression coefficient for morning PEFR on PM10 (lag zero) was -.092 l/min//*g/m3 (0.026).
29 These correspond to an estimated decrease in PEFR for an increase of SO/ig/m3 PM10 of
30 45 and 77 ml/sec respectively. Lags of one, two, and three days were also fitted, but they
31 gave smaller estimated coefficients.
April 1995 12-124 DRAFT-DO NOT QUOTE OR CITE
-------
1 Lebowitz et al. (1992) studied 30 children with a current diagnosis of asthma using
2 PEFR measurements twice daily. A total of 674 PEFR measurements were analyzed.
3 Individual activity patterns were collected. PM2 5 and PM10 samples were collected in 50%
4 of the homes. Local monitoring stations were used to measure outdoor exposure. Using a
5 random effects model, PEFR was found to be significantly lower in the morning in children
6 who lived in homes with higher concentrations of PM.
7 Johnson et al. (1982) studied lung function in children as part of the Montana Air
8 Pollution Study, designed to collect sequential pulmonary function data on children from
9 November 1979 to April 1980 at six different time points. By adding a 7th round of testing
10 on May 23, 1980, the study took advantage of the natural experiment created by the eruption
11 on May 18, 1990 of the Mt. St. Helens volcano in Washington state. Approximately
12 100 children had been measured for FVC, FEVj 0, and FEF25.75 on six earlier occasions.
13 During five of these measurement periods the 3-day TSP average was relatively low (98 to
14 154 j^g/m3), but in one period, the average was 440 jug/m3. The eruption of the volcano on
15 May 18, 1980 forced nearly everybody indoors for the following three days. Most children
16 who ventured out did so with masks on. By May 23, the air had cleared enough so that
17 children returned to school, and their pulmonary function was measured. The TSP values
18 for the four preceding days ranged from 948 to 11,054 /ig/m3. The authors used an unusual
19 method of analysis which was described in their appendix. The results indicated that there
20 was a bigger decrease in lung function on the 400 /ng/m3 day than there was on the day
21 following the high volcanic ash episode.
22 Johnson et al. (1990) studied pulmonary function in 120 3rd and 4th grade children in
23 Missoula, MT during 1978 to 1979 who were tested up to six times between October 1978
24 and May 1979. FVC, FEVj 0, and FEF25.75 were measured. TSP was monitored daily near
25 the center of the study area. RSP was measured every third day and estimated from TSP and
26 other variables on the other days. The average of the current day's pollution and the
27 previous two day's pollution was used as the estimate of exposure. Each child who had at
28 least three readings was used as his own control. Percent changes in FVC, FEVj 0, and
29 FEF25_75 on higher pollution days as compared with the same measurements on days with
30 lower pollution exposure were used as the response variable. FVC averages on days with
31 RSP 31 to 60 /Kg/m3 were decreased about 0.40% and on days with RSP > 60 were
April 1995 12-125 DRAFT-DO NOT QUOTE OR CITE
-------
1 decreased about 0.75%. Similar but smaller changes were seen in FEV10 and FEF25_75. All
2 changes were marginally significant. The authors did not mention any other pollutants.
3 Roemer et al. (1993) studied children with chronic respiratory symptoms in the
4 Netherlands. Parents of children in grades 3 to 8 in two small nonindustrial towns in the
5 Netherlands were given questionnaires about respiratory symptoms. Seventy-four of the
6 313 children with positive responses cough or shortness of breath were included in the study.
7 Peak flows were measured in the morning and evening. PM10 was measured daily using an
8 Anderson dichotomous sampler. Sulfur dioxide and NO2, and black smoke were also
9 measured. Regression coefficients for both morning and evening current day's PM10 levels
10 were significant. Lagged PM10 values were not significant. The coefficient for current days
11 PM10 with morning PEF was -0.00090 (liters/s)/0«g/m3) (0.00028). SO2 was also
12 significantly related to evening peak flow but not to morning peak flow. Black smoke was
13 not related to peak flow.
14 Bock et al. (1985) and Lioy et al. (1985) examined pulmonary function of 39 children
15 at a camp in Mendham, New Jersey during a 5-week period in July to August, 1982. Ozone
16 was continuously monitored using chemiluminescent analysis. Ambient aerosol samples were
17 collected on Teflon filters with a dichotomous sampler having a 15 /xm fractionation inlet and
18 a coarse/fine cut size of 2.5 yum (Sierra Mod 244-E). Aerosol acidity was measured by
19 strong acid (H+) content, using the pH method. Highly significant changes in peak
20 expiratory flow rate (PEFR) were found to be related to ozone exposure, as well as a
21 baseline shift in PEFR lasting approximately one week following a haze episode in which the
22 ozone exposure exceeded the NAAQS for four consecutive days that included a maximum
23 concentration of 185 ppb. There was no apparent effect of H+ on pulmonary function. The
24 authors did state, however, that the persistent effects associated with the ozone episode could
25 have been due to acid sulfates as well as, or in addition to, ozone but additional uncollected
26 data were needed to evaluate this possibility.
27 Lioy et al. (1987) and Spektor et al. (1988) measured respiratory function of 91 active
28 children who were residing at a summer camp on Fairview Lake in northwestern New Jersey
29 during a 4-week period in 1984. Continuous data were collected for ambient temperature,
30 humidity, wind speed and direction, and concentrations of ozone, H2SO4, and total sulfates.
31 Ozone was measured by U.V. absorbance, and H2SO4 and total sulfates were determined by
April 1995 12-126 DRAFT-DO NOT QUOTE OR CITE
-------
1 a flame photometric sulfate analyzer (Meloy Model 285) preceded by a programmed thermal
2 pretreatment unit. The ambient aerosol samples were collected on quartz fiber filters with a
3 dichotomous sampler having a 15 /mi fractionating inlet (PMJ5) and a coarse/fine cut-size of
4 2.5 /mi (Sierra Model 244-E). Aerosol acidity, as measured by strong acid (H+) content,
5 was determined using the pH method. The maximum values recorded for H2SO4 and
6 NH4HSO4 were 4 and 20 /^g/m3 respectively. While effects were reported as being
7 associated with exposure to ozone, no effects were found to be related to exposure to the acid
8 aerosol concentrations experienced in this study.
9 Studnicka et al. (1995) studied acidic particles in a summer camp study in southern
10 Austria between June 28 and August 28, 1991. Daily spirometry was measured in three
11 panels of children age 7 or greater with a total of 133 subjects. On site measurements were
12 taken for H"1", sulfate, ammonia, PM10, and ozone. A repeated measures linear regression
13 model was fitted using a SAS macro. Pulmonary function measurements were made using a
14 rolling-seal-type instrument which gave flow-volume tracings. From these, FEVj 0, FVC,
15 and PEFR were measured. The results from all three panels combined suggested that PM10
16 was marginally related to a decrease in FEVj 0, but was less related to FVC and PEFR.
17 Other pollutants showed less of a relationship than did PM10. The coefficient for FVC
18 suggested that an increase of 50 /ig/m3 in PM10 would result in a 66 ml (39) decrease in
19 FVC, and a 99 ml/s (99) decrease in PEFR.
20 Neas et al. (1995) studied peak expiratory flow rates in 83 children living in
21 Uniontown, Pennsylvania. PEFR rates were measured over an 87 day period during the
22 summer of 1990 using a Collins recording survey spirometer. Air pollution data was
23 collected from a monitoring site located 2 km north of the center of the town, and included
24 PM10, PM2 5, ozone, SO2, sulfate fraction, and H+. The PM2 5 values had a mean of
25 24.5 /ig/m3 with an interquartile range of 18.9. The PEFR values were analyzed using the
26 autoregressive integrated moving average procedure of SAS. The model included terms for
27 temperature, time trend, and second-order autocorrelations. The largest decreases in PEFR
28 were related to H+, but they were also related to both PM2 5 and ozone.
29
April 1995 12-127 DRAFT-DO NOT QUOTE OR CITE
-------
1 Studies of Pulmonary Function in Adults
2 Pope and Kanner (1993) studied adults in the Salt Lake Valley, using spirometric data
3 taken from the Salt Lake City Center of the Lung Health Study supported by NHLBI; 624
4 participants were selected based on presence of mild COPD and willingness to participate in
5 a 5-year smoking cessation study. Analyses were based on two initial screening visits before
6 randomization into the NHLBI Study; 399 subjects had adequate data to be in the analyses.
7 PM10 monitors were operated by the Utah State Department of Health collected 24-h
8 samples midnight to midnight. Limited monitoring of SO2, NO2, and O3 showed these
9 pollutants to be always well below their respective NAAQS and none were included in the
10 analyses. Regression analyses on change in FEVj (liters) per change in PM10 (/Kg/m3) found
11 a coefficient of -0.00058 liters/(^g/m3) (0.00022). Changes were also seen in ratio of FEV}
12 to FVC. PEF was not measured.
13 Perry et al. (1983) conducted a longitudinal study of 24 Denver area asthmatics'
14 pulmonary function, symptoms, and medication use followed daily January through March,
15 1979. Peak flows (from Mini-Wright Peak Flow Meters), symptoms, and medication use
16 were measured twice a day. Fine and coarse mass as well as sulfate and nitrate fractions
17 were available from an east and west Denver site. CO, SC^, and O3 were also measured.
18 Dichotomous, virtual impactor samplers provided daily measurements of inhaled PM (total
19 mass, sulfates, and nitrates), for coarse (2.5 to 15 jwm) and fine fractions (<2.5 ftm) with all
20 PM measures being relatively low during the study. Temperature and barometric pressure
21 were also measured. Individual subject data were analyzed separately by regression analysis.
22 The coefficients were then tested using a non-parametric Wilcoxon signed rank test. None of
23 the PM measures were associated with changes in any of the health endpoints. This study
24 had very low power, given the small sample size and lack of high PM levels.
25
26 Acute Pulmonary Function Studies Summary
27 The acute pulmonary function studies (summarized in Table 12-14) are suggestive of a
28 short term effect resulting from paniculate pollution. Peak flow rates show decreases in the
29 range of 30 to 40 ml/sec resulting from an increase of 50 jwg/m3 in PM10 or its equivalent.
30 The results appear to be larger in symptomatic groups such as asthmatics. The effects are
April 1995 12-128 DRAFT-DO NOT QUOTE OR CITE
-------
1 seen across a variety of study designs, authors, and analysis methodologies. Effects using
2 FEVj or FVC as endpoints are less consistent.
3
4 12.4 HEALTH EFFECTS FROM LONG-TERM EXPOSURE TO PM
5 12.4.1 Mortality Effects of Long-Term PM Exposures
6 The long-term effects of air pollution may be examined by considering gradual changes
7 over tune (the longitudinal study) or by contrasting spatial differences at a given point in
8 time (the cross-sectional study). Longitudinal studies require long-term changes in air
9 quality, such as those that accompany pollution abatement campaigns. Only a few such
10 studies have been published (Lipfert, 1994a); none recently. Cross-sectional studies are
11 designed to infer the accumulated long-term effects of the environment by contrasting spatial
12 differences. As with all epidemiology, such spatial gradients may only be credibly attributed
13 to air quality after the potential confounders have been controlled.
14 Mortality rates or probabilities of survival may differ by location for any of a number
15 of reasons:
1 (1). geographic variations in the presence of specific endemic diseases, such as
2 malaria or influenza.
3
4 (2). geographic variations in risk factors, which may either be associated with the
5 location per se or with the people who tend to live there, such as climate or the
6 quality of the environment, the age and racial distributions of the population, its
7 dietary and smoking habits, or occupational factors associated with local
8 industry.
9
10 (3). geographic variations in the frequencies of acute events, such as heat waves,
11 accidents, or natural disasters.
12
13 Epidemiological studies of the long-term effects of community air pollution on mortality are
14 most likely to be concerned with the second and third categories, since (aside from
15 occupational lung disease) there are no diseases specific to these pollutants.
16 The second category of geographic factors relates to chronic effects on health, and it is
17 the task of the analyst to distinguish the effects of air pollution from those of other risk
18 factors in much the same way that it is necessary to adjust for seasonal trends in time-series
April 1995 12-129 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-14. ACUTE PULMONARY FUNCTION CHANGES
J"V
s
t5J
i— *
NJ
i
t— '
O
O
£>
"ti
G'
O
2
o
^
o
1
^J
w
0
Study
Dockery et al. (1982)
School age children in
Steubenville, OH,
measured at three times
between 1978 and 1980
Dassen et al. (1986)
School age children hi
The Netherlands,
measured hi November,
1984 and January, 1985
Quackenboss et al.
(1991)
Asthmatic children aged
6 to 15 years hi Tuscon,
AZ, measured hi May
and November, 1988
Pope et al. (1991)
Study of asthmatic
children hi the Utah
Valley
Pope and Dockery
(1992)
Study of non-asthmatic
symptomatic and
asymptomatic children
in the Utah Valley
PM Type & PM Mean
No. Sites & Range
Single station up to 455 /zg/m3
measuring TSP
Six station network TSP and RSP both
measuring TSP, exceeded 200
RSP (PM10) /ig/m3
Individual
monitoring of
homes of PM2 5,
PM10
PM10 monitors in PM10 ranged from
Orem and Lindon, 11 to 195 /tg/m3
Utah
PM10 monitors in PM10 ranged from
Orem and Lindon, 11 to 195 /ig/m3
Utah
Model Type Other
& Lag pollutants
Structure measured
individual SO2
regression analyses
for each child,
coefficients pooled
across time
multiple linear SO2
regression
Random effects NO2
linear model
Weighted least SO2, NO2, ozone
squares regression
Weighted least SO2, NO2, ozone
squares regression
Weather &
Other Pollutants
Factors in model
average TSP
temperature
technician, RSP
appliance,
presence of
colds
temperature, PM25
wind speed,
dew point
low PM10
temperature
low PM10
temperature
Decrease*
(Confidence
Interval)
FVC: 8.1 ml
FEV075: 1.8ml
Note: decreases
were statistically
significant
slopes not given but
FVC, FEV^ and
PEFR were
significantly reduced
during episodes
PEFR: 375 ml/s
Note: these are
diurnal rather than
daily changes
PEFR: 55 ml/s
(24, 86)
Symptomatic
PEFR
30 ml/s
(10, 50)
Asympto-
matic PEFR
21 ml/s
(4, 38)
n
-------
TABLE 12-14 (cont'd). ACUTE PULMONARY FUNCTION CHANGES
3.
vo
o?
to
1
1— >
£
O
i>
o
2
O
O
O
H
W
0
n
H
W
Study
Koenig et al. (1993)
Study of asthmatic and
non-asthmatic elementary
school children in Seattle,
WA in 1989 and 1990
Hoek and Brunekreef
(1993)
Study of children aged 7 to
12 in Wageningen,
Netherlands
Roemer et al. (1993)
Study of children with
chronic respiratory
symptoms in The
Netherlands
Pope and Kanner (1993)
Study of adults in the Utah
Valley from 1987 to 1989
*Decreases in lung function
PM Type &
No. Sites
PM2 5 calibrated
from light
scattering
Single site
measure black
smoke. PM10 was
measured during
episodes
Single site
measure black
smoke. PM10 was
measured using an
Anderson dichot
PM10 was collected
daily from the north
Salt Lake site
PMMean
& Range
PM2 5 ranged
from 5 to
45 /ig/m3
range of PM10
was 30 to
144 /ig/m3
range of PM10
was 30 to
144 /ig/m3
PM10 daily
mean =
55 /ig/m3,
ranged from 1
to 181 /xg/m3
Model Type Other Weather &
& Lag pollutants Other
Structure measured Factors
Random effects none height,
linear regression temperature
SAS procedure SO2, NO2 day of study
AUTOREG
multiple linear SO2, NO2 none
regression
analysis
Linear regression Limited low
on difference in monitoring of temperature
PFT as a function SO2, NO2, and
of PM10 ozone
calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100
Decrease*
Pollutants (Confidence
in model Interval)
PM2 5 Asthmatics
FEVt 42 ml
(12, 73)
FVC 45 ml
(20, 70)
Non-asthmatics
FEVj 4 ml
(-7, 15)
FVC -8 ml
(-20, 3)
PM10 PEFR
41 ml/s (-8, 90)
PM10 PEFR
34 ml/s (9, 59)
PM2 5 FEVj
29 ml (7, 51)
FVC 15 ml (-
15, 45)
/ig/m3 increase in TSP.
-------
1 studies. Long-term health risk factors may be further subdivided into factors that relate to
2 the population of a given place (age, race, education, lifestyle, for example) and factors that
3 relate to the physico-chemical environment of that place (climate, air and water quality).
4 There are also likely to be interactions between these two subcategories, since places with
5 desirable environments may attract as in-migrants that portion of the population that is better
6 off economically while the disadvantaged part of the population may be forced to remain in
7 less desirable locations and in those with depressed economies.
8 The third category above is included in long-term studies because annual mortality rates
9 must also reflect the net sum of acute events that took place that year (Evans et al., 1984a).
10 If the increases in daily death rates associated with acute events are not subsequently
11 canceled by decreases (a phenomenon referred to as "harvesting"), annual rates will indicate
12 the history of these acute effects. Thus, differences in long-term mortality rates associated
13 with air pollution are likely to reflect some combination of acute and chronic effects.
14 Although both types of information are useful contributions to the overall understanding of
15 the health effects of air pollution, their distinction may be difficult if based on statistical
16 criteria alone.
17 Long-term mortality studies are considered here in two groups:
18 1. Cross-sectional studies based entirely on the characteristics of groups averaged
19 across various geopolitical units, referred to as population-based studies.
20
21 2. Prospective cohort studies based on data on individuals. The available studies of
22 this type are also cross-sectional in design, because their air pollution exposure
23 data were based on community-wide averages in much the same way as the
24 population-based studies.
25
26 Because none of the studies available for review had individual data on personal
27 exposures to air pollution, they must all be classed as "ecological," in the strictest application
28 of the definition. The population-based studies used annual mortality rates and annual
29 average air quality data, usually for coincident periods centered on decennial census years.
30 Brief considerations have been given to lagged exposures in several instances, in an attempt
31 to deduce effects of exposures over longer periods. The prospective studies consider the net
32 survival rates over a multi- year period of follow-up; various assumptions were made by the
33 different investigators about the appropriate timing of air pollution exposures. The studies
April 1995 12-132 DRAFT-DO NOT QUOTE OR CITE
-------
1 thus varied in terms of their ability to provide either a measure of lagged chronic effects or
2 an integrated measure of acute effects during a given period.
3
4 12.4.1.1 Methodological Considerations
5 Methods for analysis were considered in a general way in the Methodology discussion
6 (Section 12.2). However, there are some specific guidelines that should be considered with
7 respect to the estimation of long-term effects on the basis of spatial gradients. In general,
8 the most difficult problems are collinearity among pollutants, variable and inadequate
9 characterizations of pollutant exposure, and confounding by non-pollutant variables.
10 The study of long-term or chronic health effects of air pollution began with population-
11 based studies and became fraught with difficulty and controversy, more so than the short-
12 term studies (Smith, 1975; Lipfert, 1980; Ware et al., 1981; Ricci and Wyzga, 1983; Evans
13 et al., 1984b). The primary method of analysis involves comparing the health statistics of
14 populations of places which have had different environments over the long-term. However,
15 the comparisons are often complicated or even compromised by other differences that may
16 be related to the sources and effects of air pollution, such as industrialization or climate.
17 Cross-sectional studies often use data from only one specific year that may or may not be
18 truly representative of long-term environmental conditions.
19
20 Spatial Patterns in the United States
21 Spatial patterns of U.S. mortality rates show some well-defined trends that have existed
22 for decades (Lipfert, 1994a). Such patterns are sometimes called the "geography of
23 disease." In general, heart disease is higher east of the Mississippi and ischemic heart
24 disease shows even sharper gradients and peaks in the Northeast (part of this gradient could
25 be due to differences in diagnostic practices). Cancer death rates tend to be higher in the
26 Northeast, but not exclusively in industrialized states (Vermont, New Hampshire and Maine
27 are relatively high). Gorham et al. (1990) argue that breast cancer is higher in northern
28 latitudes because of reduced intake of vitamin D; such patterns were also seen in the former
29 Soviet Union, for example. Lung cancer deaths are more evenly distributed and include
30 some of the states with high tobacco use (Nevada, Kentucky, Virginia, but not North
April 1995 12-133 DRAFT-DO NOT QUOTE OR CITE
-------
1 Carolina). Pneumonia and influenza deaths are distributed across the country but tend to be
2 higher north of about the 36th parallel.
3 Spatial trends in air pollution have both local and regional patterns. Local patterns
4 within cities reflect the presence of primary pollutants from local sources (CO from traffic,
5 particles from industrial operations, SO2 and NO2 from combustion sources, for example).
6 There are also multi-state regional patterns in secondary pollutants, such as sulfates and other
7 fine particles in Appalachia and the "rust belt," and ozone in Southern California and along
8 the Northeast corridor from Washington to Boston. Collinearity among pollutants results
9 from common spatial patterns of their major sources.
10 Associations between air pollution and health may be suggested by coincident spatial
11 patterns; the challenge to the epidemiologist is to establish whether such associations are
12 causal or merely circumstantial, since there are also many lifestyle parameters that vary
13 spatially, such as smoking, diet, exercise and employment status. Furthermore, some of
14 these factors have much stronger effects on mortality than air pollution: for example, a
15 difference of 11 years in life expectancy according to whether all the "good" health practices
16 are followed (Belloc, 1973), and 55% higher mortality for unemployed white men, aged 25
17 to 64 (Sorlie and Rogot, 1990). (See Yeager et al. [1995] for a recent showing of the
18 significant correlation between sedentary lifestyle and coronary heart disease mortality at the
19 state level that remained after controlling for smoking and hypertension; the relative risk for
20 sedentary lifestyle was about 1.8.) Unfortunately, suitable data on these potential
21 confounders are not always available. The possibilities for confounding by regional factors
22 vary with the scale of the analysis; comparisons within regions may thus be less susceptible
23 than comparisons across the whole country. For this reason, consistency between different
24 types of studies becomes very important in considering causality.
25
26 Risk Measures
27 Most of these studies consider deaths from all causes. Some of them subtract deaths
28 from accidents, homicides, and suicides, yielding a quantity referred to by various authors as
29 "nonexternal" deaths, "deaths from all natural causes," or "all-disease deaths." Measures of
30 the risks attributed to air pollution differ by type of study and regression model. Some
April 1995 12-134 DRAFT-DO NOT QUOTE OR CITE
-------
1 studies report relative risks (mortality ratios) due to specified but arbitrary pollution
2 "reference" levels, such as 100 /xg/m3 of particulate matter or 50 ppb of ozone. These
3 figures are obtained by multiplying the regression coefficient (the relative risk per unit of
4 pollution) times the desired pollution level. This practice is convenient for comparing studies
5 of the same pollutant but is less suitable for comparing the relative effects of different
6 pollutants, because the actual relationship between pollutants in a given city may not
7 correspond with that assumed by the reference levels. Others report ordinary least-squares
8 regression coefficients in the original units of the study, such as change in annual death rate
9 per unit of pollution. These coefficients are specific to the measures used for both dependent
10 and independent variables, but may be converted to approximate log-linear coefficients or
11 relative risk by dividing by the mean value of the dependent variable.
12 One measure that is free of measurement units is the "elasticity," a term taken from
13 economics defining a nondimensional regression coefficient of y on X; (at the mean) as
14
15 es = bpc/y [12.4.1-1]
16
17 Elasticities may be expressed as decimals or in percent and offer another measure of
18 attributable risk, based on the mean values of the Xj. An elasticity of 0.04 thus corresponds
19 to a relative risk of 1.04 at the mean pollution level. Comparison of two elasticities may be
20 misleading if the mean values differ widely. Note that when the "effect" of a variable (bjXj)
21 is expressed as a percentage of the mean total response, "effect" and elasticity are
22 synonymous.
23
24 Model Specifications for Long-term Mortality Studies
25 By and large, identification of environmental influences on longevity has not been part
26 of the medical "mainstream," although epidemiological studies have included the effects of
27 both air quality and drinking water quality (mainly water hardness; see Pocock et al., 1980,
28 for example). Because of the large number of potential confounding variables in spatial
29 analysis, multiple regression has been the statistical method of choice. Models for population
30 studies may be either linear or log-linear, and some investigators have included pollutant
April 1995 12-135 DRAFT-DO NOT QUOTE OR CITE
-------
1 thresholds (hockey-stick models). For population-based studies, the dependent variable is
2 usually an annual mortality rate for the geographic unit in question. Rosenbaum and Rubin
3 (1984) argue that if age adjustment is used for the dependent variable, it must also be used
4 for any independent variable that may also exhibit age dependency (such as smoking or air
5 pollution exposure).
6 Prospective studies of individuals have featured the proportional hazards model, in
7 which the risk factors are multiplicative (these coefficients correspond to elasticities). The
8 dependent variable is thus dichotomous (alive or dead). The range in survival probability
9 among adult individuals is quite large, encompassing more than two orders of magnitude in
10 mortality rate, which indicates the very large effects of individual risk factors. Years of
11 medical research have identified some of these risk factors as genetic predisposition,
12 exposure to infectious diseases, access to medical care, and personal lifestyles, including
13 diet, exercise, and smoking habits.
14 In contrast, the variability among cities and Standard Metropolitan Statistical Area
15 (SMSA) mortality rates is relatively modest, with a coefficient of variation (CV) of about
16 17% (Lipfert, 1993a), most of which is due to differences in age distributions. This
17 reduction in variability occurs because areas as large as SMS As in the United States tend to
18 be similar in terms of their average characteristics, especially within regions; i.e., much of
19 the variability in individual risk factors is "averaged out." It is thus much easier to
20 accurately predict with a mathematical model the death rate averaged over some geopolitical
21 unit than the probability of survival of an individual over a given time. In any case, the
22 ability to accurately predict the effects of exposure to air pollution depends on the validity of
23 the model. Unfortunately, it is ineluctable that all such models are incomplete and thus
24 contain the potential for bias (Cohen, 1994).
25 As discussed in more detail above, to confound, a variable must be correlated with both
26 the dependent variable and the independent variable of interest. This limits consideration to
27 established mortality risk factors that may vary spatially in ways similar to the spatial
28 patterns of air pollution. In many cases, appropriate data on confounders are not available
29 and surrogates must be used for the actual mortality risk factors. Education is an example;
30 staying in school longer per se does not prolong life, but better educated individuals are
April 1995 12-136 DRAFT-DO NOT QUOTE OR CITE
-------
1 likely to have higher incomes and thus access to better medical care; they may also have
2 healthier personal lifestyles. However, the relationship between education and health is
3 likely to be nonlinear, as is the relationship between income and longevity (Rogot et al.,
4 1992). Greenland and Robins (1994a) point out with examples that control of potential
5 confounders "crucially hinges on adequate measurement of the potential confounders." It
6 therefore follows that the assumption of linearity may not be always be appropriate for
7 surrogate risk factors. Alcohol consumption and body mass are two of the risk factors that
8 have been shown to have non-linear relationships with mortality (see Gronbaek et al., 1994,
9 for example).
10 In cross-sectional regressions, confounding is inevitably a matter of degree, especially
11 for those pollutants that exhibit broad spatial patterns. Sulfate is a case in point; the
12 following correlations were given by Lipfert et al. (1988) among U.S. cities, ca. 1980:
13
14 % with college education -0.44
15 population change, 1970-80 -0.39
16 % below poverty 0.33
17 % Hispanic -0.22
18 % Asian -0.28
19 smoking index 0.15
20 drinking water hardness 0.00
21
22 In contrast, the correlation between SO42" and crude city mortality rate was 0.44, but after
23 taking the above variables into account in a multiple regression, the correlation dropped to
24 0.2 to 0.3, depending on the data set considered. Since this publication, state-level survey
25 data on many other behavioral risk factors has become available (Siegel et al., 1993), and
26 many of these factors are also correlated with sulfate concentrations. For example, the state-
27 level correlation with "% 65 and over with sedentary life-style" was 0.64. The spatial
28 collinearity between sulfate and these demographic/lifestyle factors is similar to that between
29 daily air pollution and weather, and presents a challenge to the analyst to separate cause from
30 circumstance.
31 The need for confounder control may also vary with the scale of the cross-sectional
32 analysis and the spatial variability of the air pollutants of concern. For example, in many
April 1995 12-137 DRAFT-DO NOT QUOTE OR CITE
-------
1 cases, ambient air quality monitors are sited near the locations of the worst air quality, near
2 point sources or in the densest part of a city, in keeping with their intended regulatory
3 function. They may thus overestimate the exposures of persons living in more distant
4 suburban areas. This is most likely to be the case for primary pollutants, such as CO, NO2,
5 SO2, and PM10 (or TSP). The opposite may be true for ozone, because of the tendency for
6 levels to be reduced near sources of NO. Secondary sulfate and other fine particles tend to
7 have much longer lifetimes and thus to be more uniformly distributed over entire states or
8 regions. Note that such differences in the reliability of exposure estimates will tend to bias
9 the regression coefficients, giving an advantage to those pollutants with smoother
10 distributions (Lipfert and Wyzga, 1995). Because the socioeconomic characteristics of the
11 population and thus their mortality risk factors are also nonrandomly distributed, especially at
12 the local level within an SMSA, collinearity may result between their actual population
13 exposures and these other mortality risk factors. More simply put, persons employed by a
14 local pollution source may tend to live closer to that source, and it may thus be difficult to
15 distinguish between their ambient exposures, their occupational exposures, and the personal
16 characteristics that led to their employment and residence there. As a result, all cross-
17 sectional studies, prospective or ecological, have the same obligation to demonstrate that
18 potential confounders have been adequately controlled.
19
20 12.4.1.2 Population-Based Cross-Sectional Mortality Studies
21 In this section, cross-sectional studies employing averages across various geopolitical
22 units (cities, SMSAs, etc.) are reviewed, in chronological order. The "ecological" label
23 applies to this group of studies without reservation, since no data on individuals are used; the
24 community- based study seeks to define the (average) community characteristics that are
25 associated with its overall average health status, in this case annual mortality rate. Extension
26 of these associations to imply causality (an individual concern) may thus require additional
27 information.
28 Studies published after 1985 are emphasized here, but it is also useful to refer to some
29 of the earlier influential studies for context. Table 12-15 summarizes some of the findings
30 from these "background" studies, which analyzed mortality from 1960 to 1974. Studies that
April 1995 U-13S DRAFT-DO NOT QUOTE OR CITE
-------
1 analyzed spatial variability in 1980 mortality are summarized in Table 12-16. Many of these
2 studies comprise a large numbers of individual regressions; the Tables indicate which ones
3 were selected for discussion here, but the numerical column headings are more convenient
4 for the discussion that follows.
5
6 Background and Critiques of Some Older Studies
1 Although there had been a few earlier intracity cross-sectional studies (Lipfert, 1994a),
8 the current "model" for the cross-sectional population-based study was introduced by Lave
9 and Seskin (1970, 1977). They published an extensive national cross-sectional regression
10 analysis and concluded that about 9% of annual U.S. metropolitan mortality (ca. 1960) was
11 associated with air pollution, considering TSP and SO42" jointly. The analysis was based on
12 multiple linear regression analysis of annual mortality rates in the major SMSAs in relation
13 to coincident annual air quality levels (as measured at city centers) and to selected other
14 explanatory variables, listed in Table 12-15. This study was the first to attempt to
15 characterize the air pollution exposure of an entire SMS A using (often fragmentary) data
16 from a single monitoring station. Previous cross-sectional studies had tried to characterize
17 city-wide averages by using multiple monitoring stations or surrogate variables such as fuel
18 use patterns (Lipfert, 1994a). Lave and Seskin also introduced the idea of using more than
19 one metric for each pollutant in the same regression; they used the annual means as well as
20 the minimum and maximum values observed during the year, thus jointly regressing 6
21 pollutant variables. This led to excessive collinearity, unstable results, and, with the ca.
22 1960 data, bias because the smaller cities (many of which were in the industrial states) often
23 recorded only 4 composite SO42" values during the year while other cities had readings every
24 two weeks. Later studies by several investigators showed that the annual mean was the
25 preferred pollution metric. As shown in the first two studies in Table 12-15, introduction of
26 a home-heating fuel variable resulted in loss of significance and reduced relative risks for
27 both pollution variables. There are other examples of this type of instability in Lave and
28 Seskin (1977).
April 1995 12-139 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-15. COMMONLY-BASED CROSS-SECTIONAL STUDIES (1960-74 MORTALITY)
v^
,_l
v§
<-"
i— >
i
£
O
O
\>
H
O'
o
lz!
o
^^
H
O
0
Source
Lave and
Seskin (1978)
Regr. 3.3-1
Lave and
Seskin (1978)
Regr. 5.2.2
Lave and
Seskin (1978)
Regr. 7.1-4
Lipfert (1984)
Regr. 4.2
Lipfert (1984)
Regr. 4.7
Lipfert (1984)
Tbl. 6, Line
10
Health
Outcome
Total
mortality
Total
mortality
Total
mortality
Total
mortality
Total
mortality
Total
mortality
Time
Period/
No. Units
1960,
117 SMSA,
USA
1960,
1 17 SMSA,
USA
1960,
112 SMSA,
USA
1970,
111 SMSA,
USA
1970,
69 SMSA,
USA
1970,
69 SMSA
PM
PM Mean
Indicators (/tg/m3)
TSP, 118
min SO4 4.7
TSP 118
min SO4 4.7
TSP 95
min SO4 3.5
TSP 96
SO4 10.9
TSP 96
SO4 10.9
non-S 80.5
TSP, 11.0
S04
PM Range/
(Std. Dev.)
(41)
(3.1)
(41)
(3/1)
(29)
(1.9)
(29)
(4.5)
(29)
(4.5)
(25)
(4.4)
Sites Mean Relative Risk1
Per City Model PM Lag Other at TSP = 100
City Pop. Type Structure Pollutants Other Factors SO4 = 15
1 447.0002 OLS, none none Pet. > Age 65, 1.050 TSP
joint Pet. Nonwhite, 1.104SO4
Pop. density,
Pet. Poor,
Population
1 447.0002 OLS, none none Pet. > Age 65, 1.019 TSP
joint Pet. nonwhite, 1 .030 SO4
Pop. density,
Pet. Poor,
Pop., Home heating
fuel
1 570,0002 OLS none none Pet. > Age 65, Pet. 1.091 TSP
joint nonwhite, 1.129SO4
Pop. density.
Pet. poor,
Population
1 989,000 OLS, none none Pet. > Age 65, Pet. 1.044 TSP
joint Afr. Amer., Pet. 1.057SO4
other nonwhite,
Pop. density,
Pet. poor, pop.
adj. cig. sales, coal,
wood, homeheat
1 989,000 OLS, none O3 same as above, with 1.052 TSP
joint water quality, 1.035 SO4
without pop.
1 989,000 OLS, none O3 Pet. > Age 65, 1.074
joint Pet. Afr. Amer., 1.019
Pet. other nonwhite,
Pop. density, Pet.
poor, adj. cig. sales,
coal, wood, home
heating, drinking
water,
pop. migration
RR.
, Confidence
Interval
(1.01-1.09)
(1.03-1.18)
(0.98-1.05)
(0.97-1.09)
(1.04-1.14)
(1.01-1.25)
(0.98-1.07)
(1.01-1.11)
(0.99-1.12)
(0.98-1.09)
(1.00-1.14)
NS
Elasticity
0.059
0.033
0.022
0.01
10.087
0.030
0.034
0.042
0.054
0.026
0.05
0.014
O
h—I
H
TO
-------
TABLE 12-15 (cont'd). COMMONLY-BASED CROSS-SECTIONAL STUDIES (1960-74 MORTALITY)
Source
Chappie and
Lave (1982)
Regr. 2-6
Chappie and
Lave (1982)
Regr. 3-6
Time
Health Period/
Outcome No. Units
Mortality 1974,
from 104 SMSA
natural
causes
Mortality 1974,
from 102 SMSA
natural
causes
PM Sites Mean
PM Mean PM Range/ Per City Model PM Lag Other
Indicators (>*g/m3) (Std. Dev.) City Pop. Type Structure Pollutants Other Factors
TSP 75 (41)
S04 9.6 (3.8)
TSP 75 (41)
S04 9.6 (3.8)
1 527,0002 OLS, none SO43
joint3
1 527,0002 OLS, none SO43
joint3
Pet. ^ Age 65, Pet.
nonwhite, pop.
density, income,
population
Pet. a Age 65, Pet.
nonwhite, pop.
density, income,
pop., tobacco sales,
alcohol sales, pet.
college grads,
industries
Relative Risk RR.
at TSP = 100, Confidence
SO4 = 15 Interval Elasticity
0.99 TSP NA
1.23SO4 NA
0.985 TSP NA
1.18SO4 NA
-0.01
0.15
-0.015
0.12
'At TSP = 100, SO4 = 15, concentration adjusted for migration.
2Median value.
Degression used minimum, maximum, and mean values for TSP and S04 in the same model; relative risks were calculated from combined elasticity for each pollutant.
N>
-------
1 Lave and Seskin also tested the hypothesis that current mortality patterns might be more
2 influenced by previous air pollution patterns than by coincident patterns. They found that
3 1969 mortality was slightly more correlated with 1960 air pollution than with 1969 pollution
4 values, although this may have been due to the "minimum" sulfate error that was only
5 present in the 1960 data base.
6 For credibility, findings on air pollution should be robust to changes in the locations
7 studied and to reasonable changes in the regression model. In the late 1970s and early
8 1980s, the Lave and Seskin model was subjected to several tests of robustness (Crocker et
9 al., 1979; Lipfert, 1978; Evans et al., 1984b). In general, the Lave and Seskin results were
10 found to lack robustness, especially for the sulfate variable, which was shown to be sensitive
11 to the inclusion of certain socioeconomic variables. Lipfert (1978) found that the Fe and
12 Mn content of TSP was more highly correlated with city mortality than was SO42~' for
13 example. He also found that TSP values lagged 10 years were about as successful as current
14 values in predicting ca. 1970 mortality.
15 Lipfert (1984) reanalyzed Lave and Seskin's 1969 total mortality data set for
16 112 SMSAs (third study in Table 12-15), using corrected data and many new independent
17 variables, including 1970 mortality (to correspond better with the socioeconomic variables
18 obtained from the 1970 Census (fourth through sixth studies in Table 12-15). The objective
19 was to provide a more rigorous analysis of the portion of the Lave and Seskin data that had
20 valid sulfate data, and to compare it with the original analysis by Lave and Seskin (1978)
21 using the same set of locations. The dependent variables were total SMS A mortality,
22 mortality by sex, and mortality for two broad age groups cut at age 65. The main
23 conclusion drawn from this analysis of 1970 SMS A mortality was that pollution effects on
24 mortality were sensitive to both model specification and the data set used. Specifically:
25 • Minimum SO4= had no advantage over mean S04= and was less significant, other
26 things being equal. When a complete model specification was used, sulfate was
27 rarely significant (as either measure) but significance occurred more often for
28 females. TSP was more important for deaths of those under 65 years of age. Non-
29 sulfate TSP (obtained by subtracting the mass equivalent of SO42" from each reading)
30 was just as important as TSP (Column 6).
31
32 • Drinking-water quality, ozone, migration, and racial distribution variables all had
33 important effects on the regressions. Only when a complete model specification was
April 1995 12-142 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-16. COMMONLY-BASED CROSS-SECTIONAL STUDIES (1980 MORTALITY)
3.
»— i
VO
^
0
H
O
CJ
o
$
§
r**
n
H
W
Source
Ozkaynak and
Thurston
(1989)
Table VI
Ozkaynak and
Thurston
(1987)
Table VD
Lipfert et al.
(1988)
Table 24
Lipfert et al.
(1988)
Table 24
Lipfert et al.
(1988)
Page 60
Lipfert (1993a)
Regr. 6.1,6.2
Health
Outcome
Total
mortality
Total
mortality
Total
mortality
Total
mortality
Total
mortality
Mortality
from
natural
causes
Time
Period/
No. Units
1980
98 SMSA
1980,
38 SMSA
1980
172-185
cities
1980
68 cities
1980
122 cities
1980
149 SMSA
PM Sites Mean
PM Mean PM Range/ Per City
Indicators (pg/m3) (Std. Dev.) City Pop.
TSP 78 (26) 1 NA
SO4 11.1 (3.4)
PM15 38 (7.3) 1 NA
PM25 20 (3.8)
Fe, 1.2 (0.61) 1 57,500
S04 9.5 (3.5)
PM15 38 (121) 1 57,500
PM25 18 (6)
TSP 88 (29) 1 about
SO4 9.0 (1.8) 60,000
TSP 68 (17) 10.6 928,000
(TSP)
SO4 9.3 (3.1)
Model1 PM Lag Other
Type Structure Pollutants Other Factors
OLS none none Pet. > Age 65,
sep. median age,
Pet. nonwhite, pop.
density,
Pet. poor, pet. w/ 4
yrs college.
OLS none none Same as above.
sep.
OLS none none Pet. > Age 65, birth
sep. rate;
Pet. Afr.-Amer. pop.
density, pet. poor;
Pet. pop. change, pet.
w/ 4 yrs. college;
Pet. Hispanic, adj.,
cig., sales; Pet. prior
res., hard water
OLS none none Same as above.
sep.
OLS 10 years none Pet. > Age 65, birth
joint rate, pet. nonwhite,
pop. density, pet.
poor, adj. cig. sales,
pet. w/ 4 yrs. college
OLS none none Pet. > Age 65, Pet.
sep. Afr.-Amer., Pet.
Hispanic, Pet. other
nonwhite, pet. poor,
pet. pop. change, adj.
cig. sales, pet. w/ 4
yrs. college, hard
water, heating degr.
days pop. density
Relative Risk2
at TSP = 100,
SO4 = 15
1.012 TSP
1.17SO4
1.059PM15
1.085PM25
1.044Fe
1.13SO4
1.036PM15
1.082 PM25
about 1 .0
1.072SO4
1.038 TSP
1.059SO4
RR.
Confidence
Interval
(0.96, 1.06)
(1.09, 1.24)
(0.95, 1.16)
(0.96, 1.21)
(1.02-1.07)
(1.06-1.20)
NS3
NS3
NS3
(1.0, 1.14)
(0.97, 1.10)
(0.99, 1.12)
Elasticity
0.01
0.086
0.045
0.068
0.041
0.071
0.027
0.059
NS
0.037
0.026
0.037
-------
TABLE 12-16 (cont'd). COMMONLY-BASED CROSS-SECTIONAL STUDIES (1980 MORTALITY)
3.
H- >
VO
0>
i— >
to
5
Q
&
T1
H
O'
O
O
H
O
O
O
n
H
w
Time PM Sites Mean
Health Period/ PM Mean PM Range/ Per City
Source Outcome No. Units Indicators (^g/m3) (Std. Dev.) City Pop.
Lipfert (1993) Mortality 1980 PM15 38 (29) 1 928,000
Regr. 13.1, from 62 SMSA
13.3 natural PM25 18 (4.5)
causes
Lipfert (1993) Mortality 1980 TSP 68 (17) 10.6 928,000
Regr. 9.1,9.3 from 62 SMSA SO4 9.3 (3.1) (TSP)
natural
causes
Lipfert (1993) Major 1980 SO4 (IP) 4.3 (2.5) 1 928,000
Regr. 13.5 CVD 62 SMSA
Lipfert (1993) Major 1980 SO4 (IP) 4.3 (2.5) 1 928,000
Regr. 12.1 CVD 62 SMSA
Lipfert (1993) COPD 1980 non-TSP* 56.4 (18) 10.6 928,000
Regr. 10.3, 149 SMSA
10.4 TSP 68.5 (17)
'All regression models used PM indicators one at a time (separate models) except as noted.
2At TSP = 100 pg/m3, SO4 = 15 /ig/m3, corrected for migration.
3NS = not statistically significant, confidence limits not available.
Model PM Lag Other
Type Structure Pollutants Other Factors
OLS none none Same as above
sep.
Log- none none Same as above
linear without other
nonwhite, heating
degr. days, pop.
density
OLS none none Same as above with
other nonwhite,
heating degree days,
pop. density
OLS none none Pet. > Age 65,
median age, pet.
nonwhite, pop.
density, pet., poor
pet. w/ 4 yrs.
college
Log- none none Pet. > Age 65, pet.
linear Afr.-Amer., Pet.
Hispanic, pop.
density, pet. poor,
adj. cig. sales
Relative Risk1 RR.
at TSP = 100, Confidence
SO4 = 15 Interval Elasticity
1.036PM15 (0.98,1.10) 0.027
1.060PM25 (0.99,1.13) 0.043
1.066 TSP (1.006, 1.13) 0.044
1.021SO4 NS 0.012
1.04SO4 NS 0.011
1.19SO4 (1.03,1.35) 0.054
1.50 TSP (1.22,1.83) 0.23
1.43 TSP (1.20,1.71) 0.25
-------
1 used did the coefficients for age, race, poverty, and smoking take on values
2 consistent with exogenous estimates; this implies that these additional variables were
3 also necessary to derive appropriate values for the air pollution variables.
4
5 • The analysis was incapable of distinguishing between linear and threshold models
6 and thus could not rule out the applicability of a threshold or hockey-stick model.
7
8 Chappie and Lave (1982) replicated the 1960 and 1969 Lave/Seskin results with data
9 for 1974 mortality using essentially their original model and found that sulfate had become
10 even more important than in previous years (relative risk was about double), in spite of
11 lower average concentrations, while TSP became insignificant (studies 7 and 8). Chappie
12 and Lave retained the multiple measures of pollutants and attempted to rebut earlier
13 criticisms about the lack of data on smoking by using retail trade sales data as a surrogate for
14 smoking behavior; however, this variable was not important in their analysis, probably
15 because it is a poor measure of actual tobacco consumption (Lipfert, 1994a).
16 In summary, at this point in the development of the methodology for population-based
17 cross-sectional studies (which was discussed in the 1982 CD and the 1986 Addendum [U.S.
18 Environmental Protection Agency, 1982a, 1986a]), it appeared that the findings of national
19 cross-sectional analyses were largely investigator-dependent. Table 12-15 shows that
20 including additional socioeconomic variables in the model reduced the apparent effects of
21 sulfate for the 1970 time period. However, the 1974 study found even larger effects of
22 sulfate, but it could not be ascertained whether this was due to the regression model used or
23 to the particular data set considered. Evans et al. (1984b) critically reviewed some of the
24 published cross-sectional mortality studies, including a quantitative comparison of regression
25 coefficients and a reanalysis of the Lave and Seskin (1978) 1960 data with new variables
26 added. For this reanalysis, the dependent variable was total 1960 mortality for 66 to 98
27 SMS As. The independent variables included percentage over 65, median age, percentage of
28 nonwhites, log of population density, percentage of college graduates, smoking index,
29 percentage of poverty, and mean sulfate concentration. Other variables explored included
30 occupational, home heating, climate and housing variables; mean values for TSP, Fe, Mn,
31 and B(a)P. In general, the sulfate variable lost statistical significance in this reanalysis; the
April 1995 12-145 DRAFT-DO NOT QUOTE OR CITE
-------
1 elasticity with respect to total mortality was about 0.03. None of the other pollutants
2 provided any better results.
3 Kim (1985) analyzed total mortality data for 1970 in an ecological analysis of 49 U.S.
4 cities. Pollutants considered were TSP and the benzene-soluble organic fraction of TSP
5 (BSO), in 5 different formats: averaged over the single years 1968, 1969, and 1970;
6 averaged for 1969 to 70, and for 1968 to 70. This analysis was intended to test for lagged
7 effects, but one might also expect the multiple-year averages to be superior because of the
8 reduction of random sampling errors. Socioeconomic/demographic (SES) variables included:
9 percent aged 65 and over, population density, percent unemployed, and percentage of
10 housing built before 1950. Four climate variables were also used: annual mean wind speed,
11 annual precipitation, and mean January and July temperatures. All climate and SES
12 variables were included in each regression, and both joint and single-pollutant models were
13 evaluated. Racial differences were not included in this model, which may have had the
14 effect of increasing the apparent climate (temperature) effects, since the black population has
15 higher mortality rates and constitutes a higher fraction of Southern cities. The wind
16 coefficients were highly nonsignificant, but positive coefficients were shown for
17 precipitation. This may reflect the lower mortality rates in the Western states, rather than a
18 harmful effect of precipitation per se.
19 Kim's lag analysis was largely inconclusive. He concluded "the effects of total
20 mortality in 1970 may be due to the air pollution in 1969, although it is not possible to
21 pinpoint a lag-effect between the time of exposure to air pollution and the time of death."
22 The TSP coefficients for the single year 1968 were the lowest by far, but the opposite was
23 true for BSO. It is possible that BSO affects different diseases (such as cancer) than TSP
24 and thus exhibits a different lag relationship. However, in spite of the advantage of three
25 times as many observations, the results for 1968 to 70 did not have larger t-values than the
26 single years. It is not possible to speculate what effect the model specification used by Kim
27 in this analysis might have had on the outcome of this lag analysis. It is clear, however, that
28 one must be careful in interpreting regression coefficients literally (e.g., the climate
29 variables) when an incomplete model specification is used.
30
April 1995 12_146 DRAFT-DO NOT QUOTE OR CITE
-------
1 Studies of 1980 SMSA Mortality
2 Ozkaynak and Thurston (1987) analyzed 1980 total mortality in 98 SMS As, using data
3 on PM15 and PM2 5 from the EPA inhalable particle (IP) monitoring network for 38 of these
4 locations. The sulfate data from this network were not used in this study. The independent
5 variables used are given in Table 12-16 (first two studies); in general, the regression
6 modeling approach was similar to that of Lave and Seskin (1970). The results presented in
7 Table 12-16 are from their "basic" regression model. Additional variables were explored,
8 including spatial correlation variables intended to examine regionality and latitude and
9 longitude variables. The sulfate measurements that Ozkaynak and Thurston used may have
10 been affected by artifacts from the high-volume sampler filters (Lipfert, 1994b); this is also
11 suggested by the fact that their mean SO42' value exceeds those of previous years and the
12 mean from the IP data set (compare studies 1 and 9 in Table 12-16).
13 Ozkaynak and Thurston (1987) ranked the importance of the various pollutants mainly
14 by relative statistical significance in separate regressions. They concluded that the results
15 were "suggestive" of an effect of particles on mortality decreasing with particle size,
16 although in the basic model only SO42" was significant. In some of the other models, PM2 5
17 was also significant and PM15, nearly so. However, if the effects are judged by elasticities
18 rather than significance levels, SO42"' PM2 5, and PM15 would be judged as equivalent, with
19 TSP ranking somewhat lower. The indicated effect of SO42~ would be reduced from an
20 elasticity of 0.086 to about 0.05 by accounting for filter artifacts (Lipfert, 1994b). Ozkaynak
21 and Thurston (1987) also used source apportionment techniques to estimate that particles
22 from coal combustion and from the metals industry appeared to be the most important.
23 The coefficients and significance levels obtained for TSP by Ozkaynak and Thurston
24 (1987) may be the result of the TSP data they used, which were based on a single monitoring
25 station in each SMSA and thus are unlikely to be fully representative of population
26 exposures. Thus, alternative interpretations of these findings are certainly possible. In
27 addition, because smoking, diet, and other socioeconomic or lifestyle variables were not
28 considered in the regression model, the pollution coefficients may have been biased. Finally,
29 this study did not specifically address the question of acute vs. chronic responses by
30 exploring lagged pollution variables.
April 1995 12-147 DRAFT-DO NOT QUOTE OR CITE
-------
1 The analysis by Lipfert et al. (1988) comprised a statistical analysis of spatial patterns
2 of 1980 U.S. central city total mortality (all causes), evaluating demographic, socioeconomic,
3 and air pollution factors as predictors (Studies 3 to 5 in Table 12-16). The advantages of
4 studying central cities vs. SMS As include potentially better measures of exposure because of
5 the smaller areas, and sufficient numbers of observations to allow analysis of subsets of
6 locations. In this study, sulfate and iron particles were significant (joint) predictors of
7 all-cause mortality in about 180 cities. If the elasticities for SO42~ were corrected to account
8 for the filter artifacts, they would be reduced by about 50% in this study. The data on
9 PM15 and PM2 5 were only available for 68 cities; neither pollutant was significant in this
10 data set but their elasticities were in the same range found for other pollutants (0.013 to
11 0.05). This study also allowed a test of lagged pollution data as a means of attempting to
12 distinguish acute from chronic responses; using 1970 TSP and SO42" data to predict 1980
13 city mortality was slightly less effective than using ca. 1980 data for these pollutants as
14 predictors.
15 Data from up to 149 metropolitan areas (mostly SMS As) were analyzed in a study of
16 the relationships between community air pollution and "excess" mortality due to various
17 causes for the year 1980 (Lipfert, 1993a). Several socioeconomic models, including the
18 model proposed by Ozkaynak and Thurston (1987), were used in cross- sectional multiple
19 regression analyses to account for non- pollution effects; the variables are listed in
20 Table 12.16 (Studies 6 to 11). Cause-of-death categories analyzed include all causes,
21 nonexternal causes (ICD9 0-800), major cardiovascular diseases (ICD9 390-448), and chronic
22 obstructive pulmonary diseases (COPD) (ICD9 490-96). The patterns for the first three
23 groupings were quite similar but differed markedly from the patterns of COPD mortality,
24 which tend to be higher in the Western U.S. Regressions were performed for these cause-of-
25 death groupings as annual mortality rates ("linear" models) and as their logarithms ("log-
26 linear" models). The original regressions used base-10 logarithms; the results have been
27 converted to natural logarithms for this review. Two different sources of measured air
28 quality data were utilized: data from the EPA AIRS database (TSP, SO4=, Mn, and ozone
29 from a long-term average isopleth map) and data from the inhalable paniculate (PM15)
30 network; the latter data (PM15,PM2 5 and SO4= from the IP filters) were only available for
April 1995 12-148 DRAFT-DO NOT QUOTE OR CITE
-------
1 63 locations. All paniculate data were averaged across all the monitoring stations available
2 for each SMSA; the TSP data were restricted to the year 1980 and were based on an average
3 of about 10 sites per SMSA.
4 The associations between mortality and air pollution were found to be dependent on the
5 socioeconomic factors included in the models, the specific locations included in the data set,
6 and the type of statistical model used, as was the case with 1970 data (Lipfert, 1984). In the
7 expanded analysis, stepwise regressions were run for each mortality variable and a
8 "parsimonious" model was developed that had statistically significant coefficients for the non-
9 pollution variables. Most of these coefficients also agreed with exogenous estimates of the
10 "correct" magnitudes. Using these models, statistically significant associations were found
11 between TSP and mortality due to non-external causes with the log-linear models evaluated,
12 but not with a linear model. Sulfates, manganese, inhalable particles (PM15), and fine
13 particles (PM2 5) were not significantly (P < 0.05) associated with mortality with any of the
14 parsimonious models, although PM2 5 and Mn were close with linear models (p=0.07) and
15 significance may have been affected by the use of smaller data sets. This study showed that
16 PM2-5 was the "strongest" paniculate variable with linear models, but that TSP performed
17 better in log-linear models. Scatter plots and quintile analyses suggested that a TSP
18 threshold might be present for COPD mortality, at around 65 /tg/m3 (annual average).
19 This study supported the previous findings of associations between TSP and premature
20 mortality and also the hypothesis that improving the accuracy of pollutant exposure data
21 tends to increase statistical significance. Similarly, the lack of significance for SO4= may
22 partly relate to the flawed measurement methods used at the time. The ambiguity between
23 linear and log-linear models probably reflects the effects of influential observations.
24
25 Cross-Sectional Mortality Studies by Age and Cause of Death
26 Only a few of the many published ecological mortality studies analyzed subgroups by
27 age and cause. Lave and Seskin (1978) used very broad age groups (0 to 14, 15 to 44, 45 to
28 64, 65+) with 1960 and 1969 data, which limited the usefulness of the analysis because of
29 the failure to account for age differences within these groupings. Lave and Seskin also
30 examined a large number of disease-specific mortality rates using 1960 and 1961 data.
April 1995 12-149 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cancers and cardiovascular diseases were associated with the flawed "minimum" sulfate
2 variable, but respiratory causes tended to be associated with TSP. Lipfert (1978) considered
3 for U.S. cities, 1969 to 71, infant mortality, ages 1 to 44, and from 45 to 85 by 10-year
4 groups. Very little significance was found below age 65; for ages 75 + , SO42"' TSP, Fe and
5 Mn were significant (one at a time). By cause of death, Lipfert (1978) considered
6 nonexternal causes, total cancers, respiratory cancer, respiratory disease (asthma, bronchitis,
7 emphysema) and all other diseases (mainly cardiovascular). Only Fe was significantly
8 associated with total cancers, only Mn with respiratory cancer, Fe was positively associated
9 with respiratory diseases but SO42" was strongly negatively associated with respiratory disease
10 mortality. Lipfert (1993a) found that PM was not significantly associated with mortality
11 from major cardiovascular causes for 1980 SMS A mortality, which implies that other causes
12 of death must be involved for this pollutant. Note that between 1960 and 1980 there were
13 major improvements in cardiovascular mortality, resulting in some changes in the geographic
14 patterns. For 1980 SMS A mortality, COPD mortality was strongly associated with TSP with
15 a variety of regression models. Significant associations were found between TSP and COPD
16 mortality for both linear and log-linear models (Study 11 in Table 12-16). When the sulfate
17 contribution to TSP was subtracted, the relationship with COPD mortality was slightly
18 strengthened. PM2 5 was a significant predictor of heart disease mortality only when the
19 regression model was restricted to the variables used by Ozkaynak and Thurston (1987).
20
21 Cross-Sectional Infant Mortality Studies
22 Bobak and Leon (1992) studied neonatal mortality (ages less than 1 month) and post-
23 neonatal mortality (ages 1 to 12 months) from 1986 to 88 in 46 administrative districts in the
24 Czech Republic, in relation to annual averages of PM10, S02, and NO2. The observations
25 comprised 121 combinations of districts and years, ranked into quintiles by mean pollution
26 level for analysis (5 districts had insufficient data). The analysis was ecological in design, in
27 that the outcome variable was the death rate per 1,000 live births and district-wide averages
28 were used as the control variables (mean income, mean savings, mean number of persons per
29 car, proportions of total births outside of marriage, and the rate of legal abortions. In the
April 1995 12-150 DRAFT-DO NOT QUOTE OR CITE
-------
1 United States, for example, infant mortality is a strong function of income or poverty status,
2 reflecting the effects of access to pre- and post-natal medical care.
3 The mean pollutant values were 68.5, 31.9, and 35.1 jig/m3 for PM10, SO2, and NO2,
4 respectively. This study appears to be based on a denser air monitoring network than many
5 of its predecessors in the U.S.; the mean population per monitor was only about 50,000 and
6 many ecological studies in the U.S. are based on values an order of magnitude higher than
7 this. Two of the three pollutants were highly correlated (R = 0.80 for SO2 vs. NO2),
8 indicating a common source (combustion). Correlations with PM10 were lower (0.15 and
9 0.26), reflecting the fact that the particle sources were more diverse and included cement
10 production. The maps presented in the paper suggested little likelihood for spatial
11 autocorrelation, although this topic was not discussed directly. Bobak and Leon (1992)
12 addressed the pollutant collinearity problem by presenting results for each pollutant alone and
13 for the combination of all three. Results were also presented with and without
14 socioeconomic adjustments.
15 The statistics used to indicate significant associations were chi-square p-values for trend
16 across the 5 quintiles, with the relative risks set to 1.0 for the lowest quintiles. Highly
17 significant trends (p < 0.01) were seen after socioeconomic adjustment for postneonatal
18 mortality only with PM10, even after including the other pollutants. Post-neonatal respiratory
19 mortality showed highly significant trends for all 3 pollutants, but only PM10 retained
20 significance (p=0.013) with all 3 pollutants. Because of the use of multiple years of data for
21 41 common locations and the strong likelihood of temporal persistence in annual average air
22 quality, the true number of degrees of freedom may be 41 rather than 121. For this reason,
23 a higher standard of association should be applied to these results (p < 0.01 rather than
24 p < 0.05). Although Bobak and Leon (1992) elected to analyze their data in terms of linear
25 responses over the entire pollutant range, their results were suggestive of a threshold at the
26 third quintile or higher (mean PM10 = 67 jug/m3).
27 It is not clear from the design of this study whether the reported effects are acute or
28 chronic. Pollution values were averaged over the same years used to aggregate deaths; thus
29 it is possible that exposure did not precede death in all cases. In any event, it may be
30 difficult to distinguish delayed acute from chronic responses for lifetimes as short as a few
April 1995 12-151 DRAFT-DO NOT QUOTE OR CITE
-------
1 months. Among the previous U.S. studies reviewed, Lave and Seskin (1978) found infant
2 mortality to be associated with TSP; Lipfert (1978) found marginal significance for the Fe
3 and Mn portions of TSP and a negative association with SO42" (ca. 1970).
4
5 Summary of Population-based Cross-Sectional Mortality Studies
6 Although most of these studies covered the entire U.S. using the basic paradigm of
7 Lave and Seskin (1970), there are major differences in the numbers of independent variables
8 considered, including the air pollutants. Most of the studies found pollutant elasticities (i.e.,
9 mean effects) of 0.02 to 0.08, although the specific pollutants associated with mortality
10 varied. However, all of these studies found at least some association between air pollution
11 and mortality on an annual average basis. There was a slight suggestion that elasticities may
12 be decreasing over time (1960 to 1980). It was not possible to determine whether the
13 mortality associations were stronger for pollution measured the same year or in previous
14 years. The finding of significant associations with metals (Lipfert, 1978, 1984, 1993) and
15 with cement plant particles (Bobak and Leon, 1992) suggests that many different types of
16 particles may be involved. Analyses by age and cause of death were limited; the most
17 consistent associations were for the elderly, especially ages 75 + , and for respiratory disease
18 mortality and TSP (but not SO4H
19
20 12.4.1.3 Prospective Mortality Studies
21 Studies considered in this section utilized data on the relative survival rates of
22 individuals, as affected by age, sex, race, smoking habits, and certain other individual risk
23 factors. This type of analysis has a substantial advantage over the population-based studies,
24 because the identification of the actual decedents allows stratification according to important
25 risk factors such as smoking. Such stratification allows tests of the hypothesis that certain
26 segments of the population may be more sensitive to air pollution than others, which is a
27 major shortcoming of population-based studies. However, analyzing individuals also entails
28 dealing with increased variability in outcome and thus requires large sample sizes if effects as
29 small as those typically found in population studies are to be detected with significance.
30 Unfortunately, none of the prospective cohort studies had data on personal exposures to air
April 1995 12-152 DRAFT-DO NOT QUOTE OR CITE
-------
1 pollution, which precludes analysis within cities or by type of exposure (primarily indoor vs.
2 outdoor, or coincident vs. accumulated, for example). In this limited sense, these studies are
3 also "ecological."
4 Although the newer prospective studies (Abbey et al., 1991, Dockery et al., 1993, and
5 Pope et al., 1995), are of most interest, brief comments on two older studies are included
6 here to allow additional comparison of effects by gender and smoking. The studies are
7 reviewed in chronological order. The main findings from the three most recent studies are
8 summarized in Table 12-17.
9
10 Comparison of Two Pennsylvania Towns
11 Morris et al. (1976) followed survival rates cohorts of volunteers in Seward (n=938)
12 and New Florence, PA, (n=732) from 1960 to 1972. Based on 1960 to 61 air quality data,
13 Seward had worse air quality for both sulfur oxides and suspended particulates. Although
14 the small sample size precluded definite conclusions about the effect of air pollution alone
15 (i.e., in nonsmokers), the data suggested a relative risk of 1.2 (p < 0.2) for mortality from
16 all causes for males with more than 20 years' residence in Seward, which would correspond
17 to a slope of about 0.005 per jig/m3 when the mortality gradient is assigned to TSP. The
18 effects of smoking and air pollution appeared to be additive. This study did not control for
19 occupation, education, other life-style variables or non-response bias.
20
21 Comparison of Two Residential Areas in Cracow, Poland
22 Nonaccidental mortality rates from random samples of two areas in Cracow were
23 compared after 10 years of follow up (1968-78) by Kryzanowski and Wojtyniak (1982). This
24 study tracked 4,355 individuals; their individual characteristics were identified (including
25 smoking habits) and used in categorical mortality models. The two residential areas were
26 characterized as: high pollution, average smoke = 180 pig/m3, SO2 = 114 /ig/m3; lower
27 pollution, average smoke = 109 jiig/m3 and SO2 = 53 jiig/m3. A total of 20 air samplers
28 was used, from 1968 to 73; the smoke samplers collected particles less than 10 jon in
29 diameter and concentrations were based on filter staining.
30
April 1995 12-153 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-17. RELATIVE MORTALITY RISKS IN SIX CITIES ADJUSTED RISKS
Adjusted Risks
Risk Factor
Residence
Portage
Topeka
Watertown
Harriman
St. Louis
Steubenville
Smoking Status
Current
Previous
No high school
education
Body mass index
of 4.5
PM2-5 Data (/ig/m3) Crude Risk
11.0 (1980-7)3 l.O1
12.5 (1980-8) 0.90
14.9(1980-5) 1.16
20.8(1980-7) 1.16
19.0 (1980-6) 1.48
29.6(1980-7) 1.51
All2
1.0
1.01
1.07
1.17
1.14
1.26
1.59
1.20
1.19
1.08
Men2
1.0
1.04
0.94
1.21
1.15
1.29
1.75
1.25
1.22
1.03
Women2
1.0
0.97
1.22
1.07
1.13
1.23
1.54
1.18
1.13
1.11
'Baseline annual crude death rate = 10.73 per thousand population
2Adjusted for age, smoking, education, and body mass
3Period of PM2 5 air monitoring
Source: Dockery et al. (1993)
1 The authors found a marginally statistically significant (p=0.05) (positive) effect of air
2 pollution on male mortality (after age adjustment and consideration of confounding variables
3 including education, density of living conditions, and occupational factors), but a marginally
4 significant (p=0.05) negative effect on female mortality. The most important factors for
5 both sexes (neglecting interactions) were smoking; occupational exposure to heat, humidity,
6 or dust; in-migration; and education. The finding that in-migrants tended to be healthier
7 (especially males) was consistent with the findings of Morris et al. (1976). However, neither
April 1995
12-154 DRAFT-DO NOT QUOTE OR CITE
-------
1 analysis is capable of distinguishing whether the effect is due to longer residence in the
2 polluted area or to selective migration of healthier people.
3 The relative risk for smoking for males was similar to that found by Morris et al.
4 (1976), as was the finding of stronger air pollution effects for males. The strong effects of
5 occupational exposure to heat and dust for both sexes are noteworthy. The effects of
6 education were seen to be about the same as those found in the U.S. by Rogot et al. (1992),
7 for males, but much stronger for females. The Cracow study should be viewed as providing
8 qualified support for the hypothesis that long-term exposure to air pollution is associated with
9 increased mortality in males, although it provides no specific information as to the harmful
10 pollutants.
11
12 California Seventh-Day Adventists.
13 Abbey et al. (1991a) described a prospective study of about 6,000 white, non-Hispanic,
14 nonsmoking, long-term California residents who were followed for 6 to 10 years, beginning
15 in 1976. The study was designed to test the use of cumulative exposure data as an
16 explanatory factor for disease incidence and chronic effects. Ambient air quality data dating
17 back to 1966 were used, and the study was restricted to those who lived within 5 miles of
18 their current residence for at least 10 years. All of the air quality monitors in the state were
19 used to create individual exposure profiles (duration of exposure to specific minimum
20 concentration levels) for each participant, by interpolating to their zip code centroids based
21 on the 3 nearest monitoring stations. Pollutant species were limited to TSP and ozone in this
22 paper; oxidant concentrations were used in the early part of the monitoring record.
23 Endpoints evaluated and the numbers of cases included: newly diagnosed cancers (incidence
24 at any site) for males, 115; any cancer site for females, 175; respiratory cancer, 17; definite
25 myocardial infarction, 62; mortality from any external cause, 845; and respiratory symptoms,
26 272. The Cox proportional hazards model was used, considering age, sex, past smoking,
27 education, and presence of definite symptoms of airway obstructive disease1 (AOD) in 1977
28 as individual risk factors, together with various exposure indices for TSP or ozone
29 Symptoms of asthma, chronic bronchitis, or emphysema.
April 1995 12-155 DRAFT-DO NOT QUOTE OR CITE
-------
1 (considered separately). Data on occupational exposures and history of high blood pressure
2 were available but were not used in the mortality model. No data were available on climate,
3 body mass, income, migration, physical activity levels or diet. Separate results by gender
4 were not reported for nonexternal mortality.
5 Of these endpoints, respiratory symptoms and female cancers (any site) were associated
6 with TSP exposure. Neither heart attacks or nonexternal mortality was associated with
7 either pollutant. The authors felt that possible errors in their estimated exposures to air
8 pollution may have contributed to the lack of significant findings, and a later version of the
9 data base include estimates of attenuation resulting from time spent indoors (Abbey et al.,
10 1993), but mortality was not considered in the 1993 paper.
11 The follow-up analysis (Abbey et al., 1995) considered exposures to SO42"' PM10
12 (estimated from site-specific regressions on TSP), PM2 5 (estimated from visibility), and
13 visibility per se (extinction coefficient). No significant associations with nonexternal
14 mortality were reported, and only high levels of TSP or PM10 were associated with AOD or
15 bronchitis symptoms.
16 This study used an unique air quality data base which was developed for the express
17 purpose of studying the effects of long-term cumulative exposures to community air pollution
18 (Abbey et al., 1991b). The technique was shown to provide reliable spatial interpolations
19 that were somewhat better for O3 than for TSP, in keeping with expectations based on the
20 regional nature of O3. However, no attention was given to temporal matching of air quality
21 and health; the studies using this data base were intended to evaluate the hypothesis that
22 health is affected by cumulative long-term pollution exposure at some undetermined time, as
23 opposed to acute or coincident exposures. Note that the data base began in 1966 and the
24 mortality follow-up began 10 years later. Because air quality generally improved during this
25 period, the highest concentrations are likely to have occurred in the earlier part of the
26 record, and thus we would not expect spatially-based correlations to also reflect the sum of
27 acute effects, as would be the case when air quality and health data are also matched in time.
28 Note that the range of air quality levels experienced in California from 1966 onward is at
29 least as large as that currently experienced in the rest of the United States, including the
30 nation's highest O3 levels, annual average TSP up to about 175 /ig/m3, and annual average
April 1995 12-156 DRAFT-DO NOT QUOTE OR CITE
-------
1 SO42~up to about 9-11 /xg/m3 (Lipfert, 1978). Thus, lack of adequate range in the pollution
2 variables does not appear to be a valid reason for the lack of statistical significance.
3 However, levels of SO2 and of certain trace metals such as Mn tend to be lower in California
4 than in the midwestern parts of the United States with larger concentrations of heavy
5 industry.
6 The finding of Abbey et al. (1991a) of no association between long-term cumulative
7 exposure to TSP or O3 and all natural-cause mortality may be interpreted as showing the
8 absence of chronic responses after 10 years but not necessarily the absence of (integrated)
9 acute responses, since coincident air pollution exposures were not considered. It is also
10 possible that the latency period for chronic effects may exceed 10 years and that additional
11 follow-up might still reveal chronic effects. The magnitudes of the other risk factors
12 considered were not given by Abbey et al. (1991), which precludes comparison with the
13 other studies.
14
15 Prospective Cohort Study in Six U.S. Cities.
16 Dockery et al. (1993) analyzed survival probabilities among 8,111 adults who were first
17 recruited in the mid-1970s in six cities in the eastern portion of the United States. The cities
18 are: Portage, WI, a small town north of Madison; Topeka, KS; a geographically-defined
19 section of St. Louis, MO; Steubenville, OH, an industrial community near the West Virginia-
20 Pennsylvania border; Watertown, MA, a western suburb of Boston; and Kingston-Harriman,
21 TN, two small towns southwest of Knoxville. This selection of locations thus comprises a
22 transect across the Northeastern and Northcentral United States, from suburban Boston,
23 through Appalachia, and into the upper Midwest.
24 The adults were white and aged 25 to 74 at enrollment. In each community, about
25 2,500 adults were selected randomly, but the final cohorts numbered 1,400 to 1,800 persons
26 in each city (Ferris et al., 1979). Follow-up periods ranged from 14 to 16 years, during
27 which from 13 to 22% of the enrollees died. Of the 1,430 death certificates, 98% were
28 located, including those for persons who had moved away and died elsewhere. However, no
29 information was given in the paper about the actual locations of death. The bulk of the
April 1995 12-157 DRAFT-DO NOT QUOTE OR CITE
-------
1 analysis was based on all-cause mortality; no mention was made of subtracting external
2 causes.
3 These cohorts have been studied extensively for respiratory health (Dockery et al.,
4 1985). Air monitoring data were obtained from routine sampling stations and from special
5 instruments set up by the research team. Individual characteristics of the members (and thus
6 of the decedents) considered included smoking habits, an index of occupational exposure,
7 body mass index, and completion of a high school education. The Cox proportional hazards
8 model was used to estimate coefficients for the individual risk factors after stratifying ny sex
9 and age (5-year groups), the effects of air pollution were evaluated in two ways: by
10 evaluating the relative risks of residence in each city relative to Portage (the city with the
11 lowest pollution levels for most indices), and by including the community-average air quality
12 levels directly in the models. Since only six different values were available for each
13 pollutant, the effective degrees of freedom are greatly reduced by this procedure.
14 Most of the air quality measures were averaged over the period of study, in an effort to
15 study long-term (chronic) responses; the specific averaging periods varied by pollutant.
16 Steubenville, Kingston-Harriman, and St. Louis were the most polluted cities and also had
17 the oldest and least educated cohorts and the heaviest rates of smoking among the six cities.
18 The index of smoking rate used in this study was pack-years, defined as the average
19 number of packs of cigarettes smoked per day times the number of years of smoking. This
20 metric is also a function of age. Current and former smokers were treated separately. This
21 smoking metric assumes that health impacts are defined by the cumulative tobacco
22 consumption rather than by the current rate of consumption; the fact that the risk per pack-
23 year was higher for former smokers (0.015 per pack year), compared to current smokers
24 (0.01 per pack year), and the finding of a risk per pack year for current smokers that
25 increased with consumption rate suggest that the current rate of smoking may also have merit
26 as a health impact index (especially if the age of starting smoking varies). The total effect of
27 smoking was thus defined as the relative risk of being a smoker plus the risk associated the
28 number of pack-years in question.
29 The index of socioeconomic status used was having less than a high school education;
30 Rogot et al. (1992) show that this index is a good measure of mortality differences due to
April 1995 12-158 DRAFT-DO NOT QUOTE OR CITE
-------
1 differences in education for white men but not for white women. For women, relative
2 mortality risk continues to increase for educational attainments less than completion of high
3 school. The index of occupational exposure to air pollution (dusts or fumes) did not take
4 into account the length or degree of exposure or the nature of the agents involved.
5 Occupational exposure to dusts or fumes was not found to be a significant risk factor; this
6 outcome may have resulted from the lack of specificity of the index used. The average
7 percentages having occupational exposure were high, ranging from 28 to 53 %, with an
8 average across all cities of 45 %.
9 The index of physiology used was the body mass index (BMI), defined as weight
10 divided by height squared (kg/m2), treated as a linear relationship. The relative risk of
11 increased body mass was similar to that found by Sandvik et al. (1993), where it was not
12 statistically significant, but other investigators have found that the relationship is U-shaped
13 rather than linear and may interact with other risk factors, especially smoking (Gronbaek
14 etal., 1994).
15 No consideration was given to possible independent effects of occupation classification,
16 other personal lifestyle variables such as diet or physical activity, migration, or income.
17 Presumably, each subject was characterized by his status at entry to the study; follow-up data
18 on possible changes in risk factors over time were not mentioned. Since the air quality data
19 used in this study were largely obtained from "private" monitoring rather than from public
20 archives, comparisons of the average levels with routine monitoring data were of some
21 interest. No serious disagreements were found, except that it might have been preferable to
22 consider peak rather than average levels of ozone, as has been done in most of the studies of
23 acute effects of ozone on mortality. However, the size-classified paniculate data began in
24 1980 while TSP data began in 1974; from 1974 to 1980 there were large reductions in TSP
25 (and probably in the size-classified particles as well), so that it appears that the size-classified
26 data are less representative of cumulative exposures than TSP. Sulfate appeared to be
27 intermediate in this regard. In this sense, there is a mismatch in time between the air
28 quality data, which were obtained after the study began, and the descriptive data on
29 individuals, which pertain to the period before entry into the study.
April 1995 12-159 DRAFT-DO NOT QUOTE OR CITE
-------
1 Based on statewide mortality data, substantial differences in survival rates would be
2 expected across this transect of the Northeastern U.S. and were observed (Table 12-17). The
3 long-term average mortality rate in Steubenville was 16.2 deaths per 1,000 person-years; in
4 Topeka, it was 9.7, yielding a range in average (crude) relative risk of 67% among the six
5 cities. After individual adjustment for age, smoking status, education, and body-mass index,
6 the range in average relative risk was reduced to 26%. The relative importance of the
7 adjustments for age, smoking, education, and body mass in determining the final ranks of the
8 cities may be seen from the table. Also, there is more scatter for men and women separately
9 than when combined, presumably because of the reduction in sample size.
10 Dockery et al. (1993) report that "mortality was more strongly associated with the
11 levels of fine, inhalable, and sulfate particles" than with the other pollutants, which they
12 attributed primarily to factors of particle size. They provided relative risk estimates and
13 confidence limits based on the differences between air quality in Steubenville and in Portage
14 for these three pollutants. However, it is relatively simple matter to estimate these
15 coefficients from the adjusted risks and pollutants levels in each of the six communities.
16 Such estimates correspond quite closely to the figures given by Dockery et al. based on
17 output from the Cox proportional hazards model. However, because there are only
18 6 degrees of freedom for the air quality data, the resulting confidence limits are considerably
19 wider than those for the risk factors having individual data. These estimates are given in
20 Table 12-18, as a means of comparing the various pollutants and combination of pollutants.
21 As in the original paper, the relative risks are based on the difference in air pollution
22 between Steubenville and Portage. The data for 1970 TSP (corresponding to a lag of about
23 12 years) were obtained from Lipfert (1978), assuming that Madison could represent
24 Portage, WI.
25 Table 12-18 shows only small differences among many pollutants, including SO2 and
26 NO2, owing in part to the strong collinearity present. Note that TSP and the coarse particle
27 variables created by subtracting PM15 from TSP and PM2 5 from PM15 were not significant,
28 suggesting that particles larger than about 15 /im in aerodynamic diameter may be less
29 important; this outcome may reflect in part greater spatial variability within the communities
30 for these measures. The non sulfate portion of PM2 5 had the tightest confidence limits
April 1995 12-160 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
(SO42" was multiplied by 1.2 before subtraction, assuming an average composition of
NH4HSO4). Note also that the estimated 1970 TSP variable performed better than the TSP
data used by Dockery et al. (ca. 1982). However, all of the differences in relative risks and
their confidence limits could have occurred due to chance, given
the availability
of only
6 observations. Dockery et al. noted that the mean ozone level varied little among cities.
This might not have been the case if some measure of peak concentration had been used
instead of the overall mean (24-hr averages). No relationship was found for aerosol acidity
(H+), but only limited data were available.
TABLE 12-18. ESTIMATED RELATIVE RISKS IN SIX U.S. CITIES
ASSOCIATED WITH A RANGE OF AIR POLLUTANTS
Regr. Standard
Species Coeff. Error Range
PM15 0.0085 (0.0026) 28.3
PM25 0.0127 (0.0034) 18.6
SO42' 0.0297 (0.0081) 8.5
TSP 0.0037 (0.0014) 55.8
TSP-PM15 0.0042 (0.0032) 27.5
PM15-PM25 0.0178 (0.0098) 9.7
PM25-SO4 0.0255 (0.0029) 8.4
PM15-SO4 0.0121 (0.0034) 18.1
SO2 0.0093 (0.0032) 19.8
NO2 0.0126 (0.0046) 15.8
1970 TSP 0.0014 (0.00044) 154.0
In comparing the most and least polluted cities, Dockery et
Rel.
Risk
1.27
1.27
1.29
1.22
1.12
1.19
1.24
1.24
1.20
1.22
1.25
al. also report
for cardiopulmonary causes (1.37, [1.11 to 1.68]) and lung cancer (1.37, [0.81
significant). The relative risk for all other causes of death was 1
.01 (0.79 to 1.
95% CIs
(n=6)
(1.04-1.56)
(1.06-1.51)
(1.06-1.56)
(0.99-1.53)
(0.88-1.43)
(0.91-1.55)
(1.16-1.32)
(1.05-1.48)
(1.01-1.43)
(1.00-1.49)
(1.03-1.50)
elevated risks
to 2.31], not
30). When
the six cities were considered individually, only Steubenville showed a statistically significant
(p < 0.05) elevated risk with respect to the least polluted city (Portage).
April 1995 12-161 DRAFT-DO NOT QUOTE OR CITE
-------
1 Comparison of the pollution risks among the various cohort subsets considered is one of
2 the most important outcomes of a study on individuals. Such comparisons must account for
3 the higher variability among subgroups, however, and the study was not capable of
4 distinguishing excess risks between subgroups less than about 18% (i.e., an excess risk of
5 1.18 cannot be distinguished from one of 1.36, for example). Although none of these
6 subgroup differences were statistically significant, the mortality risks associated with area of
7 residence (and thus air pollution) were higher for females and for smokers and the risks were
8 also higher for those occupationally exposed compared to the nonexposed. Because of
9 reduced uncertainties about their exposure to air pollution not reflected in the outdoor
10 monitoring data used in this study, it is possible that the relative risk estimates for
11 nonsmokers and the nonoccupationally exposed might be the most reliable estimates
12 (1.19 and 1.17, respectively).
13 In correspondence, Moolgavkar (1994) raised issues of residual confounding, age
14 adjustment and smoking controls. In their response, Dockery and Pope (1994) agreed that
15 confounding is a potential concern but did not address the possibility that variables other than
16 the ones they considered might be important. They dealt with the age adjustment issue
17 quantitatively and pointed out that the air pollution risk estimates were reasonably stable over
18 different subgroups by smoking status.
19 The authors of this study appear to have made the most of the available individual data
20 on some of the most important mortality risk factors. They were quite cautious in their
21 conclusions, stating only that the results suggest that fine-particulate air pollution "contributes
22 to excess mortality in certain U.S. cities." There are several other important outcomes:
23 • None of the population subgroups examined appeared to be significantly more
24 sensitive to air pollution than any other. Since the relative risks were virtually
25 unchanged by excluding subjects with hypertension and diabetes, this finding might
26 also be extended to those with pre- existing chronic diseases. This apparent
27 homogeneity of response has implications regarding the acceptability of population-
28 based studies in which such stratification is not possible.
29
30 • The implied regression coefficients are much larger (about an order of magnitude)
31 than those found in either type of cross-sectional study. This could be interpreted as
32 evidence that the chronic effects of air pollution far exceed the acute effects, or that
33 not all of the spatial confounding has been controlled. Use of linear models for non-
34 linear effects (body-mass index) and failure to control for alcohol consumption, diet,
April 1995 12-162 DRAFT-DO NOT QUOTE OR CITE
-------
1 exercise and migration may have contributed to the relatively large effects indicated
2 for air pollution.
3
4 • If the responses to air pollution truly are chronic in nature, it is logical to expect that
5 cumulative exposure would be the preferred metric (Abbey et al., 1991). Pollution
6 levels 10 years before this study began were much higher in Steubenville and St.
7 Louis, as indexed by TSP from routine monitoring networks. Estimates of previous
8 levels of fine particles are more difficult, but atmospheric visibility data suggest that
9 previous levels may have been higher in winter, but not necessarily in summer.
10 These uncertainties make it difficult to accept quantitative regression results based
11 solely on coincident monitoring data. For example, annual average TSP in 1965 in
12 Steubenville was about three times the value used by Dockery et al.
13
14 Because it seems unlikely that any of the perceived shortcomings of this study could have
15 resulted in bias sufficient to reduce the risk estimates to levels less than those found in acute
16 mortality studies, the study of Dockery et al. (1993) appears to provide support for the
17 hypothesis that the results of long-term air pollution studies must also reflect the presence of
18 acute effects on mortality as integrated over the long term, as suggested by Evans et al.
19 (1984a). It may also be concluded that support has been shown for the existence of chronic
20 effects; these two possibilities are not mutually exclusive. However, these conclusions must
21 be qualified by the realization that not all of the relevant socioeconomic factors may have
22 been properly controlled in this study.
23
24 American Cancer Society Study
25 Pope et al. (1995) analyzed 7-year survival data (1982 to 1989) for about 550,000 adult
26 volunteers obtained by the American Cancer Society (ACS). The Cox proportional hazards
27 model was used to define individual risk factors for age, sex, race, smoking (including
28 passive smoke exposure), occupational exposure, alcohol consumption, education, and body-
29 mass index. The deaths, about 39,000 in all, were assigned to geographic locations using the
30 3-digit zip codes listed at enrollment into the ACS study in 1982. Relative risks were then
31 computed for 151 metropolitan areas defined by these zip codes and were compared to the
32 corresponding air quality data, ca. 1980. The sources of air quality data used were the EPA
33 AIRS system for sulfates, as obtained from high-volume sampler filters for 1980, and the
34 Inhalable Particulate Network for fine particles (PM2 5). The latter data were obtained from
April 1995 12-163 DRAFT-DO NOT QUOTE OR CITE
-------
1 dichotomous samplers during 1979-81; Pope et al. used the values from this data base
2 reported by Lipfert et al., 1988 (this study is discussed above), but only 50 PM25 locations
3 could be matched with the death data. The correlation between the two pollutants was 0.73.
4 The sulfate values from the inhalable particle filters, which are thought to be free from
5 artifacts, were not used in this study. Causes of death considered included all causes,
6 cardiopulmonary causes (ICD-9 401-440, 460-519), lung cancer (ICD-9 162), and all other
7 causes.
8 This study took great care with the potential confounding factors for which data were
9 available. Several different measures of active smoking were considered, as was the time
10 exposed to passive smoke. The occupational exposure variable was specific to (any of)
11 asbestos, chemicals/solvents, coal or stone dusts, coal tar/pitch/asphalt, diesel exhaust, or
12 formaldehyde. The education variable was an indicator for having less than a high-school
13 education. However, alcohol use and body-mass index were considered as linear predictors
14 of survival, whereas other studies have indicated these effects to be non-linear (U or
15 J-shaped) (Doll et al., 1994; Gronbaek et al., 1994). Pope et al. (1995) did not report the
16 relative risk coefficients they obtained for these cofactors, which is unfortunate since
17 correspondence of findings for the non- pollution variables with exogenous estimates from
18 independent studies adds to the confidence in the results for the pollution variables.
19 Risk factors not considered by Pope et al. (1995) include income, employment status,
20 dietary factors, drinking water hardness and physical activity levels, all of which have been
21 shown to affect longevity (Sorlie and Rogot, 1990; Belloc, 1973; Pocock et al., 1980). In
22 addition, they did not discuss the possible influences of other air pollutants. For example,
23 Lipfert et al. (1988) found that it was not possible to separate the effects of SO2, SO42"- and
24 NOX from one another, and Lipfert (1992) found some evidence for the effects of ozone in
25 cross-sectional mortality regressions for U.S. metropolitan areas in addition to associations
26 between TSP and all-disease and COPD mortality.
27 The ACS cohort is by no means a random sample of the U.S. population; it is 94%
28 white and better educated than the general public, with a lower percentage of smokers than in
29 the Six City Study. The (crude) death rate during the 7.25 years of follow-up was just under
30 1 % per year, which is about 20% lower than expected for the white population of the
April 1995 12-164 DRAFT-DO NOT QUOTE OR CITE
-------
1 U.S. in 1985, at the average age reported by Pope et al. In contrast, the corresponding rates
2 for the Six- Cities study (Dockery et al., 1993) discussed above tended to be higher than the
3 U.S. average.
4 No mention was made of residence histories for the decedents; matching was done on
5 residence location at entry to the study. The 1979 to 1981 pollution values were assumed to
6 be representative of long-term cumulative exposures, in keeping with the objective of
7 analyzing chronic effects. However, the previous decade was one of extensive pollution
8 cleanup in most of the nation's dirtiest cities (TSP dropped by a factor of 2 in New York
9 City, for example [Ferrand, 1978]). In contrast, air quality would have remained relatively
10 constant in cities that already met the standards. Thus, it is reasonable to expect that the
11 contrast between "clean" and "dirty" cities would have been greater in 1970 than in 1980.2
12 If the excess mortality found in this study were in fact due to cumulative exposures, the
13 regression coefficients would have been biased upward (in terms of relative risk per jig/m3)
14 by using the more recent data. The typically long latency period for lung cancer (ca. 20 yr.)
15 suggests that data on prior exposures may be particularly important for this cause of death.
16 The adjusted total mortality risk ratios (computed for the range of the pollution
17 variables) were 1.15 (95% CL = 1.09 to 1.22) for sulfates and 1.17 (95% CL = 1.09 to
18 1.26) for PM2 5. When expressed as log-linear regression coefficients, these values were
19 quite similar for both pollution measures: 0.0070 (0.0014) per /ig/m3 for SO42' and 0.0064
20 (0.0015) for PM2 5, suggesting that particle chemistry may be relatively unimportant as an
21 independent risk factor (it is possible that the SO42" results have been biased high by the
22 presence of filter artifacts). Pope et al. (1995) found that the pollution coefficients were
23 reduced by 10 to 15% when variables for climate extremes were added to the model.
24 Expressed as the percentage of mortality associated with air pollution at the mean values and
25 corrected for filter artifact for SO42" using the data of Lipfert (1994), this study found mean
26 2For example, the ranges of TSP and SO42~ across the U.S. in 1970 were from 40 to 224
27 and from 3 to 28 /xg/m3, respectively (Lipfert, 1978). In 1980, these ranges decreased to 41-142
28 and 2-17 eeg/m3 (Lipfert, 1993), which suggests that the dirtiest cities became cleaner while the
29 "clean" cities stayed about the same. The change in pollution range is about a factor of 1.8.
April 1995 12-165 DRAFT-DO NOT QUOTE OR CITE
-------
1 effects of about 5% for sulfate and 12% for PM2 5. No significant excess mortality for the
2 "other" causes of death was attributed to air pollution in this study.
3 Pope et al. (1995) found very consistent pollution risks for males and females and for
4 ever-smokers and never-smokers for all-cause mortality. However, the relative risks for air
5 pollution were slightly higher for females for cardiopulmonary causes of death. The lung
6 cancer- sulfate association was only significant for males, except for male never-smokers.
7 The ACS study is unique in having attempted to control for passive smoking exposure.
8 It is unfortunate that passive smoking results were not reported and compared with the air
9 pollution risks.
10 The results of the American Cancer Society prospective study were qualitatively
11 consistent with those of the Six City study with regard to their findings for sulfates and fine
12 particles; relative standard errors were smaller, as expected because of the substantially
13 larger database. However, no other pollutants were investigated in the ACS analysis, so that
14 no further progress was made in attempting to identify the "responsible" pollutants. In
15 addition, the ACS regression coefficients were about 1/4 to 1/2 of the corresponding Six City
16 values and were much closer to the corresponding values obtained in various acute mortality
17 studies. Thus it is not clear to what extent chronic effects (as opposed to integrated acute
18 effects) are indicated by these results and to what extent the limited air quality data base used
19 was responsible for this outcome.
20
21 Summary and Conclusions from Prospective Studies
22 Table 12-19 summarizes the three newer prospective studies considered here. The
23 California and Six-City studies suffer from small sample sizes and inadequate degrees of
24 freedom, which partially offsets the specificity gained by considering individuals instead of
25 population groups. All of them may have neglected some important risk factors. The two
26 early studies not shown in this table were largely inconclusive and the studies of California
27 nonsmokers by Abbey et al. (1991, 1994) that had the best cumulative exposure estimates
28 found no significant mortality effects of previous air pollution exposure. The Six Cities and
29 ACS studies agree in their findings of strong associations between fine particles and excess
April 1995 12_i66 DRAFT-DO NOT QUOTE OR CITE
-------
H
g
£
^
5
i
s
o
ffi
o
u
r^
>
^j
1
C/3
O
C*
PUl
1— 1
s
U
H
N^J
^f
H
ji
u
u
Ji
11
a
« - 12 ° ^
> rt u
•a _, u "
to ^ II r,
"S .2 t 2 IN
M £ i
S2
I
U.
o
°
*1
cu
•3 1
si
S.
I
VI
® u
«5 . •>v
3 w &
^ ft- b
"s ^
S o
OH ^-^
g^
§ »S *w»
^
Q
2.1
3
-_ %
« ^ '3
^l|
1
3
3
Q.
£
•s 1
"8 §
nC J?
§
i/5
fS
I/}
M
2
oo
g.
'ft.
CO
H
o
u" S
*• G ••*
X gl o "O
00 C "O --^
rt «i o> rt
I
i2
X
0
Cox
proportional
hazards
i
^
z
'eo
•r. "3
t~j 3
S S
8
PL.
S 8
^ A
00 C O
M*l
-H Q « E
r* 'S
c
^ n>
oil
_g.
a 1 E 1
o 2 p 2
H E J3 -o
^-v
" 5
*!
^
< -a
Wl
1«
i 1 1 1 1
.§•
a 2
® §
H E
u ^
^* SJ °°
Q -a' o.
= i
d d
VO VO
» M?
C. C-
"i
t- o
-< «
§
» wT
x" op § S 2
S | | « d S 1
^llllll
i
I
1
ll
all
o" oo"
— —
f'l fS
12-167 DRAFT-DO NOT QUOTE OR CITE
-------
1 mortality, but it is unfortunate that the ACS study did not consider a wider range of
2 pollutants so as to also confirm the lack of importance for other pollutants. In addition, the
3 timing of the critical exposures remains an open question. It is also important that a range of
4 pollutants be considered in both chronic and acute studies, since it is possible that acute
5 effects may be exhibited by one pollutant and chronic effects by another.
6
7 12.4.1.4 Summary of Long-Term Studies
8 Previous Summaries of Cross-Sectional Studies
9 There have been many previous reviews and summaries of air pollution-mortality
10 studies, including those by Ricci and Wyzga, 1983; Lipfert, 1977, 1978, 1980, 1985;
11 International Electric Research Exchange, 1981; Evans et al., 1984b; Lave and Seskin, 1970;
12 Landau, 1978, Cooper and Hamilton, 1979; Thibodeau et al., 1980; Ware et al., 1981.
13 However, few of these adequately considered the effects of errors in estimating air pollution
14 exposures, and there are now some new results that were not available to previous reviewers.
15 For example, Thibodeau et al. (1980) reanalyzed the Lave and Seskin data set (1978) using
16 the same basic data (including the flawed "minimum" SO4= data) and variables, but
17 correcting for errors in the data and adding some statistical techniques. This type of
18 reanalysis provides confidence in the results of the original authors with respect to the
19 absence of numerical errors but cannot provide insight into robustness with respect to choices
20 of variables and observations, which can be a more important issue.
21 With respect to cross-sectional studies, Ware et al. (1981) concluded that "...The model
22 can only be approximately correct, the surrogate explanatory variables can never lead to an
23 adequate adjusted analysis, and it is impossible to separate associations of mortality rate with
24 pollutant and confounding variables. This group of studies, in our opinion, provides no
25 reliable evidence for assessing the health effects of sulfur dioxide and particulates...."
26 However, they did not support this conclusion by attempting to explain the important
27 differences among study results. Obviously, Ware et al. (1981) did not have access to the
28 recent large prospective cohort studies nor the more recent population-based cross-sectional
29 studies, which should allow some of these issues to be addressed and some of the differences
30 in study results to be explained.
April 1995 12-168 DRAFT-DO NOT QUOTE OR CITE
-------
1 Evans et al. (1984) concluded that "...important questions related to the interpretation
2 of cross-sectional studies remain unanswered. For example, it is unclear which specific
3 pollutants are responsible for any observed effect; the shape of the exposure-response
4 relationship is unresolved; and it is possible that the observed effects are due, in part, to
5 confounding or systematic misclassification...." Nevertheless, the authors were of the opinion
6 that "...the cross- sectional studies reflect a causal relationship between exposure to airborne
7 particles and premature mortality...." Several observers have arrived at this conclusion with
8 respect to the acute mortality studies; it is thus important to consider the entire body of air
9 pollution-mortality studies.
10
11 Comparison of Prospective and Population-based Cross-Sectional Study Results
12 The literature on long-term health effects of air pollution has been substantially
13 enriched by the publication of the recent prospective studies. Their ability to stratify by
14 smoking habit or occupational exposure provides valuable information not previously
15 available. These studies also provide a basis with which to evaluate the reasonableness of
16 the "ecologic assumptions" that are required in order to interpret ecological studies. In this
17 section, we consider the two types of studies on an equal footing, following the admonition
18 of Greenland and Robins (1994b) that ecological studies should not be discounted just
19 because they are ecological.
20 Table 12-20 compares regression coefficients from the two prospective studies that
21 reported significant pollution risks with corresponding estimates made on an "ecologic" basis,
22 i.e., using SMSA-wide mortality rates. Pope et al. (1995) introduced this concept by
23 comparing age-race-sex-adjusted SMS A mortality rates with their prospective findings, but
24 without adjusting the SMSA-wide values for cofactors such as smoking or education. They
25 noted the similarity in relative risk estimates between their prospective study findings and the
26 SMSA-wide "ecologic" estimates, but they did not discuss whether the risks predicted by
27 ecological studies would drop substantially if the equivalent confounding variables had been
28 considered in both types of studies. Table 12-20 also makes this comparison and goes on to
29 show how the ecologic estimates of the pollution effects diminish and become negative
30 and/or non-significant as additional cofactors are entered into the regression model. Each of
April 1995 12-169 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-20. COMPARISON OF LOG-LINEAR REGRESSION COEFFICIENTS
FROM PROSPECTIVE AND "ECOLOGIC" ANALYSES FOR U.S. METROPOLITAN
AREAS
Factors Accounted For
SO/' coeff. (SE)
FP coeff. (SE)
A. Prospective studies
1. Dockery etal. (1993)
age, sex, active smoking,
body mass, education.
2. Pope et al. (1995)
age, sex, race, active & passive3
smoking, education3, body mass3,
alcohol3, occupational exposure3
B. "Ecologic" regressions'1
1. age, race
2. age, race, smoking
3. age, race, smoking,
education
4. age, race, smoking, education,
migration
5. age, rage, smoking, education,
migration, drinking water hardness
(n=6)
0.0308 (0.011)
(n=151)
0.007 (0.0014)
(n=149)
0.0092 (0.0019)
0.0040 (0.00083)
0.0058 (0.00195)
-0.00044 (0.0021)
-0.00055 (0.0021)
(n=6)
0.0124 (0.005)
(n=50)
0.0064 (0.0015)
(n=63)
0.0048 (0.0019)
0.0048 (0.00195)
0.0018 (0.00195)
0.00012 (0.0016)
0.00035 (0.0016)
Bold factors are significant (p < 0.05).
3Significance of cofactors not stated.
bData from Lipfert, 1993.
1 these factors has been shown (by others) to exert an influence on health, and all of them
2 were significant in the ecologic model except drinking water hardness (for which t=1.6).
3 This comparison suggests that the mortality risks assigned to air pollution by the prospective
4 studies may have diminished had individual data on additional risk factors been available and
5 included in the analysis.
6 It is also interesting that introduction of the smoking variable (statewide cigarette sales)
7 into the ecologic regressions had little or no effect on the pollution coefficients, whereas the
April 1995
12-170 DRAFT-DO NOT QUOTE OR CITE
-------
1 other variables had relatively large effects (the correlation between this smoking variable and
2 SO42" was only 0.15). The relative risk corresponding to the ecologic smoking risk
3 coefficient was somewhat less than those found by the prospective studies, probably because
4 this variable is a poor surrogate for individual smoking rates.
5 Figures 12-7 to 12-9 were prepared to illustrate the overlapping confidence intervals of
6 the various studies using mortality data ca. 1980 and later. For SO42"' Figure 12-7, the two
7 prospective studies and Ozkaynak and Thurston's (1987) ecological study overlap, mainly
8 because of the very wide confidence limits of the Six City Study. However, all of these
9 studies accounted for a somewhat limited range of potential confounding variables; the 1980
10 SMSA study by Lipfert (1993) found SO42" to lose significance when additional variables
11 were entered into the model. More overlap is shown for PM2 5 (Figure 12-8), even though
12 significance was not achieved with either ecological study. Overlapping confidence intervals
13 are also seen with TSP (Figure 12-9), including the California prospective study. These
14 plots thus suggest that much of the apparent contrast among studies could be due to chance
15 variation.
16 The important contribution of the prospective studies is the proper accounting for
17 individual risk factors, mainly smoking. The question thus arises, could inadequate control
18 for smoking in an ecological study lead to an underestimate of the air pollution relationship?
19 This would require a negative correlation between smoking and air pollution. However,
20 based on state-level data, the correlations between smoking and both SO42" and PM2 5 are
21 weakly positive. Thus it does not appear that inadequate control for smoking explains the
22 difference in results. One is thus led to the conclusion that either some other factor is
23 negatively correlated with air pollution or that the prospective studies are affected by some
24 confounder that is more important at the individual level than at the community-average
25 level. Of course, much of the range in results seen in these plots could also be due to
26 chance.
April 1995 12-171 DRAFT-DO NOT QUOTE OR CITE
-------
149SMSAS
98SMSAS
151 cities, prospective
6 cities, prospective
0
0.5
1 1.5
Relative risk
2
2.5
Figure 12-7. Comparison of relative risks of exposure to 15 /tg/m3 of SO4, as
estimated by Dockery et al., 1993 (6-cities, prospective), Pope et al.,
1995 (151 cities, prospective), Ozkaynak and Thurston, 1987 (98
SMSAs, ecological), and Lipfert, 1993 (149 SMS As, ecological).
April 1995
12-172 DRAFT-DO NOT QUOTE OR CITE
-------
62 SMSAs
38SMSAS
50 cities, prospective
6 cities, prospective
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Relative risk
Figure 12-8. Comparison of relative risks of exposure to 25 /tg/m3 of PM2 5, as
estimated by Dockery et al., 1993 (6-cities, prospective), Pope et al.,
1995 (50 cities, prospective), Ozkaynak and Thurston, 1987 (38
SMSAs, ecological), and Lipfert, 1993 (62 SMSAs, ecological).
April 1995
12-173 DRAFT-DO NOT QUOTE OR CITE
-------
149SMSAs,nolag
6-cities,10-yrlag
6 cities, no lag
California, 10-yr lag
0
0,5
1 1.5
Relative risk
2
2.5
Figure 12-9. Comparison of relative risks of exposure to 100 jtg/m3 of TSP, as
estimated by Abbey et al., 1991 (California), and as estimated from
the data of Dockery et al., 1993, with and without a 10-year lag (6-
cities, prospective), and Lipfert, 1993 (149 SMSAs, ecological).
April 1995
12-174 DRAFT-DO NOT QUOTE OR CITE
-------
1 Concluding Discussion
2 Referring back to the original goals of long-term mortality studies, several questions
3 appear worthy of reconsideration:
4 1. Have potentially important confounding variables been omitted? The generic
5 criticisms of (previous) cross-sectional studies should be revisited in this context.
6 The effects of selective migration (i.e., population loss), lifestyle (diet, physical
7 fitness, etc.), drinking water hardness, and income were not investigated in any of
8 the prospective studies. The analysis protocols used by Abbey et al. (1991) and by
9 Pope et al. (1995) included possible effects of passive smoking, but results were not
10 reported for this risk factor. In view of the sensitivity of other cross-sectional
11 (population- based) studies to some of these factors, the conclusion follows that the
12 potential for biased results exists in all of the cross-sectional studies.
13
14 2. Can the most important pollutant species be identified? Abbey et al. found no
15 significance for any pollutant for all-disease mortality. In the Six City Study,
16 "importance" must be identified in terms of relative fit between residual risk and air
17 pollution across the six cities, since the aggregate risk level is the same for all
18 species. Unfortunately, with only six observations, the air pollution confidence
19 limits are quite wide and it is difficult to discriminate among the various
20 possibilities. In addition, the pollutants that seem to fit best have the shortest
21 periods of monitoring data. Comparisons between pollutants in the ACS study were
22 limited since only two pollutants were considered (and these were highly correlated)
23 and because of the inherent differences in reliability of the different data bases used
24 for each pollutant. Thus, although support exists for fine particles or sulfates as
25 important mortality predictors, other pollutants cannot be ruled out.
26
27 3. Can the studies discriminate between the effects of current and of cumulative
28 exposures, i.e.. between truly chronic effects and the integral of acute responses
29 over time? Abbey et al. only examined previous exposures, while the Six City
30 Study looked mainly at coincident exposures and the periods of record were shortest
31 for the pollutants that seem to fit the best (PM2 5, PM15, and SC^2"). The ACS
32 study was even more limited, since only one year of sulfate data was used and the
33 fine particle data were limited to a few years and a subset of locations. This
34 question must thus be considered as still open. Since the acute effects of air
35 pollution seem to be manifested much more strongly on the elderly, it might be
36 useful to stratify the prospective cohorts by age. The chronic effects of cancer and
37 heart disease might show up at earlier ages, leading to a clearer distinction between
38 the two types of responses.
39 The overarching conclusion from long-term mortality studies is that they may be easier to
40 carry out than they are to interpret. To some extent, the results are still somewhat
41 investigator-dependent, although a high degree of support has been shown for the hypothesis
April 1995 12-175 DRAFT-DO NOT QUOTE OR CITE
-------
1 that the result from a cross-sectional study for a given year reflects at least the sum of the
2 acute effects experienced during that year. However, one must also conclude that support
3 for the finding of truly chronic effects from cumulative exposures cannot be excluded.
4 In the 1982 OAQPS Staff Paper (U.S. Environmental Protection Agency, 1982b), the
5 use of cross-sectional analyses in setting standards was discouraged, based on the
6 admonitions of Ware et al. (1981) that, "these models can only be approximately correct, the
7 surrogate explanatory variables can never lead to an adequately adjusted analysis, and
8 separating associations between mortality rates and pollutant and confounding variables is
9 impossible. Thus these studies deal with unknown levels of exposure of ill-defined groups of
10 individuals to unspecified pollutants for unstated periods of time and fail to control for many
11 variables known to affect health status." Given the large number of studies that have
12 appeared in the ensuing years, this rather harsh indictment should be revisited to determine
13 what progress has been made (most of these criticisms also apply to time-series studies,
14 which should be examined in the same light).
15 "Models can only be approximately correct."
16 This is true for all of the studies reviewed above, including the prospective studies.
17 However, the most relevant question here is to what extent the estimates of pollution
18 effects have been confounded by model shortcomings.
19
20 "Surrogate explanatory variables."
21 Surrogates are heavily used in the population-based studies and less so in the
22 prospective studies, for which some individual-level data were available. This question
23 may be addressed in part in population-based studies by comparing the effects obtained
24 for a given factor (like age, race, or education) with those obtained at the individual
25 level.
26
27 "Separating collinear associations is impossible."
28 On its face, this is an overstatement for population-based cross-sectional studies,
29 while it may be more applicable to time-series studies or to prospective studies
30 involving only a few locations. For example, using 1980 data for cities (Lipfert et
31 al., 1988), SO42" was the pollutant most associated with other (nonpollutant) factors,
32 and the highest such correlation was only about 0.5. Among pollutants, however,
33 correlations tend to be higher: SO42" vs PM2 5, 0.76; PM15 vs. PM2 5, 0.61; Mn vs.
34 Fe, 0.78; PM15 vs. Fe, 0.69.
35
36 "Unknown exposures for unstated periods of time."
37 Some of the prospective studies (e.g., Abbey et al., 1991) have defined specific
38 chronic exposure levels; they were not found to be statistically significant in that
April 1995 12-176 DRAFT-DO NOT QUOTE OR CITE
-------
1 instance. Other studies have been more vague, and population-based studies in
2 general have not found important differences between concurrent and lagged
3 pollution data (see for example, Kim, 1985). Lipfert (1993) found that the TSP-
4 mortality relationship was better defined when concentrations were averaged across
5 SMSAs for the same year as the mortality data and when a migration variable was
6 included. This supports the hypothesis that the integrated sum of acute responses is
7 reflected in the long-term response.
8
9 "Ill-defined groups of individuals."
10 This deficiency was addressed directly by prospective cohort studies. Results were
11 found to be consistent for various subsets of individuals defined by smoking and
12 occupational exposures. However, population-based studies tend to find somewhat
13 unstable results when subsets defined by age and gender are analyzed (Lipfert,
14 1984).
15
16 Failure to control for known confounders.
17 All of the studies reviewed are subject to this criticism, but to varying degrees.
18 None controlled for lifestyle differences such as diet or exercise. Most of the
19 studies ignored the effects of population migration and some of the population- based
20 studies did not control for smoking. However, the extent to which these variables
21 actually confound varies with the situation and absolute levels of bivariate correlation
22 are not always a good indicator. This arises in part from the general tendency for
23 mortality to be more weakly associated with air pollution than with many of the
24 potential confounders.
25
26 Thus, it appears that, as with most epidemiology, consistency among studies of widely
27 varying design must be sought in order to respond to the shortcomings that were noted
28 earlier, since different designs have different strengths and weaknesses. Among the long-
29 term exposure studies, it is important to find consistency in terms of geographic scale, time
30 periods, pollutant levels, and regional locations. It will also be important to contrast the
31 findings from short- and long- term exposures and to examine coherence among various
32 health endpoints.
33
34 Summary of Conclusions
35 Cross-sectional studies may find a significant association between mortality and a
36 specific air pollutant for any of several reasons:
37 • The association may reflect a non-zero integral of the acute effects of that pollutant
38 over the period of study.
39
April 1995 12-177 DRAFT-DO NOT QUOTE OR CITE
-------
1 • The association may reflect a chronic effect from previous long-term exposures.
2
3 • The association may have resulted from confounding, either with another pollutant,
4 with the characteristics of the sources that produced that pollutant (occupational
5 hazards or exposures), or with human elements spatially associated with pollution
6 sources such as differential migration of the healthy, less desirable housing near
7 sources, or other socioeconomic factors.
8
9 The studies reviewed above probably all reflect some varying combinations of these
10 possibilities. It is possible that a given regression coefficient may reflect the acute effects of
11 one pollutant and the collinear portion of the chronic effects of another, for example.
12 Convincing evidence of causal associations requires demonstration of specific disease-
13 pollution combinations that are physiologically plausible. Supporting evidence may be
14 obtained by showing that the statistical relationship improves when more reliable exposure
15 data are used in the analysis.
16 Some of the prospective studies demonstrated that additional pollutant exposures
17 (cigarette smoke, occupational exposure) not reflected in the outdoor measurements lead to a
18 stronger statistical mortality relationship with the outdoor measurements. This suggests two
19 possibilities (there may be others):
20 • The indoor and outdoor exposures may reinforce each other and thus must have
21 similar physiological effects. This may provide some clues as to the most likely of
22 several collinear outdoor pollutants. The responses could be either chronic or acute.
23
24 • The indoor or occupational exposures may have created a disease state (independent
25 of the outdoor exposures) that makes the individual more susceptible to outdoor
26 pollution effects. This hypothesis suggests that the outdoor effects in the prospective
27 studies are acute, since it is unlikely that normally healthy people will experience
28 acute effects at the air pollution levels now seen in U.S. cities.
29 Distinguishing between these two scenarios will likely require additional research, probably
30 including temporal studies of long-term changes in air quality in different places.
31 At this time, the long-term studies provide support for the existence of short-term
32 increase in mortality which are not subsequently canceled by decreases below normal rates.
33 However, they do not exclude the existence of chronic effects. They provide no convincing
34 evidence as to the specific pollutant(s) involved, and they do not rule out the existence of
35 pollutant thresholds.
April 1995 12-178 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.4.2 Morbidity Effects of Long-Term PM Exposure
2 Acute exposures to PM are associated with increased reporting of respiratory symptoms
3 and with small decrements in several measures of lung function (Section 12.3.2.3). As a
4 consequence, cross-sectional studies of the relationship between long-term exposure to PM
5 (or any air pollutant) and consequent chronic effects on respiratory function and/or
6 respiratory symptoms may be limited by the inability to control for effects of recent
7 exposures on function and symptoms. Moreover, such studies are further handicapped by:
8 (1) limited or no ability to characterize accurately lifetime exposure to PM other than through
9 "area-based" ecological assignments or assignments inferred from short-term, acute
10 measurements; and (2) their inherent limited ability to characterize correctly other relevant
11 exposure histories (e.g., past histories of respiratory illnesses, passive exposure to tobacco
12 smoke products, active smoking in older subjects).
13 Longitudinal studies offer a number of obvious advantages over cross-sectional studies
14 in terms of PM exposure characterization and characterization of relevant covariates.
15 Nonetheless, to the extent to which such studies try to base their inference with regard to the
16 occurrence of long-term morbidity on the effects observed over relatively short durations of
17 cohort follow-up (e.g., incident respiratory illness in relationship to ambient PM, short-term
18 relationship between ambient PM and lung function), their result need to be accepted with
19 circumspection. These approaches do not establish long-term effects but only suggest the
20 coherence of the possibility of such long-term effects. The optimal longitudinal studies
21 would provide data on incident chronic conditions such as physician diagnosed asthma and/or
22 evidence for altered patterns of lung function growth and decline for children and adults,
23 respectively.
24
25 12.4.2.1 Respiratory Illness Studies
26 Studies of Children
27 The 1982 Criteria Document (U.S. Environmental Protection Agency, 1982a) indicated
28 that apparent quantitative relationships between air pollution and lower respiratory tract
29 illness in children were reported by Lunn et al. (1967), who studied respiratory illnesses in
30 5- and 6-year old school children living in four areas of Sheffield, England. Positive
April 1995 12-179 DRAFT-DO NOT QUOTE OR CITE
-------
1 associations were found between air pollution concentrations and both upper and lower
2 respiratory illness. Lower respiratory illness was 33 to 56% more frequent in the higher
3 pollution areas than in the low-pollution area (p < 0.005). Also, decrements in lung
4 function, measured by spirometry tests, were closely associated with respiratory disease
5 symptom rates. Lunn et al. (1970) also reported results for 11-year-old children studied in
6 1963 to 1964 that were similar to those found earlier for the younger group. These Lunn
7 et al. (1967, 1970) findings have been widely accepted (World Health Organization, 1979;
8 Holland et al., 1979; U.S. Environmental Protection Agency, 1982a,b) as valid. On the
9 basis of the results reported, it appears that increased frequency of lower respiratory
10 symptoms and decreased lung function in children may occur with long-term exposures to
11 annual BS levels in the range of 230 to 301 /*g/m3 and SO2 levels of 181 to 275 jig/m3.
12 However, these are only very approximate observed-effect levels because of uncertainties
13 associated with estimating PM mass based on BS readings. Also, it could not then be
14 concluded, based on the 1968 follow-up study, that no-effect levels were demonstrated for
15 BS levels in the range of 48 to 169 /*g/m3 because of: (1) the likely insufficient power of the
16 study to have detected small changes given the size of the population cohorts studied, and
17 (2) the lack of site-specific calibration of the BS mass readings at the time of the later (1968)
18 study. In summary, the Lunn et al. (1967) study provided the clearest evidence cited in the
19 1982 EPA criteria document (U.S. Environmental Protection Agency, 1982a) for
20 associations between both significant pulmonary function decrements and increased
21 respiratory disease illnesses in children and chronic exposure to specific ambient air levels of
22 PM and SO2.
23 Dockery et al. (1989) studied respiratory symptoms in children as part of the Harvard
24 Six Cities Study. The cities included Watertown, MA; St. Louis, MO; Portage, WI;
25 Kingston-Harriman, TN; Steubenville, OH; and Topeka, KS. A cross-sectional survey was
26 done in 1980 to 1981. The survey included questions on the presence of bronchitis, chronic
27 cough, chest illness, persistent wheeze and asthma. The analysis was restricted to 5,422 10
28 to 12-year-old white children. A centrally located air monitoring station was established in
29 each community starting in 1974 measuring TSP, SO2, NO2, and ozone. Starting in 1978
30 dichotomous samplers were used to measure PM15. Multiple logistic regression analyses
April 1995 12-180 DRAFT-DO NOT QUOTE OR CITE
-------
1 were performed on each health endpoint. The estimated relative odds of bronchitis from
2 comparing the most polluted community to the least was 2.5 (1.1 to 6.1). This corresponded
3 to a 38.7 /ig/m3 increase in the PM15 level. For chronic cough the odds ratio was 3.7 (1.0
4 to 13.5) and for chest illness it was 2.3 (0.8 to 6.7). The odds ratios corresponding to the
5 other pollutants including TSP, PM2 5, sulfate fraction, SO2, NO2, and O3 were not
6 significant, although they all were greater than 1.
7 Ware et al. (1986) had previously studied respiratory illness and symptoms in children
8 in the same six communities as Dockery et al. (1989). The earlier survey included questions
9 on the presence of bronchitis, chronic cough, chest illness, persistent wheeze and asthma.
10 The analysis was restricted to white children enrolled during one of the first three visits to
11 each city who were six through nine years of age. At least one centrally located air
12 monitoring station was established in each community starting in 1974 measuring TSP, SO2,
13 water soluble sulfate, NO2, and ozone. The cities of St. Louis, Steubenville and
14 Kingston-Harriman were divided into two regions based on exposure. Multiple logistic
15 regression coefficients were significant for cough, bronchitis, and lower respiratory illness
16 for both TSP and water soluble sulfate. The between city coefficients for TSP (A*g/m3) were
17 .0101 (.0018) for cough, .0103 (.0046) for bronchitis, and .0076 (.0035) for lower
18 respiratory illness. TSP coefficients for within city analyses tended to be negative.
19 Neas et al. (1994) analyzed a cohort of white children aged 7 to 11 from the same Six
20 Cities Study. Respiratory illness history and other background information was collected on
21 a parent-completed questionnaire administered between September, 1983 and June, 1986.
22 A stratified one-third random sample of the questionnaire respondents (300 to 350 households
23 per city) was invited to participate in a program of indoor air quality measurements. Indoor
24 air quality was measured in two consecutive 1-week sampling periods in both winter and
25 summer. Measurements included respirable particulates (PM2 5) and nitrogen dioxide.
26 Health endpoints reported on the questionnaire included shortness of breath, persistent
27 wheeze, chronic cough, bronchitis, asthma, hayfever, earache, and chest illness. Odds ratios
28 (OR) were calculated using multiple logistic regression for an increase of 30 jwg/m3 in PM2 5,
29 after adjusting for gender, age, parental education, parental history of asthma, and city.
30 Most of the health endpoints showed little effect from PM2 5 except for bronchitis
April 1995 12-181 DRAFT-DO NOT QUOTE OR CITE
-------
1 (OR = 1.18, CI = 0.99, 1.42) and any lower respiratory symptom (OR = 1.13, CI = 0.99,
2 1.30).
3 Stern et al. (1994) studied respiratory illness and lung function in five towns in
4 southwestern Ontario (Blenheim, Ridgetown, Tillsonburg, Strathroy, and Wallaceburg) and
5 five communities in south-central Saskatchewan (Esterhazy, Melville, Melfort, Weyburn, and
6 Yorkton. Self-administered parental questionnaires were distributed between October 1985
7 and March 1986. Pollution monitoring was not begun until late 1985, and included SO2,
8 NO2, and O3. Inhalable particulates were measured at each town using a high-volume Sierra
9 Instruments sampler. PM10 was measured once every six days in the Ontario towns and once
10 every three days in the Saskatchewan towns. Odds ratios were computed (presumably using
11 multiple logistic regression with a random effects model) comparing the endpoints of cough,
12 phlegm, wheeze, asthma, bronchitis, and chest illness for the Ontario towns versus the
13 Saskatchewan towns. No significant differences were found, even after adjusting for gender,
14 parental smoking, parental education, and gas cooking. Actual exposure estimates for the
15 individual towns were not used. The overall mean PM10 level in the Ontario towns was
16 23.0 jiig/m3 compared with 18.0 jig/m3 for Saskatchewan.
17
18 Studies of Adults
19 The earlier 1982 criteria document (U.S. Environmental Protection Agency, 1982a)
20 discussed one series of studies, reported on from the early 1960s to the mid-1970s, that was
21 conducted by Ferris, Anderson, and others (Ferris and Anderson, 1962; Kenline, 1962;
22 Anderson et al., 1964; Ferris et al., 1967, 1971, 1976). The initial study involved
23 comparison of three areas within a pulp-mill town (Berlin, New Hampshire). In the original
24 prevalence study (Ferris and Anderson, 1962; Anderson et al., 1964), no association was
25 found between questionnaire-determined symptoms and lung function tests assessed in the
26 winter and spring of 1961 in the three areas with differing pollution levels, after
27 standardizing for cigarette smoking. The study was later extended to compare Berlin, New
28 Hampshire, with the cleaner city of Chilliwack, British Columbia in Canada (Anderson and
29 Ferris, 1965). The prevalence of chronic respiratory disease was greater in Berlin, but the
April 1995 12-182 DRAFT-DO NOT QUOTE OR CITE
-------
1 authors concluded that this difference was due to interactions between age and smoking
2 habits within the respective populations.
3 The Berlin, New Hampshire, population was followed up in 1967 and again in 1973
4 (Ferris et al., 1971, 1976). During the period between 1961 and 1967, all measured
5 indicators of air pollution fell (e.g., TSP from about 180 /xg/m3 in 1961 to 131 ^g/m3 in
6 1967). In the 1973 follow-up, sulfation rates nearly doubled from the 1967 level (0.469 to
7 0.901 mg SO3/100 cm2 day) while TSP values fell from 131 to 80 /ig/m3. Only limited SO2
8 data were available (the mean of a series of 8-h samples for selected weeks.) During the
9 1961 to 1967 period, standardized respiratory symptom rates decreased and there was an
10 indication that lung function also improved. Between 1967 to 1973, age-sex standardized
11 respiratory symptom rates and age-sex-height standardized pulmonary function levels were
12 unchanged. Although some of the testing was done during the spring versus the summer in
13 the different comparison years, Ferris and coworkers attempted to rule out likely seasonal
14 effects by retesting some subjects in both seasons during one year and found no significant
15 differences in test results. Given that the same set of investigators, using the same
16 standardized procedures, conducted the symptom surveys and pulmonary function tests over
17 the entire course of these studies, it is unlikely that the observed health endpoint
18 improvements in the Berlin study population were due to variations in measurement
19 procedures, but rather appear to have been associated with decreases in TSP levels from
20 180 to 131 /ig/m3. The relatively small changes observed and limited aerometric data
21 available, however, argue for caution in placing much weight on these findings as
22 quantitative indices for effect or no-effect levels for health changes in adults associated with
23 chronic exposures to PM measured as TSP.
24 The earlier criteria review (U.S. Environmental Protection Agency, 1982a) also noted
25 a cross-sectional study conducted by Bouhuys et al. (1978) in Ansonia (urban) and in
26 Lebanon (rural) Connecticut towns in which differences in respiratory and pulmonary
27 function were examined in 3,056 subjects (adults and children). Bouhuys et al. (1978) found
28 no differences between Ansonia and Lebanon for chronic bronchitis prevalence rates but did
29 note that a history of bronchial asthma was highly significant for male resident of Lebanon
30 (the cleaner town) as compared to Ansonia (the higher-pollution area). No differences were
April 1995 12-183 DRAFT-DO NOT QUOTE OR CITE
-------
1 observed between the communities for pulmonary function tests adjusted for sex, age, height
2 and smoking habits. However, prevalence for three of five symptoms (cough, phlegm, and
3 plus one dyspnea) were significantly higher for adult non-smokers in Ansonia (p <0.001).
4 The mix of positive and negative health effect results found by this cross-sectional study
5 make it difficult to interpret.
6 Schwartz (1993) analyzed the NHANES data on respiratory illness diagnosed by a
7 physician. The NHANES survey was conducted from 1971 to 1974 on the
8 non-institutionalized U.S. population aged 1 to 74. The survey used a complex design and
9 the Schwartz analysis was restricted to 53 urban sampling units. Endpoints included asthma,
10 bronchitis, respiratory illness and dyspnea. EPA's SAROAD data base was used to obtain
11 data from population oriented monitors in the 53 areas. Average TSP concentrations (/ig/m3)
12 for previous years were used as the exposure measure. No other pollutants were considered.
13 Multiple logistic regression analysis was used that included terms for cigarette consumption
14 per day, former smoking, age, race, and gender. The coefficient for chronic bronchitis was
15 .0068 (.0023) and for respiratory illness it was .0058 (.0019). The coefficients were slightly
16 larger when restricted to non-smokers.
17 Yano et al. (1990) studied chronic respiratory illness in females aged 30 to 59 in two
18 cities in Japan. One city, Kanoya is 25 km from an active volcano, and Tashiro is 50 km
19 from the volcano. The winter concentrations of TSP in Kanoya average 341 /ig/m3 whereas
20 the average is 119 jwg/m3 in Tashiro. Respiratory conditions were assessed using a Japanese
21 version of the ATS-DLD questionnaire. No significant difference in rates of bronchitis,
22 asthma, wheezing, or other related illnesses were found.
23 A number of studies have been published that attempt to define chronic respiratory
24 system health effects in relationship to ambient pollutants to include PM and O3 (Hodgkin
25 et al., 1984; Euler et al., 1987, 1988; Abbey et al., 1991a,b). Among these, the series of
26 publications from the Adventist Health Smog Study (AHSMOG) (Hodgkin et al., 1984; Euler
27 et al., 1987, 1988; Abbey et al., 1991a,b) will be discussed first and as a set.
28 The basic population for these studies represents California-resident, Seventh-Day
29 Adventists aged >25 years of age who had lived 11 years or longer (as of August 1976) in
30 either a high-oxidant-polluted area (South Coast Air Basin [Los Angeles and vicinity] and a
April 1995 12-184 DRAFT-DO NOT QUOTE OR CITE
-------
1 portion of the nearby Southeast Desert Air Basin) or a low-pollution area (San Francisco or
2 San Diego). This sample was supplemented by an additional group of subjects who met the
3 11-year residence requirement but who came from low-exposure rural areas in California.
4 The total, baseline sample (March 1977) comprised 8,572 individuals, of whom
5 7,267 enrolled. From this group, 109 current smokers and 492 subjects who had lived
6 outside of the designated areas for a portion of the previous 11 years were excluded.
7 Detailed respiratory illness and occupational histories were obtained. In these studies,
8 "COPD" refers to "definite chronic bronchitis", "definite emphysema", and "definite asthma"
9 as defined by the study questionnaire. Measures of pulmonary function are not included.
10 Air monitoring data were obtained from the California ARB monitoring system.
11 Ninety-nine percent of the subjects (excluding the rural supplement) lived at a distance from
12 the nearest ARB monitoring site that was considered to provide relatively reliable
13 concentration estimates for the outdoor, ambient environment at their residence.
14 Concentrations at the monitors were interpolated to the centroid of each residential zip code
15 from the three nearest monitoring sites with the use of a 1/R2 interpolation. Subsequent
16 development of exposure indices took account of the improvements in ARB data after 1973.
1 Data were available for total oxidants, O3, TSP, SO2, NO2, CO, and SO4 (excluding 1973 to
2 1975).
3 The initial report from this study (Hodgkin et al., 1984) was summarized in the 1986
4 ozone criteria document (U.S. Environmental Protection Agency, 1986c). Based upon a
5 multiple logistic regression that adjusted for smoking, occupation, race, sex, age, and
6 education, it was estimated that residence in the South Coast Air Basin conferred a 15%
7 increase in risk for prevalent COPD. No estimates of exposure were provided, and the data
8 were considered to be of limited utility.
9 Next, Euler et al. (1988) assessed the risk of chronic respiratory disease symptoms due
10 to long-term exposure to ambient levels of TSP, oxidants, SO2, and NO2- Symptoms were
11 ascertained using the National Heart, Lung, and Blood Institute questionnaire on
12 8,572 Southern California Seventh-Day Adventists (nonsmokers—25 years and older) who
13 had lived 11 years or longer in their 1977 residential area. Tobacco smoke (active and
14 passive) and occupational exposures were assessed by questionnaires, as were lifestyle
April 1995 12-185 DRAFT-DO NOT QUOTE OR CITE
-------
1 characteristics relative to pollution exposure, such as time spent outside and residence
2 history. For each of the 7,336 participants who responded and qualified for analysis,
3 cumulative exposures to each pollutant were estimated using monthly residence zip code
4 histories and interpolated exposures from state air monitoring stations.
5 Multiple logistic regression analyses were conducted for pollutants individually and
6 together with eight covariables, including environmental tobacco smoke exposure at home
7 and at work, past smoking, occupational exposure, sex, age, race, and education.
8 Statistically significant associations with chronic respiratory symptoms were seen for
9 (1) SO2 (p = 0.03), relative risk of 1.18 for 13% of the study population with 500 h/year of
10 exposure above 0.04 ppm; (2) oxidants (p < 0.004) relative risk of 1.20 for 18% with
11 750 h/year above 0.1 ppm; and (3) TSP (p < 0.00001), relative risk of 1.22 for 25% with
3
12 750 h/year above 200 /ng/m . When these pollutant exposures were analyzed together, TSP
13 was the only one showing statistical significance (p < 0.01). Individuals working with
14 smokers for 10 years had relative risks of 1.11 and those living with a smoker for 10 years
15 had relative risks of 1.07.
16 A major improvement in the methods for assessment of exposure was presented in
17 Abbey et al (1991a). Previous exposure estimates were refined by the computation of
18 "excess concentrations" (concentration minus cutoff, summed over all relevant time periods
19 and corrected for missing data). Exposures also were corrected for time spent at work and
20 time away from residence, with estimates provided for the environments where work
21 occurred and for geographic areas away from residence. The quality of the interpolations
22 (in terms of distance of monitor from residence zip codes) also was evaluated and
23 incorporated into the estimates. Adjustments were made for the time spent indoors by
24 individuals. New indices were developed that were based on O3, rather than on total
25 oxidants. Comparison of actual versus interpolated cumulative exceedance frequencies and
26 mean concentrations at monitoring stations (1985 through 1986) for TSP and O3 were
27 assessed. The actual versus interpolated 2-y mean concentrations did not differ significantly
28 and were correlated with a Pearson correlation coefficient of 0.78 for TSP and 0.87 for O3.
29 The above estimates were applied to data that included 6 years of follow-up of the
30 study population (Abbey et al., 1991b). This analysis focused on incident occurrence of
April 1995 12_186 DRAFT-DO NOT QUOTE OR CITE
-------
1 obstructive airways disease (AOD—same definition as for COPD above). Incident symptoms
2 of AOD were significantly associated with hours above several TSP thresholds, but not with
3 hours above any O3 threshold. Incidence of definite symptoms of AOD and chronic
4 bronchitis were statistically significantly (P < 0.05) elevated for average annual hours in
5 excess of 100, 150, and 200 jtig/m3 and mean concentrations of TSP but not for 60 /-ig/m3.
6 For incidence of asthma, there was significantly elevated risks only for average annual hours
7 above thresholds of 150 and 200 jtig/m3. Relative risks for concentrations above 200 /-ig/m3
8 of TSP for bronchitis were 1.33 (95% CL = 1.07 to 1.81); and asthma was 1.74 (95% CL
9 = 1.11 to 2.92). Cumulative incidence estimates were adjusted with the use of Cox
10 proportional hazard models for the same variables noted in the original publication of
11 Hodgkin et al. (1984), as well as the presence of possible symptoms in 1977 and childhood
12 respiratory illness history. None of the analyses included both O3 and TSP thresholds.
13 No data were provided on the details of the subjects available for the prospective analysis
14 and their representativeness versus the entire base population.
15 Another analysis by Abbey et al. (1993) evaluated changes in respiratory symptom
16 severity with the TSP and O3 thresholds noted above. In this analysis, logistic regression,
17 rather than Cox proportional hazard modeling, was used to assess cumulative incidence of
18 components of the COPD/AOD complex; and multiple, linear regression was used to
19 evaluate changes in symptom severity. When O3 was considered by itself, there was a trend
20 toward an increased risk of asthma for a 1,000-h average annual increment in the OZ (10)
21 criterion (RR = 2.07, 95% CL = 0.98 to 4.89). In this analysis, there was a suggestion
22 that recent ambient O3 concentrations were more related to cumulative incidence than past
23 concentrations. Change in asthma severity score was associated significantly with the 1977
24 to 1987 average annual exceedance frequency for O3 thresholds of 10 and 12 pphm.
25 No significant effects were found for COPD or bronchitis alone. In contrast to the above
26 study of cumulative incidence, the investigators carried out an analysis in which TSP (200)
27 and OZ (10) were allowed to compete for entry into a model to evaluate asthma cumulative
28 incidence and changes in severity. In the cumulative incidence model that employed
29 exceedance frequencies (number of hours above threshold), TSP (200) entered before
30 OZ (10); when average annual mean concentrations were used, O3 entered before TSP.
April 1995 12-187 DRAFT-DO NOT QUOTE OR CITE
-------
1 From this, the authors concluded that both TSP and 03 were relevant to asthma cumulative
2 incidence. In no case did both pollutants remain significant simultaneously in the same
3 regression. No interactions were observed between TSP and O3 for either metric. A similar
4 result was observed for change in asthma severity. As in previous analyses, there was a high
5 correlation between TSP (200) and OZ (10) exceedance frequencies (0.72) and their
6 respective average annual mean concentrations (0.74).
7 Some researchers used case-control approaches to study chronic respiratory system
8 health effects in relationship to ambient pollutants such as PM. In Athens, Greece, Tzonou
9 et al. (1992) studied the relation of urban living and tobacco smoking to the development of
10 COPD. Their findings suggested that air pollution or another aspect of the urban
11 environments can be an important contribution to the development of COPD. Specific PM
12 levels were not studied. Katsouyanni et al. (1991) conducted a case-control study in Athens
13 exploring the role of smoking and outdoor air pollution and their relationship to lung cancer.
14 Air pollution levels were associated with an increased risk for lung cancer but the relative
15 risk was small and not statistically significant. Xu et al. (1989) studied air pollutants and
16 lung cancer in China, the findings of which suggested that smoking and environmental
17 pollution combined to allow for elevated rates of lung cancer mortality. In Poland,
18 Jedrychowski et al. (1990) found similar findings as the above studies. (See Chapter 11 for
19 other discussions of these studies).
20 Robertson and Ingalls (1989) studied male workers in the seven U.S. plants producing
21 carbon black. The exposures ranged from 60 to 1700 /xg/m3. The study was analyzed as a
22 matched case-control study. No differences in the time weighted average exposure were
23 found for any cancers or respiratory disease.
24 Rothman et al. (1991) reported that wildland firefighters experience a small cross-
25 seasonal decline in pulmonary function and an increase in several respiratory symptoms.
26 Hours of self-reported fire-fighting activity were used as a surrogate for fire smoke exposure.
27 At wildland fires, the concentration of a variety of pulmonary irritants, including respirable
28 particulate, acrolein and formaldehyde may exceed occupational safety and health
29 administration permissible exposure limits. Other compounds are also produced in wildland
30 fires.
April 1995 12_i88 DRAFT-DO NOT QUOTE OR CITE
-------
1 Shusterman et al. (1993) report the smoke-related disorders in Alameda County,
2 California related to the October 20, 1991 grass fire in the Oakland-Berkeley hills.
3 Bronchospaitic and irritative reactions to smoke constituted more than half of the emergency
4 department visits related to the fire. Many of these patients had a history of asthma. The
5 intensity and duration of smoke exposure ranged from mild and transient (among residents of
6 surrounding areas) to intense and prolonged (among firefighters and some residents of the
7 fire area).
8
9 Chronic Respiratory Disease Studies Summary
10 Three studies based on a similar type of questionnaire but done by two different groups
11 of researchers, provide data on chronic respiratory disease and PM. All three studies suggest
12 a chronic effect of paniculate matter on respiratory disease, but the studies suffer from the
13 usual difficulty of cross sectional studies. The effect of paniculate matter is based on
14 variations in exposure which are determined by the different number of locations. In the
15 first two studies there were six locations and in the second there were four. The results seen
16 were consistent with a paniculate matter gradient, but it is impossible to separate out the
17 effect of paniculate matter and any other factors or pollutants which have the same gradient
18 (See Table 12-21).
19
20 12.4.2.2 Pulmonary Function Studies
21 Studies of Children
22 Dockery et al. (1989) also studied lung function in children in the same six cities.
23 A cross-sectional survey was done in 1980 to 1981. Lung function was measured at the time
24 of the survey using a water filled recording spirometer. The analysis was restricted to
25 5,422 10 to 12-year white children. A centrally located air monitoring station was
26 established in each community starting in 1974 measuring TSP, SO2, NO2, and ozone.
27 Starting in 1978 dichotomous samplers were used to measure PM10. Separate regressions of
28 the adjusted city-specific pulmonary function levels on air pollution for children with and
29 without asthma or wheeze did not show any associations.
April 1995 12-189 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-21. CHRONIC RESPIRATORY DISEASE STUDIES
JO
O
O
Z
s
3
tr)
O
Study
Ware et al. (1984)
Study of respiratory
symptoms in children in 6
cities in the U.S. Survey
done 1974-1977
Dockery et al. (1989)
Study of respiratory
symptoms in children in 6
cities in the U.S. Survey
done 1980-1981
Chapman et al. (1985)
Study of persistent cough
and phlegm (bronchitis) in
adults in four Utah
communities. Survey
done in 1976.
PM Type &
No. Sites
Daily monitoring
of TSP, SO2,
NO2, and ozone at
each city
Daily monitoring
of PM15, sulfate
fraction at each
city
Daily monitoring
of TSP, and
sulfate fraction at
each city
PM Mean
& Range
City TSP means
ranged from 39
to 1 14 /ig/m3
City PM15
means ranged
from 20 to 59
/ig/m3
Previous 5 year
TSP ranged from
11 to 115 /tg/m3
Overall
Symptom
Rate
Cough, .08,
Bronchitis
.08,
Lower resp.
.19
Cough, .02 to
.09, Bronchitis
.04 to .10,
Lower resp.
.07 to .16
.02 to .05 by
city
Model
Type
&Lag
Structure
Logistic
regression
Logistic
regression
Logistic
regression
Other
pollutants
measured
S02, N02,
and ozone
SO2, NO2,
and ozone
SO2, NO2
Other
Other pollutants
Covariates in model
age, gender, none
parental
education,
maternal
smoking
age, gender, none
maternal
smoking
smoking none
Result*
(Confidence
Interval)
Cough
2.75 (1.92,
Bronchitis
2.80(1.17,
Lower resp.
2.14(1.06,
Cough
5.39(1.00,
Bronchitis
3.26(1.13,
Lower resp.
2.93 (0.75,
Mothers
1.75(1.21,
Fathers
1.94(1.16,
3.94)
7.03)
4.31)
28.6)
10.28)
11.60)
2.54)
3.25)
Neas et al. (1994)
Study of children aged
7 to 11 from six cites in
U.S. Survey done
1983-1986.
PM
2.5
Not given Not given Logistic NO2 household none
regression smoking, gas
stove, age,
gender none
none
Cough
1.08 (0.76, 1.53)
Bronchitis
1.32(0.98, 1.79)
Lower resp.
1.23(0.98, 1.55)
Estimates calculated from data tables assuming a 50 /ig/m3 increase in PM10 on 100 /ig/m3 increase in TSP.
-------
1 Ware et al. (1986) also studied lung function in children in an earlier study of the same
2 six cities. A cross-sectional survey was done between 1974 and 1977. Lung function was
3 measured at the time of the survey using a water filled recording spirometer. FEVj 0 and
4 FVC measurements were used in the analyses. Starting in 1978 dichotomous samplers were
5 used to measure PM10. Adjusted logarithms of the pulmonary function values were not
6 related to TSP concentrations. The change in FEVj 0 per 10jug/m3 change in TSP was .06%
7 (.17%) at the first examination and -0.09% (.17%) at the second examination.
8 Neas et al. (1994) analyzed a cohort of white children aged 7 to 11 from the same Six
9 Cities Study for pulmonary function. Five pulmonary function measures were studied: FVC,
10 FEVt 0, the ratio of FEVj 0 to FVC, FEF25.75, and the ratio of FEF25_75 to FVC. The
11 regression model used the logarithm of the lung function value as the dependent variable and
12 included gender, parental education, history of asthma, age, height, weight, and city as
13 covariates. No significant effects of PM2 5 on lung function could be found. The use of
14 logarithms of the dependent variables as well as the lack of overall mean lung function values
15 makes it impossible to directly compare the results of this study with other studies.
16 Stern et al. (1994) studied lung function and respiratory illness in five towns in
17 southwestern Ontario (Blenheim, Ridgetown, Tillsonburg, Strathroy, and Wallaceburg) and
18 five communities in south-central Saskatchewan (Esterhazy, Melville, Melfort, Weyburn, and
19 Yorkton. Lung function measurements were made between October 1985 and March 1986.
20 Self-administered parental questionnaires were distributed at the same time. Pollution
21 monitoring was not begun until late 1985, and included SO2, NO2, and O3. Inhalable
22 particulates were measured at each town using a high-volume Sierra Instruments sampler.
23 PM10 was measured once every six days in the Ontario towns and once every three days in
24 the Saskatchewan towns. Lung function measurements included FVC, FEV! 0, PEFR,
25 FEF25.75, and V50max. Lung function measurements were adjusted for age, gender, weight,
26 standing height, parental smoking, gas cooking, and standing height by gender interaction.
27 Ontario children had statistically significant decrements in FCV (1.7%) and FEV, 0 (1.3%)
28 compared with Saskatchewan children, but no differences were found in the flow parameters.
29 Actual exposure estimates for the individual towns were not used. The overall mean PM10
30 level in the Ontario towns was 23.0 /ig/m3 compared with 18.0 /xg/m3 for Saskatchewan.
April 1995 12-191 DRAFT-DO NOT QUOTE OR CITE
-------
1 Spektor et al. (1991) studied pulmonary function in children living in Cubatao, Brazil.
2 PMio and SO2 measurements were made at six sites in Cubatao which is located about 44 km
3 from Sao Paulo. Average annual PM10 levels ranged from 64 to 104 /xg/rn3. Pulmonary
4 function measurements were made monthly from March to November in 1988. Individual
5 regressions were performed using height, weight, and pollution was covariates. Average
6 slopes were reported for each of six schools, but no confidence intervals were given. Both
7 FEVj o and PEFR were significantly related to PM10 at the six schools. The average
8 decrease in PEFR per 50 /ig/m3 was about 100 ml/sec. This value is much larger than the
9 values seen in other studies.
10 He et al. (1993) studied lung function in children in areas of Wuhan, China, in 1988.
11 The children (aged 7 to 13 years) were from six urban and one suburban school. Pollution
12 measurements, including TSP, SO2, oxides of nitrogen, and CO, were collected by the
13 Wuhan Environmental Protection Agency Air Pollution Monitoring Network from 1981 to
14 1988. All pollutants were higher at the urban site, with TSP values averaging 251 /xg/m3, as
15 compared with 100 jig/m3 at the suburban site. The cross sectional study was conducted in
16 May and June of 1988. The hypothesis was that the relationship between lung function and
17 height would be less in the urban city. Lung function growth curves were constructed by
18 regressing FEVj 0 and FVC on height for males and females for both areas. The curves
19 were significantly steeper for the suburban children than for the urban children.
20 Arossa et al. (1987) studied lung function in approximately 2000 children in Turin,
21 Italy, during a time period when both TSP and SO2 were being reduced. Three areas of
22 Turin (central city, peripheral area, and suburban area) were studied during the winters of
23 1980 to 1981 and 1982 to 1983. Each child's respiratory health was assessed at the
24 beginning and end of the study using a questionnaire which also obtained demographic
25 information. Lung function measurements included FVC, FEVj 0, FEF25_75, and MEF50.
26 Daily SO2 and TSP measurements were available from seven monitoring sites in the area.
27 The pollution data confirmed that the large SO2 differences across areas in 1980 to 1981
28 were reduced substantially by 1982 1983. The differences in TSP remained small but
29 constant during the time period. A general linear model analysis was used to calculated
30 adjusted lung function values. From these values, individual slopes were estimated and these
April 1995 12-192 DRAFT-DO NOT QUOTE OR CITE
-------
1 became the unit of analysis. Average slopes were significantly higher within the city of
2 Turin when compared with the suburban area, suggesting to the authors that a decrease in
3 pollution (primarily SO2) resulted in an improvement of lung function.
4
5 Studies of Adults
6 Chestnut et al. (1991) analyzed the NHANES data on pulmonary function. The
7 NHANES survey was conducted from 1971 to 1974 on the non-institutionalized U.S.
8 population aged 1 to 74. The survey used a complex design and the Chestnut et al. (1991)
9 analysis was restricted to 49 urban sampling units where TSP measurements were available.
10 A subsample of 6,913 adults (aged 25 to 74) were given spirometric tests using an Ohio
11 Medical Instrument Corporation Model 800 electronic spirometer. Endpoints included FVC,
12 FEVj 0, and MMEF. EPA's SAROAD data base was used to obtain data from population
13 oriented monitors in the 49 areas. Average TSP concentrations for previous years were used
14 as the exposure measure. All individuals with reproducible results were included in a
15 multiple regression analysis that included terms for age, height, gender, ethnic group,
16 obesity, and TSP. Both a nonparametric analysis and a regression analysis suggested that
17 TSP was associated with decrease FVC at TSP levels greater than 60 ^g/m3.
18 Xu et al. (1991) studied lung function in adults in areas of Beijing, China, in 1986.
19 A stratified sampling plan over three areas with historically different pollution was used.
20 A trained interviewer obtained information on history of chest illness, respiratory symptoms,
21 cigarette smoking, occupational exposure, residential history, educational level, and type of
22 fuel used for cooking. Pulmonary function measurements were made according to guidelines
23 of the American Thoracic Society. Outdoor particulate matter (TSP) and SO2 were obtained
24 from 1981 to 1985 from World Health Organization Global Air Monitoring Stations.
25 Multiple linear regression was used to assess the impact of air pollution on FEVj 0 and FVC.
26 Highly significant decreases in FEVj 0 and FVC as a function of log(SO2) and log(TSP) were
27 found.
28 Tashkin et al. (1994) reported on the results of a long term lung function study of
29 adults living in three areas of southern California. The areas were (1) Lancaster, with
30 moderate levels of photochemical oxidants and low levels of other pollutants, (2) Glendora,
April 1995 12-193 DRAFT-DO NOT QUOTE OR CITE
-------
1 with very high levels of photochemical oxidants, sulfates, and paniculate matter, and
2 (3) Long Beach, with high levels of sulfates, oxides of nitrogen, and moderate levels of
3 sulfur oxides. A mobile lung function laboratory was used to gather pulmonary function
4 measurements and collect information on a modified NHLBI questionnaire. Residents of
5 each area were tested twice over a 5 or 6-year interval, but during the same month each
6 time. The testing schedule was as follows: (1) Lancaster, 1973 to 1974 and 1978 to 1979,
7 (2) Glendora, 1977 to 1978 and 1982 1983, and (3) Long Beach, 1974 1975 and 1980 to
8 1982. Significantly larger annual decreases in FEVj 0 were found in both Long Beach and
9 Glendora as compared with Lancaster. These results were consistent across gender, and
10 were adjusted for age, height, smoking status, and allergies. The decrease was largest in
11 Long Beach, but only slightly larger than in Glendora. Smoking showed a larger effect than
12 did area of residence. Because of the design of the study, it is impossible to attribute the
13 effects to a single pollutant, or to develop any quantitative relationships.
14
15 Chronic Pulmonary Function Studies Summary
16 The chronic pulmonary function studies are less numerous than the acute studies
17 (Table 12-22). The one study with good monitoring showed no effect from particulate
18 pollution. Cross sectional studies require very large sample sizes to detect differences
19 because the studies cannot eliminate person to person variation which is much larger than the
20 within person variation. Thus the lack of statistical significance cannot be taken as proof of
21 no effect.
22
23
24 12.5 HUMAN HEALTH EFFECTS ASSOCIATED WITH ACID
25 AEROSOL EXPOSURE
26 One key consideration in the evaluation of'PM-health effects is: Are there specific
27 chemical components of PM responsible for the noted associations between PM and human
28 health? The presence of known toxic constituents within ambient particles would add to the
29 strength of the association. Since the time of the London Fog of 1952 and other major
30 pollution episodes earlier in this century, the acidity of aerosols is one characteristic
April 1995 12-194 DRAFT-DO NOT QUOTE OR CITE
-------
d
d
o
2
9
O
cj
i
n
TABLE 12-22. CHRONIC PULMONARY FUNCTION CHANGES
Study
Neas et al. (1994)
Study of lung function in
children in 6 cities in the
U.S. Data collected from
1983-1988.
PM Type &
No. Sites
Daily monitoring
of PM2 5, sulfate
fraction at each
city
PM Mean
& Range
not given
Model Type
&Lag
Structure
Linear
regression using
logarithm of
PFT value
Other
pollutants
measured
S02, NO2, and
ozone
Weather &
Other
Factors
city, gender
parental education,
history of asthma,
age, height, weight
Pollutants
in model
PM2.5
Decrease*
(Confidence
Interval)
FVC and FEVj not
changed. Values
could not be
converted to mis.
* Decreases in lung function calculated from parameters given by author assuming a 50 jtg/m3 increase in PM10 on 100 ftg/m3 increase in TSP.
-------
1 suspected of contributing to health effects by PM air pollution. Though certainly not the
2 only PM component with potentially toxic effects, acidic aerosols have received more
3 epidemiologic study than have other PM components, to date. The epidemiologic evidence
4 available regarding acidic aerosols as a potential contributor to PM-health effects associations
5 is evaluated below.
6 Several epidemiologic studies have examined the health effects associated with ambient
7 paniculate strong acid aerosol (H+) exposures. The historical scarcity of such analyses was
8 due in large part to the absence of adequate ambient acid measurement techniques in the past
9 and to the lack of routine acid aerosol monitoring in more recent years. Some studies now
10 exists which suggests that human health effects maybe associated with exposures to ambient
11 acid aerosols, both: (1) as derived from reexamination of older, historically important data
12 on air pollution episode events hi North America and Europe, and; (2) as can be deduced
13 from recent epidemiology studies carried out in the U.S., Canada, and Europe. This section
14 concisely reviews their studies, first as they relate to acute exposure effects, and then as they
15 . pertain to chronic exposure effects. Because of the relative scarcity of direct acid aerosol
16 measurements until recent years, part of this section is also devoted to identifying studies of
17 situations in which there is good reason to suspect that high ambient acid concentrations
18 existed in the evaluated study areas. From all of these studies, the nature of any observed
19 health associations are summarized as a basis for drawing health effects conclusions, and for
20 suggesting directions for future research. The material in this review was based upon the
21 acid aerosols issue paper prepared by the U.S. Environmental Protection Agency (1989), as
22 well as more recent evidence, as appropriate.
23
24 12.5.1 Historical Evidence Evaluating the Relationship between Acid
25 Aerosols and Health Effects
26 Some of the earliest indications of associations between ambient air acid aerosols and
27 human health effects can be discerned upon reexamination of historically important air
28 pollution episode events. These include, for example, the Meuse Valley (Belgium), Donora,
29 PA (USA), and well-known London (UK) episodes, as discussed below.
30
April 1995 12-196 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.5.1.1 Meuse Valley
2 Firket (1931) described the fogs of December 1930 in the Meuse Valley in Belgium and
3 the morbidity and mortality related to them. A detailed discussion of the causes was
4 presented, and he concluded that, while multiple pollutants existed in this atmosphere, the
5 main component of the fog that caused the health effects that occurred was sulfuric acid.
6 This conclusion was based both upon consideration of the emissions in the valley, the
7 weather conditions and the aerometric chemistry required for the production of sulfuric acid.
8 Additionally, the pathophysiology seen was thought to relate to sulfuric acid exposure more
9 so than to other possible agents. More than 60 persons died from this acid fog and several
10 hundred suffered respiratory problems, with a large number becoming complicated with
11 cardiovascular insufficiency. The mortality rate during the fog was over ten times higher
12 than the normal rate. Those persons especially affected by the fog were the elderly, those
13 suffering from asthma, heart patients and the debilitated. Most children were not allowed
14 outside during the fog and few attended school. Unfortunately, no actual measurements of
15 acid aerosols in ambient air during the episode are available by which to establish clearly
16 their role in producing the observed health effects versus the relative contributions of other
17 specific pollutants.
18
19 12.5.1.2 Donora
20 Schrenk et al. (1949) reported on atmospheric pollutant exposures and the health effects
21 of the smog episode of October 1948 in Donora, PA. A total of 5,910 persons (or 42.7
22 percent) of the total population of Donora experienced some effect from the smog. The air
23 pollutant-laden fog lasted from the 28th to the 30th of October, and during a 2-week period
24 20 deaths took place, 18 of them being attributed to the fog. An extensive investigation by
25 the U.S. Public Health Service concluded that the health effects observed were mainly due to
26 an irritation of the respiratory tract. Mild upper respiratory tract symptoms were evenly
27 distributed through all age groups and, on the average, were of less than four days duration.
28 Cough was the most predominant symptom; it occurred in one-third of the population, and
29 was evenly distributed through all age groups. Dyspnea was the most frequent symptom in
30 the more severely affected, being reported by 13 percent of the population, with a steep rise
April 1995 12-197 DRAFT-DO NOT QUOTE OR CITE
-------
1 as age progressed to 55 years; above this age, more than half of the persons affected
2 complained of dyspnea.
3 It seems reasonable to state that, while no single substance can be clearly identified as
4 being responsible for the October 1948 episode, the observed health effects syndrome could
5 have likely been produced by two or more of the contaminants, i.e., sulfur dioxide and its
6 oxidation products together with particulate matter, as among the more significant
7 contaminants present. Hemeon (1955) examined the water soluble fraction of solids on a
8 filter of an electronic air cleaner operating during the smog in Donora and concluded that
9 acid salts were an important component.
10
11 12.5.1.3 London Acid Aerosol Fogs
12 Based on the mortality rate in the Meuse Valley, Firket (1931) had estimated that 3,179
13 sudden deaths would likely occur if a pollutant fog similar to that in the Meuse Valley
14 occurred in London. An estimated 4,000 deaths did later indeed occur during the London
15 Fog of December 1952, as noted by Martin (1964). During the fog of December 1952,
16 evidence of bronchial irritation, dyspnea, bronchospasm and, in some cases, cyanosis is clear
17 from hospital records and from the reports of general practitioners. There was a
18 considerable increase in sudden deaths from respiratory and cardiovascular conditions. The
19 nature of these sudden deaths remains a matter for speculation since no specific cause was
20 found at autopsy. Evidence of irritation of the respiratory tract was, however, frequently
21 found and it is not unreasonable to suppose that acute anoxia due either to bronchospasm or
22 exudate in the respiratory tract was an important factor. Also, the United Kingdom Ministry
23 of Health (1954) reported that, in the presence of moisture, aided perhaps by the surface
24 activity of minute solid particles in fog, some sulfur dioxide is oxidized to trioxide. It is
25 probable, therefore, that sulfur trioxide, dissolved as sulfuric acid in fog droplets,
26 appreciably augmented the harmful effects of sulfur dioxide and/or other particulate matter
27 species.
28 Martin and Bradley (1960) reported increases in daily total mortality among the elderly
29 and persons with preexisting respiratory or cardiac disease in relation to SO2 and PM
30 (measured as British Smoke; BS) levels in London during the winter of 1958 to 1959. The
April 1995 12-198 DRAFT-DO NOT QUOTE OR CITE
-------
1 pathological findings in 12 fatal cases and the clinical evidence of practitioners seem to
2 indicate clearly that the harmful effects of the fog were produced by the irritating action of
3 polluted air drawn into the lungs. These effects were more obvious in people who already
4 suffered from a chronic respiratory disease and whose bronchi were presumably more liable
5 to bronchospasm.
6 Waller (1963) reported that sulfuric acid was one of the pollutants considered as a
7 possible cause of the increased morbidity and mortality noted during the London fog of
8 December 1952. As noted earlier, following the 1952 pollution episode daily measurements
9 of BS and SO2 made in London starting in 1954. Concentrations of sulfuric acid, calculated
10 from net aerosol acidity, were also measured during air pollution episodes and, later, on a
11 daily basis, starting in 1963. All of these historical acid measurements must be viewed with
12 caution, since filter artifact formation is possible for these samples. For example, there was
13 no attempt to protect the sample filters from ambient SO2 or NH3, which could result in
14 excess acid formation or in acid neutralization, respectively, on the samples. No regular
15 measurements of sulfuric acid were made during the winter of 1955 to 1956, but some was
16 detected at times of high pollution. For example, Waller and Lawther (1957) detected the
17 presence of acid droplets in samples collected in January of 1956. Insufficient measurements
18 were made, however, during the rest of the winter of 1955 to 1956 to study the effects of the
19 acid aerosol present. Waller (1963) later reported measuring acid droplets in London in the
20 winter of 1958 to 1959 with mass median diameter of 0.5 /*m. Commins (1963) measured
21 particulate acid in the city of London and found concentrations especially high at times of fog
22 reaching H+ levels of 678 ji*g/m3 of air (calculated as sulfuric acid). Typical winter daily
23 concentrations were 18 /ig/m3 compared to 7 /ig/m3 in the summer. The sulfuric acid
24 content of the air in the city of London at the time could range up to 10 percent of the total
25 sulfur.
26 Acid aerosol data collected by Commins and Waller (1967) during the December 1962
27 London Fog episode, which occurred almost exactly 10 years after the 1952 episode, provide
28 some of the strongest evidence that acid aerosols were elevated during the 1950's episodes.
29 As shown in Figure 12-10, 24 h average acid concentrations reached 378 pig/m3 (as H2SO4)
30 on the peak mortality day during this later, less severe, London episode. Both BS and SO2
April 1995 12-199 DRAFT-DO NOT QUOTE OR CITE
-------
500-1
400H
OT
300-
200H
100-
1 23456789 10
0
r500
-400
-300
-200
-100
L0
O)
December, 1962
Figure 12-10. December 1962, London pollution episode.
1 were similarly elevated on these episode days, however, so it is not possible to identify
2 H2SO4 as the sole causal pollutant. Not all of the measured acids during fog episodes would
3 necessarily be respirable, reducing their health effects from that implied by the total H2SO4
4 concentration. However, these H+ data from the 1962 episode do support past anecdotal
5 evidence that elevated strong acid concentrations were present during the major London Fog
6 pollution episodes.
7 Lawther et al. (1970) reported an association between daily pollutant levels (BS and
8 SO2) and worsening of health status among a group of over 1,000 chronic bronchitis patients
9 in London during the winters of 1959 to 1960 and 1964 to 1965. A daily technique for self-
10 assessment of day-to-day change in health status was used. The concentration of acid aerosol
11 rose with that of smoke, and it is likely to have been partly responsible for health effects
12 observed in these chronic bronchitic patients. Since many patients' symptoms become worse
13 even at times of relatively low humidity, this suggests that small droplets of strong acid had
April 1995
12-200 DRAFT-DO NOT QUOTE OR CITE
-------
1 more effect than larger ones. An interesting study was also conducted on a smaller sample
2 of the patients during in the winters of 1964 to 1965 and 1967 to 1968 when pollutant levels
3 were somewhat lower than in earlier years. Approximately 50 subjects selected for their
4 susceptibility to air pollutant effects formed the cohort. Daily apparent sulfuric acid,
5 measured at St. Bartholomew Hospital Medical College, was reported as having a relatively
1 high correlation with health effects in the 1964 to 1965 winter. For 1967 to 1968, all these
2 correlation coefficients were lower, but still significant. The authors comment that the
3 patients selected must have been particularly sensitive to pollution, since from past
4 experience no correlation would have been expected with such very low levels of pollution
5 encountered by such a small group.
6 The studies discussed above suggest that mortality and morbidity effects may be
7 associated with pollutant mixes which included elevated levels of ambient air concentrations
8 of acid aerosols. The calculations and measurements of sulfuric acid levels (estimated to
9 range up to 678 ^g/m3 for a 1-h average) during some London episodes in the late fifties and
10 early sixties provide a plausible basis for hypothesizing contributions of sulfuric acid aerosols
11 to the health effects observed during those episodes.
12
13 12.5.2 Quantitative Analysis of Earlier Acid Aerosol Studies
14 12.5.2.1 London Acute Mortality and Daily Acid Aerosol Measurements
15 Thurston et al. (1989) conducted a reanalysis of the London mortality data for a multi-
16 year period in which daily direct acid aerosol measurements were made at St. Bartholomew's
17 Medical College. The data considered in this analysis include pollution and mortality records
18 collected in Greater London during winter periods (November 1 to February 29) beginning in
19 November 1963 and ending in February 1972. The air pollution data were compiled from
20 one of two sources. First, BS and S02 data (as reported in /tg/m3) were compiled as daily
21 means of seven sites run by the London County Council and spatially distributed throughout
22 London County. A second data set of BS, SO2 and aerosol acidity (calculated as /*g/m3
23 sulfuric acid) was also compiled for one central London site run by the Medical Research
24 Council Air Pollution Research unit at the St. Bartholomew's Medical College. The Greater
25 London mortality data were obtained from the London General Register Office for winter
April 1995 12-201 DRAFT-DO NOT QUOTE OR CITE
-------
1 periods (November to February) beginning in 1958, and for all days commencing in April
2 1965. Total mortality, respiratory mortality, and cardiovascular mortality were all compiled
3 daily during these periods, but only total mortality was considered in this work. The Greater
4 London population was fairly stable during the period considered in this research (1963 to
5 1972), averaging about 8 million people. The pollution and mortality data for each of the
6 nine winters of data were combined into one data set for analysis. This is reasonable in this
7 case because the period under study, late 1963 to early 1972, is subsequent to the
8 implementation of the London smoke control zones (1961 to 1963), and is therefore a period
9 of fairly constant average winter pollutant concentrations. Prior to combining the data, each
10 year's total mortality data were also prefiltered using a high-pass filter that weights the
11 mortality data in a manner very similar to the calculation of deviations from a 15-day moving
12 average of mortality, except that it eliminates the undesirable long-term cyclical fluctuations.
13 Although the filtered total mortality has largely removed slow moving fluctuations in the
14 mortality data, the winters of 1967 to 1968 and 1969 to 1970 were still slightly
15 nonstationary, probably due to influenza epidemics in those years. It may have been
16 desirable to also control for these remaining effects by considering an influenza epidemic
17 dummy variable in subsequent regression analyses of these data. The resulting data set
18 comprised a total of 921 observations of daily pollution, total mortality, and filtered total
19 mortality data for the nine-winter data set.
20 In the Thurston et al. (1989) results, the log of H2SO4 measured at the central site was
21 much more strongly correlated with raw total daily mortality than any measure of BS or SO2
22 especially when it was correlated with the next day mortality (r = 0.31). It is also clear that
23 the logarithm transformation "helps" the acid-mortality association more than is true for BS
24 or SO2. For the filtered mortality variable, however, the H2SO4 correlation is weakened
25 versus raw total mortality (e.g., r = 0.19 for log (H2SO4) with next day filtered mortality).
26 Thus, the St. Bartholomew's College H2SO4 measurements appear to be correlated with
27 Greater London mortality, especially before the mortality data are filtered for slow moving
28 fluctuations. Mortality-pollution crosscorrelation analyses indicated that mortality effects
29 usually followed pollution in time even after filtering both series (Thurston et al., 1989), a
30 basic consideration in inferring casual association.
April 1995 12-202 DRAFT-DO NOT QUOTE OR CITE
-------
1 The superiority of the log of H2SO4 concentration versus the raw H2SO4 data in
2 correlations with total mortality agrees with the previous analyses of British Smoke-total
3 mortality associations. This may imply that a "saturation" of mortality effects is indeed
4 occurring over two or more days, and that a cumulative effect of several episode days may
5 be more relevant than modeling a single day effect alone. This may be due to avertive
6 behavior, especially since episode warnings were publicized at the time of high pollution.
7 Most likely, however, the "saturation" of effects is due to the premature death of the most
8 susceptible people on prior moderate pollution days.
9 A more extensive analysis of the London total mortality and acid aerosol data was
10 conducted by Ito et al. (1993) for 1965 to 1972, when daily acid measurements were
11 available year-round and the air pollution levels were non-episodic (see Figure 12-11). BS,
12 SO2, H2SO4, and weather variables (temperature and humidity) were examined for their
13 short-term associations with daily mortality after removal of long-term components from each
14 series via prewhitening, in order to obtain "rational" crosscorrelations. Power spectra of the
15 variance of mortality, pollution, and temperature variables were employed in the
16 development of this model. Also, first order autocorrelations were found to be significant,
17 and were evaluated. Significant associations with same day and following days' mortality
18 were found for all three pollutants considered. In the most extensively controlled model the
19 winter mean pollutant effect was estimated to range from 2 to 3% of the mean 278
20 deaths/day total mortality, but all three pollutants gave similar results (for mean H2SO4 =
21 5.0 /ig/m3, SO2 = 293 /xg/m3, or BS = 72 /zg/m3) and their respective effects could not be
22 separated, due to their high intercorrelation. These models were fit to the (separate) 1962
23 London acid/mortality episode data and found to fit well, supporting the validity of such
24 deviation-derived mortality estimates.
25 Lippmann and Ito (1995) conducted a preliminary graphically-based analyses of the
26 year-round 1965 to 1972 London pollution and mortality data set that controlled for same-day
27 temperature effects by analyzing restricted temperature ranges in each season. This was
28 done to provide an alternative to more empirical approaches applied to these data in prior
29 analyses. In each season, the majority of days fell within one or two temperature ranges,
30 within which the mortality also fell within narrow ranges. Within these restricted ranges,
April 1995 12-203 DRAFT-DO NOT QUOTE OR CITE
-------
I § 5001
2 =5400
f 30300^
> 100
O 0
'.l*/^**!**^^
1965 1966 1967 1968 1969 1970 1971 1972
jg/
.8
E
0)
-------
1 analyses indicated that there were relatively strong associations between daily mortality and
2 the daily logs of the concentrations of H+ and SO2. By contrast, the mortality association
3 with BS was much weaker, especially in the winter and summer. The authors indicate that
4 more comprehensive analyses are needed, but assert that such analyses provide a useful
5 complement to model-based approaches. Things as yet not addressed by this analysis include
6 the need to control for the potential effects of prior days' extreme temperatures (i.e., lagged
7 effects), which are known to be important in winter, and the direct addressing of potential
8 temperature effects within the ranges considered. Probably the most interesting result of
9 these analyses is that the H+-mortality association is found even in the summertime, when
10 the daily H+ concentrations do not exceed approximately 10 /ig/m3, as H2SO4 (~ 200
11 nmoles H4"/ m3), which are concentrations not unlike those presently experienced during the
12 summer in the eastern United States.
13 These recent analyses by Thurston et al. (1989) and by Ito et al. (1993) of daily direct
14 acid aerosol measurements over a long span of time (1963 to 1972) in London are especially
15 important in providing more data to examine for associations between acute exposures to
16 ambient acid aerosols and mortality at H+ levels more relevant to those presently
17 experienced in North America. Also, the work of Lippmann and Ito (1995) suggests that this
18 acute H+-mortality association may exist at concentrations below 200 nmoles/m3 H+, and
19 under summertime conditions. However, as noted above, the several pollutants considered
20 were correlated over time, making a discrimination of their respective effects difficult, and
21 the historical London acid measurement techniques may or may not be directly comparable
22 with present day measurements.
23
24 12.5.3. Studies Relating Acute Health Effects to Sulfates
25 Sulfate species usually represent the principal component of particulate strong acid
26 aerosols (primarily as H2SO4 or NH4HSO4). As a result, variations in measured sulfate
27 levels have been found to represent a reasonably reliable surrogate for variations in strong
28 particulate acid aerosol levels over time at a site. However, sulfates are not necessarily as
29 useful for intercomparing aerosol particulate acidity levels between sites. This is because
30 measurements of total sulfate levels comprise not only strongly acidic sulfates, but, in fact,
April 1995 12-205 DRAFT-DO NOT QUQTE OR CITE
-------
1 are usually dominated by sulfates that are only weakly acidic (e.g. (NH4)2SO4). Moreover,
2 it has been found that local ammonia levels can diminish the ambient H+/SO4= ratio
3 experienced at a site by neutralizing the strongly acidic sulfates (Suh et al., 1994). For this
4 reason, two sites located in differing environs (e.g. urban versus suburban) may have similar
5 SO4= levels but different H+/SO4= ratios, merely because the population density around the
6 two sites is different (Ozkaynak et al., 1994). Therefore, cross-sectional studies using
7 sulfates may be limited in the insight they may provide into the potential health effects of
8 acid aerosol exposures, especially if they compare sites with differing surrounding land uses.
9 However, if two monitoring sites are in the same airshed, they will usually still be highly
10 correlated over time, as their paniculate H+ concentrations will rise and fall together from
11 day to day as regional sulfate levels rise and fall (e.g., see Thurston et al. 1994a). Thus,
12 while the surrounding land use dependence of the H+/SO4= ratio may limit somewhat the
13 usefulness of sulfates as an index of H+ differences between sites, this will not adversely
14 affect longitudinal (i.e., correlation based) studies using sulfate data as an index of paniculate
15 aerosol strong acidity.
16
17 12.5.3.1 Canadian'Hospital Admissions Related to Sulfate Acute Exposure Studies
18 Bates and Sizto (1983, 1986) reported results of an ongoing correlational study relating
19 hospital admissions in southern Ontario to air pollutant levels. Data for 1974, 1976, 1977,
20 and 1978 were discussed in the 1983 paper. The 1986 analyses evaluated data up to 1982
21 and showed: (1) no relationship between respiratory admissions and SO2 or COH in the
22 winter; (2) a complex relationship between asthma admissions and temperature in the winter;
23 and (3) a consistent relationship between respiratory (both asthma and non-asthma)
24 admissions in summer and sulfate and ozone concentrations, but not to summer COH levels.
25 However, Bates and Sizto noted that the data analyses were complicated by long-term trends
26 in respiratory disease admissions unlikely related to air pollution. They nevertheless
27 hypothesized that observed effects may be due to a mixture of oxidant and reducing
28 pollutants which produce intensely irritating gases or aerosols in the summer, but not in the
29 winter.
April 1995 12-206 DRAFT-DO NOT QUOTE OR CITE
-------
1 Bates and Sizto (1987) later studied admissions to all 79 acute-care hospitals in
2 Southern Ontario, Canada (i.e., the whole catchment area of 5.9 million people) for the
3 months of January, February, July and August for 1974 and for 1976 to 1983. Means of the
4 hourly maxima for O3, NO3, SO2, coefficient of haze (COH), and aerosol sulfates were
5 obtained from 17 stations between Windsor and Peterborough. Sulfates were measured every
6 sixth day. Total admissions and total respiratory admissions declined about 15 percent over
7 the course of the study period, but asthma admissions appeared to have risen. Evaluating the
8 asthma category of admissions is complicated by the effects of a change in International
9 Classification of Disease (ICD) coding in 1979. The analyses demonstrated that there was a
10 consistent summertime relationship between respiratory admissions (with or without asthma)
11 and sulfates, ozone, and temperature. This conclusion was strengthened by the continuing
12 lack of any association of these variables with non-respiratory conditions. The 1987 paper
13 raised the question of whether the association of increased respiratory admissions in the
14 summer in this region could be associated with ozone or sulfates. It was aerosol sulfates
15 that, in summertime, explained the highest percentage of the variance in respiratory
16 admissions; yet these were not correlated with respiratory admissions in the winter. In view
17 of this, the authors hypothesized that the observed health effects might be attributable neither
18 to ozone nor to sulfates, but to some other air pollutant species that "travel" with them over
19 the region in the summer (but not in the winter).
20 Bates and Sizto (1987) noted that recent observations had suggested the presence of
21 peaks of H+ aerosol of small particle size in this region of Canada in the summer,
22 concomitant with elevated O3 and SO4= levels. On two days in July 1986 in eastern Toronto
23 when ozone and sulfate levels were elevated, but not higher than on other days, peaks of H+
24 acid aerosol lasting for up to two hours were recorded at levels of 10 to 15 /ig/m3. The
25 particle size was small (about 0.2 /xm). Similar observations were recorded on the same
26 days by another H+ air sample operation southwest of Toronto. This raises the possibility
27 that the types of health effects noted above might be attributable neither to ozone, nor to
28 sulfates, but rather perhaps to acid aerosols. Thus, the evidence from Bates and Sizto (1983,
29 1986, 1987, 1989) neither conclusively relates sulfates nor ozone to hospital admissions.
April 1995 12-207 DRAFT-DO NOT QUOTE OR CITE
-------
1 Instead, the results appear to suggest that some other pollutant(s) may be responsible, e.g.,
2 the strong acidic aerosols that have since been measured in the region.
3 Lipfert and Hammerstrom (1992) reanalyzed the Bates and Sizto (1989) hospital
4 admissions dataset for 79 acute-care hospitals in southern Ontario, incorporating more
5 elaborate statistical methods and extending the dataset through 1985. Pollutants considered
6 included SO2, NO2, O3, SO4=, COH, and TSP. Long-wave influences were reduced by
7 using the short study periods previously employed by Bates and Sizto (e.g. July and August
8 only for summer), as well as by employing very conservative prewhitening procedures to the
9 data. Day of week effects were also controlled. In addition, the models were more
10 extensively specified, including a variety of new meteorological variables such as wind speed
11 (correlated at r=-0.5 with COH). Despite this possible model overspecification, however,
12 summerhaze pollutants (i.e. O3, SO4=, and SO2) were found to have significant associations
13 with on hospital admissions in southern Ontario. In contrast, pollution associations with
14 hospital admissions for accidental causes became nonsignificant in these models. While air
15 pollution concentrations were generally within U.S. standards, the pollutant mean effect
16 accounted for 19 to 24% of all summer respiratory admissions (mean admissions a 40/day,
17 mean S04= « 11 fJLg/m3), though the "responsible" summertime haze pollutant(s) could not
18 be discerned by the authors with certainty.
19 Burnett et al. (1994) related the number of emergency or urgent daily respiratory
20 admissions daily respiratory admissions at 168 acute care hospitals in all of Ontario during
21 1983 to 1988 to estimates of ozone and sulfates in the vicinity of each hospital. No other
22 pollutants were directly considered in this analysis, although the authors reported that SO2
23 and NO2 were only weakly correlated with SO4 in these data (r < 0.3), so these pollutants
24 were unlikely to be confounders. Daily levels of sulfates were recorded at nine monitoring
25 stations located throughout the province. Long-wave cycles in the admissions data were
26 removed using a 19-day moving average equivalent high pass filter. A random effects model,
27 wherein hospital effects were assumed random, was employed using the generalized
28 estimating equations (GEE) of Liang and Zeger (1986). After adjusting admissions data for
29 seasonal patterns, day of week effects, and individual hospital effects, positive and
30 statistically significant associations were found between hospital admissions and both ozone
April 1995 12-208 DRAFT-DO NOT QUOTE OR CITE
-------
1 and sulfates lagged 0 to 3 days. Positive associations were found in all age groups (0 to 1, 2
2 to 34, 35 to 64, and 65+). The bivariate relationship found between adjusted admissions
3 and sulfates in these data are shown in Figure 12-12. In simultaneous regressions, five
4 percent of daily respiratory admissions in the province during May to August (mean =
5 107.5/day) were found to be associated with O3 (at 50 ppb), and one percent with SO4=
6 (at 5 /ig/m3). Positive and significant air pollution associations were found for asthma,
7 chronic obstructive pulmonary disease (COPD), and infections, but not for nonrespiratory
8 (control) admissions, nor for respiratory admissions in the winter months (when people are
9 indoors and levels of these pollutants are low). While these analyses employed much more
10 sophisticated statistical methods, the results generally consistent with Bates and Sizto's prior
11 work in this region, though ozone was found to yield a larger effect than sulfates in this
12 study. The authors point out that PM2 5 and H+ are highly intercorrelated with sulfates in
13 the summer months (r > 0.8), and that one of these agents may be responsible for the health
14 effects relationships found with sulfates in this work.
15
16 12.5.3.2 Other Health Effects Related to Sulfate Exposures
17 Ostro (1987) also conducted a cross-sectional analysis of the U.S. Inhalable Particle
18 Monitoring Network airborne particulate matter dataset, but analyzed the 1979 to 1981
19 annual Health Interview Surveys (HIS) to test if there were acute morbidity associations
20 coherent with those found for mortality by Ozkaynak and Thurston (1987) during this period.
21 Ostro reported a stronger association between several measures of morbidity (work loss days,
22 restricted activity days, etc.) and lagged fine particle estimates than found with prior 2-week
23 average TSP levels in 84 U.S. cities. In this analysis, a Poisson model was employed, due
24 to the large number of zeros in the dependent variables (i.e. days with morbidity), and the
25 analyses focused on adults aged 18 to 65. Smoking was not considered in the model, since
26 not all metropolitan areas had data, but the correlation between smoking and any of the
27 pollutants was less than 0.03 and non-significant in the one-third of the HIS sample for which
28 smoking data were available. This indicates that, while presumably important to morbidity,
29 smoking is not a confounder to pollutants in such cross-sectional analyses. Ostro concluded
April 1995 12-209 DRAFT-DO NOT QUOTE OR CITE
-------
1141
CO
Q
104-
102-
•
i i i i i i i i i i
0 2 4 6 8 10 12 14 16 18 2(
Daily Average Sulphate Level (ng/rrP), Lagged One Day
Figure 12-12. Average number of adjusted respiratory admissions among all 168
hospitals by decile of the daily average sulfate level Otg/m3), lagged 1 day
(Burnett et al., 1994).
1
2
3
4
5
6
7
8
9
10
that his findings were consistent with the results of prior cross-sectional analyses reporting an
association between mortality and exposures to fine particles and sulfates.
12.5.3.3 Studies Relating Acute Health Effects to Acidic Aerosols
In recent years, a number of new studies have been conducted of acute health effects
employing direct measurements of particulate strong acid aerosols. These allow a more
direct test of the hypothesis that it is the H+ that is responsible for the sulfates-health effects
associations noted in past work.
April 1995
12-210 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.5.3.4 Acute Acidic Aerosol Exposure Studies of Children
2 Several studies have recently been carried out in the United States and Canada that
3 examine the effects of exposures to air pollutants on pulmonary function in children at
4 summer camps. Some of the available data derived from these studies allow evaluation of
5 the possible involvement of acid aerosols in the health effects observed. Furthermore, recent
6 children's diary studies have also investigated acid aerosol effects on respiratory symptoms in
7 the general population.
8
9 Studies of Pulmonary Function in Children at Summer Camp
10 Lippmann et al. (1983) studied 83 nonsmoking, middle class, healthy children (ages 8
11 to 13) during a 1980 2-week summer camp program in Indiana, PA. The children were
12 involved in camp activities which resulted in their exercising outdoors most of the time. At
13 least once, each child had height and weight measured and performed spirometry on an 8
14 liter Collins portable recording spirometer in the standing position without nose clip. During
15 the study, peak flow rates were obtained by Mini- Wright peak flow meter at the beginning of
16 the day or at lunch and adjusted for both age and height. Ambient air levels of TSP,
17 hydrogen ions, and sulfates were monitored by a high-volume sampler on the rooftop of the
18 day camp building. Ozone levels were estimated using a model that used ozone data from
19 monitoring sites located 32 and 100 km away. The hi- volume samples were collected on
20 H2SO4 treated quartz fiber filters for the determination of the concentration of H+ and total
21 suspended particulate matter (TSP). H+ was determined from filter extract using a Gran
22 titration. Peaks in acid concentration occurred on four days, when the acid values ranged
23 between 4 and 6.3 jug/m3 (if as H2SO4). On many occasions, there was no measurable
24 H2SO4 in the atmosphere. While effects were reported as being significantly associated with
25 exposure to ozone, no effects were found to be related to exposure to H2SO4 at the relatively
26 low levels observed during the study.
27 Bock et al. (1985) and Lioy et al. (1985) examined pulmonary function of 39 children
28 at a camp in Mendham, New Jersey during a 5-week period in July to August, 1982. Ozone
29 was continuously monitored using chemiluminescent analysis. Ambient aerosol samples were
30 collected on Teflon filters with a dichotomous sampler having a 15 /*m fractionation inlet and
April 1995 12-211 DRAFT-DO NOT QUOTE OR CITE
-------
1 a coarse/fine cut size of 2.5 ptm (Sierra Model 244-E). Aerosol acidity as measured by
2 strong acid (H+) content, was determined using the pH method. Highly significant changes
3 in peak expiratory flow rate (PEFR) were found to be related to ozone exposure, as well as a
4 baseline shift in PEFR lasting approximately one week following a haze episode in which the
5 O3 exposure exceeded the NAAQS for four consecutive days that included a maximum
6 concentration of 185 ppb. There was no apparent effect of H+ on pulmonary function. The
7 authors did state, however, that the persistent effects associated with the ozone episode could
8 have been due to acid sulfates as well as, or in addition to, ozone, but additional uncollected
9 data were needed to evaluate this possibility.
10 During a 4-week period in 1984, Lioy et al. (1987) and Spektor et al. (1988) measured
11 respiratory function of 91 active children who were residing at a summer camp on Fairview
12 Lake in northwestern New Jersey. Continuous data were collected for ambient temperature,
13 humidity, wind speed and direction, and concentrations of O3, H2SO4, and total sulfates were
14 determined. Ozone was measured by U.V. absorption, and H2SO4 and total sulfates were
15 alternately determined by a flame photometric sulfate analyzer (Meloy Model 285) preceded
16 by a programmed thermal pretreatment unit. The ambient aerosol samples were collected on
17 quartz fiber filters with a dichotomous sampler having a 15 fim fractionating inlet (PM15 and
18 a coarse/fine cut-size of 2.5 pm (Sierra Model 244-E). Aerosol acidity, as measured by
19 strong acid (H+) content, was determined using the pH method. The maximum values
20 recorded for H2SO4 and NH3HSO4 were 4 and 20 j«g/m3 respectively. While effects were
21 reported as being associated with exposure to ozone, no effects were found to be directly
22 related to exposure to the acid aerosol concentrations experienced in this study.
23 Raizenne et al. (1987) reported analyses of data from a study in Ontario, Canada. In
24 1983, fifty two campers (23 were asthmatics) at a summer camp were studied to examine
25 lung function performance in relation to daily pollutant concentrations. The health
26 assessment included a pre-camp clinical evaluation, a telephone administered questionnaire on
27 respiratory health, daily spirometry and symptoms measurements. Pollutants measured
28 included O3, respirable particles, sulfates, NO2, and SO2. Respirable sulfates were highly
29 variable and ranged from 10 to 26 ng/m3. Sulfate as sulfuric acid was usually very low.
30 Raizenne et al. (1989) report that O3, sulfate, and PM2 5 were associated with decrements in
April 1995 12-212 DRAFT-DO NOT QUOTE OR CITE
-------
1 lung function of children. Evidence of decrements in specific lung function indices were
2 related to current pollution levels and to a 12 to 24 h lag function for PM2 5, SO4, O3 and
3 temperature. Although both asthmatic and non-asthmatics had similar data trends, only
4 responses in the non-asthmatic group reached statistical significance. The authors note that
5 all of the air pollutants were highly correlated, and thus it was not possible to apportion
6 health effects to the individual pollutants.
7 Raizenne et al. (1989) studied 112 young girls who participated in one of three 2-week
8 camp sessions at camp Kiawa, Ontario, Canada during June to August, 1986. They
9 examined the subjects in relation to four ambient acid aerosol events (the highest H2SO4 level
10 was 47.7 /ig/m3 during one event on July 25, 1986). The influence of air pollution on lung
11 function was evaluated first by comparing responses on the day of a pollutant event (high
12 acid and ozone levels) to the mean of the responses on corresponding days of low pollutant
13 levels. For FEVj 0 there was tendency for the lung function decrements on the event day to
14 be greater than the response on the corresponding control days, except for the last event
15 (when an increase in function was observed). The largest decrements for FEVj 0 and PEFR
16 (48 to 66 mL decline for FEVj 0) were observed on the morning after the highest H2SO4
17 event, on July 25, 1986. No analyses were presented, however, that attempted to separate
18 out pollutant effects of H2SO4 from those of O3.
19 Airway hyper-responsiveness was assessed using a methacholine bronchial provocation
20 test for 96 of the subjects in the Raizenne et al. (1989) study. Children with a positive
21 response to methacholine challenge had larger decrements compared to their nonresponsive
22 counterparts. These preliminary results do not allow definitive statements to be made on the
23 susceptibility of methacholine sensitive subjects. However, there are indications in these data
24 of differential lung function profiles and responses to air pollutants in children with and
25 without airway hyper-responsiveness. Further analyses and research are indicated.
26 At the same camp, twelve young females (9 to 14 years old) performed pre- and post-
27 exercise spirometry on a day of low air pollution and at the peak of an air pollution episode.
28 Clinical interview, atopy, and methacholine airway hyper-responsiveness tests were
29 performed at the camp on the first 2 days of the study. Seven subjects had positive
30 responses to methacholine challenge (+MC) and five did not ("MC). A standardized
April 1995 12-213 DRAFT-DO NOT QUOTE OR CITE
-------
1 ergonometric physical capacity test was also administered, in which minute volume, heart
2 rates, and total work achieved were recorded. Air monitoring was performed on site and,
3 during the episode, air pollution concentrations were: O3 exceeded 130 ppb; H2SO4 exceeded
4 40 /-ig/m3 during a 1-h period. Additional discussion of the aerometric monitoring is given
5 by multivariate normal methods on the indicator of airway hyper-responsiveness. For the
6 entire group (N = 12), post exercise FVC and FEVj 0 were observed to increase on the
7 control day and decrease on the episode day. On the control day, an average 40 mL increase
8 in FVC due to exercise was observed (p < .05) for the whole group, with a 71 mL increase
9 in +MC subjects and a 17 mL increase in -MC subjects. Although not statistically
10 significant at the 10 percent level, mean FVC for the entire group was 30 ml less on the day
11 of high pollution versus low pollution, and this difference was more pronounced in -MC
12 (-65 mL) than +MC (-4 mL) subjects. The effect of exercise in the model was statistically
13 significant (p < .05), whereas the pollution day effect was not. These results suggest that
14 lung function responses to exercise differ in +MC and -MC subjects under field research
15 conditions, and that the expected normal FVC response to exercise in both groups is altered
16 during periods of elevated ambient pollution. However, no analyses were presented that
17 directly evaluated possible acid aerosol relationships to health effects.
18 It is of interest to compare results obtained in this summer camp study to findings of
19 certain controlled human exposure studies, or to other epidemiology studies. For example,
20 Spengler et al. (1989) calculated that the children in the Raizenne et al. (1989) study received
21 an average 1-h respiratory tract dose of 1050 nmoles of H+, based on a exposure model
22 which takes into account not only the concentration of exposure, but also minute ventilation
23 rate. Spengler et al. (1989) further noted that the asthmatic subjects in the human clinical
24 studies of Utell et al. (1983) and Koenig et al. (1983) had experienced an airway dose of
25 approximately 1,200 nmoles of H+, which evoked a response at reported concentrations of
26 450 /ig/m3 and 100 /^g/m3 H2SO4, respectively. These calculations suggest that, because of
27 differences in minute ventilation rates, the peak levels occurring at Camp Kiawa during an
28 ambient acid aerosol event may have produced exposures similar to those seen in clinical
29 studies of asthmatic subjects. It remains to be determined as to what extent comparable
30 C x T total respiratory tract dose(s) for H+ ions may be effective in producing pulmonary
April 1995 12-214 DRAFT-DO NOT QUOTE OR CITE
-------
1 functions decrements beyond the short exposure times employed in the controlled human
2 exposure studies or in producing other types of effects. For example, Spektor et al. (1989)
3 found that the effect of increasing the length of exposure to 100 jig/m3 sulfuric acid from one
4 to two hours increased average tracheobronchial clearance half-time from 100 to 162 percent,
5 relative to control.
6 Studnicka et al. (1995) conducted a study of the effects of air pollution on the lung
7 function of three consecutive panels of children participating in a summer camp in the
8 Austrian Alps during the summer of 1991. On-site environmental assessment consisted of
9 24-h measurements of PM10, H+, and SO^, as well as continuous measurements of O3,
10 temperature, and relative humidity. Pollen counts were sampled daily using a Burkhardt
11 spore trap. SO2 and NO2 data were obtained from routine monitoring stations located at the
12 same altitude 20 to 30 km from the camp. For 47, 45, and 41 subjects, daily FEV1, FVC,
13 and peak expiratory flow were recorded. While mean levels of ambient pollutants were
14 generally 15 % higher for Panel 1, the Panel 1 H+ concentrations averaged twice as high as
15 for the other two panels. The maximum H+ exposure (during Panel 1) was 84 nmol/m3
16 (4/ig/m3 H2SO4 equivalent). Compared with other camp studies discussed above, peak H+
17 exposure was of lesser concentration, but of longer duration. For FEV1, a significant
18 decrease of -.099 ml per nmol/m3 H+ (p = 0.01) during Panel 1. Exclusion of the first
19 5 days, or excluding the maximum H+ day did not significantly alter this result. The
20 FEV1/H+ coefficient was found to be similar (-0.74 ml per nmol/m3 H+; p = 0.28) for
21 Panel 2, but was in the opposite direction and clearly non-significant during Panel 3 (0.10 ml
22 per nmol/m3 H+; p = 0.83). The decrease in FEV1 during Panel 1 was more pronounced
23 when the mean exposure during the previous 4-d was employed (-2.99 m; FEV1 per
24 nmol/m3 H"1"; p = 0.004), suggesting greater effects from multiple-day episodes. However,
25 it is important to note that, while 03 levels were low and not significantly correlated with
26 FEV1 throughout this study, PM10 measurements showed associations of similar strength
27 with FEV1 during Panel 1 as found for H+ (rPM10_H+ = 0.94). However, in simultaneous
28 model of FEV1 on H+ with PM10, O3, and pollen in the model, the previous 4-d mean H+
29 variable's coefficient was of similar magnitude as for the single pollutant model (though the
30 coefficient SE did rise). This suggests that the H+ association with FEV1 remained, even
April 1995 12-215 DRAFT-DO NOT QUOTE OR CITE
-------
1 after controlling for other potentially confounding factors. The authors conclude that a
2 significant FEV1 decrease of 200 ml was observed in children at this camp during a summer
3 haze episode in the Austrian Alps, and the acidic PM may, therefore, be associated with
4 transient decreases in lung function in children.
5
6 Studies of Respiratory Symptoms and Pulmonary Function in Schoolchildren
1 As part of the 6-Cities study conducted by Harvard University, a cohort of
8 approximately 1800 children in grades two through five from six U.S. cities (Watertown,
9 MA; Kingston-Harriman, TN; St. Louis, MO; Portage, WI; Steubenville, OH, and; Topeka,
10 KS) was enrolled in a diary study in which parents completed a bi-weekly report on each
11 child's daily respiratory symptoms (Schwartz et al., 1994). The study extended over
12 4 school years (1984 to 1988), but data were collected for only one year in each city.
13 Environmental variables measured daily at a central site in each city included PM10, PM2 5,
14 PM2 5 sulfur, H+, H2SO4, SO2, O3, and nephelometry (a measurement of aerosol scattering
15 of light, which provides an index of sub-micron particle concentration). The H2SO4 data
16 were not analyzed in this work. The reported analysis was limited to April through August
17 in each city to reduce seasonal confounding (n = 153). Statistical analyses involved the use
18 of ordinary logistic regression, in which the logarithm of the odds of the response rate is
19 modeled as a linear function of covariates, followed by the application of logistic methods
20 incorporating corrections for autocorrelation using the GEE model proposed by Liang and
21 Zeger (1986) and Zeger and Liang (1986) for such repeated measures studies. Regressions
22 included a temperature and a temperature squared term, as well as city-specific and day of
23 week dummy variables and interaction terms for city-specific temperature terms.
24 Exploratory analyses considered pollution lags of up to 14 days. Pollutants were considered
25 individually in the regressions, and those which were significant individually were considered
26 in multiple pollutant models.
27 Lower respiratory symptoms (LRS) is defined as the reporting of at least two of:
28 cough, chest pain, phlegm, or wheeze. Analyses of daily LRS found in individual pollutant
29 regressions that PM10, PM25, PM25 sulfur (i.e., sulfates), nephelometry, SO2, and O3 were
30 all significant predictors. Of all these pollutants, PM2 5 sulfur (i.e., sulfates) and PM10
April 1995 12-216 DRAFT-DO NOT QUOTE OR CITE
-------
1 yielded the highest levels of significance (t = 3.35 and t = 3.47, respectively), suggesting
2 that it is the sulfur containing fine aerosol component which was driving the PM relationships
3 found with LRS. In the overall data analysis, aerosol acidity was not significantly associated
4 with LRS, but associations were noted for H+ above 110 nmoles/m3, with a relative odds
5 ratio of LRS estimated to be greater than 2.0 at 300 nmoles/m3 H+. Similarly, the 6-City
6 diary analysis of upper respiratory symptoms (URS, defined as any two of hoarseness, sore
7 throat, or fever) showed no consistent association with H+ until concentrations exceeded 110
8 nmoles/m3. The authors hypothesize that the acid may enhance the effects of PM10 on such
9 subjects (i.e. healthy children) if the H+ is present in high enough concentration.
10 A separate analysis of upper respiratory symptoms was also conducted using similar
11 data and methods for three of the cities only: Watertown, MA; Kingston-Harriman, TN, and;
12 St. Louis, MO (Schwartz et al., 1991b). Why these cities alone were analyzed in this work
13 is not stated. In these cities, the pollutant with the largest regression coefficient was H2SO4 ,
14 with the strongest association falling on the prior two days. Unfortunately, comparative
15 details about other pollutants are not provided in this paper. While sketchy, these results
16 provide further support of the hypothesis that ambient acid aerosols in general, and H2SO4 in
17 particular, may be associated with adverse health effects in children.
18 In a study of ambient air pollution and lung function in children reported by Neas et al.
19 (1995), a stratified sample of 83 children living in Uniontown, PA reported twice daily peak
20 expiratory flow rate (PEFR) measurements on 3,582 child-days during the summer of 1990.
21 Upon arising and before retiring, each child recorded the time, three PEFR measurements,
22 and the presence of cold, cough, or wheeze symptoms. Environmental factors were
23 monitored, including ambient temperature, O3, SO2, fine particle mass, PM10, and particle
24 strong acidity, which was measured separately during the day (8 am to 8 pm) and night.
25 Each child's maximum PEFR for each session was expressed as the deviation from their
26 mean PEFR over the study and adjusted to a standard of 300 liters/minute. The session-
27 specific average deviation was then calculated across all the children. A second order
28 autoregressive model for PEFR was developed which included a separate intercept for
29 evening measurements, trend, temperature, and 12-hour average air pollutant concentration
30 weighted by the number of hours each child spent outdoors during the previous 12-hour
April 1995 12-217 DRAFT-DO NOT QUOTE OR CITE
-------
1 period. A 12-hour exposure to a 125 nmole/m3 increment in H+ was associated with a
2 -2.5 liters/minute deviation in the group mean PEFR (95% CI = -4.2 to -0.8) and with
3 increased cough incidence (odds ratio, OR = 1.6; 95% CI = 1.1 to 2.4). It should be
4 noted, however, that H+ was highly correlated with sulfates (r = 0.92) and fine particles
5 (r = 0.86). A 30 ppb increment in ozone for 12-hours was associated with a similar
6 deviation in PEFR levels (-2.8; 95% CI = -6.7 to 1.1). However, when both O3 and H+
7 were entered into the model simultaneously, the H+ effect size was only slightly reduced and
8 remained significant. Although monitored, PM10 results were not presented for comparison.
9 The association between PEFR and particle strong acidity was observed among the 60
10 children who were reported as symptomatic on the prior symptom questionnaire (-2.5;
11 95% CI = -4.5 to -0.5). The authors concluded that summertime occurrences of elevated
12 acid aerosol and particulate sulfate pollution are associated with acute declines in peak
13 expiratory flow rates and increased incidence of cough episodes in children.
14 Overall, these recent camp and school children studies proved evidence indicating an
15 acute acidic PM effect on both children's respiratory function and symptoms. However,
16 given the usually high correlation between acidic PM and PM in general, it is difficult to
17 identify these effects solely with the acid portion of PM.
18
19 12.5.3.5. Acute Acid Aerosol Exposure Studies of Adults
20 Acute Acid Aerosol Exposures and Asthma Symptoms in Adults
21 The hypothesis that human exposures to ambient H+ concentrations are associated with
22 exacerbations of pre-existing respiratory disease was tested by a recent study of asthmatic
23 responses to airborne acid aerosols (Ostro et al., 1989, 1991). Data on daily concentrations
24 of aerosol H+, SO4, NO3, and FP, as well as gaseous SO2 and HNO3, were tested for
25 correlation with daily symptom, medication usage, and other variables for a panel of
26 207 adults with moderate to severe asthma in Denver, CO between November 1987 and
27 March 1988. However, CO and NO2, potentially confounding pollutants, were not
28 considered in the analyses. The H+ concentrations ranged from 2 to 41 neq/m3 (0.01 to 2.0
29 ng/m3 of H2SO4 equivalent), and were significantly related to both the proportion of the
30 survey respondents reporting a moderate or worse overall asthma condition, and the
April 1995 12-218 DRAFT-DO NOT QUOTE OR CITE
-------
1 proportion reporting a moderate or worse cough. However, it is important to note that these
2 concentrations are near to or below the level of detection of H+, and that, of the
3 74 H+ values used in the analysis, 47 were predicted from the observed SO4= value on that
4 day (H+ - SO4= correlation = 0.66), which is more accurately measured at such low levels.
5 PM2 5 was also highly correlated with sulfates during this study (r = 0.86). Both logit
6 models and ordinary least squares with a log pollution term, autoregressive terms, and terms
7 for trend, weekend, use of gas stove, and maximum daily temperature were modeled.
8 Of all the pollutants considered in these analyses, H+ displayed the strongest
9 associations with asthma and cough. In the first analysis, the magnitudes of effects were
10 compared by computing elasticities, or the percent change in the health effect due to a given
11 percent change in the pollutant. The results for asthma indicated elasticities with respect to
12 SO4, FP, and H+ of 0.060, 0.055 and 0.096, respectively (Ostro et al., 1989). This
13 indicates that a doubling of the concentration of H+ (from 8 to 16 nmoles/m3) would
14 increase the proportion reporting a moderate to severe asthma condition by 10 percent. In
15 their follow-up report on this study, Ostro et al. (1991) examined evidence for lagged effects,
16 and concluded that contemporaneous measures of H+ concentration provided the best
17 associations with asthma status, and that meteorological variables were not associated with
18 the health effects reported. They also examined the effects of exposure to H+, adjusting for
19 time spent outdoors, level of activity, and penetration of acid aerosol indoors. Based on the
20 adjusted exposures, the effect of H+ on cough increased 43%, suggesting that dose-response
21 estimates that do not incorporate behavioral factors affecting actual H+ exposures may
22 substantially underestimate the impact of the pollution. The associations of exposure adjusted
23 H+ with moderate to severe cough and with asthma status are shown in Figures 12-13 and
24 12-14, respectively. Although the H+ concentrations on some days had to be estimated from
25 sulfates, and potentially confounding pollutants were not considered simultaneously with
26 H+ in the model, these results allow the consideration that human exposures to present day
27 ambient H+ concentrations may be associated with exacerbations of pre-existing respiratory
28 disease.
April 1995 12-219 DRAFT-DO NOT QUOTE OR CITE
-------
0.24-
0.22-
.c 0.20H
O)
8
£
I
2
o>
•o
o
0.16H
CO
_a
o
0.14-1
0.12-
0.10-
00
o o0
o
o o
o
o
O
Note: Each point represents the average
of 212 contiguous observations.
100
1,000
Exposure-adjusted hydrogen ion (neq)
10,000
Figure 12-13. Association of moderate or severe cough with exposure-adjusted hydrogen
ion.
Source: Ostro et al. (1991).
April 1995
12-220 DRAFT-DO NOT QUOTE OR CITE
-------
0.26-
0.24-
o>
c
1
j? 0.22-
I
8>
1
2 0.20H
o
I
-------
1 12.5.4.2.2 Acute Acidic Aerosol Associations with Respiratory Hospital Admissions
2 The reported sulfate-respiratory hospital admissions associations discussed above were
3 interpreted as potentially being due to the presence of strongly acidic aerosols on high sulfate
4 days. Two follow-up studies of respiratory hospital admissions were conducted in New York
5 State and in Toronto, Ontario to directly test this hypothesis.
6 Thurston et al. (1992) analyzed unscheduled (emergency) admissions to acute care
7 hospitals in three New York State metropolitan areas during the summers of 1988 and 1989.
8 Environmental variables considered included daily 1-h maximum ozone, 24-h average sulfate,
9 and particulate strong acid aerosol (H+) concentrations, as well as daily maximum
10 temperature recorded at central sites in each community. For this study, acid aerosols were
11 sampled in residential suburbs of Buffalo, Albany, and New York City (NYC), NY. In the
12 case of NYC, the site was located well outside the urban core (in White Plains, 10 mi. north
13 of the city), so the acid levels are likely to be overestimates of the levels experienced directly
14 in the city. However, comparisons between sulfates in the White Plains site and at a site in
15 Manhattan during part of the study period showed a high correlation (r = 0.9), supporting
16 the assumption that the White Plains H+ data are indicative of particulate strong acid
17 exposures in NYC. Long wave periodicities in the data were reduced by selecting a June
18 through August study period. However, because of remaining within-season long wave
19 cycles in the data series, they were prefiltered using sine and cosine waves with annual
20 periodicities. Day of week effects were also controlled via regression. These adjustments
21 resulted in non-significant autocorrelations in the data series and also improved the pollution
22 correlations with admissions. The strongest pollutant-respiratory admissions associations
23 found by Thurston et al. (1992) were during the high pollution 1988 summer, and in the
24 most urbanized communities considered (i.e. Buffalo and New York City). Correlations
25 between the pollution data and hospital admissions for non-respiratory control diseases were
26 non-significant both before and after prefiltering. After controlling for temperature effects
27 via simultaneous regression, the summer haze pollutants (i.e. SO
-------
1 considered, but are generally low and unlikely to be highly correlated with the studied
2 pollutants during July and August in these cities. After filtering, SO4= and H+ were highly
3 correlated in these cities (e.g. r = 0.86 in Buffalo, and 0.79 in NYC during the summer of
4 1988), supporting the contention that SO4 is a useful index of H+ in such time-series
5 analyses. In regressions for the summer of 1988 for Buffalo and New York City, both
6 H+ and SO4= had similar mean effects (3 to 4% of respiratory admissions in NYC, at mean
7 H+ = 2.4 jug/m3 as H2SO4, and mean SO4= =9.3 jug/m3; and 6 to 8% in Buffalo, at mean
8 H+ = 2.2 /ig/m3 as H2SO4, and mean SO4= =9.0 /ig/m3). Ozone mean effects estimates
9 were always larger than for H+ or SO4=, but the impact of the highest day was greatest for
10 H+ in all cases. This is the case in part because H+ episodes are more extreme, relative to
11 the mean, than are O3 episodes (e.g. in Buffalo in 1988, the summer max./mean H+ = 8.5,
12 while the max./mean O3 = 2.2). Thus, the maximum H+ day in Buffalo (18.7 /ig/m3 as
13 H2SO4, or 381 nmoles H+/m3, on August 4, 1988), was estimated to be associated with
14 a 47% increase above the mean number of total respiratory admissions in this metropolitan
15 area (mean = 25/day). Thus, the H+ effects estimates reported in this work are dominated
16 by the two or three peak H+ days per year experienced in these cities (e.g. H+ > 10 /ig/m3,
17 or -200 nmoles/m3, as a 24-h average).
18 Thurston et al. (1994b) focused their analysis of respiratory hospital admissions in the
19 Toronto metropolitan area during the summers (July to August) of 1986 to 1988, when they
20 directly monitored for strong paniculate acidity (H+) pollution on a daily basis in that city.
21 This study was designed specifically to test the hypothesis that the SO4= associations found in
22 southern Ontario by Bates and Sizto were due to H+ exposures. Acid measurements were
23 made at three sites in the Toronto metropolitan area, and were found to be highly correlated
24 across sites (Thurston et al., 1994a). The H+ data from the center city site (Breadalbane St.)
25 were used for the health effects analyses, as there were a full 3 summers of data there (the
26 other two sites were not operated in 1988), and because other pollutants were measured there
27 daily, as well. The 9AM to 5PM average H+ was employed in these analyses. Long wave
28 cycles, and their associated autocorrelations, were removed by first applying an annual
29 periodicity sine-cosine fit to the data (as well as day of week dummy variables) and analyzing
30 the resulting residuals. Strong and significant positive associations with both asthma and
April 1995 12-223 DRAFT-DO NOT QUOTE OR CITE
-------
1 respiratory admissions were found for both 03 and H+, and somewhat weaker significant
2 associations with SO4=. No such associations were found for SO2 or NO2, nor for any
3 pollutant with non-respiratory control admissions. Other PM metrics examined included the
4 mass of fine particles less than 2.5 ^m in d., (FP), the mass of particles greater than 2.5 /um
5 and less than 10 urn in da (CP), PM10 (= FP+CP), TSP, and non-thoracic TSP
6 (= TSP-PM10). Temperature was only weakly correlated with respiratory admissions, and
7 became non-significant when entered in regressions with air pollution indices.
8 Simultaneous regressions and sensitivity analyses indicated that O3 and H+ were the
9 summertime haze constituents of greatest importance to respiratory and asthma admissions in
10 Toronto during these three summers. Indeed, as shown in Table 12-23, of the PM metrics
11 considered, only H+ remained significant in the respiratory admissions regression with both
12 O3 and temperature also included. The correlation of the H+ and O3 coefficients in this
13 simultaneous model was non-significant (r=-0.11), indicating that these two pollutants have
14 independent associations with respiratory admissions. As shown in Table 12-24, the 1988
15 results for Toronto are consistent with (i.e. not statistically different from) those found
16 previously for nearby Buffalo, NY (approximately 100 km to the south, across Lake
17 Ontario). As in these authors' Buffalo analysis, the maximum H+ day in Toronto (August 4,
18 1988: H+ = 391 nmoles/m3) was estimated to be associated with the greatest relative risk of
19 total respiratory and asthma admissions (1.50 and 1.53, respectively), again indicating an
20 especially large adverse respiratory effect by summertime haze air pollutants during the few
21 H+ episode days each summer. However, a sensitivity analysis eliminating the six days
22 having H+ > 100 nmoles/m3 yielded a similar, and statistically significant, H+ coefficient in
23 the total respiratory admissions regression, suggesting that the association is not limited to
24 the highest pollution days alone. The authors reviewed A.B. Hill's criteria for causality
25 (Hill, 1965), and concluded that the associations they report between summertime haze air
26 pollutants (i.e. O3 and H+) and acute exacerbations of respiratory disease (i.e. respiratory
27 hospital admissions), are causal. It is of particular interest to note that, assuming the H+ to
28 be in the form of NH4HSO4, the "effect" per /*g/m3 of mass implied by these Toronto
29 coefficients indicate that H+ is six times as potent (per /ig/m3) as non-acidic PM10.
April 1995 12-224 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-23. SIMULTANEOUS REGRESSIONS OF 1986 TO 1988 TORONTO
DAILY SUMMERTIME TOTAL RESPIRATORY ADMISSIONS ON TEMPERATURE
AND VARIOUS POLLUTION METRICS
Temp, pollutant model Pollutant Regression Coefficients
specification (adm/poll unit3) P value (one-way)
Two pollutant models
T(LGO), O3(LGO)
H+(LG1)
T(LGO), 03(LGO)
SO4=(LG1)
T(LGO), O3(LGO)
FP(LGO)
T(LGO), O3(LGO)
PM10(LGO)
T(LGO), O3(LGO)
TSP(LGO)
0.0503 ± 0.0205
0.0153 ± 0.0089
0.0508 ± 0.0207
0.0062 ± 0.0046
0.0404 ± 0.0233
0.0434 ± 0.0429
0.0388 ± 0.0241
0.0339 ± 0.0344
0.0360 ± 0.0228
0.0127 ± 0.0175
0.008
0.044
0.008
0.089
0.043
0.157
0.055
0.164
0.059
0.235
Pollution units are nmole/m3 for H+ and S04-, ppb for 03, and jig/m3 for FP, CP, PM10, TSP, and
TSP-PM10.
Source: Thurston et al. (1994b)
1 These two new studies of daily respiratory hospital admissions in New York State cities
2 and in Toronto, Ontario support the hypothesis that the summertime sulfate concentrations
3 previously found to be correlated with respiratory admissions are indeed accompanied by
4 acidic aerosols in Eastern North America. Furthermore, in these recent analyses, the
5 H+ associations with respiratory hospital admissions were found to be stronger than for
6 sulfates, or any other PM component monitored. The facts that: (1) these were studies
7 designed specifically to test the hypothesis that H+ is associated with increased respiratory
8 hospital admissions; (2) consistent results were found, both qualitatively and quantitatively
9 across these studies, and; (3) in one of them, many other pollutants and PM metrics were
10 directly intercompared with H+ in the analyses, collectively indicate that these studies
11 provide evidence that acidic aerosols may represent a component of PM which is particularly
12 associated with increases in the incidence of exacerbations in pre-existing respiratory disease.
April 1995 12-225 DRAFT-DO NOT QUOTE OR CITE
-------
I
to
10
TABLE 12-24. COMPARISON OF REGRESSIONS OF DAILY SUMMERTIME
RESPIRATORY ADMISSIONS ON POLLUTION AND TEMPERATURE IN
TORONTO, ONTARIO, AND BUFFALO, NEW YORK 1988 SUMMER
City and year
Toronto,
1988 summer
Toronto,
1988 summer
Buffalo,
1988 summer
Buffalo,
1988 summer
Respiratory admissions
category
Total respiratory
(mean = 14. I/day)
Total asthma
(mean = 9.5/day)
Total respiratory
(mean = 25.0/day)
Total asthma
(mean = 7.1 /day)
Pollutant Regression
Temp, pollutant model Coefficient (adm//xg/m3/106,
specification persons ±SE)
T(LG2), SO4=(LG1)
T(LG2), H+(LG1)
T(LG2), 03(LG1)
T(LG2), SO4=(LG1)
T(LG2), H+(LGO)
T(LG2), 03(LG1)
T(LG2), SO4=(LGO)
T(LG2), H+(LGO)
T(LG2), O3(LG2)
T(LG2), SO4=(LG1)
T(LG2), H+(LG1)
T(LG2), O3O(LG3)
0.07 ± 0.03a
0.18 ± 0.009b
0.011 ± 0.005b
0.04 ± 0.02b
0.13 ± 0.07b
0.007 ± 0.004b
0.11 ± 0.04a
0.35 ± 0.12a
0.015 ± 0.008b
0.03 ± 0.02b
0.09 ± 0.05b
0.006 ± 0.002a
Pollutant mean effect
(% ±SE)
13.3 ± 5.3
7.7 ± 3.9
26.4 ±11.8
13.0 ± 6.8
8.1 ± 4.5
25.3 ± 14.9
8.0 ± 2.7
6.4 ± 2.2
18.4 ± 9.9
7.0 ± 3.9
5.6 ± 3.3
23.9 ± 10.1
Max/mean pollutant
rel risk (+SE)
1.41 ± 0.16
1.50 ± 0.25
1.34 ± 0.15
1.40 ± 0.21
1.53 ± 0.29
1.32 ± 0.19
1.22 ± 0.12
1.47 ± 0.16
1.25 ± 0.09
1.29 ± 0.12
1.43 ± 0.26
1.25 ± 0.14
o
H
O
c!
O
a
aP<0.01 (one-way test).
bP<0.05 (one-way test).
H Source: Thurston et al. (1994b).
6
o
n
HH
s
-------
1 12.5.4.2.3 Acute Acid Aerosol Exposure Associations with Mortality
2 As discussed in the methodological discussions at the outset of this chapter, relatively
3 long records of daily mortality and pollution are required to have sufficient power to discern
4 mortality-pollution associations. Due to the dearth of sufficiently long records of
5 H+ concentration measurements (other than the historical London measurements discussed
6 previously), only one study has yet attempted to evaluate the acute mortality effects of acidic
7 aerosols.
8 Dockery et al. (1992) investigated the relationship between multiple air pollutants and
9 total daily mortality during the one year period between September 1985 and August 1986 in
10 two communities: St. Louis, MO; and Kingston/Harriman, TN and surrounding counties.
11 In the latter locale, the major population center considered is Knoxville, TN, some 50 Km
12 from the air pollution monitoring site employed. In each study area, total daily mortality
13 was related to PM10, PM25, SO2, NO2, O3, SO4=, H+, temperature, dew point, and season
14 using autoregressive Poisson models. In St. Louis, after controlling for weather and season,
15 statistically significant associations were found with both prior day's PM10 and PM2 5, but
16 not with any lags of the other pollutants considered. In the Kingston/Harriman vicinity,
17 PM10 and PM2 5 approached significance in the mortality regression, while the other
18 pollutants did not. In both cities, very similar PM10 coefficients are reported, implying a 16
19 to 17 percent increase in total mortality per lOO^g/m3 of PM10. While autocorrelation was
20 accounted for, seasonality was addressed by season indicator (dummy) variables, which may
21 not remove within-season long wave influences. However, the chief areas of concern
22 regarding this study relate to the exposure data. In both places, only one daily monitoring
23 station was employed to represent community exposure levels, and no information regarding
24 the representativeness of these sites are provided (e.g., correlations with other sites' data).
25 More importantly in the case of H+ analyses, the number of days for which pollution data
26 are available for time-series analyses is limited in this data set (e.g. only 220 days had H+
27 values at the St. Louis site). As discussed in the methodological section, it is expected that
28 roughly at least twice this number of study days are needed to be able to reliably detect PM
29 associations with mortality. Thus, in the words of the authors: "Because of the short
April 1995 12-227 DRAFT-DO NOT QUOTE OR CITE
-------
1 monitoring period for daily particulate air pollution, the power of this study to detect
2 associations was limited."
3 Thus, the only available attempt to correlate human mortality with present day ambient
4 acid aerosol concentrations was unable to find a significant association, but it is not clear to
5 what extent this result was due to the severe lack of power in the analysis (because of the
6 many fewer H+ observations than available for other pollutants). Clearly, there is a critical
7 need for present day replications of the London mortality-acid aerosol studies to be
8 conducted, in order to determine whether these London associations (dominated by
9 wintertime H+, occurring in reduction-type atmospheres) are pertinent to the U.S., where
10 acid aerosol peaks occur primarily in the summertime, in oxidation-type atmospheres.
11
12 12.5.4 Studies Relating Health Effects to Long-Term Exposure
13 A limited but growing amount of epidemiologic study data currently exist by which to
14 evaluate possible relationships between chronic exposures to ambient acid aerosols and
15 human health effects. These include one study from Japan relating effects to estimated or
16 measured acidity, and many other North American studies which relate effects to sulfate
17 levels or other surrogate measures thought to roughly parallel acid aerosol concentrations.
18 Moreover, newer epidemiologic studies, which consider measured acid aerosols, now provide
19 more direct insight into the potential chronic effects of particulate strongly acidic aerosols.
20
21 12.5.4.1 Acid Mists Exposure in Japan
22 Kitagawa (1984) examined the cause of the Yokkaichi asthma events (1960 to 1969) by
23 examining the potential for exposure to concentrated sulfuric acid mists and the location and
24 type of health effects noted. He concluded that the observed respiratory diseases were due
25 not to sulfur dioxide, but to concentrated sulfuric acid mists emitted from stacks of calciners
26 of a titanium oxide manufacturing plant located windward of the residential area. This was
27 based on the fact that the SO3/SO2 ratio of 0.48 was much higher than the normal range of
28 0.02 to 0.05. The higher ratio indicates a higher acid aerosol level. The acid particles were
29 fairly large (0.7 to 3.3 /*m) compared with acid aerosols usually seen in the United States of
30 America (see Chapter 3), but were still were in the respirable range. Between 1960 and
April 1995 12-228 DRAFT-DO NOT QUOTE OR CITE
-------
1 1969, more than six hundred patients with respiratory disease were found to have chronic
2 bronchitis, allergic asthmatic bronchitis, pulmonary emphysema and sore throat. In 1969,
3 measures of acid aerosol exposures were obtained from litmus paper measurements collected
4 near the industrial plant which showed that acid mist particles were distributed leeward of the
5 industrial plant. The author notes that the physiological effects of concentrated sulfuric acid
6 mists (per estimated mass concentration) may be quite different from that of dilute sulfuric
7 acid mists formed by atmospheric oxidation of sulfur dioxide, and that the distinction
8 between the two types of acid mists is very important. It should be noted that morbidity fell
9 markedly after the installation of electrostatic precipitators which reduced H2SO4 and other
10 paniculate matter emissions.
11
12 12.5.5.2 Studies Relating Chronic Health Effects to Sulfate Exposures
13 Franklin et al. (1985) and Stern et al. (1989) reported on a cross-sectional
14 epidemiologic study investigating the respiratory health of children in two Canadian
15 communities that was conducted in 1983 to 1984, in Tillsonburg, Ontario and Portage la
16 Prairie, Manitoba. There were no significant local sources of industrial emissions in either
17 community. Seven hundred and thirty-five children aged 7 to 12 were studied in the first
18 town and 895 in the second. Respiratory health was assessed by the measurement of the
19 forced vital capacity (FVC) and forced expiratory volume in 1 s (FEVj 0) of each child, and
20 by evaluation of respiratory symptoms and illnesses using a questionnaire self-administered
21 by the parents. While NO2 and inhalable particles (PM10) differed little between these
22 communities, SO2, SO4, and NO3 were higher in Tillsonburg. Historical data in the vicinity
23 of Tillsonburg indicate that average levels of sulfates, total nitrates and ozone (O3) did not
24 vary markedly in the 9-year period proceeding the study. The results show that Tillsonburg
25 children had statistically significantly (p < 0.001) lower levels of FVC and FEVj 0 than
26 the children in Portage la Prairie (2% and 1.7% lower, respectively). These differences
27 could not be explained by parental smoking or education, cooking or heating fuels, pollution
28 levels on the day of testing or differences in age, sex, height or weight. The differences
29 persisted when children with either cough with phlegm, asthma, wheeze, inhalant allergies or
30 hospitalization before age 2 for a chest illness were excluded from analysis. With the
April 1995 12-229 DRAFT-DO NOT QUOTE OR CITE
-------
1 exception of inhalant allergies, which occurred more frequently in Tillsonburg children, the
2 prevalence of chronic respiratory symptoms and illnesses was similar in the two
3 communities. Thus, sulfates were among the pollutants which were higher in the community
4 experiencing reduced lung function and increased inhalant allergies, while PM10 mass
5 concentrations were not different between cities.
6 Ware et al. (1986) have reported results of analyses from the ongoing Harvard study of
7 outdoor air pollution and respiratory health status of children in six eastern and midwestern
8 U.S. cities. Between 1974 and 1977, approximately 10,100 white preadolescent children
9 were enrolled in the study during three successive annual visits to the cities. On the first
10 visit, each child underwent a spirometric examination, and a parent completed a standardized
11 questionnaire regarding the child's health status and other important background information.
12 Most of the children (8,380) were seen for a second evaluation one year later.
13 Measurements of TSP, SO4, and SO2 concentrations at study-affiliated outdoor stations were
14 combined with data from other public and private monitoring sites to create a record of
15 pollutant levels in each of nine air pollution regions during a one-year period preceding each
16 evaluation, and for TSP during each child's lifetime up to the time of evaluation. Annual
17 mean TSP levels ranged from 32 to 163 ^g/m3. Sulfur dioxide levels ranged from 2.9 to
18 184 /ig/m3, and sulfate levels ranged from 4.5 to 19.3 /ig/m3.
19 Analyzing these data across all six cities, Ware et al. (1986) found that frequency of
20 chronic cough was significantly associated (p < 0.01) with the average of 24-h mean
21 concentrations of TSP, SO2, and SO4 air pollutants during the year preceding the health
22 examinations. Furthermore, rates of bronchitis and a composite measure of lower respiratory
23 illness were significantly (p < 0.05) associated with annual average particle concentrations.
24 However, within the individual cities, temporal and spatial variation in air pollutant levels
25 and symptom or illness rates were not found to be significantly associated. The history of
26 early childhood respiratory illness for lifetime residents was significantly associated with
27 average TSP levels during the first two postnatal years within cities, but not between cities.
28 Also, pulmonary function parameters (FVC and FEVj 0) were not associated with pollutant
29 concentrations during the year immediately preceding the spirometry test or, for lifetime
30 residents, with lifetime average concentrations. Ferris et al. (1986), however, reported a
April 1995 12-230 DRAFT-DO NOT QUOTE OR CITE
-------
1 small effect on lower airway function (MMEF) related to fine particle concentrations.
2 Spengler et al. (1986) report the occurrence of acid aerosol peak concentrations of 30 to
3 40 jig/m3 (1 h average) in two of the cities during recent monitoring. Overall, these results
4 appear to suggest that risk may be increased for bronchitis and some other respiratory
5 disorders in preadolescent children at moderately elevated levels of TSP, SO4, and SO2
6 concentrations, which do not appear to be consistently associated with pulmonary function
7 decrements. However, the lack of consistent significant associations between morbidity
8 endpoints and air pollution variables within individual cities argues for caution in interpreting
9 these results.
10 Dockery et al. (1989) presented further results from the cross-sectional assessment of
11 the association of air pollution with chronic respiratory health of children participating in the
12 Six Cities Study of Air Pollution and Health. Air pollution measurements collected at
13 quality-controlled monitoring stations included total suspended paniculate matter (TSP),
14 paniculate matter less than 15 pm (PM15) and 2.5 ^m (PM2 5) aerodynamic diameter, fine
15 fraction aerosol sulfate (SO^), SO2, O3, and NO2. This analysis was restricted to the 5,422
16 10 to 12 years old white children examined in the 1980 to 1981 school year. Five
17 respiratory illness and symptom responses obtained by questionnaire were considered:
18 bronchitis, cough, chest illness, wheeze, and asthma. Each symptom was analyzed using a
19 logistic regression model including sex, age, indicators of parental education, maternal
20 smoking, gas stove, and city. Reported rates of bronchitis, chronic cough, and chest illness
21 during the 1980 to 1981 school year were positively associated with all measures of
22 paniculate pollution (TSP, PM15, PM2 5, and SO4) and positively, but less strongly,
23 associated with concentrations of two of the gases (SO^ and NO2). For children experiencing
24 wheeze, the estimated relative odds (and 95% CI) for SO^ between the most and least
25 polluted cities were: 3.1 (0.6 to 16.8) for bronchitis; 2.4 (0.1 to 60.6) for chronic cough,
26 and; 2.9 (0.5 to 15.6) for chest illness. Frequency of earache also tended to be associated
27 with paniculate concentrations, but no significant associations were found with asthma,
28 persistent wheeze, hay fever, or non-respiratory illness. No associations were found between
29 pollutant concentrations and any of the pulmonary function measures considered (FVC,
30 FEVj 0, FEV0.75, and MMEF). Children with a history of wheeze or asthma had a much
April 1995 12-231 DRAFT-DO NOT QUOTE OR CITE
-------
1 higher prevalence of respiratory symptoms, and there was some evidence that the association
2 between air pollutant concentrations and symptom rates was stronger among children with
3 these markers for hyperreactive airways. Results suggest that children with hyperreactive
4 airways may be particularly susceptible to other respiratory symptoms when exposed to these
5 pollutants. The lack of statistical association between pollutant concentrations and measures
6 of both pulmonary flow and volume suggests, however, that these increased rates of illness
7 are not associated with permanent loss of pulmonary function, at least during the
8 preadolescent years. Overall, these data provide further evidence that rates of respiratory
9 illnesses and symptoms are elevated among children living in cities with high particulate
10 pollution, including sulfates, which are known to be correlated over time and across cities
11 with H+, based on direct SOJ and H+ monitoring subsequently conducted in each of these
12 cities as part of this study.
13 Dodge et al. (1985) reported on a longitudinal study of children exposed to markedly
14 different concentrations of SO2 and moderately different levels of particulate sulfate in
15 Southwestern U.S. towns. In the highest pollution area, the children were exposed to 3 h
16 peak SO2 levels exceeding 2,500 jttg/m3 and annual mean particulate sulfate levels of
17 10.1 /ig/m3. The prevalence of cough (measured by questionnaire) correlated significantly
18 with pollution levels (chi-square for trend = 5.6, p = 0.02). No significant differences
19 existed among the groups of subjects over 3 years, and pulmonary function and lung growth
20 over the study were roughly equal over all groups. The results tend to suggest that
21 intermittent high level exposure to SO2, in the presence of moderate particulate sulfate levels,
22 produced evidence of bronchial irritation (increased cough), but no chronic effect on lung
23 function or lung function growth. These results suggest a bronchitis - H+ relationship,
24 assuming that SO2 or sulfates are indicative of acidic aerosols.
25 Chapman et al. (1985) report the results of a survey done in early 1976 that measured
26 the prevalence of persistent cough and phlegm among 5,623 young adults in four Utah
27 communities. The communities were stratified to represent a gradient of sulfur oxides
28 exposure. Community specific annual mean SO2 levels had been 11, 18, 36, and 115 ^g/m3
29 during the five years prior to the survey. The corresponding annual mean sulfate levels were
30 5,7, 8, and 14 /jg/m3. No gradients for TSP or suspended nitrates were observed. The
April 1995 12-232 DRAFT-DO NOT QUOTE OR CITE
-------
1 analyses were made using multiple logistic regression, in order to adjust for confounding
2 factors such as smoking, age and education. Persistent cough and phlegm rates in fathers
3 were about 8 percent in the high SO2/SO4 exposure community, versus about 3 percent in the
4 other communities. For mothers, the rates in the high SO2/SO4 exposure community were
5 about 4 percent, as opposed to about 2 percent in the other communities. Both differences
6 were statistically significant, suggesting that communities with higher SO2 and SO4 pollution,
7 which may be associated with higher H+ concentrations, experience chronically higher
8 respiratory symptom rates in adults.
9 Stern et al. (1994) reported on a Canadian survey assessing the effects to transported
10 acidic pollution on the respiratory health of children, regional differences in respiratory
11 symptoms and lung function parameters. A cohort of about 4,000 Canadian school children,
12 aged 7 to 11 years, residing in five rural communities in southwestern Ontario (high
13 exposure area) and in five rural communities in central Saskatchewan (low exposure area)
14 were examined. Respiratory health status was assessed through the use of parent-completed
15 questionnaires and standard pulmonary function tests performed by the children in the
16 schools. The levels of paniculate sulfates and nitrates varied little among communities
17 within each region, but sulfate means did differ between regions, with annual average sulfate
18 readings for 1980 of 1.9 pg/m3 and 6.6 fig/m3 in Saskatchewan and Ontario, respectively.
19 There were no significant differences in PM10 between these regions, however. After
20 adjusting for the effects of age, sex, parental smoking, parental education and gas cooking,
21 no differences in the prevalence of chronic cough, chronic phlegm, persistent wheeze,
22 current asthma, bronchitis in the past year, or any chest illness that kept a child home for
23 3 or more days in the previous year most days and nights were observed. This differs with
24 the results of the Harvard Six City Study (Dockery et al., 1989), which Stern et al. (1994)
25 conclude may be due to a threshold of effects for chronic air pollution and respiratory
26 symptoms effects. There were no regional differences in PEFR, FEF25.75, FEF75.85,
27 Vmax5O, and Vmax25. However, statistically significant decrements of 1.7% in FVC and
28 1.3% in FEVj o were observed in Ontario children, as compared with those in Saskatchewan,
29 after adjusting for age, sex, weight, standing height, parental smoking, and gas cooking.
30 These results are noted to be similar to those reported by Schwartz (1989), but not with the
April 1995 12-233 DRAFT-DO NOT QUOTE OR CITE
-------
1 Six-Cites results (Dockery et al., 1989). It is hypothesized that this new study had greater
2 power to detect such effects because the areas being contrasted are more similar, other than
3 with respect to air pollution. The authors conclude that statistically significant decrement in
4 the pulmonary volume parameters, FVC and FEVj 0, of preadolescent children residing in
5 rural southwestern Ontario are associated with moderately elevated ambient concentrations of
6 sulfates and ozone.
7 Schenker et al. (1983b) studied 5,557 adult women in a rural area of western
8 Pennsylvania using respiratory disease questionnaires. Air pollution data (including SO2, but
9 not particulate matter measurements) were derived from 17 air monitoring sites and stratified
10 in an effort to define low, medium and high pollution areas. The four-year means (1975 to
11 1978) of SO2 in each stratum were 62, 66, and 99 /xg/m3, respectively. Respiratory
12 symptom rates were modeled using multiple logistic regression, which controlled for several
13 potentially confounding factors, including smoking. A model was used to estimate air
14 pollutant concentrations at population-weighted centroids of 36 study districts. The relative
15 risk (odds ratio) of "wheeze most days or nights" in nonsmokers residing in the high and
16 medium pollution areas was 1.58 and 1.26 (p < 0.02) respectively, as compared with the
17 low pollution area. For residents living in the same location for at least five years, these
18 relative risks were 1.95 and 1.40 (p < 0.01). Also, the increased risk of grade 3 dyspnea in
19 nonsmokers was associated with SO2 levels (p < 0.11). However, no significant association
20 was observed between cough or phlegm and air pollution variables. The results of this study
21 suggest that wheezing may be associated with SO2 levels, but these results must be viewed
22 with caution, since the gradient between areas was small and there were no particle or other
23 pollutant measures. Lippmann (1985) suggested that it was plausible that the effects in this
24 study are associated with submicrometer acid aerosol which deposits primarily in small
25 airways, rather than with S02 levels.
26 Jedrychowski and Krzyzanowski (1989) related SO2 and PM levels to increased rates of
27 chronic phlegm, cough and wheezing in females living in and near Cracow, Poland. The
28 authors suggest that the effects may have been due to hydrogen ions, but no direct
29 measurements were available.
April 1995 12-234 DRAFT-DO NOT QUOTE OR CITE
-------
1 Several authors (Lave and Seskin, 1972, 1977; Chappie and Lave, 1982; Mendelsohn
2 and Orcutt, 1979; Lipfert, 1984; Ozkaynak and Spengler, 1985; Ozkaynak and Thurston,
3 1987) have related annual mortality rates in U.S. Metropolitan Statistical Areas (MSA's) to
4 sulfate and other pollution measurements using aggregate population cross-sectional analyses.
5 There are significant problems and inconsistencies in results obtained across many of these
6 analyses, as reviewed extensively by the U.S. Environmental Protection Agency (1986,
7 1982). For example, Lave and Seskin (1977) reported that mortality rates were correlated
8 with sulfates. Lipfert (1984), reanalyzing the same data, found that it was not possible to
9 conclude whether sulfates or paniculate matter had a statistically significant effect on total
10 mortality. These studies are reviewed in more detail in Section 12.4.1, but are included
11 again in this section because of their relevance to acid aerosol epidemiology.
12 In one of the more extensive of these analyses, employing a variety of model
13 specifications and controls for possible confounding, Ozkaynak and Spengler (1985),
14 Ozkaynak et al. (1986), and Ozkaynak and Thurston (1987) used more sophisticated
15 statistical approaches in an effort to improve upon some of the previous analyses of mortality
16 and morbidity associations with air pollution in U.S. cities. The principal findings concern
17 cross-sectional analysis of the 1980 U.S. vital statistics and available air pollution data bases
18 for sulfates, and fine, inhalable and total suspended particles. In these analyses, using
19 multiple regression methods, the association between various particle measures and 1980 total
20 mortality were estimated for 98 and 38 SMSA subsets by incorporating information on
21 particle size relationships and on a set of socioeconomic variables to control for potential
22 confounding. Issues of model misspecification and spatial autocorrelation of the residuals
23 were also investigated. Results from the various regression analyses indicated the importance
24 of considering particle size, composition, and source information in modeling of PM-related
25 health effects. In particular, particle exposure measures related to the respirable and/or toxic
26 fraction of the aerosols, such as FP (fine particles) and sulfates were the most consistently
27 and significantly associated with the reported (annual) cross-sectional mortality rates. On the
28 other hand, particle mass measures that included coarse particles (e.g., TSP and IP) were
29 often found to be nonsignificant predictors of total mortality. In addition, an analysis of
30 source-related fine particle trace element components for the 38 SMSA set found the
April 1995 12-235 DRAFT-DO NOT QUOTE OR CITE
-------
1 strongest mortality associations with industrial and combustion-related components of the fine
2 aerosol, but not with soil-derived particles. Thus, these analyses indicated that sulfate
3 containing fine combustion-related particles were most closely associated with mortality.
4 The Ozkaynak and Thurston (1987) results noted above for analysis of 1980 U.S.
5 mortality provide an interesting overall contrast to the findings of Lipfert (1984) for 1969 to
6 1970 U.S. mortality data, and to the findings of Lipfert et al. (1988) for the 1980 U.S.
7 mortality data. In particular, whereas Lipfert found TSP coefficients to be most consistently
8 statistically significant (although varying widely depending upon model specifications,
9 explanatory variables included, etc.), Ozkaynak and Thurston (1987) found particle mass
10 measures, including coarse particles (TSP, IP), often to be nonsignificant predictors of total
11 mortality. Also, whereas Lipfert found the sulfate coefficients to be even more unstable than
12 the TSP associations with mortality (and questioned the credibility of the sulfate coefficients),
13 Ozkaynak and Thurston (1987) found that particle exposure measures related to the respirable
14 or toxic fraction of the aerosols (e.g., FP or sulfates) to be most consistently and
15 significantly associated with annual cross-sectional mortality rates. They estimated a range
16 of paniculate matter-total mortality mean effects of 4 to 9% of total U.S. mortality, when
17 sulfates were used as the PM metric. When Lipfert (1988) conducted a reanalysis of the
18 1980 cross-sectional dataset, and added many more controllers for confounding (e.g. for
19 smoking, water hardness and sulfate artifact), he also reports a significant sulfate coefficient
20 having an elasticity of 2.8 to 13%, which is not statistically different from that reported by
21 Ozkaynak and Thurston (see Lipfert and Morris, 1991, and; Thurston and Ozkaynak, 1992
22 for discussion). Thus, while results vary somewhat across studies, most cross-sectional
23 analyses of the 1960, 1970, and 1980 are supportive of an association between chronic
24 sulfate exposure and increased human mortality.
25 Taken as a whole, these various analyses are suggestive of mortality and morbidity
26 associations with the sulfate fraction of fine particles found in contemporary American urban
27 airsheds. Without nationwide measurements of airborne acidity, however, it is difficult to
28 evaluate the relative contribution of acid aerosols within these fine particle sulfates to the
29 reported health effects.
30
April 1995 12-236 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.5.4.3 Studies Relating Chronic Health Effects to Acid Aerosols
2 In an hypothesis generating discussion, Speizer (1989) presented city-specific bronchitis
3 prevalence rates from the six cities. While no direct aerosol acidity measurements were
4 actually made during or before the 1980/81 school year (when the children were examined),
5 Speizer (1989) utilized pollution data that Spengler et al. (1989) gathered in
6 Kingston/Harriman and St. Louis from December 1985 through September 1986 and in
7 Steubenville and Portage from November 1986 to early September 1987. His plot of
8 bronchitis prevalence as a function of PM15 is presented in Figure 12-15. Additional H+
9 concentration data from Watertown, MA and Topeka, KS have since been published by
10 Dockery (1993), and all these data are included in the updated version of Speizer's H+ plot
11 presented in Figure 12-16. It should be noted that these points may contain unaddressed
12 bronchitis variation due to factors other than pollution. For example, illness and
13 hospitalization rates are known to vary across areas, independent of health status factors
14 (Wennberg, 1987; McPherson et al., 1982). Thus, the relationship of bronchitis rates with
15 pollution in these preliminary analyses must be considered as being only suggestive.
16 However, as seen in these figures, when the city-specific bronchitis rates are plotted against
17 mean H+ concentrations, instead of PM15, there is a relative shift in the ordering of the
18 cities which suggests a better correlation of bronchitis prevalence with H+ than with PM15.
19 Damokosh et al. (1993) has analyzed the 6-City children's bronchitis data, more
20 thoroughly, it is an abstract incorporating controls for confounding variables, and has added
21 a seventh locale, Kanahwa County, WV to the analysis. In that county, PM10, PM2 5, and
22 H+ were measured from 1987 to 1988 during the collection of data on the respiratory health
23 status of 7,910 children in third through fifth grade. As in the 6-City study, respiratory
24 health status was assessed in Kanahwa County via a parent completed questionnaire. Nine
25 indicators of asthmatic and bronchitic symptom reports were considered. A two-stage
26 logistic regression analysis was used, adjusting for maternal smoking and education, race,
27 and any unexplained variation in symptom rates between the cities. Significant associations
28 were found between summer mean H+ and chronic bronchitis and related symptoms (cough,
29 phlegm, and chest illness). The estimated relative odds for bronchitic symptoms associated
30 with the lowest mean value of particle strong acidity (15.7 nmoles/m3) to the highest
April 1995 12-237 DRAFT-DO NOT QUOTE OR CITE
-------
11
10
8
a>
(0
ol
V)
I
CD
8
6
10
K
20
30
40
50
60
PM
15
Figure 12-15. Bronchitis in the last year, children 10 to 12 years of age in 6 U.S. cities,
by PM15. (P = Portage, WI; T = Topeka, KS; W = Watertown, MA;
K = Kingston, TN; L = St. Louis, MO; S = Steubenville, OH.)
Source: Speizer (1989).
April 1995
12-238 DRAFT-DO NOT QUOTE OR CITE
-------
a
to
o
o
2
s
o
a
§
o
*
n
h-H
S
O
I
S'
00
2. CO
n
*§ S t^
i 8 § T
s § " ?
"*• ^ O ^p
C/3 B SJ ^
11 11" 3
55 S'S O
CD
S H-
-' §2,
^ a .,
r g.
ii d
C/3
% Bronchitis Prevalence
ro
ro
o
3
CO
CO
I
cn
O)
oo
I
CO
II II
p p
CO LL
W 00
X
+
CO
-------
1 (57.8 nmoles/m3) was 2.4 (95% CI: 1.9 to 3.2). No associations were found for asthma or
2 asthma related symptoms (doctor diagnosed asthma, chronic wheeze, and wheeze with attacks
3 of shortness of breath). However, equivalent results were found with other particle mass
4 measurements highly correlated with aerosol acidity.
5 As a follow-up to the 6-City study, the relationship of respiratory symptom/illness
6 reporting with chronic exposures to acidic aerosols was tested among a cohort of
7 schoolchildren in 24 rural and suburban communities in the United States and Canada
8 (Dockery et al., 1993). Ambient air pollution concentrations were measured for one year in
9 each community. Annual mean paniculate strong acidity concentrations ranged from 0.5 to
10 52 nmoles/m3 across the 24 communities. Questionnaires were completed by the parents of
11 15,523 schoolchildren 8 to 12 years of age. Both bronchitic symptoms, (reports of
12 bronchitis, cough, or phlegm) and asthmatic symptoms, (reports of asthma, shortness of
13 breath with wheeze) or persistent wheeze, were considered separately. City-specific
14 reporting rates were first calculated after adjustment for the effects of gender, age, parental
15 asthma, parental education, and parental allergies. Associations with ambient air pollution
16 were then evaluated. Bronchitic symptoms were associated with particulate strong acidity:
17 relative odds 1.7 (95% CI: 1.1 to 2.5) across the range of exposures. Increased reporting of
18 bronchial symptoms were also associated with other measures of particulate air pollution
19 including sulfate - relative odds 1.7 (95% CI: 1.1 to 2.4). However, associations of
20 asthmatic symptom reports with any of the air pollutants, including particulate acidity, were
21 not statistically significant. Stratified analyses did not show any evidence that asthmatics or
22 other potentially sensitive groups of children had a greater response to particulate acidity.
23 Raizenne et al. (1993) report results from the 24-City follow-up to the Six Cities study
24 designed by researchers at Harvard University to specifically examine the health effects in
25 children of living in regions having periods of elevated ambient acidic air pollution
26 (24 communities in the U.S. and Canada, 8 sites/year, 3 years) . Parents of children between
27 the ages of 8 to 12 completed a questionnaire and provided consent for their child to perform
28 a standardized forced expiratory maneuver on one occasion between October and May. Air
29 and meteorological monitoring was performed in each community for the year preceding the
30 pulmonary function tests. The annual mean particle strong acidity (H+) ranged from 0.5 to
April 1995 12-240 DRAFT-DO NOT QUOTE OR CITE
-------
1 52 nmoles/m3, PM10 ranged from 18 to 35 /ig/m3, and PM2-1 from 6 to 21 fig/m3. Annual
2 H+ was more highly correlated with PM2.1 (r = .72) and SO4 (r = .91) than with PM10
3 (r = 0.29). FVC and FEVj measurements of 10,753 Caucasian children in 22 communities
4 were used in the analyses. A two-stage logistic regression analysis was used adjusting for
5 age, sex, height, weight, sex-height interaction and parental history of asthma. The reported
6 effect estimates were expressed in terms of 52 nmoles/m3 difference in H+. The results
7 indicated that residing in high particle strong acidity regions was associated, on average, with
8 a 3.41% (95% CI 4.72, 2.09) and a 2.95% (95% CI 4.36, 1.52) lower than predicted FVC
9 and FEV! 0, respectively. For children with a measured FVC less than or equal to 85% of
10 predicted, the odds ratio for lower lung function was 2.5 (95% CI 1.7, 3.6) across the range
11 of H+ exposures. Assuming that these exposures reflect lifetime exposure of the children in
12 this study, the data suggest that long-term exposure to ambient particle acidity may have a
13 deleterious effect on normal lung growth, development, and function.
14 As discussed in detail earlier in this chapter, Dockery et al. (1993) reported the results
15 of a prospective cohort study in which the effects of air pollution on mortality were
16 estimated, controlling for individual risk factors. Survival analysis, including Cox
17 proportional-hazards regression modeling, was conducted with data from a 14 to 16 year
18 mortality follow-up of 8,111 adults in six U.S. cities. After adjusting for smoking and other
19 risk factors, statistically significant associations were found between air pollution and
20 mortality. Using inhalable particles, fine particles, or sulfates as the indicator of pollution
21 all gave similar results: an adjusted mortality-rate ratio for the most polluted city as
22 compared to the least polluted city of 1.26 (95% CI = 1.08 to 1.47). Weaker mortality
23 associations were found with H+ in this analysis. However, this analysis did not
24 demonstrate that H+ was not a factor in the pollution related excess mortality, as the H+
25 data employed was not appropriate for such an analysis. Of the pollutant data considered,
26 the H+ was the most limited. Less than one year of H+ data were collected in each city,
27 near the end of this study, and this was used to characterize lifetime exposures of adult study
28 participants. This seems especially inappropriate in Steubenville, OH where the industrial
29 (e.g. steel mill) pollution levels declined during the course of the study, as the steel industry
30 in the valley declined. Indeed, in Steubenville, OH, the H+ data were only collected from
April 1995 12-241 DRAFT-DO NOT QUOTE OR CITE
-------
1 mid-October, 1986 through early September, 1987 (Spengler et al., 1989). In contrast, the
2 inhalable particle, fine particle, and sulfate data employed in each city were more
3 representative, having been collected earlier and over roughly a five to six year period in
4 these cities. Thus, the inability of this study to find a statistically significant correlation
5 between H+ and mortality (relative to sulfates and fine particles) may be due in large part to
6 the fact that the limited H+ data employed were not sufficient for this application.
7
8 12.5.4.4 Chronic Exposure Effects in Occupational Studies
9 The last remaining type of information considered here concerns the effects of chronic
10 exposures to acid aerosols in occupational settings. Such studies are discussed mainly in
11 order to provide some perspective on the variety of health effects associated with acid
12 aerosol exposures, albeit at extremely high concentrations not likely to occur in ambient air.
13 Gamble et al. (1984a) studied pulmonary function and respiratory symptoms in 225
14 workers in five lead battery acid plants. This acute effect study obtained personal samples of
15 H2SO4 taken over the shift. Most personal samples were less than 1 mg/m3 H2SO4, and
16 mass median aerodynamic diameter of H2SO4 averaged about 5 /mi. The authors concluded
17 that exposure to sulfuric acid mist at these plants showed no significant association with
18 symptoms or with acute effect on pulmonary function. The ability of the body to neutralize
19 acidity of H2SO4 was considered as one factor is this outcome. Additionally, the authors
20 speculated that tolerance to H2SO4 may develop in workers habitually exposed.
21 In a related study of chronic effects of sulfuric acid on the respiratory system and teeth,
22 Gamble et al. (1984b) measured in the same workers respiratory symptoms, pulmonary
23 function, chest radiographs, and tooth erosion. Concentrations measured at the time of the
24 study were usually 1 mg/m3 or less. Exposure to the concentration of acid mist showed no
25 significant association with cough, phlegm, dyspnea, wheezing, most measures of pulmonary
26 function, and abnormal chest radiographs. Tooth etching and erosion were strongly related
27 to acid exposure. The authors noted that the absence of a marked effect of acid exposure on
28 respiratory symptoms and pulmonary function may be due to the size of the acid particles.
29 The range of the mass median diameter in the 5 plants was 2.6 to 10 /mi, which is much
30 larger than (typically submicrometer) ambient H+ aerosols. Moreover, the relative humidity
April 1995 12-242 DRAFT-DO NOT QUOTE OR CITE
-------
1 of the lung may cause at least a doubling of particle size, especially in the lower size range.
2 Thus, most acid particles may be deposited in the upper respiratory tract, and many may not
3 even reach the lung. Finally, the authors note that the lack of any convincing finding in this
4 study relating to the acute respiratory symptoms is not completely unexpected, due to the
5 relatively low exposure (< 1 mg/m3) compared to previous occupational studies.
6 Williams (1970) studied sickness absence and ventilatory capacity of workers exposed
7 to high concentrations of sulfuric acid mist in the forming department of a battery factory
8 (location not stated). Based on 38 observations made on two days, the forming department
9 had a mean H2SO4 concentration of 1.4 mg/m3, ranging from a trace to 6.1 /ng/m3- In a
10 different forming department, the mass median diameter of the acid particles was 14 ^tm.
11 Compared with control groups, men exposed to the high concentrations of sulfuric acid mist
12 in the forming department had slight increases in respiratory disease, particularly bronchitis.
13 There was no evidence of increased lower respiratory disease, which might be explained by
14 the large particle size. After adjusting for circadian variations, there was no evidence of
15 decreased ventilatory function.
16 Beaumont et al. (1987) studied mortality patterns in 1,165 workers exposed to sulfuric
17 acid and other acid mists in steel-pickling operations. Workplace monitoring during the
18 1970's indicated worker personal exposures to average 190 /ng/m3 H2SO4. However, as
19 discussed for battery plant operations, the particle size of these mists tend to be larger than
20 ambient acid aerosols, so not all is likely to be respirable. Standardized mortality ratio
21 (SMR) analysis of the full "any acid exposure" cohort (n = 1,165), with the use of U.S.
22 death rates as a standard, showed that lung cancer was significantly elevated, with a
23 mortality ratio of 1.64 (95%CI = 1.14 to 2.28, based on 35 observed deaths). The lung
24 cancer mortality ratio for workers exposed only to sulfuric acid (n = 722) was lower (SMR
25 = 1.39), but further restriction to the time 20 years and more from first employment in a job
26 with probably daily sulfuric acid exposure ( — 0.2 mg/m3) yielded a mortality ratio of 1.93
27 (95% CI = 1.10 to 3.13). An excess lung cancer risk was also seen in workers exposed to
28 acids other than sulfuric acid (SMR = 2.24; 95% CI = 1.02 to 2.46). When comparison
29 was made to other steel workers (rather than to the U.S. general population) to control for
30 socio-economic and life-style factors such as smoking, the largest lung cancer excess was
April 1995 12-243 DRAFT-DO NOT QUOTE OR CITE
-------
1 again seen in workers exposed to acids other than sulfuric acid (SMR = 2.00; 95% CI =
2 1.06 to 3.78). However, the smaller rate ratios may have been partly due to the restriction
3 of this sub-analysis to white males, which excluded the higher excess lung cancer risk in
4 nonwhite males. Adjustment for potential differences in smoking habits showed that
5 increased smoking was unlikely to have entirely explained the increased risk. Mortality from
6 causes of death other than lung cancer was unremarkable, with the exception of significantly
7 lower rates for deaths due to digestive system diseases. These results suggest that chronic
8 acid aerosol exposures may promote lung cancer at high concentrations, perhaps via chronic
9 irritation of respiratory tissues, or by some other mechanism (e.g. by affecting clearance
10 rates in the lung).
11
12 12.5.5 Summary of Studies on Acid Aerosols
13 Historical and present-day evidence suggest that there can be both acute and chronic
14 effects by strongly acidic PM on human health. Evidence from historical pollution for
15 episodes, notably the London Fog episodes of the 1950's and early 1960's, indicate that
16 extremely elevated daily acid aerosol concentrations (on the order of 400 /xg/m3 as H2SO4, or
17 roughly 8,000 nmoles/m3 H+) may be associated with excess acute human mortality when
18 present as a co-pollutant with elevated concentrations of PM and SO2. In addition, Thurston
19 et al. (1989) and Ito et al. (1993) both found significant associations between acid aerosols
20 and mortality in London during non-episode pollution levels (< 30 ptg/m3 as H2SO4, or
21 < approximately 600 nmoles/m3 H+), though these associations could not be separated from
22 those for BS or SO2. The only attempts to-date to associate present-day levels of acidic
23 aerosols with acute and chronic mortality (Dockery et al., 1992; Dockery et al., 1993,
24 respectively) were unable to do so, but there may not have been a sufficiently long series of
25 H+ data to detect H+ associations. There is a critical need for present day replications of
26 the London mortality-acid aerosol studies to be conducted, however, in order to determine if
27 the London wintertime associations (which occurred in reduction-type atmospheres) are
28 pertinent to present-day U.S. conditions, in which acid aerosol peaks occur primarily in the
29 summer months (in oxidation-type atmospheres).
April 1995 12-244 DRAFT-DO NOT QUOTE OR CITE
-------
1 Increased hospital admissions for respiratory causes were also documented during the
2 London Fog episode of 1952, and this association has now been observed under present-day
3 conditions, as well. Thurston et al. (1992) and Thurston et al. (1994b) have noted
4 associations between ambient acidic aerosols and summertime respiratory hospital admissions
5 in both New York State and Toronto, Canada, respectively, even after controlling for
6 potentially confounding temperature effects. In the latter of these studies, significant
7 independent H+ effects remained even after simultaneously considering the other major
8 co-pollutant, O3, in the regression model. In the Toronto analysis, the increase in
9 respiratory hospital admissions associated with + was indicated to be roughly six times that
10 for non-acidic PM10 (per unit mass). In these studies, H+ effects were estimated to be the
11 largest during acid aerosol episodes (H+ > 10 jwg/m3 as H2SO4, or =200 nmoles/m3 H+),
12 which occur roughly 2 to 3 times per year in eastern North America. These studies provide
13 evidence that present-day strongly acidic aerosols may represent a portion of PM which is
14 particularly associated with significant acute respiratory disease health effects in the general
15 public.
16 Results from recent acute symptoms and lung function studies of healthy children
17 indicate the potential for acute acidic PM effects in this population. While the 6-City study
18 of diaries kept by parents of children's respiratory and other illness did not demonstrate H+
19 associations with lower respiratory symptoms except at H+ above 110 moles/m3 (Dockery
20 et al., 1994), upper respiratory symptoms in two of the cities were found to be most strongly
21 associated with daily measurements of H2SO4 (Schwartz, et al., 1991). Recent summer
22 camp and school children studies of lung function have also indicated significant associations
23 between acute exposures to acidic PM and decreases in the lung function of children
24 independent of those associated with O3 (Studnicka et al., 1995; Neas et al., 1995).
25 Studies of the effects of chronic H+ exposures on children's respiratory health and lung
26 function are generally consistent with effects as a result of chronic H+ exposure.
27 Preliminary analyses of bronchitis prevalence rates as reported across the 6-City study locales
28 were found to be more closely associated with average H+ concentrations than with PM in
29 general (Speizer, 1989). A follow-up analysis of these cities and a seventh locality which
30 controlled the analysis for maternal smoking and education and for race, suggested
April 1995 12-245 DRAFT-DO NOT QUOTE OR CITE
-------
1 associations between summertime average H+ and chronic bronchitic and related symptoms
2 (Damokosh et al., 1993). The relative odds of bronchitic symptoms with the highest acid
3 concentration (58 nmoles/m3 H+) versus the lowest concentration (16 nmoles/m3) was
4 2.4 (95% CI: 1.9 to 3.2). Furthermore, in a follow-up study of children in 24 U.S. and
5 Canadian communities (Dockery et al., 1993a) in which the analysis was adjusted for the
6 effects of gender, age, parental asthma, parental education, and parental allergies, bronchitic
7 symptoms were confirmed to be significantly associated with strongly acidic PM (relative
8 odds = 1.7, 95% CI: 1.1 to 2.4). It was also found in the 24-Cities study that mean FVC
9 and FEVj 0 were lower in locales having high particle strong acidity (Raizenne et al., 1993).
10 Thus, chronic exposures to strongly acidic PM may have effects on measures of respiratory
11 health in children.
12
13
14 12.6 DISCUSSION
15 12.6.1 Introduction and Basis for Study Evaluation
16 The epidemiologic studies of human health effects related to PM exposure play a
17 particularly important role because there is somewhat less supporting information on
18 exposure-response information from toxicological or clinical studies compared to other
19 criteria pollutants. We have therefore paid much greater attention to methodological issues
20 in the studies that have been reviewed in this epidemiology chapter. Various health
21 endpoints have been used in these studies, including respiratory function measures,
22 respiratory symptom reports, hospital admissions, total non-accidental mortality, and
23 mortality classified by medical cause of death such as respiratory or cardiovascular
24 classifications. Each health outcome has many causes other than air pollution, and no
25 specific air pollutant can be uniquely associated with a specific outcome, including PM and
26 its components. Subject-specific (personal) exposure to PM or to other air pollutants is
27 unmeasured in almost all of the studies, and exposure to PM, to other pollutants, or even to
28 weather variables, is only estimated from one or a few monitoring sites in a large
29 metropolitan area or region. Demographic information can be used with either longitudinal
30 studies, prospective studies, or cross-sectional studies, but age is the only individual subject
April 1995 12-246 DRAFT-DO NOT QUOTE OR CITE
-------
1 variable that has been used in almost all studies. Other personal variables can be obtained in
2 prospective studies. Comparisons across different cities must be adjusted for demographic
3 and climatologic differences, and usually are in cross-sectional studies. Studies of acute
4 responses to air pollutants, whether measured by respiratory function indices, respiratory
5 symptoms, hospital admissions, or mortality, have been compared by various formal or
6 informal meta-analytic techniques (Schwartz, 1992a, 1994c; Dockery and Pope, 1994b), but
7 there has so far been no effort to adjust the results of the metaanalyses for quantitative
8 differences among study groups or for differences in data-analytic methodologies.
9 Differences in results may also be attributed to differences in methods of data analysis.
10 Many investigators of acute effects have expressed concerns about data analysis issues,
11 including problems of non-Gaussian data distributions, mis-specification of exposure-response
12 relationships, inappropriate adjustments for covariates such as weather and co-pollutants,
13 mis-specification of the temporal lag structure between exposure and response, and
14 inadequate adjustment for seasonality and other random or systematic long-wave (low
15 frequency) effects. However, it has become clear that very similar estimates of the effects of
16 PM can be obtained for a wide range of alternative data analysis methods. Many models
17 seem to work well, but all adequate models for short-term effects must be adjusted for
18 seasonality, for long-term and transient irregular events such as influenza epidemics, must
19 include adjustments for auto- and cross-correlation structure when necessary, must be
20 examined for sensitivity to distributional assumptions such as Poisson or hyper-Poisson
21 variability and, if not based on demonstrably robust methods, must be carefully examined for
22 sensitivity to unusual values among either predictor or response data. Some models used by
23 different investigators have not met all of these criteria.
24
25 12.6.1.1 Differences Among Study Results
26 What is more disturbing is that, using ostensibly similar data sets, different
27 investigators of acute mortality effects have come up with different estimates of PM effect
28 size or statistical significance. There are at least two possible reasons for this. The first is
29 that there may be some genuine confounders of PM effects on human health. In some
30 studies, under some meteorological or seasonal conditions, co-pollutants will be emitted by
April 1995 12-247 DRAFT-DO NOT QUOTE OR CITE
-------
1 some of the same sources as emit PM, so that there will be a close intrinsic relationship
2 between PM and some other pollutants. This may also extend to certain meteorological
3 variables, which may be related both to atmospheric dispersion of all outdoor pollutants and
4 to pollutant emissions rates. For example, an extremely hot day in summer may be
5 associated with increased use of electrical power for air conditioning (increasing emissions of
6 PM and other pollutants such as SO2 from local electricity generating plants that burn fossil
7 fuels) and, also, with increased motor vehicle use as people travel to less uncomfortable
8 locations (increasing vehicle-generated pollutants from gasoline and other motor vehicle
9 fuels, including O3, CO, and NO2). Primary gaseous pollutants may become secondary
10 atmospheric sources of certain PM components, such as sulfates and nitrates. While there
11 are a number of statistical diagnostics for intrinsic confounding, and even a few adequate
12 methods for partially resolving seriously confounded predictors of response, these have rarely
13 been used. Analyses in which only a single pollutant is used to predict a health effect are not
14 wholly satisfactory without confirmation by multi-pollutant analyses, adjusted for
15 confounding insofar as possible. In this regard, comparison across different studies,
16 including those in which each potentially confounding factor is or is not present, may be
17 needed to assess the effects of PM in the absence of detailed technical assessments of
18 sensitivity to intrinsically confounded variables.
19 The second reason why different investigators may come up with different results for
20 acute mortality is much more profound. In the absence of generally acceptable mechanistic
21 relationships among potentially confounding variables, and in the absence of generally
22 acceptable specifications for the exposure-response relationships for PM, for co-pollutants,
23 and for weather, all modelling is data-driven and empirical. This has led almost all
24 investigators into extensive model specification searches, in which numerous alternative
25 models may be fitted to the same data or to subsets of the same data set until a "best fitting"
26 or "statistically significant" model is obtained. It has long been known (Learner, 1978) that
27 data-driven model specification searches can seriously distort the actual significance level of
28 the regression coefficients in ordinary linear regression models with independent Gaussian
29 errors, and by extension we expect the same problem in Poisson and hyper-Poisson
30 exponential regression models with complicated correlation structures. This is similar to the
April 1995 12_248 DRAFT-DO NOT QUOTE OR CITE
-------
1 better-known "multiple comparisons" problem, in which all possible subsets of a set of
2 hypothesis tests in a linear analysis of (co)variance could be tested, with a corresponding
3 artificial inflation of the statistical significance of the whole ensemble of tests. However, the
4 complicated model specification searches that have produced the models reported in the
5 published PM epidemiologic studies have a hypothetically limitless number of alternative
6 specifications.
7 In evaluating numerous model specifications, on the one hand, a model specification
8 search may be extended until some combination of correlation model or lag structure,
9 adjustments for time trends, season, co-pollutants, and weather produces a model in which
10 the study response data are fitted well and the PM coefficient is "statistically significant".
11 Statistical significance for a PM coefficient means that either an asymptotic confidence
12 interval or a more exact likelihood ratio-based confidence interval for the effect does not
13 cover the null value (0 for effect size, 1 for relative risk). Or, on the other hand, the
14 specification search may proceed towards the goal of establishing that some other pollutant in
15 the model is a statistically significant predictor of changes in mortality rates or hospital
16 admission rates (etc.) or that some combination of meteorological variables can fit the
17 observed health effects data when the PM coefficient is not statistically significant. This
18 could provide the basis of an argument that some factor(s) other than PM are accounting for
19 the observed effects. Because of the confounding that exists between PM and other variables
20 that may be used in the models, there may be many substantial points of similarity between
21 the models with a significant PM effect and those without a significant PM effect, at least in
22 some cities during some years. There may thus be little internal basis for choosing between
23 two models, one with a significant PM effect and another, using similar specifications in
24 many ways, without a significant PM effect.
25 There are several ways in which the indeterminacy of the models from different studies
26 of the same data set could be resolved. The first method, and hi many ways the best, is to
27 see which of the competing models does the best job of predicting new information. Since
28 new information is not readily at hand, a more realistic method would be "internal cross-
29 validation". The model would be fitted to one subset of the data and then the parameters
30 derived from the model based on one part of the data would be used to predict the other part.
April 1995 12-249 DRAFT-DO NOT QUOTE OR CITE
-------
1 In time series analysis, the use of the first part of the series to predict the last part of the
2 series is known as "postdiction", to distinguish the exercise from a genuine forecast or
3 prediction in which the future observations and their predictors are in fact unknown.
4 A related approach would be to use the PM and co-pollutant models derived from one group
5 of cities to estimate health effects in another group of cities, where "pre-models" specific to
6 each of the second group of cities are used to adjust mortality rates for all non-pollution
7 variables such as meteorological variables. In practice, we are not aware of any efforts to
8 assess the predictive validity of any of the models, either in an absolute sense or relative to a
9 competing model.
10
11 12.6.1.2 Importance of Comparisons Across Different Cities
12 We are therefore limited to evaluating models reported in different studies on the basis
13 of comparisons of results for different geographic sites (cities, SMSA's, etc.) or during
14 different periods of time. If the estimated PM effect is similar in magnitude across a range
15 of different cities, differing in location, climate, co-pollutant inventories, demographics, or
16 other relevant factors, we may argue that these effect estimates are relatively robust with
17 respect to exact specifications of different models. This is discussed in more detail in
18 Section 12.6.3.
19 Similarly, weather is an important confounding factor. Adjustments for meteorological
20 variables may differ substantially from one study to another. It is easier to compare effect
21 size estimates from studies with similar adjustment methods. However, there are likely to be
22 real differences among cities that complicate the use of weather effect models found at one
23 location to adjust for weather effects on human health in another location. This is
24 particularly likely to affect adjustments made for extreme weather conditions, whether
25 defined by a threshold for a temperature effect or by a weather-related synoptic category. It
26 is, in any event, easier to identify a quantitative PM relationship during non-extreme weather
27 conditions, or during non-offensive synoptic categories. Studies in which the size of the PM
28 exposure-response relationship was estimated for non-extreme weather conditions, or for
29 which appropriate adjustments were made in the analysis, are also accorded higher weight
30 than those without such distinctions.
April 1995 12-250 DRAFT-DO NOT QUOTE OR CITE
-------
1 Finally, there is a question about how the effect size estimates in different cities should
2 be combined, or whether there should be a combined estimate. Combined estimates using
3 meta-analytic techniques have been published (Schwartz, 1994c; Dockery and Pope, 1994b),
4 and additional meta-analyses for the more recent studies may be useful. However, there is a
5 possibility that real differences exist among PM effect sizes in different communities. The
6 differences may be due to differences in sub-populations, in pre-existing health status, in
7 acclimatization to weather conditions, or to effects of other unmeasured air pollutants. If the
8 differences among communities are substantial, it may be preferable to treat the PM effect on
9 health outcome as a random effect across communities, even though the reasons for the
10 differences are potentially explainable, but unknown at present.
11
12 12.6.2 Sensitivity of PM Effects to Model Specification in Individual
13 Studies
14 12.6.2.1 Model Specification for Acute Mortality Studies
15 Many different statistical models have been used to interpret short-term mortality and
16 morbidity studies. The model specifications and methods used to interpret the long-term
17 studies are generally different from those used in analyzing the short-term studies. It is often
18 difficult to compare estimates of PM effect in different studies when the estimates of effect
19 size are obtained by different methods. Differences in effect size estimates may then occur
20 because of differences in modelling approach as well as any real differences in response to
21 PM exposure.
22 Many of the papers reviewed in this chapter provide enough information to assess the
23 authors' choice of their "best" model, which we have reported in the summary tables. An
24 extensive discussion of alternative modelling approaches for short-term exposure studies was
25 already evident in earlier papers, such as the analyses of BS in London in the 1960's (Ostro,
26 1984; Thurston et al., 1989; Schwartz and Marcus, 1990; Ito, 1990), KM in Los Angeles
27 (Shumway et al., 1988; Kinney and Ozkaynak, 1991), and COH in Santa Clara (Fairley,
28 1989). More recent work has moved in some substantially different directions, recognizing
29 the non-Gaussian nature of discrete data such as daily death counts and hospital admissions,
30 and incorporating a growing variety of data-driven non-parametric or semi-parametric models
April 1995 12-251 DRAFT-DO NOT QUOTE OR CITE
-------
1 for PM and other Covariates. We will here discuss the more recent studies, emphasizing
2 those studies in which PM10 or TSP are used as PM indicators.
3
4 Model Specification for the Utah Valley Mortality Study (Pope et al, 1992)
5 One of the most comprehensive assessments of alternative model specifications was
6 presented by Pope in a report for presentation at the EPA-sponsored workshop on daily
7 mortality data and short-term changes in PM exposure in November, 1994 (Pope, 1994).
8 The results of these additional analyses of the Utah Valley study were described briefly in
9 Section 12.3.1. We will present the results graphically, with a view towards resolving model
10 specification issues. For each comparison, we will exhibit a sequence of three graphs
11 illustrating the results for total (non-accidental) mortality, for death from respiratory causes,
12 and for death from cardiovascular causes. The horizontal bars show the 95 percent
13 confidence limits for relative risk (denoted RR) corresponding to 50 /zg/m3 in PM10.
14 Figures 12-17a through 12-17c show the RR estimates for Poisson regression models.
15 The RR for PM quintiles given in the published paper (Pope et al., 1991) is denoted Model
16 0. The next group, Models 1 through 5, show the results of fitting increasingly adjusted
17 parametric models, from those with only a linear PM10 effect (Model 1), and subsequently
18 adding adjustments for time trend (Model 2), temperature (Model 3), humidity (Model 4),
19 and operation of the mill (Model 5) to the preceding model. The relative risk for total
20 mortality (Figure 12-17a) was little affected in Models 1 and 2, but dropped somewhat after
21 temperature was included (Model 3). The relative risk for respiratory mortality
22 (Figure 12-17b) was less affected by temperature, but shifted upward after humidity was
23 added (Model 4). Cardiovascular mortality (Figure 12-17c), like total mortality, also
24 dropped slightly after temperature was added to the model. The relative risk for the next
25 four models (Model 6 through 9) are parallel to Models 2 through 5, except that a non-
26 parametric smoothing function LOESS was used to model time trend, temperature, and
27 humidity respectively in Models 6, 7, and 8; a dummy variable for mill operation was added
28 in Model 9. Model 13 is the same as Model 8 without adjusting for time trend by a LOESS
29 fit on day of the study. In general, RR using at least one LOESS smoother provided a
April 1995 12-252 DRAFT-DO NOT QUOTE OR CITE
-------
I
g
(a) All Causes
(b) Respiratory Causes
(c) Cardiovascular Causes
o
o
1
1
a
13
9
Loess
Models 8
7
6
5
4
Poisson
Models 3
2
1
Published 0~
Model
0.
—
f-.-.
35 1.05 1.15 1.25 1.35 1.45
Relative Risk per 50 \igln? PMto
O.J
)5 1.05 1.15 1.25 1.35 1.45
Relative Risk per 50 ng/nt3 PMio
1 Lower 95% CL • Relative Risk I Upper 95% CL
.+ 1
i ^ 1
^ i
i „ — -^ 1
i • 1
0.95 1.05 1.15 1.25 1.35 1.4
Relative Risk per 50 ^g/nf PMio
Figure 12-17a-c. Relative risk of mortality for PM10 in Utah Valley, as a function of several parametric and semi-parametric
models of time, temperature, and dewpoint: (a) all causes; (b) respiratory causes; (c) cardiovascular causes.
Q
-------
somewhat higher RR for total mortality against PM10 in the Utah Valley study, but the
1 difference in RR among these Poisson models is small. RR for respiratory mortality
2 increased as each smoothed covariate was added, but never rose much beyond that for the
3 published model. LOESS smoothers had little effect on RR for cardiovascular mortality.
4 The next group of model comparisons is shown as Figures 12-18a through 12-18c.
5 These compare several parametric Poisson models with the analogous Gaussian ordinary least
6 squares (OLS) linear models for mortality. Even though the distributional assumptions for a
7 Gaussian distribution fail utterly, the regression coefficients and calculated RR are not very
8 different than the analogous estimates from Poisson regression models.
9 Figures 12-19a through 12-19c show the effects of separating the annual data into
10 segments, here called "summer" (April to September) and "winter" (October to March). The
11 RR for total mortality (Figure 12-19a), respiratory mortality (Figure 12-19b), and
12 cardiovascular mortality (Figure 12-19c) are all statistically significant on an annual basis,
13 while differing substantially in magnitude. Most of this effect is seen to occur from the
14 winter months when the PM10 concentrations were highest, whether or not the mill was
15 operating, based on Model 13 in which temperature and humidity effects were adjusted using
16 LOESS smoothing. The relative risk and its estimated uncertainty for all three mortality
17 endpoints is nearly the same using whole year data as when using winter data alone. While
18 PM level are generally much lower during the summer half of the year than in the winter
19 half, the summer RR estimates are higher than the winter RR estimates, but not significantly
20 different. However, the smaller range of summer PM values results in much larger
21 uncertainty about the summer RR than the winter RR. This illustrates a general problem in
22 subsetting the data by year, season, or month: the increased specificity of RR estimates for
23 subsets of data is usually offset by the loss of precision in the estimates. In general, small
24 increases in uncertainty of subset data RR estimates compared to whole data set RR estimates
25 occur only for the subset(s) of the data that are most influential in establishing the whole data
26 set RR estimate, such as the "winter" subset in this Utah Valley study.
27 Figures 12-20a through 12-20c extend these analyses to assessing the effect of a co-
28 pollutant, ozone. Including either daily average ozone concentration or maximum one-hour
29 ozone concentration as predictors of the three mortality endpoints leaves the PM10 RR
April 1995 12-254 DRAFT-DO NOT QUOTE OR CITE
-------
o
i
3
CO
O
&
1
(0
i
o
^j
s*""'
V)
i
o
S
2
5.
8
rr
^
° *" S3
^* S^
^ °- «M «
eo_. W v
jo c 52 ^^
W ^ § ^
•*! ll
-'5 in.
^^ .52 ^J tu
•si i!
c o
1- JS
So) c3
>• S
° It
BO *o g
T~ S fl —
IO SE ^^V
o> O t*>
§J 1 '€e
-£ = § I
"^S *« «
^ * 1 S °
in m a ^ S
"1 1 =3 S
" E • > ^
" I - -S ^
s S 2 « "S
P O SP ^ 5<
in 1 "" TJ
8 liJ gl
U) . _.
IO b V
*>»!
'-if 1 ^
21 ll
S
-------
>a
(a) All Causes
\ *
ON
o
O
1
(b) Respiratory Causes
(c) Cardiovascular Causes
April_Sept
Oct_March
Year
0.
t-
-• — i
m
m
5 1.0 1.5 2.0 2.5
Relative Risk per 50 \iglm3
i • i
-•H
*
3.0 0.5 1.0 1.5 2.0 2.5 3.0 0.5 1.0 1.5 2.0 2.5 3.0
Relative Risk per 50 jig/m3 Relative Risk per 50 ng/m3
1 Lower 95% CL • Relative Risk 1 Upper 95% CL
m
o
Figure 12-19a-c. Relative risk of mortality for PM10 in Utah Valley, as a function of season: (a) all causes; (b) respiratory
causes; and (c) cardiovascular causes.
n
-------
o
CO
§
CO
O
Jl
i
>
0
i
o
"o"
•
<
i <
•
• «
• •
1 T
CO
•owE
"cvi Bs
a.
-S|
.•£ cc
_>
*- DC
in
d
I
CO
0
B
2
'o.
CO
-------
1 estimate nearly unchanged from the summer PM10 RR estimate obtained without including
2 ozone as a predictor. Summer RR estimates for all models, with or without ozone, are
3 somewhat larger than the winter or whole-year RR estimate for PM, and have much greater
4 uncertainty. It may be argued that this indicates little confounding of the estimated PM
5 effect with an estimated ozone effect, and by implication little potential for confounding with
6 other pollutants generated by combustion of fossil fuels by mobile sources, at least in this
7 study.
8 Figures 12-21a through 12-21c show that specification of PM averaging time may be a
9 critical component of the modelling exercise. Moving averages of 4, 5, or 6 days would
10 provide very similar estimates of a statistically significant PM effect on total mortality
11 (Figure 12-21a) or respiratory mortality (Figure 12-21b). The 5-day moving average used by
12 Pope in most analyses gave the better prediction of cardiovascular mortality (Figure 12-21c).
13
14 Model Specification for the Santiago, Chile, Mortality Study (Ostro, 1995)
15 Many model specifications were evaluated by Ostro et al. (1995). The study was
16 discussed in Section 12.3.1. Model specification tests were designed to systematically
17 examine important issues, and results were reported in detail. The results are shown
18 graphically. Figure 12-22a shows the RR estimates and large-sample confidence intervals for
19 10 different Poisson regression models. We have recalculated the RR values in Table 3 of
20 their paper to a base of 50 ug/m3 for models that are linear in average or maximum PM10, or
21 for a change from 100 to 150 ug/m3 for their logarithms. Inclusion of temperature-related
22 variables reduced RR slightly, from about 1.13 to about 1.10. Inclusion of additional
23 dummy variables for year, quarter, and day of week had little effect on RR, but adding
24 variables for quarter and month reduced RR to about 1.05, which was still statistically
25 significant. Figure 12-22b shows the results of an additional sensitivity tests controlling
26 seasonality in a variety of different ways. The results are somewhat parallel to those of the
27 Utah Valley study discussed above, but with somewhat smaller values. Summer and winter
28 coefficients were very similar, but the RR effect was not quite statistically significant in
29 summer using a two-tailed test with alpha = 0.05. All other model specifications showed a
30 significant PM10 effect. The RR of the effect increased somewhat when the coldest days
April 1995 12-258 DRAFT-DO NOT QUOTE OR CITE
-------
HUD HO aiono ION oa-uvna 6$z-zi
£661
a
3
89
o ^ri
1 B-
8 S
^-s <*
S3.
»s
U
| *'
t/i 1—15
S 2
s»» ^
a^
^^o
^5'
&9 ^^
t!
85 ^
W 65
£ i
*9
s s
C 89
!
2,
r+
ffi
!
I
Moving Average
I
n
o
— *• o
ro -* p
+ + T
u ro _L p
p
CO
CD P
91
3J ^
"
co
i'
p
CO
cp_ p
_,,
3) ^
at
x"
TJ _i
s N
1 fc-
3co
w
+ + + +
t ¥ ¥ + ?
01 *k w ro -•• p
.,,!.•
T
_
_
( 1
1 I
I 1
1 1
II "
-*•
'
•
•L
j J
(1
II < *
o (1 «• 1
•^
X -^
s
0
0)
(0
-------
(a) Different Models
(b) Copollutants in Models
(c) Moving Averages and Lag Times
S
basic OLS
basic Poisson
summer only
winter only
omit 1% cold days
omit 5% cold days
omit 10% cold days
add trig functions
36 months, all year
Gen. Additive Models
basic OLS, 36 mos
n, 36 mos
al + ozone
S, 36 mos
r~
i
/^
^/
\,
' /"
I I ,'
. /
i i
0.95 1.0 1.0S 1.1 1.15
Relative Risk per 115 |ig/m3 PM,0
O
O
0.9 0.95 1.0 1.05 1.1 1.15
Relative Risk per 115 ng/m3 PM,0
| I Lower 95% CL • Relative Risk I Upper95%Cl.|
0.9 1.0 1.1 1.2 1.3
Relative Risk per 115 (ig/m3 PM,0
I
w
S
n
Figure 12-22a-c. Relative risk of total mortality for PM10 in Santiago, Chile, as a function of (a) different models; (b) models
for co-pollutants; and (c) moving averages and lag times.
Source: Ostro et al. (1995).
-------
1 were omitted. Including additional trigonometric terms, or including 36 dummy variables for
2 combinations of year and month reduced the RR for PM, but did not eliminate PM as a
3 significant contributor to total mortality. Control of seasonality by use of a generalized
4 additive model to adjust for time effects gave a somewhat larger RR for PM10, with small
5 uncertainty.
6 Figure 12-22c evaluates a number of lag and moving average models for PM. The
7 relative risks corresponding to each term have been recalculated from the regression
8 coefficients (denoted b) in their Table 7, for a basis of 100 to 150 ug/m3, by the formula
9
10 RR = exp( b * log(150/100)),
11
12 with confidence limits estimated analogously. All of the PM effects are statistically
13 significant, with the exception of the 3-day lag term in the 4-day polynomial distributed lag
14 (PDL) model. The 0-day and 2-day single lag models and the 3-day and 4-day moving
15 average models perform almost as well at predicting total mortality as does the PDL model,
16 of which they are each a special case.
17
18 Model Specification for the St. Louis and Eastern Tennessee Mortality Studies
19 (Dockery et al., 1992)
20 The daily mortality data for St. Louis and for eastern Tennessee analyzed by Dockery
21 et al. (1992) were discussed in Section 12.3.1. Additional results contributing to the analysis
22 were presented by Dockery in a report for presentation at the EPA-sponsored workshop on
23 daily mortality data and short-term changes in PM exposure in November, 1994 (Dockery,
24 1994c). Figure 12-23a,b illustrates the sensitivity of the PM10 RR to the lag tune or moving
25 average model in the Poisson regression for St. Louis total mortality, and Figure 12-24a,b
26 shows the analogous plot for the eastern Tennessee area. Models were fitted for lags from 0
27 to 4 days, and for the lagged moving average from the two preceding days. The lag 1 and 2
28 RR estimates for St. Louis, and the lagged 2-day moving average were statistically
29 significant for the St. Louis mortality series, but no PM indicator had a statistically
30 significant RR for PM10 in the eastern Tennessee mortality series even though the RR
31 estimates were numerically very similar. Longer-term moving averages were not evaluated,
April 1995 12-261 DRAFT-DO NOT QUOTE OR CITE
-------
I
LA
(a) PM10 Relative Risk
Relative Risk per 50 ng/m3 PM10
(b) PM2.5 Relative Risk
Relative Risk per 25 ng/rrr PM25
I Lower 95% CL • Relative Risk I Upper 95% CL
8
o
Figure 12-23a-b. Relative risk of total mortality for particulate matter in St. Louis, as a function of moving average and lag
times: (a) PM10; (b)PM2 5.
-------
HIO HO aiono ION oa-iavna
9661
Moving Average Model
CO
01
O
o
00
o
to
p
bo
33
(D
tt P
I"0
33
to _^
^ •_
en
O -L
!"
CO
"D ^
N)
In
N)
Ul
3D
(D
-------
1 but the effects of PM10 would probably have been much smaller than the RR calculated using
2 the average of 1- and 2-day lagged PM. As in the Utah Valley and Santiago studies, PM lag
3 structure needed to be identified in order to obtain a significant PM effect.
4
5 Model Specification for the New York City Respiratory Mortality Study
6 (Thurston and Kinney, 1995)
1 Thurston and Kinney compared several Gaussian OLS time series models with a
8 Poisson regression model, using respiratory mortality data for New York City for
9 1972-1975. Time series were done using both unfiltered mortality and pollution data, and
10 filtered mortality and pollution time series using a 19-day moving average. Analyses were
1 done using year-round unfiltered OLS, April-September OLS, April-September filtered OLS,
2 April-September adjustments by sines and cosines, and April-September Poisson regression
3 adjusted with sines and cosines. During the April-September ozone season, the unfiltered
4 OLS model showed a strong significant COH effect, but the COH effect size decreased to
5 small and nonsignificant values when the filtered or detrended analyses were performed. The
6 ozone effect size decreased somewhat from the unfiltered OLS analysis, but was similar in
7 magnitude and statistically significant using filtered or detrended OLS, or Poisson regression
8 models.
9
10 Model Specification for the Los Angeles Mortality Studies
11 (Kinney et al, 1995; Ito et al, 1995)
12 Kinney et al. (1995) have discussed a number of important model specification issues
13 for an air pollution time series model. Figure 12-25a,b, taken from their paper, shows the
14 RR estimates for 100 ug/m3 PM10, with alternative methods to control for temporal cycles.
15 In general, most such adjustments for seasonal cycles using dummy variables or Fourier
16 series (sines and cosines) reduced the RR slightly. Subsetting the data into winter and
17 summer groups increased the uncertainty, but did not greatly affect the RR estimate.
18 However, the summer-only RR adjusted with 4 sine/cosine terms was larger than the
19 unadjusted annual RR, and statistically significant. Figure 12-25b shows the results of
20 including co-pollutants, O3 and CO. Including O3 in the model, along with PM10, did not
21 change the RR for PM, but increased its uncertainty slightly so that the RR for PM was now
22 only marginally significant (two-tailed test, p < 0.05). Including CO in the model reduced
April 1995 12-264 DRAFT-DO NOT QUOTE OR CITE
-------
(a) Seasonal Model
(b) Model for Copollutants
No control
Seasonal dummy variables h
4 sine/cosine waves h
10 sine/cosine waves
Wintor I
Summer + 4 sine/cosine
0.9
i • _ )
i • i
• 1
— • 1
• 1
• i
i
• i
i
• i
i
1
i
1 1.1 1.2
Relative Risk
PM10 only
Ozone only \
CO only
L
PM10 +03,
i
PM10 + CO |_
•„, ,. i
— • — 1
i • i
• „. ._! BM
T r MIO
Ij r\ ......
i uzone
1 (-.111
1 PMio
• I CO
0.9 0.95 1 1.05 1.1 1.15
Relative Risk
to
H
6
o
z
3
O
O
H
W
i
o
H-1
H
W
Figure 12-25a,b. Relative risk of total mortality for PM10 in Los Angeles, as a function of (a) seasonal model; and (b) models
including Copollutants.
-------
1 the RR for PM, which was also less significant. CO and O3 were too highly correlated to
2 use in a three-pollutant model.
3 Ito et al. (1995) have evaluated alternative model specifications for combining data
4 from a network of urban monitoring stations, when one station collects data daily and others
5 at an irregular schedule such as once-every-6-days with different days at different stations.
6 While an important subject, this is not the primary source of concern about possible model
1 mis-specification. The optimal use of monitoring data distributed over space and time is
2 more likely to appear as a problem in exposure measurement error arising when any
3 surrogate is used instead of the actual individual exposure.
4
5 Model Specification for the Chicago Mortality Studies
6 (Ito et al., 1995; Styer et al., 1995)
7 Styer et. al. evaluated several alternative models for the Chicago PM10 study discussed
8 in Section 12.3.1. These models assess the effects of dividing data by season. Figure 12-26
9 shows the RR for total elderly mortality per 50 ug/m3 of PM10 in ten different models.
10 Model 0 is their basic best-fitting model using all of the data and assuming a common PM
11 effect for all seasons. The next eight models deal with pairs of model specifications for PM
12 in each season. Models 1,4,6,8 are based on a single model using all of the data with
13 dummy variables for each season that allows separate PM effects in fall, spring, winter, and
14 summer respectively. Models 2, 5, 7, and 9 are similar models fitted independently using
15 subsets of the data for each season. Model 3 is also a separate model for elderly mortality in
16 fall, similar to Model 2 except that the moving average for PM is 5 days, whereas all of the
17 other models used 3-day moving averages. In general, the RR for each season did not show
18 large differences when different estimation methods were used, but there were large
19 differences among seasons in these analyses. The only statistically significant RR were fall
20 and spring. The winter and summer seasons had RR for PM that did not differ significantly
21 from 1.0.
22 Ito et al. (1995) also evaluated alternative model specifications for combining data from
23 a network of urban monitoring stations in Chicago. Relative risks for models with daily
April 1995 12-266 DRAFT-DO NOT QUOTE OR CITE
-------
s
Winter -
d
s
Summer -
d
s
Spring -
d
s5
Fall s
d
Whole Year d
i
i • i
• , ]/ .
i ^
i A
-
. — i 1
r i
1 _ .A — -1
I
i A j
1 W 1
J
\
\
I-A-. 1
^r 1
1 1 1 1
0.9 9.5 1.0 1.05 1.1 1.15 1.2
Relative Risk per 50 ng/m3 PM10
I Lower 95% CL • Relative Risk I Upper 95% CL
Figure 12-26. Relative risk of total mortality for PM10 in Chicago as a function of the
model for co-pollutants.
Source: Styer et al., 1995.
1
2
3
4
5
6
PM10 were statistically significant using any of several alternative averaging models, such as
averaging from all non-missing sites or averaging from all sites using regression-imputed
PM10 for missing sites. Data from some individual sites also gave significant PM effects,
but models using every-6-day data were generally not significant, typically because the
estimated RR had greater uncertainty when only 1/6 as many data were available.
April 1995
12-267 DRAFT-DO NOT QUOTE OR CITE
-------
1 Model Specification for the Steubenville Mortality Studies (Schwartz and Dockery, 1992b;
1 Moolgavkar et al 1995a)
3 Two papers have assessed alternative treatments of a single data base, air pollution and
4 mortality data from Steubenville for 1974-1984, and more recently 1981-1988 (Moolgavkar
5 et al. 1995a). The initial analyses by Schwartz and Dockery (1992b) evaluated several
6 Poisson regression model specifications, including a basic model with mean temperature and
7 dewpoint (same day and lagged one day), and seasonal indicators. Neither same-day nor
8 lagged temperature and dewpoint were statistically significant, nor the square of these
9 variables, nor indicators for hot days (> 70 °F). However, humidity measured by mean
10 dewpoint temperature was nearly statistically significant at the 0.05 level, and an indicator
11 for days that were both hot (> 70°) and humid (dewpoint > 65°) was a statistically
12 significant predictor of mortality. Sensitivity analyses included putting both average of same-
13 day and previous day TSP and SO2 in the model, omitting weather and season variables,
14 including year of the study as either a random effect or as a fixed effect (no year was
15 significant), and including an autocorrelation structure. As expected, including SO2 reduced
16 the TSP effect, but the decrease was small, with RR for TSP decreased from 1.04 without
17 including SO2 to 1.03 per 100 ug/m3 when SO2 was included, and the SO2 coefficient was
18 not significant whereas the TSP coefficient was still statistically significant. As shown in
19 Figure 12-27a, these had little effect on the estimated relative risk for 100 ug/m3 TSP. This
20 paper also demonstrated the use of TSP quartiles for displaying a relationship between the
21 PM indicator and adjusted mortality or morbidity. However, TSP was used as a continuous
22 covariate in the models because the grouping of continuous measurements into groups or
23 categories must involve a loss of information, whether large or small.
24 Moolgavkar et al. (1995a) evaluated a number of Poisson regression models, with
25 particular emphasis on seasonal subsets of the data. The whole-year models analogous to
26 those of Schwartz and Dockery (1992b) are also shown in Figure 12-27a. The results are
27 close to those of Schwartz and Dockery, but are not identical. The RR for 100 ug/m3 TSP
28 are somewhat smaller, but the decrease is only from about 1.032 to 1.025 when SO2 is
29 included in the model. These coefficients are for what Moolgavkar et al. define as the
30 "restricted" mortality data set, which consists of deaths in Steubenville of people who resided
April 1995 12-268 DRAFT-DO NOT QUOTE OR CITE
-------
aionD ION OQ-MVHQ
S66I
?
§
CL
!
a
r
K-
I
a
"8
5S 3d
?» sr
so **
w £3
B £•
fa
s*
^ o
~ >-»9
41
5 e.
^1
S
H
C/}
^d
1
I
I
CO CO
O
CO
-
.
§-.
5
a
f
3
3
I
Whole Year Summer Fall
§8 8 8 8 8
O
%
37
«
i °
37
w
H Oi
8
T7
(
)
1
(
I
<
> 1
5
7?
3.
O
9.
8?
O J M
Whole Year Spring Winter ^
8 8 8 8 8 8 |
-i—:—i—:—i———i-
%
8
e.
5s
3
O
01
-------
1 there. This is comparable to the data set used by Schwartz and Dockery in this study, and
2 by Schwartz or Dockery and their associates in many other studies. The argument for use of
3 the "restricted" mortality data is that community-based air monitors provide better exposure
4 indicators for people who live in the community most of the time, as opposed to commuters
5 or to other visitors who die in the community. Also, since many metropolitan areas contain
6 medical facilities that may be better equipped than those in more remote areas, it is possible
7 that some excess number of the deaths in elderly or ill patients transported from the more
8 remote areas occur in urban centers such as Steubenville. Moolgavkar et al. (1995a) also
9 show results for analyses of "full" mortality data, which includes individuals who did not
10 reside in the location at which they died.
11 It is clear that season-to-season effects are present in these data. Schwartz and Dockery
12 found that winter and spring mortality was significantly higher than summer and fall
13 mortality. Moolgavkar separated the analyses by season. He found that whole-year RR for
14 TSP was nearly the same as RR in the separate summer and fall models, with or without SO2
15 in the model, and nearly the same in the spring model when SO2 was included. However,
16 TSP coefficients were higher in the winter, and in the spring model when SO2 was not
17 included. In fact, as shown in Figure 12-27b, the RR for TSP increased slightly in the winter
18 model when SO2 was included.
19 There is a possibility that the weather model used by Schwartz and Dockery, and by
20 Moolgavkar et al., are not adequate to remove all of the seasonal effects. It is possible that
21 additional variance reduction could have been achieved with the use of additional weather
22 data, emphasizing more extreme conditions than the very moderate cutpoints of temperature
23 and dewpoint, since temperature extremes are known to have effects on mortality (Kalkstein
24 et al., 1991, 1995; Kunst et al., 1993). Variables used by other investigators, such as
25 barometric pressure, could have been tested. The flexibility of the model to fit nonlinear
26 relationships could be improved by the use of nonparametric or semi-parametric models, and
27 classifying data by synoptic weather category may provide a useful alternative approach to
28 evaluating the interaction between season and weather.
29 Moolgavkar found that the TSP coefficients were not statistically significant (two-tailed
30 tests at 0.05 level) in any season except winter, nor in the whole-year model, when SO2 was
31 included in the model. However, the season-by-season TSP coefficients were not tested in a
April 1995 12-270 DRAFT-DO NOT QUOTE OR CITE
-------
1 whole-year model. Part of the non-significance may be attributable to the fact that
2 confidence intervals for a regression parameter in a separate seasonal model, with about 1/4
3 of the data in a whole-year model, may be on the order of twice (= reciprocal square root of
4 1/4) as wide as the confidence interval for the corresponding season-by-pollutant regression
5 coefficient in a whole-year model, everything else being equal.
6 The method of adjusting the mortality series for weather effects and for other time-
7 related effects (detrending) may be important in explaining why the RR estimates for TSP
8 vary seasonally and why those derived by Moolgavkar et al. are quantitatively different from
9 those derived by Schwartz and Dockery (1992b), even though the differences are small in
10 these studies. There may exist some residual confounding with weather, since other studies
11 have found that substantial adjustment of weather by use of temperature and dewpoint
12 categories, or nonparametric smoothers of temperature, humidity, and time can effectively
13 eliminate seasonal variations in residuals and in PM effect. Even so, the estimated TSP
14 effect on RR of mortality is positive in most seasons, even in Steubenville models including
15 the collinear co-pollutant SO2. No adjustments were made for other pollutants such as CO,
16 NOX, and O3.
17 These analyses of the Steubenville data set are primarily useful for demonstrating the
18 results of different data analysis strategies and methods, since the PM indicator was TSP, not
19 PM10. These analyses have shown the desirability of adequately adjusting the analysis of
20 pollution effects for weather and for long-term and medium-term time trends and variations.
21 When co-pollutants were evaluated, it was evident that only part of the TSP effect could be
22 attributed to SO2- Differences in RR of TSP between analyses presented in the two papers
23 are not regarded as large.
24
25 Model Specification for Philadelphia Mortality Studies (Schwartz and Dockery, 1992a; Li
26 and Roth, 1995; Moolgavkar et al 1995; Wyzga and Lipfert, 1995)
27 Several papers have recently appeared that allow assessment of alternative treatments of
28 a single data base, the air pollution and mortality data from Philadelphia for the years 1973
29 to 1980, and more recently 1981 to 1988 (Moolgavkar et al. 1995b). The initial analyses by
30 Schwartz and Dockery (1992a) evaluated several Poisson regression model specifications,
31 including a basic model with mean temperature and dewpoint (lagged one day), winter season
32 temperature (same day), and an indicator for hot days (> 80 F). Sensitivity analyses
April 1995 12-271 DRAFT-DO NOT QUOTE OR CITE
-------
1 included putting both average of same-day and previous day TSP and SO2 in the model,
2 stratifying analyses as above or below median SO2 level (18 ppb), omitting weather and
3 season variables, and including day of week. As shown in Figure 12-28, these had little
4 effect on the estimated relative risk for 100 /ig/m3 TSP. RR for mortality in the elderly was
5 greater than for other age groups. A more detailed assessment of the age structure was
6 presented by Schwartz (1994), showing clearly that there was increased mortality in ages
7 65 to 74, and again higher at ages 75 + . There was also significantly increased mortality at
8 ages 5 to 14 years, based on a small number of cases. This paper also demonstrated the use
9 of TSP quantiles for displaying a relationship between the PM indicator and adjusted
10 mortality or morbidity. However, TSP was used as a continuous covariate in the models
11 because the grouping of continuous measurements into groups or categories must involve a
12 loss of information, whether large or small.
13 Li and Roth (1995) reanalyzed these data, but only reported results in the form of
14 t-statistics. A wide range of model specifications were tested, although some models (such
15 as those using deviations from mean values for day of year, or from monthly average) appear
16 to assume an unrealistic level of seasonal recurrence of air pollution and weather effects.
17 The model most directly comparable to the Poisson regression models used by Schwartz and
18 Dockery (1992a). The autoregressive model they denoted AR(6) was somewhat comparable
19 to models tested in the London analyses of Schwartz and Marcus (1990). The models with
20 residual deviations of mortality from 7-, 15-, or 29-day moving averages did not have
21 comparably filtered predictors on the "right" side of the prediction equation, however, so the
22 regression coefficients are not readily interpretable as predictions of mortality deviations
23 from mean pollution levels. The Poisson log models that are most comparable to those used
24 by other investigators involved comparisons of model specifications for averaging times.
25 The results only indicate statistical significance by use of statistics, not effect size in any
26 form more useful in epidemiologic studies (Greenland et al., 1986). In a model that includes
27 TSP, SO2, and O3, statistical significance of TSP is clearly highest with the moving average
28 of 0+1 day lags, and diminishes sharply for all pollutants when longer lags are included.
29 Models with longer weather averages are also more predictive. The lower significance of the
30 TSP term may be related to the fact that it may have greater exposure measurement error
April 1995 12-272 DRAFT-DO NOT QUOTE OR CITE
-------
Schwartz and Dockery Study
Moolgavkar whole-year model
Model 1 when age < 65
Model 1 for age 65+
Model 1 + day of week
Model 1 w/o weather
Model 1 when SO2 > 18ppb
Model 1 when SO2 < 18ppb
Model 1 + SO2
Model 1: hot day, mean temp, winter temp,
mean dewpoint, year, 4 seasons, linear trend
•._„. _ .„„ 1
1*1
1 A 1
1 A 1
1* •
— 1 1
0.95 1.00 1.05 1.10 1.15
Relative Risk per 100 ng/m3 TSP
I l Lower 95% CL • Relative Risk ' Upper 95% CL|
Figure 12-28. Relative risk of total mortality for TSP in Philadelphia.
Sources: Schwartz and Dockery 1992a; Moolgavkar et al. 1995b.
1 than the gaseous pollutants. The models being evaluated in this paper have not been adjusted
2 for collinearity even though there are some fairly strong collinearities in the data, such as
3 between TSP and SO2, and between temperature and ozone, so that inclusion of several
4 collinear variables is almost certain to greatly inflate the variance and thus to reduce the
5 statistical significance of many of the regression coefficients.
6 Moolgavkar et al. (1995b) evaluated a number of Poisson regression models, with
7 particular emphasis on seasonal subsets of the data. These are shown in Figure 12-29a-d. It
8 is clear that season-to-season effects are present in these models. The models were adjusted
9 for weather and time trend by using quintiles of temperature and indicators of year. There is
10 a possibility that the weather model is not adequate to remove all of the seasonal effects.
11 Subdividing the temperature range by quintiles will result in three or four closely spaced
April 1995
12-273 DRAFT-DO NOT QUOTE OR CITE
-------
SO2 quintiles
+ Ozone
BothS02
and Ozone
No Ozone,
+ SO2
NoSO2
or Ozone
(a) Mool
gavkar: Spring
1 ,, _,..j
(b) Mool
1
gavkar: Summer
1 1 i
0.90 0.95 1.00 1.05 1.10 1.15 0.90 0.95 1.00 1.05 1.10 1.15
Relative Risk per 100 ng/m3 TSP Relative Risk per 100 ng/m3 TSP
SO2 quintiles
+ Ozone
Both SO 2
and Ozone
No Ozone,
+ SO2
NoSO2
or Ozone
(C) Mool!
pavkar: Fall
1 1
(d) Mool!
1
javkar: Winter
1 i,,
0.90 0.95 1.00 1.05 1.10 1.15 0.90 0.95 1.00 1.05 1.10 1.15
Relative Risk per 100 ng/m3 TSP Relative Risk per 100 jig/m3 TSP
Lower 95% CL • Relative Risk I Upper 95% CLI
Figure 12-29a-d. Relative risk of total mortality for TSP in Philadelphia, in the (a)
spring; (b) summer; (c) fall; and (d) winter.
Source: Moolgavkar et al., 1995.
April 1995
12-274 DRAFT-DO NOT QUOTE OR CITE
-------
1 quintiles corresponding to moderate temperatures which have little effect on mortality, and
2 will not adequately take into account temperature extremes. Quintiles of temperature are not
3 given here, but Li and Roth (1995) report values of maximum temperature at the 10th, 25th,
4 50th, 75th, and 90th percentiles as 37, 48, 63, 78, and 84 degrees F; (Schwartz and Dockery
5 (1992a) report mean temperature percentiles corresponding to 25, 36, 52, 66, and 73 degrees
6 F, respectively. This paper finds a significant effect for the highest quintile in the summer,
7 and the lowest quintile in the other seasons. This suggests that additional variance reduction
8 could have been achieved with the use of additional weather data, emphasizing more extreme
9 conditions than the 20th and 80th percentiles, and possibly including information on dewpoint
10 or barometric pressure, as used by other investigators. In general, replacing numeric data by
11 grouped equivalents such as quintile classes involves some loss of information. The loss of
12 information may be acceptable if there is a corresponding increase in the flexibility of the
13 model to fit nonlinear relationships, but in this instance the loss of information may be
14 substantial since extreme temperatures are known to have a quantifiable and increasing
15 relationship with mortality as the temperatures become more extreme (Kunst et al., 1993).
16 The method of adjusting the mortality series for weather effects and for other time-related
17 effects (detrending) may be important in explaining why the RR estimates for TSP vary
18 seasonally and are quantitatively different from those derived by Schwartz and Dockery
19 (1992a). There may exist some residual confounding with weather, since other investigators
20 have found that substantial adjustment of weather by use of temperature and dewpoint
21 categories, or nonparametric smoothers of temperature, humidity, and time can effectively
22 eliminate seasonal variations in residuals and in PM effect. Even so, the estimated TSP
23 effect on RR of mortality is positive in most seasons, even in models including collinear co-
24 pollutants SO2 and O3.
25 Neither the Schwartz and Dockery (1992a) study nor the Moolgavkar et al. (1995b)
26 study allows a complete assessment of the actual role of co-pollutants as confounders of a
27 PM effect. While SO2 is not as strongly correlated with temperature as is O3, it is also
28 subject to weather conditions that affect atmospheric dispersion along with TSP. Therefore,
29 if there is an incorrect assignment of weather effects on mortality, some part of the mortality
30 that could have been explained with weather-related variables will probably be allocated to
31 various other predictors of mortality used in the models, especially TSP and the co-
April 1995 12-275 DRAFT-DO NOT QUOTE OR CITE
-------
1 pollutants. The development of a predictive model for mortality using weather and other
2 time-varying covariates would probably have required use of humidity, since humidity along
3 with temperature had been predictive of mortality in earlier studies such as London (Schwartz
4 and Marcus, 1990) and Steubenville (Schwartz and Dockery, 1992b). If confounding can be
5 explained by an unobserved (or in this case, unused) covariate, then omission of humidity
6 from any of the models in the Moolgavkar et al. (1995b) study is certainly another candidate
7 explanation for the differences in results between these papers. However, both papers may
8 also have provided an inadequate adjustment for other medium-term effects on a scale longer
9 than a day and shorter than a season or quarter, such as epidemics. The use of
10 nonparametric smoothers such as LOESS, or GAM models of time, would allow subtraction
11 of such trends. Even a simple alternative, such as including a dummy variable for every
12 month in every year (96 parameters for the 1973-1980 series; another 96 parameters for the
13 1981-1988 series) would probably have greatly improved the ability of these analyses to
14 evaluate short-term responses to short-term changes in air pollution and weather. The
15 parameters that relate mortality to pollution and weather over intervals of a few days were
16 likely the same or similar over periods of some years, and would require only a few more
17 parameters. In view of these questions, we regard potential confounding among TSP, SO2,
18 and summer ozone in Philadelphia that was identified in the Moolgavkar et al. (1995b) study
19 as possible, but not yet proven.
20 Another unresolved issue is that TSP may have relatively large exposure measurement
21 error than the gaseous pollutants. Since TSP includes large particles, there is more
22 association of TSP levels with local sources and transport near the air pollution monitors, and
23 a weaker correlation with TSP at other monitors than is the case of smaller particles. In
24 particular, TSP would be expected to show less correlation within the Philadelphia region
25 than would PM10, and even less yet than would PM2 5 across the region. Therefore, TSP
26 may be less predictive of individual PM exposure than the gaseous pollutants in Philadelphia.
27 Since variables with larger exposure measurement error are more likely to show attenuated
28 effects (bias towards smaller RR) than covariates with smaller measurement errors, it is at
29 least possible that SO2 may appear to be a more important predictor of pollution-related
30 mortality than does TSP. There does not seem to be any way to evaluate these possibilities
31 from the published reports.
April 1995 12-276 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wyzga and Lipfert (1995) also reanalyzed the Philadelphia time series data for 1973 to
2 1990, using Gaussian OLS regression models with time-lagged predictors. In view of the
3 moderately large number of deaths per day (21 deaths at ages less than 65 years, 34.5 deaths
4 at ages 65 and older), the OLS regression coefficients are probably sufficiently accurate
5 approximations to regression coefficients estimated from Poisson regression models. They
6 evaluated model specifications for daily mortality, log mortality, and deviations of mortality
7 from 15-day moving averages. The regression models were adjusted for maximum
8 temperature dummy variables in 6 categories, winter season, daily changes in barometric
9 pressure, and time trend. Maximum hourly ozone was used as a co-pollutant. RR estimates
10 for TSP were calculated using regression coefficients and standard errors in their Table 3,
11 plus data from their Figures 14 and 20. Figure 12-30a shows RR for ages < 65 years,
12 Figure 12-30b for ages 65+ years, for all days (N = 2380) and for N=390 hot days
13 (maximum temperature at least 85 degrees), for different averaging times. The largest and
14 most significant estimates of TSP effect, measured as deviations from 15-day moving
15 averages, are in the elderly, especially on hot days. For the elderly on hot days, the TSP
16 effect is nearly the same for averaging times from 2 to 5 days on hot days, but the 0+1 day
17 moving average has only slightly greater statistical significance than thee 0+1+2+3 and
18 0+1+2+3+4 day averages. When all days are considered, the RR for TSP is only half as
19 large, and statistically significant only for 0+1 day TSP averages. For deaths at age <65,
20 none of the all-day TSP were significant; on hot days, the 0+1 average TSP was nearly
21 significant, and the 0+1+2 day average TSP effect nearly as large, but other RR estimates
22 were much smaller. The estimates were not calculated using filtered pollution series, but the
23 moving averages of TSP had some of the same effect of removing long-term trends and
24 effects. These estimates are in general similar to those found by Schwartz and Dockery, but
25 larger differences were found for other model specifications. This paper did not attempt to
26 include SO2 as a covariate, since TSP was clearly collinear with SO2.
27 These analyses of the Philadelphia data set are primarily useful for demonstrating the
28 results of different data analysis strategies and methods, since the PM indicator was TSP, not
29 PM10. These analyses have shown the desirability of adequately adjusting the analysis of
30 pollution effects for weather and for long-term and medium-term time trends and variations.
April 1995 12-277 DRAFT-DO NOT QUOTE OR CITE
-------
>
•g (a) Age < 65
H- ' .
\O A.
^o
to 3
>. Q 2
0 I
.E 1
(D
1 °
I 4.
Cu
flj ^4
!Tt Q o
C^* f
\ — fc
" 1
0
1
6
o
1
o
3 Figure 12-30a,b.
w
Q Source: Wyzga and
3
h
•
9
1 ^
H
•
1
1 •
»l
M
.8 0.9 1.0
Relative Risk per
Relative risk of mortalit
(a) Age < 65; (b) Age :
Lipfert, 1995.
(b) Age 65+
1
H
I_
L
h
I • 1
I • *
1
• I
•H
' '
1.1 1.2 1.3 0.9 1.0 1.1 1.2 1.3 1.4
100 ng/m3 TSP Relative Risk per 100 ng/m3 TSP
1 Lower 95% CL • Relative Risk 1 Upper 95% CL
y for TSP in Philadelphia, as a function of age, averaging tune, and temperat
> 65.
-------
1 When co-pollutants were evaluated, it was evident that only part of the TSP effect could be
2 attributed to ozone, and that the ozone effect was more nearly confounded with temperature
3 and season than with TSP. However, there was a substantial degree of confounding between
4 TSP and SO2 effects, which could be separated in some analyses but not in all analyses. The
5 best averaging time for pollution was 0+1 days, but longer averages seemed useful in
6 estimating RR among the elderly during hot weather.
7
1 Model Specification for Other Mortality Studies
2 Other studies on acute mortality have evaluated alternative model specifications. A
3 number of OLS and time series regression models for COH in Santa Clara were compared by
4 Fairley (1990). Mortality studies by Schwartz for Detroit (Schwartz 1991) and Birmingham
5 (1993a) evaluated other regression and time series approaches. These are not reported in as
6 much detail as the studies cited here, and the Detroit study also uses estimates for PM.
7
1 12.6.2.2. Model Specification for Morbidity Studies
2 There have been a large number of recent studies on hospital admissions related to PM
3 (Schwartz, 1993b, 1994b, 1994e, 1995a, 1995b, 1995c). These have used generally similar
4 strategies for evaluating alternative Poisson regression models. The basic model includes a
5 set of variables for temperature and dewpoint (usually in 6 to 8 categories), linear and
6 quadratic time trends, indicators or dummies for each month in each year (so that no
7 assumptions need to be made about recurrent seasonal or monthly effects), and the PM
8 indicator. Alternative model specifications usually include: (1) piecewise cubic spline
9 functions for time trend, temperature, and dewpoint; (2) generalized additive models (GAM)
10 for time trend, temperature, and dewpoint; (3) basic model, excluding all non-attainment
11 days (PM10 > 150 ug/m3, or ozone > 120 ppb, etc.); (4) basic model without hot days; (5)
12 extended range of lag times or moving averages; (6) basic model plus co-pollutants.
13 Differences in RR for PM among most specifications is small. RR estimates from the GAM
14 method tend to be higher than most other specifications, but the conclusions about RR are
15 fairly insensitive to alternative specifications. Since there have been no studies that disagree
16 with these conclusions, we will not review these in detail, since the assessments are in many
17 ways similar to those for the acute mortality studies.
April 1995 12-279 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.6.2.3. Model Specification Issues: Conclusions
2 Published research articles have provided a substantial amount of evidence about the
3 consequences of different model specifications for short-term and long-term models. The
4 short-term studies have been generally consistent across many different kinds of model
5 specifications. While it is tempting to conclude that differences in model specification
6 produce only minor differences in estimates of PM effects, that is clearly not the case. The
7 general concordance of PM effects, particularly in analyses of short-term mortality studies, is
8 a consequence of certain appropriate choices in modelling strategy that most authors have
9 adopted but the results are not dictated by the use or misuse of any specific model. Some
10 features of the more useful models are described below.
11
12 12.6.3 Other Methodological Issues for Epidemiology Studies
13 The issues in air pollution epidemiology for PM are similar to those of many other
14 pollutants. No single air pollutant, nor any mixture such as PM or an identifiable component
15 of PM, is uniquely related to a specific health outcome and individual exposure
16 measurements are generally lacking, with exposure to PM typically measured at only one site
17 in an urban or regional airshed or, at most, at a few widely spaced sites. U.S. studies of
18 acute mortality typically depend on combining three data bases: (1) mortality data tapes
19 provided by the National Center for Health Statistics (NCHS); (2) air pollution data sets for
20 urban areas, accessed through the Aerometric Information Retrieval System (AIRS) network;
21 and (3) meteorological data for urban areas and smaller SMSA's, obtained from the National
22 Climatic Data Center (NCDC). Hospital admissions data involve a more diverse set of
23 sources. Merging the data sets has not always been a straight-forward task, and attempts to
24 replicate results have sometimes been complicated by the fact that different investigators have
25 used different approaches to creating a merged data set for subsequent analyses. As a simple
26 example, the PM10 monitoring data for Chicago consists of every-day monitoring at one site
27 and every-6-days monitoring at up to eight other sites. In that ease, different investigators
28 may calculate different PM10 concentrations according to how the data from the intermittent
29 monitoring sites are combined with data from the every-day site. In this section, we will
30 discuss specific methodology issues encountered in the studies reviewed earlier.
31
April 1995 12-280 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.6.3.1 Participate Matter Exposure Characterization
2 PM10 measures the inhalable particles better than TSP. The NAAQS are specified by
3 PM10 concentrations, which were not generally available before 1986. PM10 is also a better
4 index of ambient fine particle exposure than TSP because it is more uniformly distributed in
5 an urban area or region than TSP. Since fine particles from outside can also penetrate
6 indoors and constitute a major fraction of indoor air concentrations, PM10 is also likely to be
7 a better index of indoor air exposure to fine particles than TSP. Currently, data on AIRS do
8 not allow discrimination among important components of PM10, including fine particles,
9 coarse particles, or sulfates. In the absence of any clearly demonstrated mechanistic
10 relationship between PM components (by size, composition, or source) and specific health
11 endpoints, there is little a-priori reason to believe that health endpoints related to PM should
12 not be predicted well in different studies by different PM indices. The indices include PM10,
13 fine particles (defined as PM2 5), coarse fraction of PM10 particles (defined as PM10 -
14 PM2 5), or surrogates that may be more closely correlated with fine particles than to coarse
15 particles, including sulfates, SO2, and even TSP. Results presented by Dockery and Pope
16 (1994) suggest that PM2 5 may be a more appropriate "proxy" of exposure to particles that
17 are predictive of health effects.
18 While PM2 5 particles are more likely to be uniformly distributed within an urban
19 airshed, and are more likely to penetrate indoors so that outdoor ambient fine particle
20 concentration becomes a better predictor of total fine particle exposure than ambient coarse
21 particle concentration does for total coarse particle exposure, it is not clear that coarse
22 particle concentration can be neglected in terms of its health effects. While sulfates are a
23 significant part of fine particle levels in some places, they may be of more limited value as
24 an indicator of a toxic component of PM10 due to measurement artifacts (filter artifacts).
25 The usefulness of sulfate data may also be limited because high levels are reached only in the
26 eastern and southeastern U.S., mainly during the summer. Information on other components
27 of PM10, including acidity, metal ions, and organic components, is often not available.
28 Similarly, data deficiencies exist for most co-pollutants. In studies where SO2 is a good
29 proxy for PM10, it may be difficult to assign effects to one or the other without evaluating
30 the relationships linking the two, since SO2 is the source of some fraction of particle
31 sulfates.
April 1995 12-281 DRAFT-DO NOT QUOTE OR CITE
-------
1 Chapter 7 on Human Exposure to PM indicates that variations in ambient PM
2 concentration may have limited influence on the variation of individual personal PM
3 exposures as measured by personal monitors. However, although individual exposures may
4 not be well correlated with ambient PM concentrations measured at representative fixed
5 ambient monitoring site(s), the mean personal exposure of people in a community is expected
6 to be better correlated to the ambient concentration. In terms of community air pollution, a
7 properly sited ambient PM measurement is reasonably related to the mean personal PM
8 exposure of the community, although it will not be a good indicator of any single individuals'
9 daily PM exposure. The important consideration here is that the ambient monitors be
10 properly sited. This would have to be evaluated study by study, which can be difficult or
11 impossible if data has not been reported in the study. There must be limits to the
12 acceptability to using a monitor for daily level changes in both distance from the population
13 and terrain between population and monitoring site.
14 Data are available at more than one monitoring site in a few studies, including
15 Birmingham AL (Schwartz, 1993, 1994), Utah Valley (Pope et al., 1992, 1994), Los
16 Angeles (Kinney et al., 1991, 1995), San Francisco Bay area (Fairley, 1994), Philadelphia
17 (Wilson et al., 1995), and Chicago (Ito et al., 1994, 1995; Styer et al., 1995). While PM10
18 varies from place to place, with a decreasing correlation across a metropolitan area,
19 measurements are well correlated up to a few kilometers. FP (PM2 5) is particularly well
20 correlated across a metropolitan region.
21
22 Exposure Relevance
23 The majority of the PM data used in the PM/mortality literature are daily observations,
24 rather than the standard every-6th-day observations. The ambient daily mean PM levels
25 reported in these PM/mortality studies of U.S. cities range from 28 /ig/m3 (St. Louis, MO)
26 to 58 /ig/m3 (Los Angeles, CA) for PM10; 76 /xg/m3 (Cincinnati, OH) to 111 /ig/m3
27 (Steubenville, OH) for TSP. Other PM indices in the literature include COH in Santa Clara
28 County, CA; and KM (mean = 25) in Los Angeles County, CA. The data description
29 reported for these PM indices indicate a generally skewed distribution, and the maximum
30 daily values deviate about 50 to 150 /xg/m3 from these means. The current 24-h standard,
31 150 /xg/rn3, is rarely exceeded in these communities. Many of these communities studied
April 1995 12-282 DRAFT-DO NOT QUOTE OR CITE
-------
1 were urban, but the PM levels observed appeared to be representative of metropolitan areas
2 where substantial fraction of U.S. population reside.
3
4 Size and Chemistries
5 In theory, since TSP includes particle sizes (da < 50 /zm) that exceed those having
6 thoracic deposition (da < 10-15 /mi), it is expected that TSP would be a less reliable
7 measure of particulate matter for health effects analyses. However, comparison of the
8 significance of the PM regression coefficients in the recent U.S. PM/mortality studies do not
9 show systematically lower significance for TSP than PM10. This may be because, so long as
10 TSP levels fluctuate together with smaller particles over time, TSP may still be a reasonable
11 surrogate for inhalable particles. The error introduced by large particles depends on their
12 availability, and therefore, is site-specific.
13 In most of the PM/mortality studies, only one PM index was employed. An exception
14 is the study conducted in St. Louis, Mo. and Kingston, TN (Dockery et al., 1992). In this
15 study, PM10, PM25, sulfates, and aerosol acidity were available. The regression results
16 indicate that, for both cities, PM10 showed the most significant mortality associations, and
17 the significance declined as the size of the index decreased. However, the sample size of
18 this study was relatively small (n = 300; PM10 coefficient t-ratio = 2.17 in St. Louis, and
19 1.07 in Kingston), and the sample size for the aerosol acidity was even smaller (n = 200).
20 Furthermore, we currently do not know the extent to which the measurement errors of these
21 different PM measures affect PM/mortality significance. Thus, it is as yet premature to
22 relate the significance of various PM measures to size or chemistry specific causality from
23 this study. This inability, in a sense, to distinguish the possible difference in health effects
24 from particles of various size and chemistries that fluctuate together, is one of the
25 shortcomings of existing time-series studies. Cross-sectional studies reported more
26 significant mortality associations for fine particles (PM2 5: Dockery et al., 1993; sulfates:
27 Ozkaynak and Thurston, 1987). However, significant PM/mortality associations have also
28 been reported in areas where summertime sulfates are not the major component of PM (e.g.,
29 winter analysis of Santa Clara, CA; Los Angeles, CA). All the PM measures in the U.S.
30 studies do include some type of combustion source originated particles (e.g., automobile
31 emissions in Los Angeles, sulfates in the eastern U.S.). Overall, PM composition vary
April 1995 12-283 DRAFT-DO NOT QUOTE OR CITE
-------
1 widely, not only between sites, but also over time at a single site. This represents a major
2 challenge to any attempts to quantify the health implication of PM.
3
4 12.6.3.2 Exposure-Response Functions, Including Thresholds
5 A PM threshold for mortality is difficult to detect because of small numbers of deaths
6 (especially when broken down by age group and cause of death), and because the observed
7 PM concentration is only a surrogate for exposure. In general, the threshold question has
8 not been extensively examined. Other model specification issues that have had little
9 examination include non-linear transformations of pollutant variables, interactions among
10 pollutant variables, and interactions between meteorological variables and pollutants.
11
12 Thresholds
13 The question of threshold is critical when extrapolation of effects from episodes to non-
14 episode (i.e., more usual) condition is attempted. Also, a key question of interest is whether
15 effects extend below the current standard.
16 Many of recent U.S. PM/mortality studies have reported PM/mortality "exposure-
17 response" curves of the data, after controlling for weather and seasonal variables (Schwartz,
18 1992, 1993, 1994; Pope et al., 1992). Some of these are shown in Figures 12-31 and 12-32.
19 They were constructed either by using quintile or quartile indicator variables in the
20 regression, or by nonparametric smoothing (in the Generalized Additive Models), both of
21 which should allow for possible non-linear relationships. In all the figures presented, a
22 generally monotonic increase in mortality, as PM increases, is suggested. However, a search
23 for a threshold from these results is difficult because of the distribution of the data. For
24 example, in the plots of the quintile (or quartile) PM versus relative risk, the resolution of
25 the shape of slope is determined by the number (5 or 4) of indicator categories. The lowest
26 quintile (or quartile) could be higher than a potential threshold level (e.g., the lowest quintile
27 of TSP was about 50 jig/rn3 in Philadelphia). Because estimation of more stable coefficients
28 require greater number of cases, even a large dataset may not allow smaller data division
29 than quintiles. Thus, from these results, we cannot determine if any threshold exists below
30 about 50 /ig/m3 for TSP or 20 /^g/m3 for PM10. Nonparametric smoothing of relative risk
31 versus PM can, in theory, allow greater resolution of the shape. However, the stability of
April 1995 12-284 DRAFT-DO NOT QUOTE OR CITE
-------
Philadelphia, PA
Utah Valley, UT
.c
cs
8
"5
to
ir
>
CO
CD
_c
CO
CD
Q
•5
co
ir
CD
'•§
CD
DC
l.W
1.03
1.02
1.01
1.00
0.99
A
-
_
A A
A
A
i i
JZ
to
Q
"5
CO
ir
.1
to
CD
cc
0 50 100 150
Total Suspended Particulates
Steubenville. OH
1.06
1.04
1.02
1.00
0.98
i i
_c
08
CD
Q
"5
CO
ir
CD
'•§
CD
DC
0 100 200 300
TSP
_c
I
O
\s
CO
ir
.1
to
CD
DC
1.20
1.10
1.00
0.90
A
•
A
A
A
i i i i
0 20 40 60 80 100
PM,0 Concentration
Detroit, Ml
1.03
1.02
1.01
1.00
0.99
•
A
A
A
• A
i i i
60 80 100 120 140
Predicted TSP
St. Louis, MO (A)
Eastern TN (V )
1.10
1.05
1.00
np.<>
V
^ V ^^
^^"^ A
^^X'^
^^^ T
•^^^^ A
i i i
10 20 30 40 50
PM,0
Figure 12-31 Relative risk of mortality in five studies, where the particulate matter
index is either total suspended particulates (TSP) or PM10, in units of
/tg/m3 grouped in quintiles. Figures are based on those in the original
papers; note differences in scale.
April 1995
12-285 DRAFT-DO NOT QUOTE OR CITE
-------
Birmingham, AL
1.15
•5 1.10
1.05
o
DC
1.00 1
20 40 60 80 100 120 140
PM10
Cincinnati, OH
1.10
1.05
1.00
20 40 60 80 100 120 140
Total Suspended Particulates
Philadelphia, PA
1.10-
1'08'
1.06'
1.04-
1.02-
1.00
0 50 100 150
Total Suspended Paniculate Matter
(ng/rrf)
Figure
12-32. Smoothed non-parametric estimate of relative risk of mortality in three
studies, where the particulate matter index is either total suspended
particulates (TSP) or PM10, in units of jig/m3.
1 the results also depends on the weights of neighborhood and the interval of the PM, or
2 "span", used to compute each segment of the curve (these parameters are not described in
3 these publications). Again, these smoothed curves, as with the quintile approach, cannot
4 describe the shape of the curve where data do not exist. For example, the smoothed curve
5 shown for Cincinnati, OH, appears to suggest a threshold around 40 /xg/m3 of TSP, but the
6 distribution (25th percentile TSP = 53 /xg/m3) indicates that there are not enough data points
7 below 40 jig/m3 to obtain stable curve shape below this level. Thus, while these figures do
8 collectively suggest a linear-like PM/mortality relationship, any examination of a threshold
9 level is limited by the data. Other studies did not consider or present graphical examination
10 of the possible shape of any exposure/response relationship, and, thus, the results could have
11 been constrained by the functional form specified in the regression model.
12 It is possible that there is a causal effect of airborne PM, but rather than altering the
13 long term average mortality rate, peaks in exposure simply advance the date of death of
14 otherwise terminally ill subjects. The terms "mortality displacement" or "harvesting" have
15 generally been applied to this hypothesis. Under this scenario, lowering particulate matter
16 concentrations might grant a few extra days life to a small part of the population, but have
17 no effect on the general mortality rate. It is obviously extremely important for policy
18 making purposes to resolve whether this is indeed the case.
April 1995
12-286 DRAFT-DO NOT QUOTE OR CITE
-------
1 Although the possibility has been discussed in several of the papers reviewed, only one
2 (Spix et al., in press) seems to have offered a serious test of the hypothesis. They point out
3 that the effect of harvesting should be to induce a negative effect on the autocorrelation,
4 since " a high number of deaths on one day may leave a smaller number of vulnerable
5 individuals at risk of dying on succeeding days." They further suggest that the magnitude of
6 this effect should be proportional to the excess deaths due to pollution. Hence, they test the
7 hypothesis by adding an interaction between the pollutant level on that day and the last k
8 days mortality deviation from the expected value, where the expected value is based on a
9 previously fitted model including trend, season, and influenza epidemics. A negative
10 estimate for this interaction term would be interpreted as evidence for this phenomenon.
11 Applying this test to data from Erfurt, East Germany, Spix et al. found a weak effect for
12 suspended particles in the expected direction (nominal one-tailed p = 0.07 ignoring the
13 multiple testing for k): the RR comparing the 5th and 95th percentiles of the exposure
14 distribution was 1.51 if the previous 18 days mortality was above expected, 1.26 if it was
15 below expected. A somewhat similar approach, examining mortality displacement from
16 summer heat waves, has been described by Kalkstein (1994).
17 The statistical properties of this test merit further research. However, a full
18 investigation of the performance of the test in realistic settings with the more sophisticated
19 time series and GEE methods, including estimation of the harvesting parameter k is beyond
20 the scope of this assessment.
21 In addition to statistical research, further epidemiologic research is warranted to better
22 characterize the excess deaths in terms of age, cause of death, hospitalization status, prior
23 morbidity, etc. It may be necessary to develop a multistage model, with recruitment of
24 individuals from a healthy stage through one or more stages of morbidity until they reach a
25 susceptible stage at which acute air pollution exposure may cause deaths.
26
27 12.6.3.3 Adjustments for Seasonally, Time Lags, and Correlation Structure
28 Trends, long-term and medium-term recurrent or cyclical effects, and effects of
29 medium-term non-recurrent or random events such as influenza epidemics are removed from
30 the data so that short-term responses to short-term changes in PM concentration can be
31 detected without confounding or interference from longer-term effects. For Gaussian time
April 1995 12-287 DRAFT-DO NOT QUOTE OR CITE
-------
1 series models, this can usually be done well by filtering. However, filtering has the potential
2 to remove longer-term effects of PM exposure, and therefore may underestimate the true PM
3 effect. For example, death may occur from PM exposure during the first few days after
4 exposure because the PM exposure may exacerbate pulmonary insufficiency in individuals
5 whose respiratory capacity has already been compromised, especially the elderly and the ill.
6 This may also contribute to excess short-term cardiovascular mortality. However, if PM
7 exposure also compromises the immune system, the exposed individual may succumb to an
8 infectious disease some weeks after the PM exposure, an effect that would be more likely to
9 be cancelled out by application of filtering or other detrending techniques. Detrending could
10 also be done by using regressors that are functions of the time or day of study. Candidate
11 regressors are Fourier series (sums of sine and cosine terms), polynomial functions of time,
12 dummy variables for year, season or quarter, month, or day of week. Fourier series are
13 mathematically convenient, but require many terms in order to fit asymmetric seasonal
14 variations, and cannot include random year-to-year differences in seasonal effects. Dummy
15 variables for year, season, and month provide a great deal of flexibility, but may still be too
16 "rough" in that such models allow abrupt changes between December 31, 1980 and January
17 1, 1981, between June 30 and July 1, etc. Non-parametric smoothers such as spline
18 functions, LOESS smoothers, and generalized additive models are often good choices, but as
19 with any other detrending procedure, the scale or span of the smooth detrender determines
20 what medium-term effects are removed from the model.
21 A number of short-term studies have provided reasonable control over time-related
22 exogenous changes. The use of tapered high-pass filters in Gaussian time series models, in
23 connection with linear time trends or dummy variables for season or day of week, has been
24 demonstrated in numerous papers, for example, among acute mortality studies: (Shumway et
25 al., 1988; Schwartz and Marcus, 1990; Kinney and Ozkaynak, 1991; Ito et al., 1993;
26 Thurston et al., 1995; Kinney et al. 1995; Ito et al., 1995).
27 Time-series models for non-Gaussian data (e.g., Poisson) require additional
28 development. Various methods of accounting for longitudinal correlations have been
29 proposed. One approach depends on filtering using either time or frequency domain methods
30 followed by regressions based on the independence assumption. Another approach builds
31 complicated correlation models for the residuals, and a third combines the foregoing. All
April 1995 12-288 DRAFT-DO NOT QUOTE OR CITE
-------
1 require care to avoid over-modeling that eliminates or reduces the opportunity to find the
2 signal of interest. Random effects models are conceptually straightforward and recent
3 developments make them relatively easy to develop. But, they cannot represent auto-
4 regressive and other models where correlations depend on separation in time. Though one
5 can put in lagged dependent variables to account for auto-regression, unless these are pre-
6 whitened (represented as residuals from predicted), models may be deceptive. In models
7 where defining residuals is a problem (e.g., logistic regression with 0/1 data), pre-whitening
8 is a problem.
9 It is somewhat a matter of choice whether the analyst used covariates to remove
10 spurious correlations or to build a covariance structure. In either approach, care must be
11 taken to avoid inappropriately adjusting-away the effects of interest. For example, if filters
12 are used for short-term patterns, or high-frequency sine and cosine effects are pulled from
13 explanatory and dependent variables, they can remove the PM effect. There is no objective
14 way to determine how far to go with these adjustments, and care is needed to remove
15 confounders but retain power to detect an air pollution effect.
16 Filtering needs to be done on both sides of the equation. It may be better to include
17 moving averages explicitly in covariate adjustments than to use filters. Unlike the acute
18 mortality studies where some studies found that same-day or previous-day concentrations
19 gave the best fit and other studies found that 3-day to 5-day moving averages were more
20 predictive, the hospitalization studies found the strongest association with the current day's
21 PM measurement. Only a few studies found strong effects with lagged measures. These
22 studies found the greatest effect with the previous day's PM measurement, as well as an
23 effect with the current day's PM measurement.
24 The analyst needs to worry about temporal correlation after pulling out fixed effects.
25 Properly done, relatively little residual correlation should remain, but models should allow
26 for the possibility. After filtering one still needs to include correlated error structures.
27 Filtering can induce correlations and, more generally even filtered regressions need to be
28 robust.
29
30
April 1995 12-289 DRAFT-DO NOT QUOTE OR CITE
-------
1 12.6.3.4 Adjustments for Meteorological Variables and Other Confounders
2 There has been relatively little progress in developing a systematic approach to the use
3 of weather-related variables in daily mortality or morbidity studies. A variety of ad hoc
4 procedures have been used. While various statistical methods for adjusting daily mortality or
5 morbidity time series for weather effects appear to be successful on a case-by-case basis,
6 there is little understanding of how to do this systematically in a way that appropriately
7 characterizes current knowledge about the relationship between weather, weather changes,
8 and changes in mortality. The empirical adjustments used in most studies are made with
9 little theoretical basis and may be arguable for that reason alone. It is clear that the effects
10 of some variables, such as temperature, are intrinsically nonlinear, and that it may be more
11 useful to define the likelihood of excess weather-related mortality by the presence of clusters
12 of related meteorological variables, such as the synoptic classes suggested by Kalkstein et al.
13 (1994). While the synoptic class approach appears promising, it has so far been applied to
14 relatively few cities, and may require further modification to be applicable in a general health
15 effects modelling framework. The problem is that meteorological variables are confounded
16 with other pollutants as well as with PM, so that any misspecifications of the relationship
17 between health effects and weather can provide a distorted set of residual effects to be
18 modelled using air pollution variables. A causal or mechanistic model could be useful in
19 relating weather, season, pollutant emissions, pollutant concentrations, behavior as it affects
20 exposure, and health endpoints. Remarkably, weather continues to be significantly related to
21 mortality and other health effects, in spite of increasing use of air conditioning.
22 One interesting possibility in the use of synoptic categories has been demonstrated by
23 Kalkstein et al. (1994). They showed that during the most offensive synoptic weather
24 category, there may be little detectable relationship between PM and excess mortality since
25 most of the excess is attributable to weather. During non-offensive weather categories,
26 however, the excess mortality attributable to PM is readily detected since the weather effect
27 is much smaller and there is a quantitative dose-response relationship between PM and excess
28 mortality, as shown earlier in Figure 12-2 for Philadelphia.
29 Weather/climate control between studies has been discussed by Schwartz (1994a,b),
30 Dockery and Pope (1994) and others as a qualitative issue rather than as a formal numerical
31 evaluation. These papers present global comparisons of RR between cities studied that are
April 1995 12-290 DRAFT-DO NOT QUOTE OR CITE
-------
1 labeled as warm or cold cities, based on longer term mean temperature. Since the actual
2 study analysis looked at day to day changes, long-term comparisons of means may not be as
3 informative or appropriate to examine in such a global manner. First, it is not clear that the
4 classification of a city as a warm or cold climate is correct. This dichotomy does not
5 consider moderate climates in a continuum as a factor, so the comparison may not be
6 appropriate. Second, the mortality in the studies is examined on a daily basis as is the
7 temperature. Mean comparison over several months of temperature is an inappropriate
8 control for the study design. Appendix 12A discusses aspects of these concerns in more
9 detail.
10
11 Infectious Disease (Influenza)
12 Control for confounding with infectious diseases (influenza) by focusing on epidemics,
13 rather than the day to day variation of the infectious disease, may not be the most appropriate
14 model. Pollution may peak in one season and disease epidemics may peak in another season.
15 Statistical adjustment may control for epidemics, but may not adequately account for the day
16 to day variation below the epidemic level. The PM mortality effect has a small RR. The
17 influenza factor which may control for epidemics may be small also. Influenza effects may
18 be modeled more accurately using estimates of weekly or daily variation along with pollution
19 and mortality.
20
21 Confounding: Is It a Real Problem?
22 In developing criteria for assessing epidemiologic studies, we have paid a great deal of
23 attention to the potential confounding of PM effects on human health with the effects of
24 other agents that are associated with PM. Confounding has both conceptual and technical
25 aspects. We will first discuss some of the conceptual aspects.
26 There are three distinct options by which an analyst can deal with confounding in an
27 epidemiology study: (1) control; (2) avoid; or (3) adjust by analysis. It is obviously
28 preferable to control confounding by designing a study in such a way that all of the potential
29 confounding effects are anticipated and avoided. If confounding is unavoidable, then all
30 levels of the nominal causal agent (PM) and its confounding factors should be included in the
31 study, preferably in a balanced design so as to simplify the analyses of the data. Since the
April 1995 12-291 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM studies are all observational studies, study design rarely allows a representative sampling
2 of all levels of all factors. For example, in a city or region where there are large stationary
3 sources that bum fossil fuels containing sulfur,, both PM and SO2 are likely to be high at the
4 same time or low at the same time, being governed by similar patterns of generation and
5 dispersion. Likewise, if mobile sources burning fossil fuel are the primary source of PM in
6 a region, then PM during the summer is likely to be associated with some or all of the
7 following factors: high temperatures, low wind speed, high concentrations of ozone, CO, and
8 airborne nitrates. Therefore, avoiding situations in which confounding occurs is not usually
9 an option.
10 However, there are some situations in which certain kinds of confounding are
11 minimized. One example occurred in the Utah Valley studies. During the year that the mill
12 was closed due to a strike, PM emissions from the mill were greatly reduced, but not quite
13 eliminated since the coke ovens were banked during the closure, and not shut down. The
14 years before and after the closure were years with high PM10 concentrations and typical
15 weather. The year during the closure had generally typical seasonal weather, but much
16 lower PM10 levels. Hence, confounding between PM10 and weather was relatively minimal
17 during the study. Other studies by Pope et al. (1989, 1991) in surrounding counties showed
18 little evidence of any change in the incidence of respiratory infections during the year of
19 closure, so that confounding of winter health effects with epidemics of respiratory infection
20 seems unlikely. Other pollutants were at low levels even when the mill was operating,
21 particularly SO2. Summer levels of ozone were high enough to merit covariate adjustment,
22 but had little effect on the estimated RR for various health effects of PM10.
23 In general, the potential for confounding of PM effects with the effects of other air
24 pollutants is regionally distributed, with sulfates forming a higher percentage of particle mass
25 in areas of the eastern U.S. and Canada, and nitrates a larger percentage than sulfates in the
26 western U.S. and Canada. Thus, the potential for confounding with SO4= and with SO2 is
27 greater in studies in eastern states, and the potential for confounding of PM effects with
28 effects of NOX, and (presumably) with other air pollutants such as CO and O3 that are
29 generated largely by mobile sources, is greater in the western states. Likewise, there is
30 some confounding of health effects of PM with health effects from weather, since weather
31 conditions may affect both generation of PM and its atmospheric dispersion (that is,
April 1995 12-292 DRAFT-DO NOT QUOTE OR CITE
-------
1 concentration). For this reason, it may also be helpful to take a multi-city or multi-study
2 perspective in comparing the effects of potential confounding variables on RR for PM.
3 Schwartz (1994cd, 1995abc) has emphasized a multi-study and multi-endpoint
4 perspective from several points of view. We believe that comparisons of study results across
5 different studies is very useful, but the approach still leaves some unresolved questions about
6 confounding. For example, a completely factorial design controlling for effects of weather
7 and co-pollutants might require finding studies in both "hot" and "cold" cities, in "wet" and
8 "dry" cities, in cities with "high" SO2 and "low" SO2, with "high" O3 and "low" O3. Thus,
9 even a simple factorial design would require comparisons of at least 24 = 16 cities, counties
10 or SMSA's. Since the variables used in describing the cities are numeric, combining the
11 results would be more appropriately done using a "meta-regression" in the same sense as in
12 the cross-sectional analyses done for the long-term exposure studies, rather than a "meta-
13 analysis". Meta-analyses are discussed below. In general, there have not been enough
14 reported studies to do this "meta-regression". There are also problems in defining levels of
15 weather effects, since Kalkstein et al. (1994) have shown that thresholds for excess mortality
16 from high temperatures are different in different cities. That is, a "high" temperature in
17 Minneapolis-St. Paul or Seattle, may not have the same effect in Birmingham or Los
18 Angeles, and that differences may depend on other weather variables and on climate
19 conditions. This approach also shares another concern about population-based cross-sectional
20 studies, that populations in different cities are demographically different in ways that affect
21 population-based health outcomes. Even the measure of effect size that we have used for
22 most of our comparisons, relative risk of health outcome for PM or other factors, is relative
23 to a base rate for the health outcome that one would expect to differ somewhat among
24 different populations in different cities.
25 Avoidance of confounding is also possible for some co-pollutants. Gaseous chemical
26 compounds such as SO2, CO, and O3 are likely to have very similar effects in different
27 conditions, everything else (such as temperature and humidity) being equal. When levels of
28 these pollutants are very low, such as SO2 in most western studies, there is virtually no
29 chance that these pollutants have a causal effect on health endpoints such as mortality and
30 hospital admissions. While such effects cannot be absolutely excluded, the fact that they are
April 1995 12-293 DRAFT-DO NOT QUOTE OR CITE
-------
1 often found at levels very far below the NAAQS should control their contribution to some
2 extent.
3 In spite of these concerns, the general similarity of RR estimates for acute mortality in
4 different studies and the large differences in potential confounding variables among the
5 studies, along with the similarity of RR to that found hi studies where confounding effects
6 seem relatively minimal, adds a great deal of credibility to the conclusion that the PM
7 mortality effects are real, and similar in many locations, even if their magnitude is small and
8 somewhat uncertain. This is not to say that there is no confounding with co-pollutants,
9 particularly where pollutants such as SO2 are generated by the same process that generates
10 PM. Differences in RR for hospital admissions are somewhat greater, possibly reflecting
11 differences in demographic factors or regional differences in hospital admissions criteria, but
12 for similar reasons these estimates are not so seriously confounded in every study as to
13 preclude concluding that, in some studies, there are real increases in hospital admissions
14 rates for the elderly, and for certain classes of respiratory and cardiovascular conditions.
15
16 Control of Confounding By Covariate Adjustment
17 For most of the short-term studies, there is some unavoidable confounding with co-
18 pollutants, with weather, and possibly with other medium-term and long-term events such as
19 epidemics and seasons. Different model specifications of some studies in Section 12.3 were
20 compared at length in Section 12.6.2. Weather variables and temporal variations over tunes
21 longer than a few weeks can be adequately modeled using any of several approaches
22 discussed above, such as polynomials, sinusoids, indicator variables for each month and
23 year, indicators of synoptic climatological categories, nonparametric smoothers or
24 generalized additive models, or high-pass filtering for Gaussian models. Careful examination
25 of residuals for Poisson or Gaussian models have found that a large number of alternative
26 models can provide regression residuals or Poisson expectations apart from air pollution
27 variables that are independent of season, so that seasonal subsetting of time series data in
28 short-term studies may not be necessary for adequately adjusted models. Sometimes, as in
29 analyses of the London mortality series (U.S. EPA, 1986; Schwartz and Marcus, 1990), only
30 seasonal monitoring data are available, but one should not make a virtue of necessity by
31 subsetting time series, since statistical tests to detect PM effects of the magnitude currently
April 1995 12-294 DRAFT-DO NOT QUOTE OR CITE
-------
1 observed in the U.S. require long series of data, roughly at least 800 values. Apart from
2 this sample size requirement, different methods for adjusting for weather and time trends
3 provided adequate levels of adjustment to control for these factors. In addition to controlling
4 confounding with air pollution, it is also important to fit very good models for weather and
5 time trends in time series data, however, so as to help reduce residual variability in daily
6 response data to the limiting or irreducible Poisson minimum variance, which is equal to the
7 expected number on that day.
8
9 12.6.3.5 Adjustments for Co-pollutants
10 Not all studies contain data on the major co-pollutants, and a wide variety of
11 approaches has been used to assess the importance of these co-pollutants as predictors of
12 health effects that compete with PM in terms of explanatory power. Studies in which no
13 other co-pollutant is assessed probably over-estimate the PM effect, but the use of a large
14 number of more or less closely related pollutants to predict the health outcome almost
15 guarantees that the statistical significance and size of the PM effect will be under-estimated.
16 So far, none of these studies have used effective diagnostic techniques or alternative methods
17 for dealing with correlated (e.g. multicollinear) predictor data.
18
19 Co-pollutants
20 Earlier discussion has indicated that other pollutants such as SO2 are factors that play a
21 role in modifying the relationship between PM and mortality when they are incorporated into
22 models examining these relationships such that the RR is usually smaller. Other pollutants
23 such as O3 and CO also need to be considered. Indeed as more studies incorporate these
24 other pollutants into the studies, concern for the role they play becomes more important.
25 This applies to hospitalization studies where possible relationships with CO may be evident.
26 The biological plausibility of CO and sudden death is established. The earlier major air
27 pollution episode events in London involved relatively high levels of CO (Commins and
28 Waller, 1967). Section 12.6.2 conducted an intense examination of the roles of copollutants
29 with a focus on SO2 but also O3 and CO to determine what roles these copollutants play and
30 what summary statements are possible to allow conclusions about PM effects to be stronger.
April 1995 12-295 DRAFT-DO NOT QUOTE OR CITE
-------
1 One of the more difficult problems in interpreting the analyses of the studies discussed
2 here is that of separating the effects of several air pollutants. These pollutants are often
3 fairly highly correlated, and the correlation is often causal, in that several pollutants may be
4 emitted by the same mix of sources in a community, or that one pollutant is a precursor to
5 another pollutant or to a component of that pollutant, such as the fractions of sulfates and
6 nitrates in PM that are secondary pollutants formed from SO^ or NOX. There have been a
7 number of studies in which several different model specifications were tested, involving PM
8 as the only air pollutant, versus PM and other pollutants used jointly in the model. In many
9 studies, such as TSP in Philadelphia (Schwartz and Dockery, 1992a) there was little effect of
10 SO2 on the RR for TSP, whereas other authors have found that SO2 appeared to modify the
11 TSP effect in some seasons, using a similar approach and data set, but with less
12 comprehensive adjustment for weather variables and time trends. There are two ways in
13 multi-pollutant models can cause differences in interpretation from a single-pollutant model:
14 (1) the correlation between PM and the other pollutant(s) is (are) sufficiently high that the
15 effect of health outcome attributable is shared among the pollutants and the individual RR for
16 any one pollutant may be seriously biased. Measurement error in pollutants or other
17 covariates may also bias the result, not necessarily towards the null, and the most poorly
18 measured exposure covariate is usually the one that is driven towards no effect; (2)
19 parameter variance estimates are seriously inflated among the entire group of nearly collinear
20 covariates, increasing estimated standard errors and the width of the confidence intervals for
21 the RR estimates and thereby also attenuating their apparent statistical significance.
22 Collinearity diagnostics have been developed for Gaussian OLS regression models
23 (Belsley et al., 1980) and are implemented in most modern statistical programs. Analogous
24 methods for Gaussian, logistic, or Poisson time series models are less well developed. Most
25 programs allow calculation of the correlation coefficient between estimates of regression
26 parameters (denoted B) based on the asymptotic covariance matrix. However, as noted in
27 Table 13-6, correlation of the B's was given in only two out of ten studies relating acute
28 mortality to PM10. Pollutants with similar patterns and effects can be identified by B-
29 correlation values close to -1. Numeric diagnostics for confounding of co-pollutants could be
30 easily included in reports of long-term studies, many of which use Gaussian OLS linear or
31 nonlinear regression methods for which these diagnostics are readily calculated.
April 1995 12-296 DRAFT-DO NOT QUOTE OR CITE
-------
1 Some investigators have noted that similarity of PM regression coefficients in single-
2 and multiple-pollutant models is sufficient to show that PM is not confounded with the other
3 pollutants. This is not the whole story, since there is a possibility that the B coefficient or
4 RR for PM is unchanged, but the confidence limits are much wider because of the variance
5 inflation of the parameter estimate for collinear pollutants. When the RR estimate for PM is
6 relatively unchanged and there is little increase in the width of the confidence interval, then
7 one can say there is little evidence of confounding. This has been done in a number of
8 analyses discussed in this section, for example in the Utah Valley mortality study as shown
9 in Figure 12-19. The RR estimates for the summer season and the width of the confidence
10 intervals for PM10 are similar without ozone in the model, with daily average ozone, or with
11 maximum daily one-hour ozone as the co-pollutant. The summer PM coefficient, with or
12 without ozone, is similar to the winter value, when ozone levels were so low as to have little
13 probable effect on mortality, which illustrates both covariate adjustment and confounder
14 avoidance strategies in the same study.
15 There is some question about whether the confounding of certain co-pollutants such as
16 PM and SO2 should be regarded as true confounding when one pollutant is part of a causal
17 pathway from pollution source to pollution monitor (Rothman, 1986). Our assessment of
18 probable causal pathways in a hypothetical multivariate model relating source emissions,
19 weather, air pollution, and health outcomes is shown in Figure 12-33. This could serve as a
20 framework for a statistical analysis in which the direct and indirect effects of air pollutants
21 and other factors could be disentangled using substantive scientific hypotheses and data.
22 In summary, confounding by weather and by time effects can be adjusted statistically so
23 as to remove a substantial amount of confounding, but possibly at the expense of reducing
24 the estimated PM effect by attributing it to weather or longer-term time effects not related to
25 short-term PM exposure. Confounding by co-pollutants sometimes cannot be avoided, but
26 should be diagnosed and reported more completely than in most studies now available. In
27 studies where sensitivity analyses demonstrate that including other pollutants in the model
28 causes little change in either the RR estimate for PM or on the width of the confidence
29 interval for the PM effect, one may conclude that the model is not seriously confounded by
30 co-pollutants. Since a number of mortality and morbidity studies have shown that the PM
31 effect on health is not sensitive to other pollutants, we may conclude that the PM effects in
April 1995 12-297 DRAFT-DO NOT QUOTE OR CITE
-------
T - TEMPERATURE
RH - RELATIVE HUMIDITY
BP - BAROMETRIC PRESSURE
WS - WIND SPEED
WEATHER
T, RH, BP, WS
ELECTRIC
POWER
USE
MORTALITY
MOTOR
VEHICLE
USE
AGE
GENDER
RACE
EDUCATION
DAY OF
WEEK
Figure 12-33. A conceptual model of sources and pathways for air pollution health
effects such as mortality, including a causal model of potential
confounding by co-pollutans.
1 these studies are real. This adds some credibility to the claim that a significant PM effect
2 exists in the remaining studies where PM is statistically significant in a model without other
3 pollutants, though similar in magnitude to the PM effect found in other studies with less co-
4 pollutant confounding, but is not statistically significant when other pollutants are included in
5 the model. This then provides a basis for the meta-analyses discussed below.
6
7 12.6.3.6 Ecological Study Design
8 Most of the studies considered are ecologic in design. Even in the daily longitudinal
9 studies, individuals are grouped by region, SMSA, or catchment area for hospital admissions,
10 and all are assumed to have exposure to PM and other covariates characterized by a single
April 1995
12-298 DRAFT-DO NOT QUOTE OR CITE
-------
1 numerical value for the area on that day. The "ecological fallacy" refers to the biases
2 inherent in making individual-level predictions from aggregate-level data. However, such
3 studies are often used because of the availability of data bases for air pollution, weather, and
4 mortality or hospital admissions on a daily basis. Relative risk estimates for individuals
5 should therefore be regarded as subject to much uncertainty, even for age-specific sub-
6 populations, in the absence of subject-specific exposure and covariate data. Recent additions
7 to the NCHS mortality data base, including demographic information such as educational
8 attainment, may allow better resolution of the effects of socio-demographic covariates.
9 While residential location might improve estimates of exposure in communities with several
10 monitoring sites, there would still be considerable uncertainty about individual non-residential
11 exposures in the absence of information about daily activity. Better individual-level exposure
12 information would still be needed to reduce the substantial uncertainties about exposure.
13
14 12.6,3.7 Measurement Error
15 While there has been much discussion about the effects of measurement error,
16 particularly with respect to exposure misclassification, few suggestions have been made as to
17 how to deal with this question.
18 There have been few quantitative assessments of errors in measurements of paniculate
19 matter or other copollutants. There are at least two major components of these errors.
20
21 (1) Instrument error: Errors in measurement of pollutant levels at the point of
22 measurement.
23
24 (2) Proxy error: Error in using levels at a point (even if correctly measured) as the
25 levels to which study population members are exposed.
26
27 For studies of chronic effects, another potentially important problem is sometimes dealt
28 with under the heading of "exposure definition":
29
30 (3) Construct error: Error in using a particular exposure summary other than the
31 biologically relevant exposure (for example, using time-weighted average level
32 when only time above a critical threshold is biologically relevant). This is also
33 encountered in constructing moving averages for short-term studies.
34
35 It is often assumed that any measurement error is nondifferential, and that consequently
36 any bias produced by the error would be towards the null. Neither assumption is necessarily
April 1995 12-299 DRAFT-DO NOT QUOTE OR CITE
-------
1 correct. There are several possible scenarios under which proxy measurement errors will be
2 differential. For example, suppose monitor readings in low-pollution, low-mortality areas
3 tend to understate exposure more than in high-pollution, high mortality areas because many
4 residents of low-mortality areas commute to jobs in high-mortality areas. Then measurement
5 errors will be differentially higher for low-mortality populations (and among noncases in an
6 individual-level study based on these measurements and areas).
7 Contrary to popular treatments, nondifferential error does not guarantee that the
8 resulting bias in effect estimates is towards the null. In ecologic designs, nondifferential
9 error in individual-level exposure measurements can easily produce very large bias away
10 from the null. In individual-level designs, nondifferential error may produce bias away from
11 the null if errors are interdependent or if the dependence of measured on true levels is not
12 monotonic. Interdependence of errors seem likely. For example, wind patterns would
13 induce correlated proxy errors in all atmospheric pollutants. Effects of confounder errors
14 can be in either direction, whether or not the errors are nondifferential. Under the best of
15 circumstances the only predictable effect of nondifferential confounder errors is that they will
16 tend to leave the exposure effect estimates partially confounded.
17 In summary, there has been no evidence presented that measurement errors are
18 nondifferential. Even if there were such evidence, it would not imply that the biases
19 produced by the errors are toward the null. Bias due to measurement error can be profound.
20
21 12.6.4 Assessment Issues for Epidemiology Studies
22 12.6.4.1 Significance of Health Effects/Relevancy
23 The "relative risks" derived from the regression coefficients in recent short-term
24 PM/mortality studies appear to be consistently "small" (i.e., 1.04-1.05 per 100 /*g/m3
25 increase in PM10), compared (at a face value) to the relative risks in other types of studies.
26 In cancer epidemiology, for example, some (Shapiro, 1994) consider a relative risk of 1.7 as
27 weak support, "at most", for a causal inference. However, much lower RR estimates of 1.2
28 to 1.3 have been regarded as sufficient for establishing a presumption of a causal relationship
29 for health effects from environmental pollutants in recent EPA studies on environmental
30 tobacco smoke (U.S. Environmental Protection Agency, 1992) and nitrogen oxides (U.S.
31 Environmental Protection Agency, 1993).
April 1995 12-300 DRAFT-DO NOT QUOTE OR CITE
-------
1 The fact that a relationship is weak, or that an effect is small, does not mean that the
2 relationship is not causal. As Rothman (1986, pp. 17-18) points out, "By 'strength of
3 association', Hill [1965] means the magnitude of the ratio of incidence rates. Hill's
4 argument is essentially that the strong associations are more likely to be causal than weak
5 associations because if they were due to confounding or some other bias, the biasing
6 association would have to be even stronger and would therefore presumably be evident.
7 Weak associations, on the other hand, are more likely to be explained by undetected biases.
8 Nevertheless, the fact that an association is weak does not rule out a causal connection."
9 Many of the studies cited in this chapter included substantial assessments of the effects of
10 potential confounding factors, particularly age group, identifiable cause of death or hospital
11 admission, weather or climate, and the levels of co-pollutants. In some cases, potentially
12 confounding factors were either not present or present at such levels as to have an ignorable
13 effect on the health outcome. Even when potential confounders were present, it was often
14 possible to carry out a statistical adjustment for the confounder, with the PM effect size
15 estimated with and without the potential confounder in the model. The PM effect size
16 estimates and their statistical uncertainty in many studies showed little sensitivity to the
17 adjustment for confounding variables. In a few other studies, there was substantial
18 confounding with some co-pollutants such as SO2 or O3, but estimates of RR for PM without
19 inclusion of the confounders in the statistical concentration-effect model used in these studies
20 were quantitatively similar to RR estimates from other studies where confounding was either
21 avoided or was shown statistically to have little effect. This bears out the comment by
22 Rothman (1986, p. 18) that "... the strength of an association is not a biologically consistent
23 feature, but rather a characteristic that depends on the relative prevalence of other causes,"
24 which here includes confounders such as weather and co-pollutants.
25 However, these two types of relative risks are not directly comparable. The "relative
26 risk" estimates used in these short-term PM exposure studies are not only "acute" in their
27 exposure/response relationship, but also represent "indirect" cause of deaths. A healthy
28 person does not develop respiratory disease and die from an exposure to 100 pig/m3 PM10 in
29 one day. The causal hypothesis is that people with chronic respiratory or cardiovascular
30 diseases, who may be near death from the preexisting conditions, are pushed toward death
31 prematurely by the additional stress on the respiratory system imposed by an increased level
April 1995 12-301 DRAFT-DO NOT QUOTE OR CITE
-------
1 of air pollution. This is in contrast to a cancer risk from exposures to a chemical, through
2 which a perfectly healthy person may develop cancer and die at the age of 50, when the
3 person may otherwise have lived up to 70 years old. This difference may be obvious to the
4 researchers analyzing these data, but needs to be clarified when such "risk estimates" are
5 communicated to people who are not familiar with this field.
6 With this difference taken into consideration, there are several reasons why we may be
7 concerned about the estimated "relative risks":
8
9 • The apparent "relative risk" estimates are often calculated for the entire death
10 categories. Cause-specific "relative risk" estimates are often greater than for total
11 mortality (e.g., in Pope et al.'s Utah study, the excess relative risk calculated for the
12 respiratory category mortality was 43% as opposed to 16% for total mortality). If
13 susceptible populations were defined and categorized, for example by age, the risk
14 estimate would be even higher than for the general population.
15
16 • The apparent "relative risk" tacitly assumes a baseline death population in which all
17 are subject to the change in PM exposures. It is likely that this is not the case.
18 An unknown fraction of the population are not subject to the change in exposure
19 levels of outdoor PM, thereby causing an underestimation of the risk of those
20 actually exposed.
21
22 • There may be a downward bias in the estimated PM/mortality regression coefficients
23 (and, therefore, in the estimated relative risk) due to the PM measurement errors.
24 The extent of this bias is not known.
25
26 • The extent of prematurity of the deaths, which may range from days to years, is not
27 known.
28
29 12.6.4.2 Biological Mechanisms
30 Most of speculation on the biological mechanism of PM mortality effects were made in
31 the earlier major air pollution episodes. According to Firket's report (1936) on the fog
32 episode of Meuse Valley in 1930, the autopsies with microscopical examinations found local
33 and superficial irritation of the mucus membrane of the respiratory ducts and the inhalation
34 of fine particles of soot in the pulmonary alveoli. The chemists concluded that "the SO2 in
35 the presence of oxidation catalysts such as ferric and zinc oxide, must have been partly
36 transformed to sulfuric acid". The discussion of the report suggested sulfuric acid to be "the
37 most probable cause" of deaths. In the 1952 London fog episode (United Kingdom Ministry
38 of Health, 1954), the association of the air pollution and the observed increase in deaths,
April 1995 12-302 DRAFT-DO NOT QUOTE OR CITE
-------
1 estimated to be 4,000 excess deaths, was rather obvious. The report suggested "it is
2 probable that sulphur trioxide dissolved as sulphuric acid in fog droplets, appreciably
3 reinforced the harmful effects of sulphur dioxide." One immediate cause of death was
4 speculated to be acute anoxia from bronchospasm.
5 Such extreme observations do not exist in recent PM/mortality studies. There are some
6 speculations regarding possible mechanisms, identifying specific chemical components
7 responsible for the effects such as acid aerosols. The pattern that does appear to resemble
8 the past episodes in these more recent observational studies is the age and cause specificity of
9 the deaths associated with PM. Both cardiovascular and respiratory deaths hi the elderly
10 population increased in the 1952 London episode. The estimated relative risks for these
11 categories were found to be disproportionately higher and more significant in the analysis of
12 Philadelphia (Schwartz, 1994). Other cause specific analyses (e.g., Fairley, 1990; Pope et
13 al., 1992; Schwartz, 1994b) also reported higher estimated relative risks for respiratory and
14 cardiovascular categories than total or other categories. While the excess deaths in the
15 cardiovascular category, which was also apparent in past episodes, do not provide direct
16 information on possible causal mechanisms, the analysis of contributing causes (Schwartz,
17 1994h) appears to suggest that the respiratory illness is contributing to the deaths of people
18 with cardiovascular conditions. If a person has been suffering from a major cardiovascular
19 disease, that person's death may be still categorized as cardiovascular, even if the respiratory
20 condition causes the death. Such misclassification may also occur for other categories (e.g.,
21 cancer). More analyses using the contributing cause of deaths are needed to further
22 characterize such mechanisms.
23
24 12.6.4.3 Coherence
25 The strength of an association and the consistence of an association can be evaluated by
26 looking across the data base. The repeated observation by different persons, in different
27 places, circumstances and tune, and the consistency with other known facts of an association
28 help demonstrate the strength of that association. This is related to the Bates (1992)
29 discussion on the Question of Coherence. One can look for interrelationships between
30 different health indices to provide a stronger and more consistent synthesis of available
31 information. The various findings that support a picture of coherence would provide a
April 1995 12-303 DRAFT-DO NOT QUOTE OR CITE
-------
1 stronger case with quantitative studies as opposed to qualitative studies. Some studies may
2 be inappropriate to use in such a discussion, and the quality of each study should be
3 considered. Bates (1992) states that the difficulty with discussing any index of internal
4 coherence is that this requires a series of judgements on the reliability of the individual
5 findings and observations. The outcome of a coherence discussion then is a qualitative
6 presentation in the end, not quantitative. Thus, coherence cannot be formally measured.
7 Bates (1992) noted that the strength of different health indices are important as are
8 difficulties in assessing exposure. Bates (1992) suggests three areas to look for coherence:
9 (1) within epidemiological data, (2) between epidemiological and animal toxicological data,
10 and (3) between epidemiological, controlled human and animal data. Only the first is
11 considered in this chapter. The other two are more appropriately discussed in the synthesis
12 chapter (Chapter 13).
13 Coherence by its nature considers biological relationships of exposure to health
14 outcome. In looking for coherence one should compare outcomes that look at similar time
15 frames—for example, daily hospitalizations compared to daily mortality rather than monthly
16 hospitalizations. Underlying mechanisms related to different effects should also be
17 considered. Biologic mechanisms underlying a small acute PFT reduction in children in
18 relation to a non-episodic PM exposure, for example, are unlikely to explain the acute basis
19 for a change in the mortality rate of the sub-population group of individuals over 65 years of
20 age in relation to a similiar PM exposure. Also, evidence for coherence is strongest when
21 reported increases in morbidity endpoints are obtained for the same population segment (e.g.,
22 the elderly) in the same geographic locale as where PM-mortality effects were demonstrated.
23 Most of the morbidity endpoints measured in available PM studies generally do not speak
24 directly to coherence with a change in the mortality rate.
25 The various PFT and respiratory symptoms and disease endpoints studied do indicate
26 that PM has a relationship with a continuum of health outcomes. However, the underlying
27 mechanisms between the health outcomes studied may be different. The principal health
28 outcome for which coherence is desirable is mortality, the death rate in a population. Of the
29 various morbidity outcomes studied and discussed earlier in this chapter, the outcome
30 potentially most related to mortality is hospitalization for respiratory or cardiovascular causes
31 hi the older age group (i.e., > 65 years old). In a qualitative sense, the hospitalization
April 1995 12-304 DRAFT-DO NOT QUOTE OR CITE
-------
1 studies reviewed in the chapter support this notion. The mortality studies suggest that these
2 specific causes provide stronger relationships (i.e., RR) than total mortality in that age
3 group. While a much larger increase in hospitalizations than deaths may have been
4 expected, this is not evident in the data. The population of decedents may be fundamentally
5 different from the underlying population of hospital users (Bates, 1992). The relationship
6 between health outcomes may be more complex than a simple approach would suggest.
7
8 12.6.5 Meta-analyses and Other Methods for Synthesis of Studies
9 12.6.5.1 Background
10 Several reports have appeared in which results from different studies have been
11 combined, formally or informally, to present an overall effect size estimate for acute health
12 effects. For example, a synthesis of daily mortality studies for seven cities was published by
13 Schwartz (1992). The seven cities included four TSP studies (Steubenville, Philadelphia,
14 Detroit, Minneapolis) and three PM10 studies (St. Louis, Eastern Tennessee, Utah Valley).
15 The daily mortality studies were further analyzed by Schwartz in a later paper (1994), which
16 added studies from New York, from Birmingham, Alabama, later London studies (1959 to
17 1972), and a study in Athens, Greece. The RR estimates were combined in formal
18 quantitative meta-analyses, using either unweighted RR estimates, or using a smaller set of
19 estimates weighted by inverse of the estimation variance of the RR coefficient from studies in
20 which the standard error was reported. Several methods were used, and several subsets of
21 the data were tested according as to whether or not the study city was "warm" or the TSP
22 coefficient was adjusted for copollutants.
23 A recent paper by Dockery and Pope (1994) extends the research synthesis to a variety
24 of health outcomes, including hospital admissions studies and respiratory function tests. This
25 paper is also based on conversion of different PM measures to an equivalent PM by applying
26 a scaling factor: 1.0 for PM15 and BS, 0.55 for TSP, 4 for sulfates (SO4), 1/0.60 for PM2.5,
27 and 1/0.55 for COH. This synthesis paper uses eight cities for total mortality, four cities for
28 respiratory mortality and for cardiovascular mortality, three cities for hospital admissions for
29 respiratory symptoms, four studies for asthma admissions, and combines three cities with
30 different reasons for emergency room visits. The paper examines the effects of PM on
31 exacerbation of asthma by combining results of two cities for bronchodilator use, and
April 1995 12-305 DRAFT-DO NOT QUOTE OR CITE
-------
1 combining three studies for asthmatic attacks. Pulmonary function tests are synthesized from
2 four studies for Forced Expired Volume (FEVj and FEV0 75), and six studies for Peak
3 Expiratory Flow (PEF daily, weekly, or longer). Respiratory symptom results are divided
4 into combining six studies reporting lower respiratory symptom results, upper respiratory
5 symptom results, and six studies reporting cough symptom results. The authors conclude
6 that these results demonstrate a coherence of effects across a range of related health
7 outcomes, and a consistency of effects across independent studies by different investigators in
8 different settings.
9 The synthesis of the epidemiologic evidence in this document presents some unusual
10 problems. Many of the studies showing mortality and morbidity effects are based on
11 relatively small increases estimated with great precision resulting from sophisticated analyses
12 of long series of frequent events. As a result, relative risks (or odds ratios) of 1.06 are
13 common and often statistically significant. A value of 1.06 would indicate that mortality (or
14 morbidity) is increased by 6% when PM10 is increased a specified amount (usually
15 50 ftg/m3). Traditionally, relative risks less than 1.5 were considered to be of questionable
16 biological meaning. Although relative risks near 1.06 are not large in magnitude, they may
17 represent a large net effect because the events are so common. The question remains: are
18 these effects real or are they an artifact of the analysis?
19 A careful review of the analysis techniques in Section 12.6.3 suggests that similar
20 results are obtained as long as similar covariates and independent variables are included in
21 the analysis. There are remaining questions about the accuracy of the variances and the
22 assumptions upon which they are based. Even allowing for these problems, the estimated
23 regression coefficients are consistently estimating the correct quantities although the exact
24 p-values may be slightly in error.
25 The results do not appear to depend heavily on the form in which covariates were
26 included in the model. Analyses that included the known covariates such as temperature and
27 season usually gave similar results. The one factor which appeared to make a consistent
28 difference was the inclusion of one or more copollutant(s) in addition to particulate matter.
29 The inclusion of SO2 tend to reduce the effect of particulate matter in most analyses, while
30 O3 generally had less of an impact on PM regression coefficients. This would be expected
April 1995 12-306 DRAFT-DO NOT QUOTE OR CITE
-------
1 because O3 tends to be less correlated with PM than does SO2. Although the PM coefficients
2 were reduced by the inclusion of SO2, most remained statistically significant.
3 One unresolved question is the possibility that the effects seen were the result of some
4 covariate which, had it been included, would have reduced the PM coefficients to a non-
5 significant level. Although this is always a concern with epidemiologic studies, the concern
6 is often dismissed as improbable when the relative risks are large as 1.5 or 2.0. When the
7 relative risks are less than 1.1, the question is of greater concern. There have been serious
8 attempts by several authors to find such covariates, but none have so far been discovered.
9
10 12.6.5.2 Meta-analyses Using Studies Reviewed in This Document
11 In order to compare the results of the various studies relating acute exposure to PM to
12 excess mortality, we selected studies that satisfied certain criteria: (1) the study has been
13 published or is in press; (2) the study used PM10 or TSP as an index of paniculate matter
14 exposure; and (3) the study included adequate adjustments for seasonality, weather, other
15 effects. The first criterion was imposed to provide adequate access to a description of study
16 data, methods, and results; and the second so as to restrict consideration to studies with
17 pollutants for which EPA has extensive air monitoring data. Even here, analyses were
18 performed separately for PM10 studies and for TSP studies, so as to avoid having to make
19 any assumptions about site-specific calibrations of one PM concentration or index into
20 another. It may be possible to extend the meta-analyses to a wider range of studies when
21 methods are developed for assessing the uncertainty associated with generic versus city-
22 specific calibrations of one PM index to another.
23 The results of the analyses have been standardized for purposes of comparison. All of
24 the acute exposure studies used Poisson or equivalent regression methods with the expected
25 mortality an exponential function of a linear combination of predictors, or with the logarithm
26 of the mortality rate as a linear combination of predictors including the PM index. This
27 means that the relative risk (RR) - the fractional increase in the mortality rate relative to a
28 baseline value without pollution, everything else being equal - can be expressed in terms of
29 changes per unit of pollution. The base unit for change in risk was chosen differently for
30 each pollutant. For PM10 studies, the effect was the odds ratio for mortality corresponding
31 to an increase of 50 jig/m3 in PM10. Other ranges have been used in published papers, most
April 1995 12-307 DRAFT-DO NOT QUOTE OR CITE
-------
1 commonly 10 or 100 /-ig/m3. We selected 50 pig/m3 because it is closer to the range of
2 values in various morbidity studies, whereas the range in mortality studies usually is larger
3 than 100 /ig/m3. Since the range of values in TSP studies is typically much larger than in
4 PM10 studies, we used 100 jig/m3 as the base unit for TSP studies of mortality.
5 The basic data on effects size estimates, in appropriate units, are shown earlier in
6 Tables 12-3 and 12-5. Note that the confidence intervals derived in the various papers are
7 not always symmetric about the estimated RR. The data could be naturally sorted into five
8 distinct groups:
9
10 • Estimates of PM10 effect on RR, not adjusted for copollutants, lags <2 d (4 studies,
11 4 cities)
12 • Estimates of PM10 effect on RR, not adjusted for copollutants, lags >2 d (6 studies,
13 6 cities)
14 • Estimates of TSP effect on RR, not adjusted for copollutants (4 studies, 3 cities)
15 • Estimates of PM10 effect on RR, adjusted for copollutants (3 studies, 3 cities)
16 • Estimates of TSP effect on RR, adjusted for SO2 (3 studies, 2 cities)
17 • Estimates of PM10 effects on RR short, long lags (3 studies, 3 cities)
18
19 There are presently no methods for using results of different analyses of the same data set,
20 such as the two studies on Steubenville (Schwartz and Dockery, 1992; Moolgavkar et al.,
21 1995a). (For this assessment, we report results using each separately.)
22 The meta-analysis methods were similar to those used in the Nitrogen Oxides criteria
23 document (US EPA 1994; Hasselblad et al. 1994). Differences among studies are regarded
24 as random effects. Results are shown in Figures 12-34 through 12-39 and Table 12-25. The
25 relative risk for PM10 exposure averaged <2 days is estimated as 1.032 per 50 /ig/m3 PM10,
26 with a 95% confidence interval of 1.025 to 1.038 per 50 />ig/m3 PM10. There is overall
27 evidence of an effect, even though one of the four studies in Figure 12-34 is not significant.
28 The relative risk for PM10 exposure with longer averaging times, 3 to 5 d, is estimated as
29 1.064 with 95% confidence interval of 1.047 to 1.082. In Figure 12-35, one study is
30 negative and another marginally significant. The combined estimate for TSP effect in
31 Figure 12-36 depends on which study is used for the Steubenville estimate; with the Schwartz
April 1995 12-308 DRAFT-DO NOT QUOTE OR CITE
-------
Touloumi Athens
Ozkaynak Toronto
Kinney Los Angeles
Ostro Santiago
Combined
ns
rto
es
go
ed
W 1
1 • 1
0.99 1.01 1.03 1.05
Relative Risk per 50 ng/m3 PM,0
I Lower 95% CL • Relative Risk I Upper 95% CL
Figure 12-34. Summary of studies used in a combined EPA meta-analysis of PM10
effect on mortality with short averaging tunes (0-1 days), and co-
pollutants in model.
Dockery St. Louis
Dockery E. Tenn.
Pope Utah
Schwartz Birmh.
Ostro Santiago
Styer Chicago
Combined
, ,.,_... ,+ ,, i
h*H
.9 1.0 1.1 1.2 1.3
Relative Risk per 50 iig/m3 PM^
1 Lower 95% CL * Relative Risk I Upper 95% CL
Figure 12-35. Summary of studies used in a combined EPA meta-analysis of PM10
effects on mortality with longer averaging tunes (3-5 d), and no co-
pollutants in model.
April 1995
12-309 DRAFT-DO NOT QUOTE OR CITE
-------
Schwartz Cinti
Schwartz Phila
Schwartz Steub
Moolg. Steub
Combined (Schwartz)
Combined (Moolgav.)
i-
1.0 1.02 1.04 1.06 1.08 1.1
Relative Risk per 100 ng/m3 TSP
I Lower 95% CL • Relative Risk I Upper 95% CL
Figure 12-36. Summary of studies used in a combined EPA meta-analysis of TSP
effects on mortality, with no co-pollutants in model.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
study, the effect is 1.051 per 100 jug/m3 TSP, whereas with the Moolgavkar study, the
estimate is 1.050, but is less certain. However, none of these studies included SO2, the most
probable confounding co-pollutant. The analogous estimates for a TSP effect with
copollutants in the model is less significant across three studies, RR = 1.018 with a 95%
confidence interval from 1.007 to 1.029 per 50 /ig/m3 PM10, as shown in Figure 12-37.
Also, Figure 12-38 shows that when SO2 is included in the model, estimated PM10 effects
still remain significant. The overall EPA meta-analyses results are summarized in
Table 12-25 and Figure 12-39.
We conclude that there is a short-term increase in mortality in response to acute PM
exposures. This appears to be at least partly confounded with other pollutants, especially
SO2, but even with SO2 included in the model the effect is on the order of 1 to 5% increase
in relative risk per 100 /ig/m3 TSP. This is probably a minimum estimate of effect size.
If SO2 is in fact a proxy for fine particle exposure through the SO2 to sulfate to fine particle
pathway, then adjusting for SO2 may overcontrol the estimate of PM effect, which could be
as large as 1 to 5% per 100 /ug/m3 TSP, or 2 to 6% per 50 /ig/m3 PM10. This also
April 1995
12-310 DRAFT-DO NOT QUOTE OR CITE
-------
Schwartz Phila
Schwartz Steub
Moolg. Steub
Combined (Schwartz)
Combined (Moolgav.)
0.98 1.0 1.02 1.04 1.06 1.08 1.1
Relative Risk per 100 ng/m3 TSP
I Lower 95% CL • Relative Risk I Upper 95% CL
Figure 12-37. Summary of studies used in combined EPA meta-analysis of TSP
effects on mortality, with SO2 in the model.
Touloumi Athens
Kinney Los Angeles
Ito Chicago
Combined
Figure 12-38.
IS
>s
o
d
1 1
1 * •
1 • 1
0.99 1.00 1.01 1.02 1.03 1.04 1.(
Relative Risk per 50 |ig/m3 PM10
1 Lower 95% CL • Relative Risk I Upper 95% CL
Summary of studies used in a combined EPA meta-analysis of PM10,
effects on mortality, with other pollutants in the model.
April 1995
12-311 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12-25. COMBINED ESTIMATES OF RELATIVE RISK OF INCREASED
MORTALITY FROM ACUTE EXPOSURE TO AIR POLLUTANTS
Pollutant
Increment
Model
Averaging
Time
Relative Risk
Estimate Per
Increment
95 Percent
Confidence
Limits
PMio
PM10
TSP
TSP
PM10
TSP
TSP
50 /ig/m3
50 pig/m3
100 /zg/rn3
100 fjLg/m3
50 /ig/m3
100 /ig/m3
100 /ig/m3
No copollutant
No copollutant
NoSO2
NoSO2
+copollutants
+SO2
+SO2
0-1 days
3-5 days
1.031
1.064
1.051
1.050
1.018
1.038
1.030
1.025to 1.038
1.047 to 1.082
1.035 to 1.067
1.029 to 1.072
1.007 to 1.029
1.016 to 1.059
1.008 to 1.053
Deluding Schwartz Steubenville study.
2Including Moolgavkar Steubenville study.
1 depends on PM10 averaging times, with a 3% increase for averages of current and preceding
2 day PM10 and 6% effect for 3 to 5 day moving averages.
3 These analyses suggest that there is an identifiable effect of PM exposure on increases
4 in acute mortality, even when characterized by TSP, a relatively insensitive index of
5 respirable particle concentration. The role of SO2 as a possible proxy for fine particle
6 exposure remains to be clarified. It is also not possible to overlook the potential confounding
7 effects of other pollutants such as O3 and NO2.
8
9 12.6.5.3 Discussion
10 In general, there appears to be a range of acute health responses to air pollution
11 exposure as characterized by some PM indicator. Dockery and Pope (1992) have stated that
12 "It is ... presumptuous to assign these adverse health effects solely to the mass concentration
13 of particulates. ... Many health effects of particles are thought to reflect the combined action
14 of the diverse components of the pollutant mix." Since pollutant mixes and exposed
15 populations differ from one location to another, it is more probable that there are real
16 differences among different studies.
17 Several approaches to estimating a combined PM effect as a weighted average of study-
18 specific effects may be considered: (1) regard each effect size estimate as a measurement in
April 1995
12-312 DRAFT-DO NOT QUOTE OR CITE
-------
PM10only, <2d lag
PM10only, >2d lag
PM10 with other pol.
1.0 1.02 1.04 1.06 1.08
Relative Risk per 50 ug/m3 PM10
I Lower 95% CL • Relative Risk I Upper 95% CL
Figure 12-39. Summary of PM10 effects on mortality.
1 an ecological study and adjust for differences in effect size among cities as a function of
2 differences in climate, mixture of other air pollutants, and differences in demographic
3 characteristics; (2) carry out multiple comparisons of effect size estimates and group together
4 those estimates that are not significantly different; (3) perform combined analyses in which
5 the PM effect size parameter(s) are constrained to be equal in different data sets.
6 With the first approach (1), it may be possible to model the differences in PM effect
7 size estimates by multiple regression on known quantitative differences in climate,
8 copollutant mix, and population. This would require a "meta-regression" in which some
9 assumptions would need to be made about the relationship between PM effect size and the
10 inter-study variables that distinguish different cities, adding yet another layer of uncertainty
11 about model specification. It would not be feasible to carry out this analysis unless there
12 were a large enough number of studies, since multiple linear regression models do not
13 perform well unless there are several times as many data values (effect size estimates from
14 different studies) as there are variables that are used for adjustment.
15 With approach (2), each effect size estimate for which there was an attached standard
16 error estimate would be compared with each other effect size estimate, as if each effect size
April 1995
12-313 DRAFT-DO NOT QUOTE OR CITE
-------
1 estimate was a separate group mean in an analysis of variance. The effect size estimates
2 would then be grouped into clusters in which the cluster members (studies) were not
3 statistically different from each other, although some methods allow for the possibility of
4 partially overlapping clusters. A variety of multiple comparison procedures are available,
5 using either methods based on normally distributed data or more robust methods (e.g.,
6 Tamhane and Hochberg, 1987). Some comparisons of a multiple hypothesis testing approach
7 with a metaanalysis approach are described by Westfall and Young (1993), who prefer
8 computer-intensive resampling methods such as bootstrap estimation or permutation testing
9 that may not be feasible unless raw data were available. Conventional multiple testing
10 methods can be done without raw data when standard error estimates are available, and may
11 be especially suitable when there are only a few effect size estimates.
12 With alternative approach (3), it is essential that raw data be available. It is unlikely
13 that raw data for all studies of any specified health outcome could be assembled within a
14 short period of time, and even then it would likely take months to conduct such an analysis
15 adequately.
16 The formal metaanalytic methods used to combine effect size estimates for acute
17 mortality (Schwartz, 1994c) or for a variety of health outcomes (Dockery and Pope, 1994)
18 could possibly be improved by including more information when weighing the studies, as we
19 have suggested above. There are still many unresolved questions about how the synthesis of
20 PM health effects data from different studies should be carried out.
21
22
23 12.7 SUMMARY AND CONCLUSIONS
24 The time-series mortality studies reviewed in this and past PM criteria documents
25 provide strong evidence that ambient air pollution can cause increases in daily human
26 mortality. Recent studies provide confirmation that such effects occur at routine ambient
27 levels, extending to 24 h concentrations below 150 /xg/m3 (the level of the present U.S. air
28 quality standards). Furthermore, these new PM studies are consistent with the hypothesis
29 that PM is a causal agent in the mortality impacts of air pollution. One of the more
30 important findings is that longer averaging times (3 to 5 day moving averages) predicted
31 larger and more significant effects on total, respiratory, or cardiovascular mortality in many
April 1995 12-314 DRAFT-DO NOT QUOTE OR CITE
-------
1 studies than did PM concentrations on the present or preceding day. Overall, the PM10
2 relative risk estimates derived from the most recent PM10 total mortality studies suggest a
3 24-h average 50 pg/m3 PM10 acute exposure effect of the order of RR = 1.025 to 1.05 in
4 the general population, with higher relative risks indicated for the elderly sub-population and
5 for those with pre-existing respiratory conditions, both of which represent sub-populations
6 especially at risk to the mortality implications of acute exposures to air pollution, including
7 PM. Results are very similar over a range of specifications of statistical models used in the
8 analyses, and are not artifacts of the methods by which the date were analyzed.
9 There is an indication among these various analyses that children may be more
10 susceptible to the mortality effects of air pollution exposure than the population in general,
11 but it is difficult, given the limited and somewhat conflicting results available at this time, to
12 ascribe any such association to PM pollution in particular. This is an area where further
13 research is clearly needed to broaden the base upon which to assess the potential for PM to
14 increase mortality among children.
15 Hospitalization data can provide a measure of the morbidity status of a community
16 during a specified time frame. Hospitalization data specific for respiratory illness diagnosis
17 or more specifically for COPD and pneumonia give a measure of the respiratory status.
18 Such studies provide an outcome measure that relates to mortality studies for total and
19 specified respiratory measures. Tables 12-9 through 12-12 summarize these studies. These
20 studies associate hospitalization data with measure of PM. Some of the same factors and
21 concerns related to the mortality studies are an issue for these studies also.
22 Both COPD and Pneumonia hospitalization studies show moderate but statistically
23 significant relative risks in the range of 1.06 to 1.25 resulting from an increase of 50 /ig/m3
24 in PM10 or its equivalent. There is a possible suggestion of a relationship to heart disease,
25 but the evidence is very inconclusive. The admission studies of respiratory disease show a
26 similar effect. The hospitalization studies in general use very similar analysis methodologies
27 and the majority of the papers are written by a single author. Overall, these studies are
28 indicative of health outcome related to PM. They are also supportive of the mortality
29 studies, especially with the more specific diagnosis relationships.
30 Schwartz (1995c) reviewed the hospital admission and mortality studies of paniculate
31 matter and ozone. The hospitalization results were based on the studies of Thurston et al,
April 1995 12-315 DRAFT-DO NOT QUOTE OR CITE
-------
1 (1992), Schwartz, 1994a, Burnett et al, 1994, Schwartz et al, 1995, Schwartz, 1994b, Sunyer
2 et al, 1993, Schwartz, 1994c, and Burnett et al, 1995. Summary tables for all respiratory
3 admissions showed relative risks ranging from 1.10 to 1.20 per 100 jug/m3 PM10 (or
4 equivalently, 1.05 to 1.10 per 50 /ig/m3 PM10). Summary tables for COPD admissions
5 showed relative risks ranging from 1.15 to 1.57 per 100 /ig/m3 PM10 (or equivalently, 1.07
6 to 1.25 per 50 /*g/m3 PM10). Schwartz (1995c) argues that because there is no significant
7 heterogeneity in the relative risks across studies that
8
9 "This suggests that confounding by other pollutants or weather is not the source of
10 these associations, since the coincident weather patterns and levels of other
11 pollutants varied greatly across the studies. In particular, studies in the western
12 United States (Spokane, Tacoma) had very low levels of sulfur dioxide, and much
13 less humidity than [sic] in the eastern United States locations."
14
15 However, tests for homogeneity are known to have very little power against specific
16 alternatives, and so this conclusion may not be appropriate. Analyses of individual studies
17 have identified situations in which the effects of confounding are minimal as discussed in
18 Section 12.6.2.
19 The hospitalization studies usually compared daily fluctuations in admissions about a
20 long term (e.g., 19 day) moving average. These fluctuations were regressed on PM
21 estimates that were for the time period immediately preceding or concurrent with the
22 admissions. Some authors considered lags up to 5 days, but the best predictor for
23 hospitalization usually was the most recent exposure. Some morbidity outcomes associated
24 with hospitalization may appropriately be associated with concurrent admission, while others
25 may require several days for development to end in an admission. The exposure-response
26 lag periods is not yet well examined.
27 Acute respiratory illness include several different endpoints, but the majority of authors
28 reported results on at least two of (1) upper respiratory illness, (2) lower respiratory illness,
29 or (3) cough (See Table 12-13). The results for upper respiratory illness are very
30 inconsistent: two studies estimate a relative risk near 1.00 whereas four others obtain
31 estimates between 1.14 and 1.55. These relative risks are all estimated for an increase of 50
April 1995 12-316 DRAFT-DO NOT QUOTE OR CITE
-------
in PM10 or its equivalent. The relative risks for lower respiratory illness are spread
2 between 1.01 and 2.03, but all are positive. The relative risks for cough include two below
3 1.0 and go as high as 1.51. All of these are suggestive of an effect. Whereas the hospital
4 admission studies were all done in a similar manner and resulted in very similar results,
5 these studies are done with very different designs and give very inconsistent results.
6 The acute pulmonary function studies are suggestive of a short term effect resulting
7 from particulate pollution. Peak flow rates in children show decreases in the range of 30 to
8 40 ml/sec resulting from an increase of 50 /*g/m3 in PM10 or its equivalent. The results
9 appear to be larger in symptomatic groups such as asthmatics. The effects are seen across a
10 variety of study designs, authors, and analysis methodologies. Effects using FEVj or FVC
11 as endpoints are less consistent (See Table 12-14).
12 Long-term exposure to air pollution was studied by use of cross-sectional studies,
13 comparing rates of mortality or morbidity at a point in time against differences in annual
14 average pollutant concentrations. Most older studies were population-based cross-section
15 studies. These studies used outcome rates for entire cities or SMSA's. Several recent
16 prospective cohort-based cross-sectional studies allow use of subject-specific information
17 about other health risk factors, such as cigarette smoking or occupational exposure; and
18 subject-specific outcome measures. The relative risk shows some sensitivity to model
19 specification. For models of 1980 mortality from all natural causes, the RR from separate
20 OLS regression models using TSP, PM15, PM2 5 or SO4 as PM indicators all showed a
21 positive test but statistically, non-significant effect. The PM15 RR is 1.036 at PM15 = 50
22 /ig/m3 (95% confidence interval 0.98 to 1.10), whereas a log-linear model for the same 62
23 SMS A found a larger and statistically significant RR for TSP of 1.066 (95% confidence
24 interval 1.006 to 1.13 at TSP = 100 j«g/m3). The relative risk of major cardiovascular
25 disease (CVD) for sulfate inhaled particles was 1.19 at SO4 = 15 Mg/m3 (interval 1.03 to
26 1.35) when adjusted for one set of demographic coraviates, but smaller and not significant
27 after adjustment with a larger set of covariates. The relative risk of COPD for TSP at TSP
28 = 100 /ig/m3 or for non-sulfur TSP was highly significant, 1.50 and 1.43 with confidence
29 intervals (1.22, 1.83) and (1.20, 1.71), respectively.
30 Although most of these studies covered the entire U.S. using the basic paradigm of
31 Lave and Seskin (1970), there are major differences in the numbers of independent variables
April 1995 12-317 DRAFT-DO NOT QUOTE OR CITE
-------
1 considered, including the air pollutants. Most of the studies found pollutant elasticities (i.e.,
2 mean effects) of 0.02 to 0.08, although the specific pollutants associated with mortality
3 varied. However, all of these studies found at least some association between air pollution
4 and mortality on an annual average basis. There was a slight suggestion that elasticities may
5 be decreasing over time (1960 to 1980). It was not possible to determine whether the
6 mortality associations were stronger for pollution measured the same year or in previous
7 years. The finding of significant associations with metals (Lipfert, 1978, 1984, 1993) and
8 with cement plant particles (Bobak and Leon, 1992) suggests that many different types of
9 particles may be involved. Analyses by age and cause of death were limited; the most
10 consistent associations were for the elderly, especially ages 75 + , and for respiratory disease
11 mortality and TSP (but not SO42').
12 Table 12-19 summarizes the three newer prospective studies considered here. The
13 California and Six-City studies suffer from small sample sizes and inadequate degrees of
14 freedom, which partially offsets the specificity gained by considering individuals instead of
15 population groups. All of them may have neglected some important risk factors. The two
16 early studies not shown in this table were largely inconclusive and the studies of California
17 nonsmokers by Abbey et al. (1991, 1994) that had the best cumulative exposure estimates
18 found no significant mortality effects of previous air pollution exposure. The Six Cities and
19 ACS studies agree in their findings of strong associations between fine particles and excess
20 mortality, but it is unfortunate that the ACS study did not consider a wider range of
21 pollutants so as to also confirm the lack of importance for other pollutants. In addition, the
22 timing of the critical exposures remains an open question. It is also important that a range of
23 pollutants be considered in both chronic and acute studies, since it is possible that acute
24 effects may be exhibited by one pollutant and chronic effects by another.
25 The RR estimates for total mortality are large and highly significant in the Six-Cities
26 study. With their 95 percent confidence intervals, the RR for 50 ^ig/m3 PM15 is 1.42 (1.16,
27 2.01), the RR for 25 /*g/m3 PM25 is 1.31 (1.11, 1.68), and the RR for 15 /xg/m3 SO4 is
28 1.46 (1.16, 2.16). The estimates for total mortality in the ACS study are much smaller, but
29 also much more precise, 1.17 for 25 /xg/m3 PM25 (RR 1.09, 1.26), and 1.10 for 15 /zg/m3
30 SO4 (RR 1.06, 1.16). Both studies used Cox regression models and were adjusted for rather
31 similar sets of individual covariates. In each case, however, caution must be applied in use
April 1995 12-318 DRAFT-DO NOT QUOTE OR CITE
-------
1 of the stated quantitative risk estimates, given that the life-long cumulative exposures of the
2 study cohorts (especially in the dirtiest cities) included much higher PM exposures than those
3 indexed by the more current PM measurements used to estimate the chronic PM exposures of
4 the study cohorts. Thus, lower risk estimates very likely apply.
5 Cross-sectional studies may find a significant association between mortality and a
6 specific air pollutant for any of several reasons:
7 • The association may reflect a non-zero integral of the acute effects of that
8 pollutant over the period of study.
9
10 • The association may reflect a chronic effect from previous long-term exposures.
11
12 • The association may have resulted from confounding, either with another
13 pollutant, with the characteristics of the sources that produced that pollutant
14 (occupational hazards or exposures), or with human elements spatially associated
15 with pollution sources such as differential migration of the healthy, less desirable
16 housing near sources, or other socioeconomic factors.
17
18 The studies reviewed above probably all reflect some varying combinations of these
19 possibilities. It is possible that a given regression coefficient may reflect the acute effects of
20 one pollutant and the collinear portion of the chronic effects of another, for example.
21 Convincing evidence of causal associations requires demonstration of specific disease-
22 pollution combinations that are physiologically plausible. Supporting evidence may be
23 obtained by showing that the statistical relationship improves when more reliable exposure
24 data are used in the analysis.
25 Some of the prospective studies demonstrated that including additional pollutant
26 exposures in a statistical model (cigarette smoke, occupational exposure) not reflected in the
27 outdoor measurements lead to a stronger statistical mortality relationship with the outdoor
28 measurements. This suggests two possibilities (there may be others):
29 • The indoor and outdoor exposures may reinforce each other and thus must have
30 similar physiological effects. This may provide some clues as to the most likely
31 of several collinear outdoor pollutants. The responses could be either chronic or
32 acute.
33
34 • The indoor or occupational exposures may have created a disease state
35 (independent of the outdoor exposures) that makes the individual more
36 susceptible to outdoor pollution effects. This hypothesis suggests that the
37 outdoor effects in the prospective studies are acute, since it is unlikely that
April 1995 12-319 DRAFT-DO NOT QUOTE OR CITE
-------
1 normally healthy people will experience acute effects at the air pollution levels
2 now seen in U.S. cities.
3
4 Distinguishing between these two scenarios will likely require additional research, probably
5 including temporal studies of long-term changes in air quality hi different places.
6 At this tune, the long-term studies provide support for the existence of short-term
7 increase hi mortality which are not subsequently canceled by decreases below normal rates.
8 However, they do not exclude the existence of chronic effects. They provide no convincing
9 evidence as to the specific pollutant(s) involved, and they do not rule out the existence of
10 pollutant thresholds.
11 Three studies based on a similar type of questionnaire but done by two different groups
12 of researchers, provide data on chronic respiratory disease and PM. All three studies suggest
13 a chronic effect of paniculate matter on respiratory disease, but the studies suffer from the
14 usual difficulty of cross sectional studies. The effect of paniculate matter is based on
15 variations in exposure which are determined by the different number of locations. In the
16 first two studies there were six locations and in the second there were four. The results seen
17 were consistent with a paniculate matter gradient, but it is impossible to separate out the
18 effect of paniculate matter and any other factors or pollutants which have the same gradient
19 (See Table 12-21).
20 The chronic pulmonary function studies are less numerous than the acute studies
21 (Table 12-22). The one study with good monitoring showed no effect from paniculate
22 pollution. Cross sectional studies require very large sample sizes to detect differences
23 because the studies cannot eliminate person to person variation which is much larger than the
24 within person variation. Thus the lack of statistical significance cannot be taken as proof of
25 no effect.
26 Historical and present-day evidence suggest that there can be both acute and chronic
27 effects by strongly acidic PM on human health. Evidence from historical pollution for
28 episodes, notably the London Fog episodes of the 1950's and early 1960's, indicate that
29 extremely elevated daily acid aerosol concentrations (on the order of 400 /xg/m3 as H2SO4, or
30 roughly 8,000 nmoles/m3 H+) may be associated with excess acute human mortality when
31 present as a co-pollutant with elevated concentrations of PM and SC^. In addition, Thurston
April 1995 12-320 DRAFT-DO NOT QUOTE OR CITE
-------
1 et al. (1989) and Ito et al. (1993) both found significant associations between acid aerosols
2 and mortality in London during non-episode pollution levels (< 30 /ig/m3 as H2SO4, or
3 < approximately 600 nmoles/m3 H+), though these associations could not be separated from
4 those for BS or SO2. The only attempts to-date to associate present-day levels of acidic
5 aerosols with acute and chronic mortality (Dockery et al., 1992; Dockery et al., 1993b,
6 respectively) were unable to do so, but there may not have been a sufficiently long series of
7 H+ data to detect H+ associations. There is a critical need for present day replications of
8 the London mortality-acid aerosol studies to be conducted, however, in order to determine if
9 the London wintertime associations (which occurred in reduction-type atmospheres) are
10 pertinent to present-day U.S. conditions, in which acid aerosol peaks occur primarily in the
11 summer months (in oxidation-type atmospheres).
12 Taken as a whole, several analyses are suggestive of mortality and morbidity
13 associations with the sulfate fraction of fine particles found in contemporary American urban
14 airsheds. Without nationwide measurements of airborne acidity, however, it is difficult to
15 evaluate the relative contribution of acid aerosols within these fine particle sulfates to the
16 reported health effects.
17 Increased hospital admissions for respiratory causes were also documented during the
18 London Fog episode of 1952, and this association has now been observed under present-day
19 conditions, as well. Thurston et al. (1992) and Thurston et al. (1994) have noted
20 associations between ambient acidic aerosols and summertime respiratory hospital admissions
21 in both New York State and Toronto, Canada, respectively, even after controlling for
22 potentially confounding temperature effects. In the latter of these studies, significant
23 independent H+ effects remained even after simultaneously considering the other major
24 co-pollutant, O3, in the regression model. In the Toronto analysis, the increase in
25 respiratory hospital admissions associated with H+ was indicated to be roughly six times that
26 for non-acidic PM10 (per unit mass). In these studies, H+ effects were estimated to be the
27 largest during acid aerosol episodes (H+ S: 10 ^g/m3 as H2SO4, or =200 nmoles/m3 H+),
28 which occur roughly 2 to 3 times per year in eastern North America. These studies provide
29 evidence that present-day strongly acidic aerosols may represent a portion of PM which is
April 1995 12-321 DRAFT-DO NOT QUOTE OR CITE
-------
1 particularly associated with significant acute respiratory disease health effects in the general
2 public.
3 Results from recent acute symptoms and lung function studies of healthy children
4 indicate the potential for acute acidic PM effects in this population. While the 6-City study
5 of diaries kept by parents of children's respiratory and other illness did not demonstrate H+
6 associations with lower respiratory symptoms except at H+ above 110 moles/m3 (Dockery
7 et al., 1994), upper respiratory symptoms in two of the cities were found to be most strongly
8 associated with daily measurements of H2SO4 (Schwartz, et al., 1991). Recent summer
9 camp and school children studies of lung function have also indicated significant associations
10 between acute exposures to acidic PM and decreases in the lung function of children
11 independent of those associated with O3 (Studnicka et al., 1995; Neas et al., 1995).
12 Studies of the effects of chronic H+ exposures on children's respiratory health and lung
13 function are generally consistent with effects as a result of chronic H+ exposure.
14 Preliminary analyses of bronchitis prevalence rates as reported across the 6-City study locales
15 were found to be more closely associated with average H+ concentrations than with PM in
16 general (Speizer, 1989). A follow-up analysis of these cities and a seventh locality which
17 controlled the analysis for maternal smoking and education and for race, suggested
18 associations between summertime average H+ and chronic bronchitic and related symptoms
19 (Damokosh et al., 1993). The relative odds of bronchitic symptoms with the highest acid
20 concentration (58 nmoles/m3 H+) versus the lowest concentration (16 nmoles/m3) was
21 2.4 (95% CI: 1.9 to 3.2). Furthermore, in a follow-up study of children in 24 U.S. and
22 Canadian communities (Dockery et al., 1993a) in which the analysis was adjusted for the
23 effects of gender, age, parental asthma, parental education, and parental allergies, bronchitic
24 symptoms were confirmed to be significantly associated with strongly acidic PM (relative
25 odds = 1.7, 95% CI: 1.1 to 2.4). It was also found in the 24-Cities study that mean FVC
26 and FEVj 0 were lower in locales having high particle strong acidity (Raizenne et al., 1993).
27 Thus, chronic exposures to strongly acidic PM may have effects on measures of respiratory
28 health in children. However, given the usually high correlation between acidic PM and PM
29 in general, it is difficult to identify these effects solely with the acid portion of PM.
April 1995 12-322 DRAFT-DO NOT QUOTE OR CITE
-------
1 There is some uncertainty about human exposure to acid aerosols. Even though
2 ambient acid aerosols may easily penetrate into inside air, household sources of ammonia
3 may cause rapid reduction of interior acidity, so that total exposure to acidic particles may be
4 proportionately smaller than exposure to non-acidic particles. However, even a relatively
5 brief exposure to ambient acid aerosols outside the house could conceivably have detectable
6 biological effects. This could explain a role for acidity effects in children, but may be less
7 adequate to explain PM effects in adults who may spend even less time outdoors than
8 children.
April 1995 12-323 DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Abbey, D. E.; Mills, P. K.; Petersen, F. F.; Beeson, W. L. (1991a) Long-term ambient concentrations of total
4 suspended particulates and oxidants as related to incidence of chronic disease in California Seventh-Day
5 Adventists. Environ. Health Perspect. 94: 43-50.
6
7 Abbey, D. E.; Moore, J.; Petersen, F.; Beeson, L. (1991b) Estimating cumulative ambient concentrations of air
8 pollutants: description and precision of methods used for an epidemiological study. Arch. Environ. Health
9 46: 281-287.
10
11 Abbey, D. E.; Petersen, F.; Mills, P. K.; Beeson, W. L. (1993) Long-term ambient concentrations of total
12 suspended particulates, ozone, and sulfur dioxide and respiratory symptoms in a nonsmoking population.
13 Arch. Environ. Health 48: 33-46.
14
15 Abbey, D. E.; Lebowitz, M. D.; Mills, P. K.; Petersen, F. F.; Beeson, W. L.; Burchette, R. J. (1995)
16 Long-term ambient concentrations of particulates and oxidants and development of chronic disease in a
17 cohort of nonsmoking California residents. Inhalation Toxicol. 7: 19-34.
18
19 Akaike, H. (1973) Information theory and an extension of the maximum likelihood principle. In: Petrov, B. N.;
20 Csaki, F., eds. 2nd International symposium on information theory. Budapest, Hungary: Akademiai
21 Kiado.
22
23 American Conference of Governmental Industrial Hygienists. (1993) 1993—1994 threshold limit values for
24 chemical substances and physical agents and biological exposure indices. Cincinnati, OH: American
25 Conference of Governmental Industrial Hygienists, Inc.; p. 31.
26
27 American Lung Association. (1994) The perils of particulates: an estimation of populations-at-risk of adverse
28 health consequences from paniculate matter in areas with paniculate matter levels above the National
29 Ambient Air Quality Standards of the Clean Air Act and the State of California's Air Quality Standard.
30 New York, NY: American Lung Association.
31
32 American Thoracic Society. (1962) Definitions and classification of chronic bronchitis, asthma, and pulmonary
33 emphysema. Am. Rev. Respir. Dis. 85: 762-768.
34
35 American Thoracic Society. (1969) Standards for epidemiologic surveys in chronic respiratory disease. New
36 York, NY: National Tuberculosis and Respiratory Disease Association.
37
38 American Thoracic Society. (1991) Lung function testing: selection of reference values and interpretative
39 strategies Am. Rev. Respir. Dis. 144: 1202-1218.
40
41 Anderson, D. O.; Ferris, B. G., Jr. (1965) Air pollution levels and chronic respiratory disease. Arch. Environ.
42 Health 10: 307-311.
43
44 Anderson, D. O.; Ferris, B. G., Jr.; Zickmantel, R. (1964) Levels of air pollution and respiratory disease in
45 Berlin, New Hampshire. Am. Rev. Respir. Dis. 90: 877-887.
46
47 Annesi, I.; Kauffmann, F. (1986) Is respiratory mucus hypersecretion really an innocent disorder?: a 22-year
48 mortality survey of 1,061 working men. Am. Rev. Respir. Dis. 134: 688-693.
49
50 Archer, V. E. (1990) Air pollution and fatal lung disease in three Utah counties. Arch. Environ. Health 45:
51 325-334.
52
April 1995 12-324 DRAFT-DO NOT QUOTE OR CITE
-------
1 Arossa, W.; Spinaci, S.; Bugiani, M.; Natale, P.; Bucca, C.; de Candussio, G. (1987) Changes in lung function
2 of children after an air pollution decrease. Arch. Environ. Health 42: 170-174.
3
4 Ashley, D. J. B. (1967) The distribution of lung cancer and bronchitis in England and Wales. Br. J. Cancer 21:
5 243-259.
6
7 Auliciems, A.; Frost, D. (1989) Temperature and cardiovascular deaths in Montreal. Int. J. Biometeorol. 33:
8 151-156.
9
10 Ayres, J.; Fleming, D.; Williams, M.; Mclnnes, G. (1989) Measurement of respiratory morbidity in general
11 practice in the United Kingdom during the acid transport event of January 1985. In: Symposium on health
12 effects of acid aerosols; October 1987; Research Triangle Park, NC. Environ. Health Perspect. 79:
13 83-88.
14
15 Bailey, D. L. R.; Clayton, P. (1982) The measurement of suspended particle and total carbon concentrations in
16 the atmosphere using standard smoke shade methods. Atmos. Environ. 16: 2683-2690.
17
18 Baker-Blocker, A. (1982) Winter weather and cardiovascular mortality in Minneapolis-St. Paul. Am. J. Public
19 Health 72: 261-265.
20
21 Barker, W. H.; Mullooly, J. P. (1981) Underestimation of the role of pneumonia and influenza in causing excess
22 mortality. Am. J. Public Health 71: 643-645.
23
24 Bates, D. V. (1983) Ecclesiastes reread. Am. Rev. Respir. Dis. 128: 383-384.
25
26 Bates, D. V. (1992) Health indices of the adverse effects of air pollution: the question of coherence. Environ.
27 Res. 59: 336-349.
28
29 Bates, D. V.; Sizto, R. (1983) Relationship between air pollutant levels and hospital admissions in Southern
30 Ontario. Can. J. Public Health 74: 117-122.
31
32 Bates, D. V.; Sizto, R. (1986) A study of hospital admissions and air pollutants in southern Ontario. In: Lee, S.
33 D.; Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols: research, risk assessment and control
34 strategies, proceedings of the second U.S.-Dutch international symposium; May 1985; Williamsburg,
35 VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 767-777.
36
37 Bates, D. V.; Sizto, R. (1987) Air pollution and hospital admissions in southern Ontario: the acid summer haze
38 effect. Environ. Res. 43: 317-331.
39
40 Bates, D. V.; Sizto, R. (1989) The Ontario Air Pollution study: identification of the causative agent. Environ.
41 Health Perspect. 79: 69-72.
42
43 Beard, C. M.; Yunginger, J. W.; Reed, C. E.; O'Connell, E. J.; Silverstein, M. D. (1992) Interobserver
44 variability in medical record review: an epidemiological study of asthma. J. Clin. Epidemiol. 45:
45 1013-1020.
46
47 Beaumont, J. J.; Leveton, J.; Knox, K.; Bloom, T.; McQuiston, T.; Young, M.; Goldsmith, R.; Steenland, N.
48 K.; Brown, D. P.; Halpern, W. E. (1987) Lung cancer mortality in workers exposed to sulfuric acid mist
49 and other acid mists. JNCI J. Natl. Cancer Inst. 79: 911-921.
50
51 Belloc, N. D. (1973) Relationship of health practices and mortality. Prev. Med. 2: 67-81.
52
April 1995 12-325 DRAFT-DO NOT QUOTE OR CITE
-------
1 Belsley, D. A.; Kuh, E.; Welsch, R. E. (1980) Regression diagnostics: identifying influential data and sources of
2 collinearity. New York, NY: John Wiley & Sons, Inc. (Bradley, R. A.; Kendall, D. G.; Hunter, J. S.;
3 Watson, G. S., eds. Wiley series in probability and mathematical statistics).
4
5 Binder, R. E.; Mitchell, C. A.; Hosein, H. R.; Bouhuys, A. (1976) Importance of the indoor environment in air
6 pollution exposure. Arch. Environ. Health 31: 277-279.
7
8 Blakeslee, S. (1992) Birthdays: a matter of life and death. New York Times September 22.
9
10 Blakeslee, S. (1993) Beliefs reported to shorten life. New York Times November 6.
11
12 Bloomfield, P. (1976) Fourier analysis of time series: an introduction. New York, NY: John Wiley and Sons.
13
14 Bobak, M.; Leon, D. A. (1992) Air pollution and infant mortality in the Czech Republic, 1986-1988. Lancet
15 (8826): 1010-1014.
16
17 Bock, N.; Lippmann, M.; Lioy, P.; Munoz, A.; Speizer, F. E. (1985) The effects of ozone on the pulmonary
18 function of children. In: Lee, S. D., ed. Evaluation of the scientific basis for ozone/oxidants standards:
19 proceedings of an APCA international specialty conference; November 1984; Houston, TX. Pittsburgh,
20 PA: Air Pollution Control Association; pp. 297-308. (APCA international specialty conference
21 transactions: TR-4).
22
23 Born, W.; Happ, M. P.; Dallas, A.; Reardon, C.; Kubo, R.; Shinnick, T.; Brennan, P.; O'Brien, R. (1990)
24 Recognition of heat shock proteins and y8 cell function. Immunol. Today 11: 40-43.
25
26 Bouhuys, A.; Beck, G. J.; Schoenberg, J. B. (1978) Do present levels of air pollution outdoors affect respiratory
27 health? Nature (London) 276: 466-471.
28
29 Boussin, G.; Leloup, M.; Cayla, F.; Giroux, M.; Pous, J. (1990) Symptomes respiratoires chez les enfants d'age
30 scolaire a toulouse. Influence de la pollution atmospherique [Respiratory symptoms in schoolchildren in
31 Toulouse. Influence of the air pollution]. Arch. Fr. Pediatr. 47: 351-356.
32
33 Box, G. E. P.; Draper, N. R. (1987) Empirical model-building and response surfaces. New York, NY: John
34 Wiley and Sons.
35
36 Box, G. E. P.; Jenkins, G. M. (1979) Time series analysis: forecasting and control. San Francisco, CA:
37 Holden-Day.
38
39 Brackbill, R.; Frazier, T.; Shilling, S. (1988) Smoking characteristics of US workers, 1978-1980. Am. J. Ind.
40 Med. 13: 5-41.
41
42 Braun-Fahrlander, C.; Ackermann-Liebrich, U.; Schwartz, J.; Gnehm, H. P.; Rutishauser, M.; Wanner, H. U.
43 (1992) Air pollution and respiratory symptoms in preschool children. Am. Rev. Respir. Dis. 145: 42-47.
44
45 Breiman, L.; Friedman, J. H.; Olshen, R. A.; Stone, C. J. (1984) Classification and regression trees. Belmont,
46 CA: Wadsworth International Group. (Bickel, P. J.; Cleveland, W. S.; Dudley, R. M., eds. The
47 Wadsworth statistics/probability series).
48
49 Britten, N.; Davies, J. M. C.; Colley, J. R. T. (1987) Early respiratory experience and subsequent cough and
50 peak expiratory flow rate in 36 year old men and women. Br. Med. J. 294: 1317-1320
51
April 1995 12-326 DRAFT-DO NOT QUOTE OR CITE
-------
1 Bnmekreef, B.; Kinney, P. L.; Ware, J. H.; Dockery, D.; Speizer, F. E.; Spengler, J. D.; Ferris, B. G., Jr.
2 (1991) Sensitive subgroups and normal variation in pulmonary function response to air pollution episodes.
3 Environ. Health Perspect. 90: 189-193.
4
5 Brunekreef, B.; Hoek, G.; Breugelmans, O.; Leentvaar, M. (1994) Respiratory effects of low-level
6 photochemical air pollution in amateur cyclists. Am. J. Respir. Crit. Care Med. 150: in press.
7
8 Buffmgton, J.; Chapman, L. E.; Schmeltz, L. M.; Kendal, A. P. (1993) Do family physicians make good
9 sentinels for influenza? Arch. Fam. Med. 2: 859-865.
10
11 Buffler, P. A.; Cooper, S. P.; Stinnett, S.; Contant, C.; Shirts, S.; Hardy, R. J.; Agu, V.; Gehan, B.; Burau,
12 K. (1988) Air pollution and lung cancer mortality in Harris County, Texas, 1979-1981. Am. J.
13 Epidemiol. 128: 683-699.
14
15 Bull, G. M. (1973) Meteorological correlates with myocardial and cerebral infarction and respiratory disease. Br.
16 J. Prev. Soc. Med. 27: 108-113.
17
18 Bull, G. M.; Morton, J. (1975) Seasonal and short-term relationships of temperature with deaths from myocardial
19 and cerebral infarction. Age Ageing 4: 19-31.
20
21 Bultena, G. L. (1969) Health patterns of aged migrant retirees. J. Am. Geriatr. Soc. 17: 1127-1131.
22
23 Burnett, R. T.; Dales, R. E.; Raizenne, M. E.; Krewski, D.; Summers, P. W.; Roberts, G. R.; Raad-Young,
24 M.; Dann, T.; Brook, J. (1994) Effects of low ambient levels of ozone and sulfates on the frequency of
25 respiratory admissions to Ontario hospitals. Environ. Res. 65: 172-194.
26
27 Burrows, B.; Lebowitz, M. D. (1975) Characteristics of chronic bronchitis in a warm, dry region. Am. Rev.
28 Respir. Dis..112: 365-370.
29
30 Candas, V. (1987) Adaptation to extreme environments: thermophysiological changes in man during humid heat
31 acclimation. In: Dejours, P., ed. Comparative physiology of environmental adaptations, v. 2. Basel,
32 Switzerland: Karger; pp. 76-93.
33
34 Carr, W.; Zeitel, L.; Weiss, K. (1992) Variations in asthma hospitalizations and deaths in New York City. Am.
35 J. Public Health 82: 59-65.
36
37 Cass, G. R.; Conklin, M. H.; Shah, J. J.; Huntzicker, J. J.; Macias, E. S. (1984) Elemental carbon
38 concentrations: estimation of an historical data base. Atmos. Environ. 18: 153-162.
39
40 Cavender, J. H.; Cox, W. M.; Georgevich, M.; Huey, N. A.; Jutze, G. A.; Zimmer, C. E. (1971) Interstate
41 surveillance project: measurement of air pollution using static monitors. Research Triangle Park, NC:
42 U.S. Environmental Protection Agency, Air Pollution Control Office; publication no. APTD-0666.
43 Available from: NTIS, Springfield, VA; PB-200519/BA.
44
45 Chanock, R. M.; Parrott, R. H. (1965) Acute respiratory disease in infancy and childhood: present understanding
46 and prospects for prevention. Pediatrics 36: 21-39.
47
48 Chanock, R. M.; Mclntosh, K.; Murphy, B. R.; Parrott, R. H. (1989) Respiratory syncytial virus. In: Evans, A.
49 S., ed. Viral infections of humans: epidemiology and control. 3rd ed. New York, NY: Plenum
50 Publishing Corporation; pp. 525-544.
51
April 1995 12-327 DRAFT-DO NOT QUOTE OR CITE
-------
1 Chapman, R. S.; Henderson, F. W.; Clyde, W. A., Jr.; Collier, A. M.; Denny, F. W. (1981) The epidemiology
2 of tracheobronchitis inpediatric practice. Am. J. Epidemiol. 114: 786-797.
J
4 Chapman, R. S.; Calafiore, D. C.; Hasselblad, V. (1985) Prevalence of persistent cough and phlegm hi young
5 adults hi relation to long-term ambient sulfur oxide exposure. Am. Rev. Respir. Dis. 132: 261-267.
6
7 Chapman, R. S.; Mumford, J. L.; Harris, D. B.; He, X.; Jiang, W.; Yang, R. (1988) The epidemiology of lung
8 cancer in Xuan Wei, China: current progress, issues, and research strategies. Arch. Environ. Health 43:
9 180-185.
10
11 Chapman, L. E.; Tipple, M. A.; Schmeltz, L. M.; Good, S. E.; Regnery, H. L.; Kendal, A. P.; Gary, H. E.,
12 Jr.; Cox, N. J. (1992) Influenza—United States, 1989-90 and 1990-91 seasons. Morb. Mortal. Wkly.
13 Rep. 41(SS-3): 35-46.
14
15 Chappie, M.; Lave, L. (1982) The health effects of air pollution: a reanalysis. J. Urban Econ. 12: 346-376.
16
17 Charlton, A.; Blair, V. (1989) Absence from school related to children's and parental smoking habits. Br. Med.
18 J. 298: 90-92.
19
20 Charpin, D.; Kleisbauer, J. P.; Fondarai, J.; Graland, B.; Viala, A.; Gouezo, F. (1988) Respiratory symptoms
21 and air pollution changes in children: the Gardanne coal-basin study. Arch. Environ. Health 43: 22-27.
22
23 Chestnut, L. G.; Schwartz, J.; Savitz, D. A.; Burchfiel, C. M. (1991) Pulmonary function and ambient
24 paniculate matter: epidemiological evidence from NHANES I. Arch. Environ. Health 46: 135-144.
25
26 Chestnut, L.; Trabka, E.; Smith, J. (1993) Differences in summer mortality due to climate in U.S. cities. U.S.
27 Environmental Protection Agency, Climate Change Division.
28
29 Chinn, S.; Florey, C. du V.; Baldwin, I. G.; Gorgol, M. (1981) The relation of mortality in England and Wales
30 1969-73 to measurements of air pollution. J. Epidemiol. Commun. Health 35: 174-179.
31
32 Choi, K.; Thacker, S. B. (1981) An evaluation of influenza mortality surveillance, 1962-1979:1. time series
33 forecasts of expected pneumonia and influenza deaths. Am. J. Epidemiol. 113: 215-226.
34
35 Choi, K.; Thacker, S. B. (1981) An evaluation of influenza mortality surveillance, 1962-1979: II. percentage of
36 pneumonia and influenza deaths as an indicator of influenza activity. Am. J. Epidemiol. 113: 227-235
37
38 Christiansen, G. B.; Degen, C. G. (1980) Air pollution and mortality rates: a note on Lave and Seskin's pooling
39 of cross-sectional and time-series data. J. Environ. Econ. Mgmt. 7: 149-155.
40
41 Clark, R. R. (1982) The error-in-variables problem hi the logistic regression model [dissertation]. Chapel Hill,
42 NC: University of North Carolina.
43
44 Cleaveland, R. B.; Cleaveland, W. S.; McRae, J. E.; Terpenning, I. (1990) STL: a seasonal trend decomposition
45 procedure based on Loess. J. Off. Stat. 6: 3-73.
46
47 Cliff, A. D.; Ord, J. K. (1981) Spatial processes: models and applications. London, United Kingdom: Pion.
49 Clifton, M.; Kerridge, D.; Moulds, W.; Pemberton, J.; Donaghue, J. K. (1959) The reliability of air pollution
50 measurements in relation to the siting of instruments. Int. J. Air Pollut. 2: 188-196.
51
April 1995 12-328 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cochrane, D.; Orcutt, G. H. (1949) Application of least squares regression to relationships containing
2 auto-correlated error terms. J. Am. Stat. Assoc. 44: 32-43.
3
4 Cohen, A. (1989) Comparison of correlated correlations. Stat. Med. 8: 1485-1495.
5
6 Cohen, B. L. (1990) Ecological versus case-control studies for testing a linear-no threshold dose-response
7 relationship. Int. J. Epiderniol. 19: 680-684.
8
9 Cohen, B. L. (1992) Percentage of lifetime spent in area of residence at time of death. Environ. Res. 57:
10 208-211.
11
12 Cohen, B. L. (1994) Invited commentary: in defense of ecologic studies for testing a linear-no threshold theory.
13 Am. J. Epidemiol. 139: 765-768.
14
15 Colley, J. R. T.; Holland, W. W.; Leeder, S. R.; Corkhill, R. T. (1976) Respiratory function of infants in
16 relation to subsequent respiratory disease: an epidemiological study. Bull. Eur. Physiopathol. Respir. 12:
17 651-657.
18
19 Commins, B. T. (1963) Determination of paniculate acid in town air. Analyst (London) 88: 364-367.
20
21 Commins, B. T.; Waller, R. E. (1967) Observations from a ten-year-study of pollution at a site in the city of
22 London. Atmos. Environ. 1: 49-68.
23
24 Commission of the European Communities. (1993) Study designs. Luxembourg: Office for Official Publications
25 of the European Communities; report no. EUR 15095 EN. (Katsouyanni, K., ed. Air pollution
26 epidemiology reports series no. 4).
27
28 Cooper, D. E.; Hamilton, W. C. (1979) Mortality and sulfates—the phantom connection? Min. Congr. J.:
29 January.
30
31 Coutant, R. W. (1977) Effect of environmental variables on collection of atmospheric sulfate. Environ. Sci.
32 Technol. 11: 873-878.
33
34 Cox, C. (1987) Threshold dose-response models in toxicology. Biometrics 43: 511-523.
35
36 Crane, J.; Pearce, N.; Burgess, C.; Woodman, K.; Robson, B.; Beasley, R. (1992) Markers of risk of asthma
37 death or readmission in the 12 months following a hospital admission for asthma. Int. J. Epidemiol. 21:
38 737-744.
39
40 Craxford, S. R.; Clifton, M.; Weatherly, P. M. (1967) Smoke and sulphur dioxide in Great Britain: distribution
41 and changes. In: Proceedings of the 1st international clean air congress; October 1966; London, United
42 Kingdom. London, United Kingdom: National Society for Clean Air; pp. 213-216.
43
44 Crews, D. E.; Stamler, J.; Dyer, A. (1991) Conditions other than underlying cause of death listed on death
45 certificates provide additional useful information for epidemiologic research. Epidemiology 2: 271-275.
46
47 Crocker, T. D.; Schulze, W.; Ben-David, S.; Kneese, A. V. (1979) Methods development for assessing air
48 pollution control benefits: volume I, experiments in the economics of air pollution epidemiology.
49 Washington, DC: U.S. Environmental Protection Agency, Office of Health and Ecological Effects; EPA
50 report no. EPA-600/5-79-001a. Available from: NTIS, Springfield, VA; PB-293615.
51
April 1995 12-329 DRAFT-DO NOT QUOTE OR CITE
-------
1 Cramp, K. S. (1984a) A new method for determining allowable daily intakes. Fundam. Appl. Toxicol. 4:
2 854-871.
3
4 Cramp, K. S. (1984b) Mechanisms leading to dose-response models. In: Ficci, P. F., ed. Principles of health
5 risk assessment. Englewood Clifts, NJ: Prentice-Hall, Inc.; pp. 235-277.
6
7 Cramp, K. S.; Howe, R. B. (1985) A review of methods for calculating statistical confidence limits in low dose
8 extrapolation. In: Clayson, D. B.; Krewski, D.; Munro, I., eds. Toxicological risk assessment: v. I,
9 biological and statistical criteria. Boca Raton, FL: CRC Press, Inc.; pp. 187-203.
10
11 de Leeuw, F. A. A. M.; van Rheineck Leyssius, H. J. (1989) Long-range transport modeling of air pollution
12 episodes. In: Symposium on the health effects of acid aerosols; October 1987; Research Triangle Park,
13 NC. Environ. Health Perspect. 79: 53-59.
14
15 Daly, C. (1954) Air pollution and bronchitis. Br. Med. J. 2: 687-688.
16
17 Damokosh, A. I.; Spengler, J. D.; Dockery, D. W.; Ware, J. H.; Speizer, F. E. (1993) Effects of acidic
18 particles on respiratory symptoms in 7 US communities. Am. Rev. Respir. Dis. 147: A632.
19
20 Dassen, W.; Branekreef, B.; Hoek, G.; Hofschreuder, P.; Staatsen, B.; De Groot, H.; Schouten, E.; Biersteker,
21 K. (1986) Decline in children's pulmonary function during an air pollution episode. J. Air Pollut. Control
22 Assoc. 36: 1223-1227.
23
24 Davis, R. E.; Kalkstein, L. S. (1990) Development of an automated spatial synoptic climatological classification.
25 Int. J. Climatol. 10: 769-794.
26
27 Delfmo, R. J.; Becklake, M. R.; Hanley, J. A. (1994) The relationship of urgent hospital admissions for
28 respiratory illnesses to photochemical air pollution levels in Montreal. Environ. Res. 67: 1-19
29
30 Delfmo, R. J.; Becklake, M. R.; Hanley, J. A.; Singh, B. (1994) Estimation of unmeasured paniculate air
31 pollution data for an epidemiological study of daily respiratory morbidity. Environ. Res. 67: 20-38
32
33 Denny, F. W.; Clyde, W. A. (1986) Acute lower respiratory tract infections in nonhospitalized children. J.
34 Pediatr. (St. Louis) 108: 635-646.
35
36 Denny, F. W.; Murphy, T. F.; Clyde, W. A., Jr.; Collier, A. M.; Henderson, F. W. (1983) Croup: an 11-year
37 study in a pediatric practice. Pediatrics 71: 871-876.
38
39 Derriennic, F.; Richardson, S.; Mollie, A.; Lellouch, J. (1989) Short-term effects of sulphur dioxide pollution on
40 mortality in two French cities. Int. J. Epidemiol. 18: 186-197.
41
42 Diamond, J. (1991) Pearl Harbor and the emperor's physiologists. Nat. History (12): 2-7.
43
44 Dinman, B. D. (1972) "Non-concept" of "no-threshold": chemicals in the environment. Science (Washington
45 DC) 175: 495-497.
46
47 Dockery, D. W. (1988) Comments on acid aerosols issues paper [submitted at the SAB meeting to discuss the
48 February draft]. Boston, MA: Harvard University, School of Public Health
49
50 Dockery, D. W. (1993) Epidemiological study design for investigating respiratory health effects of complex air
51 pollution mixtures. Environ. Health Perspect. 101(suppl. 4): 187-191.
52
April 1995 12-330 DRAFT-DO NOT QUOTE OR CITE
-------
1 Dockery, D. W.; Pope, C. A., III. (1994) Air pollution and mortality: the authors reply [letter]. N. Engl. J.
2 Med. 330: 1238.
3
4 Dockery, D. W.; Pope, C. A., III. (1994) Acute respiratory effects of paniculate air pollution. Annu. Rev.
5 Public Health 15: 107-132.
6
7 Dockery, D. W.; Schwartz, J. (1992) The authors' response to Waller and Swan. Am. J. Epidemiol. 135: 23-25.
8
9 Dockery, D. W.; Ware, J. H.; Ferris, B. G., Jr.; Speizer, F. E.; Cook, N. R.; Herman, S. M. (1982) Change
10 in pulmonary function in children associated with air pollution episodes. J. Air Pollut. Control Assoc. 32:
11 937-942.
12
13 Dockery, D. W.; Berkey, C. S.; Ware, J. H.; Speizer, F. E.; Ferris, B. G., Jr. (1983) Distribution of forced
14 vital capacity and forced expiratory volume in one second in children 6 to 11 years of age. Am. Rev.
15 Respir. Dis. 128: 405-412.
16
17 Dockery, D. W.; Ware, J. H.; Ferris, B. G., Jr.; Glicksberg, D. S.; Fay, M. E.; Spiro, A., Ill; Speizer, F. E.
18 (1985) Distribution of forced expiratory volume in one second and forced vital capacity in healthy, white,
19 adult never-smokers in six U.S. cities. Am. Rev. Respir. Dis. 131: 511-520.
20
21 Dockery, D. W.; Speizer, F. E.; Ferris, B. G., Jr.; Ware, J. H.; Louis, T. A.; Spiro, A., III. (1988)
22 Cumulative and reversible effects of lifetime smoking on simple tests of lung function in adults. Am.
23 Rev. Respir. Dis. 137: 286-292.
24
25 Dockery, D. W.; Speizer, F. E.; Stram, D. O.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr. (1989) Effects
26 of inhalable particles on respiratory health of children. Am. Rev. Respir. Dis. 139: 587-594.
27
28 Dockery, D. W.; Schwartz, J.; Spengler, J. D. (1992) Air pollution and daily mortality: associations with
29 particulates and acid aerosols. Environ. Res. 59: 362-373.
30
31 Dockery, D. W.; Damokosh, A. I.; Neas, L. M.; Raizenne, M.; Spengler, J. D.; Koutrakis, P.; Ware, J. H.;
32 Speizer, F. E. (1993) Health effects of acid aerosols on North American children: respiratory symptoms
33 and illness. Am. Rev. Respir. Dis. 147: A633.
34
35 Dockery, D. W.; Pope, C. A., Ill; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G., Jr.;
36 Speizer, F. E. (1993) An association between air pollution and mortality in six U.S. cities. N. Engl. J.
37 Med. 329: 1753-1759.
38
39 Dodge, R.; Solomon, P.; Moyers, J.; Hayes, C. (1985) A longitudinal study of children exposed to sulfur
40 oxides. Am. J. Epidemiol. 121: 720-736.
41
42 Dodge, R.; Cline, M. G.; Burrows, B. (1986) Comparisons of asthma, emphysema, and chronic bronchitis
43 diagnoses in a general population sample. Am. Rev. Respir. Dis. 133: 981-986.
44
45 Doll, R.; Peto, R.; Wheatley, K.; Gray, R.; Sutherland, I. (1994a) Mortality in relation to smoking: 40 years'
46 observations on male British doctors. Br. Med. J. 309: 901-911.
47
48 Doll, R.; Peto, R.; Hall, E.; Wheatley, K.; Gray, R. (1994b) Mortality in relation to consumption of alcohol: 13
49 years' observations on male British doctors. Br. Med. J. 309: 911-918.
50
51 Draper, N. R.; Smith, H. (1966) Applied regression analysis. New York, NY: John Wiley & Sons, Inc.
52
April 1995 12-331 DRAFT-DO NOT QUOTE OR CITE
-------
1 Duclos, P.; Sanderson, L. M.; Lipsett, M. (1990) The 1987 forest fire disaster in California: assessment of
2 emergency room visits. Arch. Environ. Health 45: 53-58.
3
4 Dusseldorf, A.; Kruize, H.; Brunekreef, B.; Hofschreuder, P.; de Meer, G.; van Oudvorst, A. B. (1994) Acute
5 effects of PM10 and airborne iron on respiratory health: a panel study among adults living near a steel
6 industry in the Netherlands. Am. J. Respir. Crit. Care Med.: submitted.
7
8 Eder, B. K.; Davis, J. M.; Bloomfield, P. (1994) An automated classification scheme designed to better elucidate
9 the dependence of ozone on meteorology. J. Appl. Meteorol. 33: 1182-1199.
10
11 El-Sadik, Y. M.; Osman, H. A.; El-Gazzar, R. M. (1972) Exposure to sulfuric acid in manufacture of storage
12 batteries. J. Occup. Med. 14: 224-226.
13
14 Elliot, T. (1992) Stress proteins. In: Roitt, I. M.; Delves, P. J., eds. Encyclopedia of immunology, v. 3.
15 London, United Kingdom: Academic Press; pp. 1402-1405.
16
17 Ellis, F. P. (1972) Mortality from heat illness and heat-aggravated illness in the United States. Environ. Res. 5:
18 1-58.
19
20 Ellis, F. P.; Nelson, F. (1978) Mortality in the elderly in a heat wave in New York City, August 1975. Environ.
21 Res. 15: 504-512.
22
23 Euler, G. L.; Abbey, D. E.; Magie, A. R.; Hodgkin, J. E. (1987) Chronic obstructive pulmonary disease
24 symptom effects of long-term cumulative exposure to ambient levels of total suspended particulates and
25 sulfur dioxide in California Seventh-Day Adventist residents. Arch. Environ. Health 42: 213-222
26
27 Euler, G. L.; Abbey, D. E.; Hodgkin, J. E.; Magie, A. R. (1988) Chronic obstructive pulmonary disease
28 symptom effects of long-term cumulative exposure to ambient levels of total oxidants and nitrogen
29 dioxide in California Seventh-day Adventist residents. Arch. Environ. Health 43: 279-285
30
31 Evans, J. S.; Kinney, P. L.; Koehler, J. L.; Cooper, D. W. (1984) The relationship between cross-sectional and
32 time series studies. J. Air Pollut. Control Assoc. 34: 551-553.
33
34 Evans, J. S.; Tosteson, T.; Kinney, P. L. (1984) Cross-sectional mortality studies and air pollution risk
35 assessment. Environ. Int. 10: 55-83.
36
37 Faich, G.; Rose, R. (1979) Blizzard morbidity and mortality: Rhode Island, 1978. Am. J. Public Health 69:
38 1050-1052.
39
40 Fairley, D. (1990) The relationship of daily mortality to suspended particulates in Santa Clara county, 1980-86.
41 Environ. Health Perspect. 89: 159-168.
42
43 Federal Register. (1987) Revisions to the national ambient air quality standards for paniculate matter. F. R. (July
44 1) 52: 24634-24669.
45
46 Federal Register. (1989) Silica, crystalline—cristobalite. F. R. (January 19) 54: 2521-2523.
47
48 Fedson, D. S.; Wajda, A.; Nicol, J. P.; Roos, L. L. (1992) Disparity between influenza vaccination rates and
49 risks for influenza-associated hospital discharge and death in Manitoba in 1982-1983. Ann. Intern. Med
50 116:550-555.
51
52 Ferris, B. G. (1978) Epidemiology standardization project. Am. Rev. Respir. Dis. 118: 1-120.
April 1995 12-332 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ferris, B. G., Jr.; Anderson, D. O. (1962) The prevalence of chronic respiratory disease in a New Hampshire
2 town. Am. Rev. Respir. Dis. 86: 165-185.
3
4 Ferris, B. G., Jr.; Burgess, W. A.; Worcester, J. (1967) Prevalence of chronic respiratory disease in a pulp mill
5 and a paper mill in the United States. Br. J. Ind. Med. 24: 26-37.
6
7 Ferris, B. G., Jr.; Higgins, I. T. T.; Higgins, M. W.; Peters, J. M.; Van Ganse, W. F.; Goldman, M. D.
8 (1971) Chronic nonspecific respiratory disease, Berlin, New Hampshire, 1961-1967: a cross-sectional
9 study. Am. Rev. Respir. Dis. 104: 232-244.
10
11 Ferris, B. G., Jr.; Chen, H.; Puleo, S.; Murphy, R. L. H., Jr. (1976) Chronic nonspecific respiratory disease in
12 Berlin, New Hampshire, 1967 to 1973: a further follow-up study. Am. Rev. Respir. Dis. 113: 475-485.
13
14 Ferris, B. G., Jr.; Speizer, F. E.; Spengler, J. D.; Dockery, D.; Bishop, Y. M. M.; Wolfson, M.; Humble, C.
15 (1979) Effects of sulfur oxides and respirable particles on human health: methodology and demography of
16 populations in study. Am. Rev. Respir. Dis. 120: 767-779.
17
18 Ferris, B. G., Jr.; Ware, J. H.; Spengler, J. D.; Dockery, D. W.; Speizer, F. E. (1986) The Harvard six-cities
19 study. In: Lee, S. D.; Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols: research, risk
20 assessment and control strategies: proceedings of the second U.S.-Dutch international symposium; May
21 1985; Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 721-730.
22
23 Firket, M. (1931) Sur les causes des accidents survenus dans la vallee de la Meuse, lors des brouillards de
24 decembre 1930 [The causes of accidents which occurred in the Meuse Valley during the fogs of
25 December 1930]. Bull. Acad. R. Med. Belg. ll[ser. 5]: 683-741.
26
27 Firket, J. (1936) Fog along the Meuse Valley. Trans. Faraday Soc. 32: 1192-1197.
28
29 Fleisher, J. M.; Nayeri, K. (1991) Re: "Mortality and air pollution in London: a time series analysis" [letter].
30 Am. J. Epidemiol. 133: 631-632.
31
32 Fleiss, J. L.; Shrout, P. E. (1977) The effects of measurement errors on some multivariate procedures. Am. J.
33 Public Health 67: 1188-1191.
34
35 Fleming, D. M.; Crombie, D. L.; Norbury, C. A.; Cross, K. W. (1990) Observations on the influenza epidemic
36 of November/December 1989. Br. J. Gen. Pract. 40: 495-497.
37
38 Forastiere, F.; Corbo, G. M.; Pistelli, R.; Michelozzi, P.; Agabiti, N.; Brancato, G.; Ciappi, G.; Perucci, C.
39 A. (1994) Bronchial responsiveness in children living in areas with different air pollution levels. Arch.
40 Environ. Health 49: 111-118.
41
42 Franklin, C. A.; Burnett, R. T.; Paolini, R. J. P.; Raizenne, M. E. (1985) Health risks from acid rain: a
43 Canadian perspective. Environ. Health Perspect. 63: 155-168.
44
45 Frost, D. B.; Auliciems, A. (1993) Myocardial infarct death, the population at risk, and temperature habituation.
46 Int. J. Biometeorol. 37: 46-51.
47
48 Fuller, W. A. (1987) Measurement error models. New York, NY: John Wiley & Sons.
49
50 Gail, M. H. (1985) Adjusting for covariates that have the same distribution in exposed and unexposed cohorts.
51 In: Moolgavkar, S. H.; Prentice, R. L., eds. Modern statistical methods in chronic disease epidemiology.
52 New York, NY: John Wiley and Sons; pp. 3-18.
April 1995 12-333 DRAFT-DO NOT QUOTE OR CITE
-------
1 Gamble, J.; Jones, W.; Hancock, J. (1984a) Epidemiological-environmental study of lead acid battery workers:
2 II. acute effects of sulfuric acid on the respiratory system. Environ. Res. 35: 11-29.
3
4 Gamble, J.; Jones, W.; Hancock, J.; Meckstroth, R. L. (1984b) Epidemiological-environmental study of lead
5 acid battery workers: III. chronic effects of sulfuric acid on the respiratory system and teeth. Environ.
6 Res. 35: 30-52.
7
8 Gardner, M. J.; Crawford, M. D.; Morris, J. N. (1969) Patterns of mortality in middle and early old age in the
9 county boroughs of England and Wales. Br. J. Prev. Soc. Med. 23: 133-140.
10
11 Gergen, P. J.; Weiss, K. B. (1990) Changing patterns of asthma hospitalization among children: 1979 to 1987.
12 JAMA J. Am. Med. Assoc. 264: 1688-1692.
13
14 Gerstman, B. B.; Bosco, L. A.; Tomita, D. K. (1993) Trends in the prevalence of asthma hospitalization in the
15 5- to 14-year-old Michigan Medicaid population, 1980 to 1986. J. Allergy Clin. Immunol. 91: 838-843.
16
17 Gladen, B.; Rogan, W. J. (1979) Misclassification and the design of environmental studies. Am. J. Epidemiol.
18 109: 607-616.
19
20 Glass, R. T.; Zack, M. M. J. (1979) Increase in deaths from ischemic heart disease after blizzards. Lancet
21 (8114): 485-487.
22
23 Glezen, W. P. (1982) Serious morbidity and mortality associated with influenza epidemics. Epidemiol. Rev. 4:
24 25-44.
25
26 Glezen, W. P. (1989) Antecedents of chronic and recurrent lung disease: childhood respiratory trouble. Am.
27 Rev. Respir. Dis. 140: 873-874.
28
29 Glezen, W. P.; Denny, F. W. (1973) Epidemiology of acute lower respiratory disease in children. N. Engl. J.
30 Med. 288: 498-505.
31
32 Glezen, W. P.; Payne, A. A.; Snyder, D. N.; Downs, T. D. (1982) Mortality and influenza. J. Infect. Dis. 146:
33 313-321.
34
35 Gold, D. R.; Tager, I. B.; Weiss, S. T.; Tosteson, T. D.; Speizer, F. E. (1989) Acute lower respiratory illness
36 in childhood as a predictor of lung function and chronic respiratory symptoms. Am. Rev. Respir. Dis.
37 140: 877-884.
38
39 Gong, H., Jr. (1992) Health effects of air pollution: a review of clinical studies. Clin. Chest Med. 13: 201-214.
40
41 Goodman, L. A. (1953) Ecological regressions and behavior of individuals. Am. Soc. Rev. 13: 663-664.
42
43 Goren, A. I.; Goldsmith, J. R. (1986) Epidemiology of childhood respiratory disease in Israel. Eur. J.
44 Epidemiol. 2: 139-150.
45
46 Gorham, E. D.; Garland, F. C.; Garland, C. F. (1990) Sunlight and breast cancer incidence in the USSR. Int. J.
47 Epidemiol. 19: 820-824.
48
49 Greenland, S.; Morganstern, H. (1989) Ecological bias, confounding, and effect modification. Int. J. Epidemiol.
50 18: 269-274.
51
April 1995 12-334 DRAFT-DO NOT QUOTE OR CITE
-------
1 Greenland, S.; Robins, J. (1994) Invited commentary: ecologic studies—biases, misconceptions, and
2 counterexamples. Am. J. Epidemiol. 139: 747-760.
3
4 Greenland, S.; Robins, J. (1994) Accepting the limits of ecologic studies: Drs. Greenland and Robins reply to
5 Drs. Piantadosi and Cohen. Am. J. Epidemiol. 139: 769-771.
6
7 Gr^nbsk, M.; Deis, A.; S^rensen, T. I. A.; Becker, U.; Borch-Johnsen, K.; Miiller, C.; Schnohr, P.; Jensen,
8 G. (1994) Influence of sex, age, body mass index, and smoking on alcohol intake and mortality. Br.
9 Med. J. 308: 302-306.
10
11 Gross, J.; Goldsmith, J. R.; Zangwill, L.; Lerman, S. (1984) Monitoring of hospital emergency room visits as a
12 method for detecting health effects of environmental exposures. Sci. Total Environ. 32: 289-302.
13
14 Hall, C. B. (1981) Respiratory syncytial virus. In: Feigin, R. D.; Cherry, J. D., eds. Textbook of pediatric
15 infectious diseases. Philadelphia, PA: Saunders; pp. 1247-1267.
16
17 Hammerstrom, T.; Silvers, A.; Roth, N.; Lipfert, F. (1991) Statistical issues in the analysis of risks of acute air
18 pollution exposure. Palo Alto, Electric Power Research Institute.
19
20 Hasselblad, V.; Creason, J. P.; Nelson, C. J. (1976) Regression using "hockey stick" function. Research
21 Triangle Park, NC: U.S. Environmental Protection Agency, Health Effects Research Laboratory; EPA
22 report no. EPA-600/1-76-024. Available from: NTIS, Springfield, VA; PB-253 576.
23
24 Hasselblad, V.; Humble, C. G.; Graham, M. G.; Anderson, H. S. (1981) Indoor environmental determinants of
25 lung function in children. Am. Rev. Respir. Dis. 123: 479-485.
26
27 Hasselblad, V.; Eddy, D. M.; Kotchmar, D. J. (1992) Synthesis of environmental evidence: nitrogen dioxide
28 epidemiology studies. J. Air Waste Manage. Assoc. 42: 662-671.
29
30 Hastie, T.; Tibshirani, R. (1990) Generalized additive models. London, United Kingdom: Chapman and Hall.
31
32 Hatzakis, A.; Katsouyanni, K.; Kalandidi, A.; Day, N.; Trichopoulos, D. (1986) Short-term effects of air
33 pollution on mortality in Athens. Int. J. Epidemiol. 15: 73-81.
34
35 Haugh, L. D.; Box, G. E. P. (1977) Identification of dynamic regression (distributed lag) models connecting two
36 time series. JASA, Journal of the American Statistical Association 72: 121-130.
37
38 Hausman, J. A.; Ostro, B. D.; Wise, D. A. (1984) Air pollution and lost work. Cambridge, MA: National
39 Bureau of Economic Research; NBER working paper no. 1263.
40
41 Hawkins-Bell, L.; Rankin, J. T., Jr. (1994) Heat-related deaths—Philadelphia and United States, 1993-1994.
42 Morb. Mortal. Wkly. Rep. 43: 453-455.
43
44 He, Q.-C.; Lioy, P. J.; Wilson, W. E.; Chapman, R. S. (1993) Effects of air pollution on children's pulmonary
45 function in urban and suburban areas of Wuhan, People's Republic of China. Arch. Environ. Health 48:
46 382-391.
47
48 Heat Wave Preparedness Task Force. (1994) Heat Wave Preparedness Task Force status report. Presented to:
49 Lingel, M. S., Lieutenant Governor. Harrisburg, PA: Pennsylvania Emergency Management Council.
50
April 1995 12-335 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hefflin, B. J.; Jalaludin, B.; McClure, E.; Cobb, N.; Johnson, C. A.; Jecha, L.; Etzel, R. A. (1994)
2 Surveillance for dust storms and respiratory diseases in Washington State, 1991. Arch. Environ. Health
3 49: 170-174.
4
5 Hemeon, W. C. L. (1955) The estimation of health hazards from air pollution. AMA Arch. Ind. Health 11:
6 397-402.
7
8 Henderson, F. W.; Clyde, W. A., Jr.; Collier, A. M.; Denny, F. W.; Senior, R. J.; Sheaffer, C. I.; Conley,
9 W. G., Ill; Christian, R. M. (1979) The etiologic and epidemiologic spectrum of bronchiolitis in
10 pediatric practice. J. Pediatr. (St. Louis) 95: 183-190.
11
12 Henderson, F. W.; Collier, A. M.; Clyde, W. A., Jr.; Denny, F. W. (1979) Respiratory-syncytial-virus
13 infections, reinfections and immunity: a prospective, longitudinal study in young children. N. Engl. J.
14 Med. 300: 530-534.
15
16 Hill, A. B. (1965) The environment and disease: association or causation? Proc. R. Soc. Med. 58: 295-300.
17
18 Hochberg, Y.; Tamhane, A. C. (1987) Multiple comparison procedures. New York, NY: John Wiley & Sons,
19 Inc.
20
21 Hodge, W. T.; Nicodemus, M. L. (1980) Human biometeorology: a selected biography. NOAA technical
22 memorandum EDIS NCC-4.
23
24 Hodgkin, J. E.; Abbey, D. E.; Euler, G. L.; Magie, A. R. (1984) COPD prevalence in nonsmokers in high and
25 low photochemical air pollution areas. Chest 86: 830-838.
26
27 Hoek, G. (1992) Acute effects of ambient air pollution episodes on respiratory health of children [thesis].
28 Wageningen, The Netherlands: Agricultural University of Wageningen.
29
30 Hoek, G.; Brunekreef, B. (1993) Acute effects of a winter air pollution episode on pulmonary function and
31 respiratory symptoms of children. Arch. Environ. Health 48: 328-335.
32
33 Hoek, G.; Brunekreef, B. (1994) Effects of low-level winter air pollution concentrations on respiratory health of
34 Dutch children. Environ. Res. 64: 136-150.
35
36 Hoek, G.; Brunekreef, B. (1995) Effect of photochemical air pollution on acute respiratory symptoms in children.
37 Am. J. Respir. Crit. Care Med. 151: 27-32.
38
39 Holland, W. W.; Bennett, A. E.; Cameron, I. R.; Florey, C. du V.; Leeder, S. R.; Schilling, R. S. F.; Swan,
40 A. V.; Waller, R. E. (1979) Health effects of particulate pollution: reappraising the evidence. Am. J.
41 Epidemiol. 110:525-659.
42
43 Horowitz, M.; Samueloff, S. (1987) Circulation under extreme heat load. In: Dejours, P., ed.Comparative
44 physiology of environmental adaptations, v. 2. Basel, Switzerland: Karger; pp. 94-106.
45
46 Hosein, H. R.; Mitchell, C. A.; Bouhuys, A. (1977a) Evaluation of outdoor air quality in rural and urban
47 communities. Arch. Environ. Health 32: 4-13.
48
49 Hosein, H. R.; Mitchell, C. A.; Bouhuys, A. (1977b) Daily variations in air quality. Arch. Environ. Health 32:
50 14-21.
51
April 1995 12-336 DRAFT-DO NOT QUOTE OR CITE
-------
1 Hudson, D. J. (1966) Fitting segmented curves whose join points have to be estimated. J. Am. Stat. Assoc. 61:
2 1097-1129.
3
4 Humphreys, N., ed. (1985) Vital statistics: a memorial volume of selections from the reports and writings of
5 William Fair. London, United Kingdom: Sanitary Institute of Great Britain; p. 164.
6
7 Hunter, J. E.; Schmidt, F. L. (1989) Methods of meta-analysis: correcting error & bias in research findings.
8 Thousand Oaks, CA: Sage Publications, Inc.
9
10 Infante-Rivard, C. (1993) Childhood asthma and indoor environmental risk factors. Am. J. Epidemiol. 137:
11 834-844.
12
13 Intergovenmental Panel on Climate Change. (1995) Assessing the health impacts of climate change. IPCC impacts
14 assessment; ch. 2C: in press.
15
16 International Electric Research Exchange. (1981) Effects of SO2 and its derivatives on health and ecology: v. I,
17 human health. Palo Alto, CA: Electrical Power Research Institute.
18
19 Ito, K. (1990) An examination of the role of aerosol acidity in historical London, England daily mortality [Ph.D.
20 dissertation]. Syracuse, NY: New York University.
21
22 Ito, K.; Thurston, G. D.; Hayes, C.; Lippmann, M. (1993) Associations of London, England, daily mortality
23 with paniculate matter, sulfur dioxide, and acidic aerosol pollution. Arch. Environ. Health 48: 213-220.
24
25 Ito, K.; Kinney, P.; Thurston, G. D. (1995) Variations in PM10 concentrations within two metropolitan areas
26 and their implications to health effects analyses. Inhalation Toxicol.: accepted.
27
28 Jedrychowski, W.; Krzyzanowski, M. (1989) Ventilatory lung function and chronic chest symptoms among the
29 inhabitants of urban areas with various levels of acid aerosols: prospective study in Cracow. In:
30 Symposium on the health effects of acid aerosols; October 1987; Research Triangle Park, NC. Environ.
31 Health Perspect. 79: 101-107.
32
33 Jedrychowski, W.; Becher, H.; Wahrendorf, J.; Basa-Cierpialek, Z. (1990) A case-control study of lung cancer
34 with special reference to the effect of air pollution in Poland. J. Epidemiol. Commun. Health 44:
35 114-120.
36
37 Jendritzky, G. (1991) Selected questions of topical interest in human bioclimatology. Int. J. Biometeorol. 35:
38 139-150.
39
40 Jendritzky, G.; Bucher, K. (1993) Medical-meteorological fundamentals and their utilization in Germany. In:
41 Maarouf, A. R., ed.Proceedings of the weather and health workshop; November 1992; Ottawa, ON,
42 Canada. Ottawa, ON, Canada: Environment Canada; pp. 42-59.
43
44 Johnson, K. G.; Loftsgaarden, D. O.; Gideon, R. A. (1982) The effects of Mount St. Helens volcanic ash on the
45 pulmonary function of 120 elementary school children. Am. Rev. Respir. Dis. 126: 1066-1069.
46
47 Johnson, K. G.; Gideon, R. A.; Loftsgaarden, D. 0. (1990) Montana Air Pollution Study: children's health
48 effects. J. Off. Stat. 5: 391-407.
49
50 Jollis, J. G.; Ancukiewicz, M.; DeLong, E. R.; Pryor, D. B.; Muhlbaier, L. H.; Mark, D. B. (1993)
51 Discordance of databases designed for claims payment versus clinical information systems: implications
52 for outcomes research. Ann. Intern. Med. 119: 844-850.
April 1995 12-337 DRAFT-DO NOT QUOTE OR CITE
-------
1 Jones, T. S.; Liang, A. P.; Kilbourne, E. M.; Griffin, M. R.; Patriarca, P. A.; Wassilak, S. G. F.; Mullan, R.
2 J.; Herrick, R. F.; Donnell, H. D., Jr.; Choi, K.; Thacker, S. B. (1982) Morbidity and mortality
3 associated with the July 1980 heat wave in St. Louis and Kansas City, Mo. JAMA J. Am. Med. Assoc.
4 247: 3327-3331.
5
6 Kalkstein, L. S. (1988) The impacts of predicted climate change on human mortality. Publications in climatology,
7 no. 41.
8
9 Kalkstein, L. S. (1991) A new approach to evaluate the impact of climate on human mortality. Environ. Health
10 Perspect. 96: 145-150.
11
12 Kalkstein, L. S. (1993) Direct impacts in cities. Lancet 342: 1397-1399.
13
14 Kalkstein, L. S.; Davis, R. E. (1989) Weather and human mortality: an evaluation of demographic and
15 inter-regional responses in the United States. Ann. Assoc. Am. Geographers 79: 44-64.
16
17 Kalkstein, L. S.; Smoyer, K. E. (1993) The impact of climate change on human health: some inernational
18 implications. Experientia 49: 969-979.
19
20 Kalkstein, L. S.; Barthel, C. D.; Ye, H.; Smoyer, K.; Cheng, S.; Greene, J. S.; Nichols, M. C. (1994) The
21 differential impacts of weather and pollution on human mortality. Newark, DE: University of Delaware,
22 Center for Climatic Research.
23
24 Kalkstein, L. S.; Barthel, C. D.; Ye, H.; Smoyer, S.; Cheng, S.; Greene, J. S.; Nichols, M. C. (1995) The
25 differential impacts of weather and pollution on human mortality. Washington, DC: U.S. Environmental
26 Protection Agency, Office of Research and Development: in press.
27
28 Kamat, S. R.; Patil, J. D.; Gregart, J.; Dalai, N.; Deshpande, J. M.; Hardikar, P. (1992) Air pollution related
29 respiratory morbidity in central and north-eastern Bombay. J. Assoc. Physicians India 40: 588-593.
30
31 Katsouyanni, K.; Hatzakis, A.; Kalandidi, A.; Trichopoulos, D. (1990a) Short-term effects of atmospheric
32 pollution on mortality in Athens. Arch. Hellen. Med. 7: 126-132.
33
34 Katsouyanni, K.; Karakatsani, A.; Messari, I.; Touloumi, G.; Hatzakis, A.; Kalandidi, A.; Trichopoulos, D.
35 (1990b) Air pollution and cause specific mortality in Athens. J. Epidemiol. Commun. Health 44:
36 321-324.
37
38 Katsouyanni, K.; Trichopoulos, D.; Kalandidi, A.; Tomos, P.; Riboli, E. (1991) A case-control study of air
39 pollution and tobacco smoking in lung cancer among women in Athens. Prev. Med. 20: 271-278.
40
41 Katsouyanni, K.; Pantazopoulou, A.; Touloumi, G.; Tselepidaki, I.; Moustris, K.; Asimakopoulos, D.;
42 Poulopoulou, G.; Trichopoulos, D. (1993) Evidence for interaction between air pollution and high
43 temperature in the causation of excess mortality. Arch. Environ. Health 48: 235-242.
44
45 Kelsey, K. T.; Xia, F.; Bodell, W. J.; Spengler, J. D.; Christiani, D. C.; Dockery, D. W.; Liber, H. L. (1994)
46 Genotoxicity to human cells induced by air particulates isolated during the Kuwait oil fires. Environ. Res.
47 64: 18-25.
48
49 Kenline, P. A. (1962) In quest of clean air for Berlin, New Hampshire. Cincinnati, OH: U.S. Department of
50 Health, Education, and Welfare, Public Health Service; technical report no. SEC TR A62-9.
51
April 1995 12-338 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kerigan, A. T.; Goldsmith, C. H.; Pengelly, L. D. (1986) A three-year cohort study of the role of environmental
2 factors in the respiratory health of children in Hamilton, Ontario: epidemiologic survey design, methods,
3 and description of cohort. Am. Rev. Respir. Dis. 133: 987-993.
4
5 Kilbourne, E. M. (1989) Heat waves. In: Gregg, M. B., ed. The public health consequences of disasters 1989.
6 Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service; pp. 51-61.
7
8 Kilbourne, E. M. (1989) Illness due to thermal extremes. In: Last, J. M.; Wallace, R. B., eds. 13th ed.
9 Maxcy-Rosenau-Last public health and preventive medicine. Norwalk, CT: Appleton & Lange; pp.
10 491-501.
11
12 Kilbourne, E. M.; Choi, K.; Jones, T. S.; Thacker, S. B. (1982) Risk factors for heatstroke: a case-control
13 study. JAMA J. Am. Med. Assoc. 247: 3332-3336.
14
15 Kim, Y. S. (1985) Air pollution, climate, socioeconomic status and total mortality in the United States. Sci. Total
16 Environ. 42: 245-256.
17
18 Kinney, P. L.; Ozkaynak, H. (1991) Associations of daily mortality and air pollution in Los Angeles County.
19 Environ. Res. 54: 99-120.
20
21 Kinney, P. L.; Ito, K.; Thurston, G. D. (1995) A sensitivity analysis of mortality/PM10 associations in Los
22 Angeles. Inhalation Toxicol. 7: 59-69.
23
24 Kitagawa, T. (1984) Cause analysis of the Yokkaichi asthma episode in Japan. J. Air Pollut. Control Assoc. 34:
25 743-746.
26
27 Kleinbaum, D. G.; Kupper, L. L.; Morgenstern, H. (1982) Epidemiologic research: principles and quantitative
28 methods. Belmont, CA: Lifetime Learning Publications.
29
30 Kleinman, J. C.; DeGruttola, V. G.; Cohen, B. B.; Madans, J. H. (1981) Regional and urban-suburban
31 differentials in coronary heart disease mortality and risk factor prevalence. J. Chronic Dis. 34: 11-19.
32
33 Kleinman, M. T.; Bailey, R. M.; Whynot, J. D.; Anderson, K. R.; Linn, W. S.; Hackney, J. D. (1985)
34 Controlled exposure to a mixture of SO2, NO2, and particulate air pollutants: effects on human
35 pulmonary function and respiratory symptoms. Arch. Environ. Health 40: 197-201.
36
37 Klerman, L. V. (1988) School absence—a health perspective. Pediatr. Clin. N. Am. 35: 1253-1269.
38
39 Koenig, J. Q.; Pierson, W. E.; Horike, M. (1983) The effects of inhaled sulfuric acid on pulmonary function in
40 adolescent asthmatics. Am. Rev. Respir. Dis. 128: 221-225.
41
42 Koenig, J. Q.; Pierson, W. E.; Horike, M.; Frank, R. (1983) A comparison of the pulmonary effects of 0.5 ppm
43 versus 1.0 ppm sulfur dioxide plus sodium chloride droplets in asthmatic adolescents. J. Toxicol.
44 Environ. Health 11: 129-139.
45
46 Koenig, J. Q.; Larson, T. V.; Hanley, Q. S.; Rebolledo, V.; Dumler, K.; Checkoway, H.; Wang, S.-Z.; Lin,
47 D.; Pierson, W. E. (1993) Pulmonary function changes in children associated with fine particulate
48 matter. Environ. Res. 63: 26-38.
49
50 Kornguth, M. L. (1990) School illnesses: who's absent and why? Pediatr. Nurs. 16: 95-99.
51
April 1995 12-339 DRAFT-DO NOT QUOTE OR CITE
-------
1 Kotchmar, D. J.; McMullen, T.; Hasselblad, V. (1987) Adequacy of a single monitoring site for defining mean
2 outdoor concentrations of fine particles in a demarcated residential community. JAPCA 37: 377-381.
3
4 Kryzanowski, M.; Wojtyniak, B. (1982) Ten-year mortality in a sample of an adult population in relation to air
5 pollution. J. Epidemiol. Commun. Health 36: 262-268.
6
7 Krzyzanowski, M.; Wojtyniak, B. (1991) Air pollution and daily mortality in Cracow. Public Health Rev. 19:
8 73-81.
9
10 Kunst, A. E.; Looman, C. W. N.; Mackenbach, J. P. (1993) Outdoor air temperature and mortality in the
11 Netherlands: a time-series analysis. Am. J. Epidemiol. 137: 331-341.
12
13 Kupper, L. L. (1984) Effects of the use of unreliable surrogate variables on the validity of epidemiologic
14 research studies. Am. J. Epidemiol. 120: 643-648.
15
16 Landau, E. (1978) The nation's health. March.
17
18 Larsen, U. (1990a) The effects of monthly temperature fluctuations on mortality in the United States from 1921
19 to 1985. Int. J. Biometeorol. 34: 136-145.
20
21 Larsen, U. (1990b) Short-term fluctuations in death by cause, temperature, and income in the United States
22 1930-1985. Soc. Biol. 37: 172-187.
23
24 Lave, L. B.; Seskin, E. P. (1970) Air pollution and human health. The quantitative effect, with an estimate of
25 the dollar benefit of pollution abatement, is considered. Science (Washington, DC) 169: 723-733.
26
27 Lave, L. B.; Seskin, E. P. (1972) Air pollution, climate, and home heating: their effects on U.S. mortality rates.
28 Am. J. Public Health 62: 909-916.
29
30 Lave, L. B.; Seskin, E. P. (1977) Air pollution and human health. Baltimore, MD: The Johns Hopkins
31 University Press.
32
33 Lawther, P. J.; Waller, R. E.; Henderson, M. (1970) Air pollution and exacerbations of bronchitis. Thorax 25:
34 525-539.
35
36 Leaderer, B. P.; Samet, J. M. (1994) Passive smoking and adults: new evidence for adverse effects. Am. J.
37 Respir. Crit. Care Med. 150: 1216-1218.
38
39 Learner, E. E. (1978) Specification searches: ad hoc inference with nonexperimental data. New York, NY: John
40 Wiley & Sons. (Bradley, R. A.; Hunter, J. S.; Kendall, D. G.; Watson, G. S., eds. Wiley series in
41 probability and mathematical statistics).
42
43 Lebowitz, M. D. (1971) A stimulus response method for the analysis of environmental and health events related
44 in time. Environ. Lett. 2: 23-24.
45
46 Lebowitz, M. D. (1973) A comparative analysis of the stimulus-response relationship between mortality and air
47 pollution-weather. Environ. Res. 6: 106-118.
48
49 Lebowitz, M. D.; Holberg, C. J. (1987) Physiological changes in pulmonary development related to passive
50 smoking and its interaction with active smoking. Toxicol. Lett. 35: 101-105.
51
April 1995 12-340 DRAFT-DO NOT QUOTE OR CITE
-------
1 Lebowitz, M. D.; O'Rourke, M. K.; Dodge, R.; Holberg, C. J.; Gorman, G.; Hoshaw, R. W.; Pinnas, J. L.;
2 Barbee, R. A.; Sneller, M. R. (1982) The adverse health effects of biological aerosols, other aerosols,
3 and indoor microclimate on asthmatics and nonasthmatics. Environ. Int. 8: 375-380.
4
5 Lebowitz, M. D.; Quackenboss, J. J.; Krzyzanowski, M.; O'Rourke, M. K.; Hayes, C. (1992) Multipollutant
6 exposures and health responses to paniculate matter. Arch. Environ. Health 47: 71-75.
7
8 Lebret, E. (1987) Errors hi exposure variables and their role in obscuring exposure-response relations: the case
9 of indoor exposure to nitrogen dioxide and fine particles. In: Seifert, B.; Esdom, H.; Fischer, M.;
10 Rueden, H.; Wegner, J., eds. Indoor air '87: proceedings of the 4th international conference on indoor
11 air quality and climate, v. 1, volatile organic compounds, combustion gases, particles and fibres,
12 microbiological agents; August; Berlin, Federal Republic of Germany. Berlin, Federal Republic of
13 Germany: Institute for Water, Soil and Air Hygiene; pp. 288-292.
14
15 Leuenberger, P.; Schwartz, J.; Ackermann-Liebrich, U.; Blaser, K.; Bolognini, G.; Bongard, J. P.; Brandli, O.;
16 Braun, P.; Bron, C.; Brutsche, M.; Domenighetti, G.; Elsasser, S.; Guldimann, P.; Hollenstein, C.;
17 Hufschmid, P.; Karrer, W.; Keller, R.; Keller-Wossidlo, H.; Kunzli, N.; Luthi, J. C.; Martin, B. W.;
18 Medici, T.; Perruchoud, A. P.; Radaelli, A.; Schindler, C.; Shoeni, M. H.; Solari, G.; Tschopp, J. M.;
19 Villiger, B. Wuthrich, B.; Zellweger, J. P.; Zemp, E. (1994) Passive smoking exposure in adults and
20 chronic respiratory symptoms (SAPALDIA study). Am. J. Respir. Crit. Care Med. 150: 1221-1228.
21
22 Levine, R. V. (1990) The pace of life. Am. Sci. 78: 450-459.
23
24 Li, Y.; Roth, H. D. (1995) Daily mortality analysis by using different regression models in Philadelphia County,
25 1973-1990. Inhalation Toxicol. 7: 45-58.
26
27 Liang, K.-Y.; Zeger, S. L. (1986) Longitudinal data analysis using generalized linear models. Biometrika 73:
28 13-22.
29
30 Liang, K.-Y.; Zeger, S. L. (1993) Regression analysis for correlated data. Annu. Rev. Public Health 14: 43-68.
31
32 Liberates, P.; Link, B. G.; Kelsey, J. L. (1988) The measurement of social class in epidemiology. Epidemiol.
33 Rev. 10: 87-121.
34
35 Lioy, P. J.; Lippmann, M. (1986) Measurement of exposure to acidic sulfur aerosols. In: Lee, S. D.; Schneider,
36 T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols: research, risk assessment and control strategies,
37 proceedings of the second U.S.-Dutch international symposium; May 1985; Williamsburg, VA. Chelsea,
38 MI: Lewis Publishers, Inc.; pp. 743-752.
39
40 Lioy, P. J.; Vollmuth, T. A.; Lippmann, M. (1985) Persistence of peak flow decrement in children following
41 ozone exposures exceeding the national ambient air quality standard. J. Air Pollut. Control Assoc. 35:
42 1068-1071.
43
44 Lioy, P. J.; Spektor, D.; Thurston, G.; Citak, K.; Lippmann, M.; Bock, N.; Speizer, F. E.; Hayes, C. (1987)
45 The design considerations for ozone and acid aerosol exposure and health investigations: the Fairview
46 Lake summer camp—photochemical smog case study. Environ. Int. 13: 271-283.
47
48 Lipfert, F. W. (1978) The association of human mortality with air pollution: statistical analyses by region, by
49 age, and by cause of death [Ph. D. dissertation]. Cincinnati, OH: Union Graduate School. Available
50 from: University Microfilms, Ann Arbor, MI.
51
April 1995 12-341 DRAFT-DO NOT QUOTE OR CITE
-------
1 Lipfert, F. W. (1980a) Differential mortality and the environment: the challenge of multicollinearity in
2 cross-sectional studies. Energy Sys. Policy 3: 367-400.
3
4 Lipfert, F. W. (1980b) Sulfur oxides, particulates, and human mortality: synopsis of statistical correlations. J.
5 Air Pollut. Control Assoc. 30: 366-371.
6
7 Lipfert, F. W. (1984) Air pollution and mortality: specification searches using SMSA-based data. J. Environ.
8 Econ. Manage. 11: 208-243.
9
10 Lipfert, F. W. (1985) Mortality and air pollution: is there a meaningful connection? Environ. Sci. Technol. 19:
11 764-770.
12
13 Lipfert, F. W. (1993a) Community air pollution and mortality: analysis of 1980 data from US metropolitan areas.
14 I. Paniculate air pollution. Upton, NY: U.S. Department of Energy, Brookhaven National Laboratory;
15 report no. BNL 48446-R.
16
17 Lipfert, F. W. (1993b) A critical review of studies of the association between demands for hospital services and
18 air pollution. Environ. Health Perspect. 101(suppl. 2): 229-268.
19
20 Lipfert, F. W. (1994a) Air pollution and community health. New York, NY: Van Nostrand Reinhold.
21
22 Lipfert, F. W. (1994b) Application of spatial filtering techniques to cross-sectional analysis of air
23 pollution-mortality relationships. Am. J. Respir. Crit. Care Med. 149: A661.
24
25 Lipfert, F. W. (1994b) A simulation study of the effect of measurement error on the determination of empirical
26 dose-response functions [draft]. Palo Alto, CA: Electric Power Research Institute
27
28 Lipfert, F. W. (1994c) Filter artifacts associated with paniculate measurements: recent evidence and effects on
29 statistical relationships. Atmos. Environ. 28: 3233-3249.
30
31 Lipfert, F. W.; Hammerstrom, T. (1992) Temporal patterns in air pollution and hospital admissions. Environ.
32 Res. 59: 374-399.
33
34 Lipfert, F. W.; Morris, S. C. (1991) Air pollution benefit-cost assessment. Science (Washington, DC) 253: 606.
35
36 Lipfert, F. W.; Wyzga, R. E. (1992) Observational studies of air pollution health effects: are the temporal
37 patterns consistent? Presented at: 9th World clean air conference; September; Montreal, Canada, paper
38 no. IU-21A.09.
39
40 Lipfert, F. W.; Wyzga, R. E. (1994) Uncertainties in identifying "responsible" pollutants in observational
41 epidemiology studies. Presented at: Colloquium on paniculate air pollution and human mortality and
42 morbidity; January; Irvine, CA. Inhalation Toxicol.: in press.
43
44 Lipfert, F. W.; Malone, R. G.; Daum, M. L.; Mendell, N. R.; Yang, C.-C. (1988) A statistical study of the
45 macroepidemiology of air pollution and total mortality. Upton, NY: Brookhaven National Laboratory;
46 report no. BNL-52122.
47
48 Lippmann, M. (1985) Airborne acidity: estimates of exposure and human health effects. Environ. Health
49 Perspect. 63: 63-70.
50
51 Lippmann, M.; Ito, K. (1995) Separating the effects of temperature and season on daily mortality from those of
52 air pollution in London: 1965-1972. Inhalation Toxicol. 7: 85-97.
April 1995 12-342 DRAFT-DO NOT QUOTE OR CITE
-------
1 Lippmann, M.; Schlesinger, R. B. (1984) Interspecies comparisons of particle deposition and mucociliary
2 clearance in tracheobronchial airways. J. Toxicol. Environ. Health 13: 441-469.
3
4 Lippmann, M.; Yeates, D. B.; Albert, R. E. (1980) Deposition, retention, and clearance of inhaled particles. Br.
5 J. Ind. Med. 37: 337-362.
6
7 Lippmann, M.; Schlesinger, R. B.; Leikauf, G.; Spektor, D.; Albert, R. E. (1982) Effects of sulphuric acid
8 aerosols on respiratory tract airways. In: Walton, W. H., ed. Inhaled particles V: proceedings of an
9 international symposium organized by the British Occupational Hygiene Society; September 1980;
10 Cardiff, Wales. Ann. Occup. Hyg. 26: 677-690.
11
12 Lippmann, M.; Lioy, P. J.; Leikauf, G.; Green, K. B.; Baxter, D.; Morandi, M.; Pasternack, B. S.; Fife, D.;
13 Speizer, F. E. (1983) Effects of ozone on the pulmonary function of children. In: Lee, S. D.; Mustafa,
14 M. G.; Mehlman, M. A., eds. International symposium on the biomedical effects of ozone and related
15 photochemical oxidants; March 1982; Pinehurst, NC. Princeton, NJ: Princeton Scientific Publishers,
16 Inc.; pp. 423-446. (Advances in modern environmental toxicology: v. 5).
17
18 Lippmann, M.; Gearhart, J. M.; Schlesinger, R. B. (1987) Basis for a particle size-selective TLV for sulfuric
19 acid aerosols. Appl. Ind. Hyg. 2: 188-199.
20
21 Louis, T. A. (1991) Assessing, accommodating, and interpreting the influences of heterogeneity. Environ. Health
22 Perspect. 90: 215-222.
23
24 Love, G. J.; Lan, S.-P.; Shy, C. M.; Riggan, W. B. (1982) Acute respiratory illness in families exposed to
25 nitrogen dioxide ambient air pollution in Chattanooga, Tennessee. Arch. Environ. Health 37: 75-80.
26
27 Lowry, W. P. (1988) Atmospheric ecology for designers and planners. McMinniville, OR: Peavine Publications.
28
29 Lunn, J. E.; Knowelden, J.; Handyside, A. J. (1967) Patterns of respiratory illness in Sheffield infant
30 schoolchildren. Br. J. Prev. Soc. Med. 21: 7-16.
31
32 Lunn, J. E.; Knowelden, J.; Roe, J. W. (1970) Patterns of respiratory illness in Sheffield junior schoolchildren:
33 a follow-up study. Br. J. Prev. Soc. Med. 24: 223-228.
34
35 Lyon, J. L.; Mori, M.; Gao, R. (1995) Is there a causal association between excess mortality and exposure to
36 PMIO ^r pollution? Additional analyses by location, year, season and cause of death. Inhalation Toxicol.:
37 in press.
38
39 MacMahon, B.; Pugh, T. F. (1970) Epidemiology: principles and methods. Boston, MA: Little, Brown and Co.;
40 pp. 57-72.
41
42 MacVean, D. W.; Franzen, D. K.; Keefe, T. J.; Bennett, B. W. (1986) Airborne particle concentration and
43 meteorologic conditions associated with pneumonia incidence in feedlot cattle. Am. J. Vet. Res. 47:
44 2676-2682.
45
46 Macfarlane, A. (1977) Daily mortality and environment in English conurbations 1: air pollution, low
47 temperature, and influenza in Greater London. Br. J. Prev. Soc. Med. 31: 54-61.
48
49 Mackenbach, J. P.; Looman, C. W. N.; Kunst, A. E. (1993) Air pollution, lagged effects of temperature, and
50 mortality: The Netherlands 1979-87. J. Epidemiol. Commun. Health 47: 121-126.
51
52 Marmor, M. (1975) Heat wave mortality in New York City, 1949 to 1970. Arch. Environ. Health 30: 130-136.
April 1995 12-343 DRAFT-DO NOT QUOTE OR CITE
-------
1 Martin, A. E. (1964) Mortality and morbidity statistics and air pollution. Proc. R. Soc. Med. 57: 969-975.
2
3 Martin, A. E.; Bradley, W. H. (1960) Mortality, fog and atmospheric pollution: an investigation during the
4 winter of 1958-59. Mon. Bull. Minist. Health Public Health Lab. Serv. (GB) 19: 56-73.
5
6 Martinez, F. D.; Morgan, W. J.; Wright, A. L.; Holberg, C. J.; Taussig, L. M. (1988) Diminished lung
7 function as a predisposing factor for wheezing respiratory illness in infants. N. Engl. J. Med. 319:
8 1112-1117.
9
10 Martinez, B. F.; Annest, J. L.; Kilbourne, E. M.; Kirk, M. L.; Lui, K.-J.; Smith, S. M. (1989) Geographic
11 distribution of heat-related deaths among elderly persons: use of county-level dot maps for injury
12 surveillance and epidemiologic research. JAMA J. Am. Med. Assoc. 262: 2246-2250.
13
14 Martinez, F. D.; Taussig, L. M.; Morgan, W. J. (1990) Infants with upper respiratory illnesses have significant
15 reductions in maximal expiratory flow. Pediatr. Pulmonol. 9: 91-95.
16
17 Martinez, F. D.; Morgan, W. J.; Wright, A. L.; Holberg, C.; Taussig, L. M.; Group Health Medical
18 Associates. (1991) Initial airway function is a risk factor for recurrent wheezing respiratory illnesses
19 during the first three years of life. Am. Rev. Respir. Dis. 143: 312-316.
20
21 Martinez, F. D.; Wright, A. L.; Taussig, L. M.; Holberg, C. J.; Halonen, M.; Morgan, W. J.; Group Health
22 Medical Associates. (1995) Asthma and wheezing in the first six years of life. N. Engl. J. Med. 332:
23 133-138.
24
25 Mazumdar, S.; Sussman, N. (1983) Relationships of air pollution to health: results from the Pittsburgh study.
26 Arch. Environ. Health 38: 17-24.
27
28 Mazumdar, S.; Schimmel, H.; Higgins, I. (1981) Daily mortality, smoke and SO2 in London, England 1959 to
29 1972. In: Frederick, E. R., ed. A specialty conference on: the proposed SOX and paniculate standard;
30 September 1980; Atlanta, GA. Pittsburgh, PA: Air Pollution Control Association; pp. 219-239.
31
32 Mazumdar, S.; Schimmel, H.; Higgins, I. T. T. (1982) Relation of daily mortality to air pollution: an analysis of
33 14 London winters, 1958/59-1971/72. Arch. Environ. Health 37: 213-220.
34
35 McConnochie, K. M.; Hall, C. B.; Barker, W. H. (1988) Lower respiratory tract illness in the first two years of
36 life: epidemiologic patterns and costs in a suburban pediatric practice. Am. J. Public Health 78: 34-39.
37
38 McCullagh, D.; Nelder, J. A. (1983) Generalized linear models. London, United Kingdom: Chapman and Hall.
39
40 McCullagh, P.; Nelder, J. A. (1989) Generalized linear models. 2nd ed. London, United Kingdom: Chapman
41 and Hall.
42
43 McPherson, K.; Wennberg, J. E.; Hovind, O. B.; Clifford, P. (1982) Small-area variations in the use of
44 common surgical procedures: an international comparison of New England, England, and Norway. N.
45 Engl. J. Med. 307: 1310-1314.
46
47 Mendelsohn, R.; Orcutt, G. (1979) An empirical analysis of air pollution dose-response curves. J. Environ.
48 Econ. Manage. 6: 85-106.
49
50 Merchandani, H.; Pace, P.; Hameli, A. Z.; Pressler, R.; Tobin, J. G.; Davis, J. H.; Melton, V. (1993)
51 Heat-related deaths—United States, 1993. Morb. Mortal. Wkly. Rep. 42: 558-560.
52
April 1995 12-344 DRAFT-DO NOT QUOTE OR CITE
-------
1 Meyer, W. B. (1991) Urban heat island and urban health: early American perspectives. Prof. Geograph. 43:
2 38-48.
3
4 Miettinen, O. S.; Cook, E. F. (1981) Confounding: essence and detection. Am. J. Epidemiol. 114: 593-603.
5
6 Mills, P. K.; Abbey, D.; Beeson, W. L.; Petersen, F. (1991) Ambient air pollution and cancer in California
7 Seventh-day Adventists. Arch. Environ. Health 46: 271-280.
8
9 Mitchell, C. A.; Schilling, R. S. F.; Bouhuys, A. (1976) Community studies of lung disease in Connecticut:
10 organization and methods. Am. J. Epidemiol. 103: 212-225.
11
12 Moolgavkar, S. H. (1994) Air pollution and mortality [letter]. N. Engl. J. Med. 330: 1237-1238.
13
14 Moolgavkar, S. H.; Luebeck, E. G.; Hall, T. A.; Anderson, E. L. (1995a) Paniculate air pollution, sulfur
15 dioxide, and daily mortality: a reanalysis of the Steubenville data. Inhalation Toxicol. 7: 35-44.
16
17 Moolgavkar, S. H.; Luebeck, E. G.; Hall, T. A.; Anderson, E. L. (1995b) Air pollution and daily mortality in
18 Philadelphia. J. Exposure Anal. Environ. Epidemiol.: submitted
19
20 Morgan, W. J.; Taussig, L. M. (1984) The chronic bronchitis complex in children. Pediatr. Clin. N. Am. 31:
21 851-864.
22
23 Morgenstern, H. (1982) Uses of ecologic analysis in epidemiologic research. Am. J. Public Health 72:
24 1336-1344.
25
26 Morris, S. C.; Shapiro, M. A.; Waller, J. H. (1976) Adult mortality in two communities with widely different
27 air pollution levels. Arch. Environ. Health 31: 248-254.
28
29 Mueller, P. K.; Hidy, G. M. (1983) The sulfate regional experiment (SURE): report of findings. Palo Alto, CA:
30 Electric Power Research Institute; EPRI report no. EA-1901. 3v.
31
32 Murphy, T. F.; Henderson, F. W.; Clyde, W. A., Jr.; Collier, A. M.; Denny, F. W. (1981) Pneumonia: an
33 eleven-year study in a pediatric practice. Am. J. Epidemiol. 113: 12-21.
34
35 National Air Pollution Control Administration. (1969) Air quality criteria for paniculate matter. Washington, DC:
36 U.S. Department of Health, Education, and Welfare, Public Health Service; NAPCA publication no.
37 AP-49. Available from: NTIS, Springfield, VA; PB-190251/BA.
38
39 National Center for Health Statistics. (1978) Standardized micro-data tape transcripts. Washington, DC: U.S.
40 Department of Health, Education, and Welfare.
41
42 National Center for Health Statistics. (1993) National Hospital Discharge Survey: annual summary, 1991.
43 Hyattsville, MD: U.S. Department of Health and Human Services, Public Health Service, Centers for
44 Disease Control; DHHS publication no. (PHS)93-1775. (Series 13: Data from the National Health
45 Survey, no. 114).
46
47 National Center for Health Statistics. (1994) Detailed diagnoses and procedures, National Hospital Discharge
48 Survey, 1992. Hyattsville, MD: U.S. Department of Health and Human Services, Public Health Service,
49 Centers for Disease Control and Prevention; DHHS publication no. (PHS) 94-1779. (Series 13: Data
50 from the National Health Survey, no. 118).
51
April 1995 12-345 DRAFT-DO NOT QUOTE OR CITE
-------
1 National Center for Health Statistics. (1994) Mortality surveillance system charts. Mon. Vital Stat. Rep. 43(5):
2 6-7, 1994.
3
4 National Institute for Occupational Safety and Health. (1990) NIOSH pocket guide to chemical hazards.
5 Cincinnati, OH: U.S. Department of Health and Human Services, Public Health Service; pp. 194-197;
6 report no. DHHS (NIOSH) publication no. 90-117.
7
8 National Institutes of Health. (1991) Guidelines for the diagnosis and management of asthma. Bethesda, MD:
9 U.S. Department of Health and Human Services, National Heart, Lung, and Blood Institute, National
10 Asthma Education Program; publication no. 91-3042.
11
12 Neas, L. M.; Dockery, D. W.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr.; Speizer, F. E. (1994)
13 Concentration of indoor paniculate matter as a determinant of respiratory health in children. Am. J.
14 Epidemiol. 139: 1088-1099.
15
16 Neas, L. M.; Dockery, D. W.; Koutrakis, P.; Tollerud, D. J.; Speizer, F. E. (1995) The association of ambient
17 air pollution with twice daily peak expiratory flow rate measurements in children. Am. J. Epidemiol.
18 141: 111-122.
19
20 Osborne, M. L.; Vollmer, W. M.; Buist, A. S. (1992) Diagnostic accuracy of asthma within a health
21 maintenance organization. J. Clin. Epidemiol. 45: 403-411.
22
23 Ostro, B. D. (1983) The effects of air pollution on work loss and morbidity. J. Environ. Econ. Manage. 10:
24 371-382.
25
26 Ostro, B. (1984) A search for a threshold in the relationship of air pollution to mortality: a reanalysis of data on
27 London winters. Environ. Health Perspect. 58: 397-399.
28
29 Ostro, B. D. (1987) Air pollution and morbidity revisited: a specification test. J. Environ. Econ. Manage. 14:
30 87-98.
31
32 Ostro, B. D. (1989) Estimating the risks of smoking, air pollution, and passive smoke on acute respiratory
33 conditions. Risk Anal. 9: 189-196.
34
35 Ostro, B. D. (1990) Associations between morbidity and alternative measures of paniculate matter. Risk Anal.
36 10: 421-427.
37
38 Ostro, B. (1993) The association of air pollution and mortality: examining the case for inference. Arch. Environ.
39 Health 48: 336-342.
40
41 Ostro, B. D.; Rothschild, S. (1989) Air pollution and acute respiratory morbidity: an observational study of
42 multiple pollutants. Environ. Res. 50: 238-247.
43
44 Ostro, B.; Lipsett, M.; Wiener, M.; Seiner, J. C. (1989) A panel study of the effect of acid aerosols on
45 asthmatics. Presented at: 82nd annual meeting and exhibition of the Air and Waste Management
46 Association; June; Anaheim, CA. Pittsburgh, PA: Air & Waste Management Association; paper no.
47 89-94.1.
48
49 Ostro, B. D.; Lipsett, M. J.; Wiener, M. B.; Seiner, J. C. (1991) Asthmatic responses to airborne acid aerosols.
50 Am. J. Public Health 81: 694-702.
51
April 1995 12-346 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ostro, B. D.; Lipsett, M. J.; Mann, J. K.; Krupnick, A.; Harrington, W. (1993) Air pollution and respiratory
2 morbidity among adults in Southern California. Am. J. Epidemiol. 137: 691-700.
3
4 Ostro, B. D.; Lipsett, M. J.; Mann, J. K.; Wiener, M. B.; Seiner, J. (1994) Indoor air pollution and asthma:
5 results from a panel study. Am. J. Respir. Crit. Care Med. 149: 1400-1406.
6
7 Ostro, B.; Sanchez, J. M.; Aranda, C.; Eskeland, G. S. (1995a) Air pollution and mortality: results of a study of
8 Santiago, Chile. J. Exposure Anal. Environ. Epidemiol.: submitted.
9
10 Ostro, B. D.; Lipsett, M. J.; Mann, J. K.; Braxton-Owens, H.; White, M. C. (1995b) Air pollution and asthma
11 exacerbations among African-American children in Los Angeles. Inhalation Toxicol.: in press.
12
13 Ozkaynak, H.; Spengler, J. D. (1985) Analysis of health effects resulting from population exposures to acid
14 precipitation precursors. Environ. Health Perspect. 63: 45-55.
15 Ozkaynak, H.; Thurston, G. D. (1987) Associations between 1980 U.S. mortality rates and alternative measures
16 of airborne particle concentration. Risk Anal. 7: 449-461.
17
18 Ozkaynak, H.; Schatz, A. D.; Thurston, G. D.; Isaacs, R. G.; Husar, R. B. (1985) Relationships between
19 aerosol extinction coefficients derived from airport visual range observations and alternative measures of
20 airborne particle mass. J. Air Pollut. Control Assoc. 35: 1176-1185.
21
22 Ozkaynak, H.; Spengler, J. D.; Garsd, A.; Thurston, G. D. (1986) Assessment of population health risks
23 resulting from exposures to airborne particles. In: Lee, S. D.; Schneider, T.; Grant, L. D.; Verkerk, P.
24 J., eds. Aerosols: research, risk assessment and control strategies, proceedings of the second U.S.-Dutch
25 international symposium; May 1985; Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp.
26 1067-1080.
27
28 Ozkaynak, H.; Xue, J.; Severance, P.; Burnett, R.; Raizenne, M. (1994) Associations between daily mortality,
29 ozone, and paniculate air pollution in Toronto, Canada. Presented at: Colloquim on paniculate air
30 pollution and human mortality and morbidity; January; Irvine, CA. Irvine, CA: University of California
31 at Irvine, Air Pollution Health Effects Laboratory; report no. 94-02.
32
33 Paffenbarger, R. S., Jr.; Hyde, R. T.; Wing, A. L.; Lee, I.-M.; Jung, D. L.; Kampert, J. B. (1993) The
34 association of changes in physical-activity level and other lifestyle characteristics with mortality among
35 men. N. Engl. J. Med. 328: 538-545.
36
37 Parcel, G. S.; Oilman, S. C.; Nader, P. R.; Bunce, H. (1979) A comparison of absentee rates of elementary
38 schoolchildren with asthma and nonasthmatic schoolmates. Pediatrics 64: 878-881.
39
40 Patrick, C. H. (1980) Health and migration of the elderly. Res. Aging 2: 233-241.
41
42 Pearlman, M. E.; Finklea, J. F.; Creason, J. P.; Shy, C. M.; Young, M. M.; Horton, R. J. M. (1971) Nitrogen
43 dioxide and lower respiratory illness. Pediatrics 47: 391-398.
44
45 Penna, M. L. F.; Duchiade, M. P. (1991) Contamination del aire y mortalidad infantil por neumonia [Air
46 pollution and infant mortality due to pneumonia]. Bol. Of. Sank. Panam. 110: 199-207.
47
48 Perry, G. B.; Chai, H.; Dickey, D. W.; Jones, R. H.; Kinsman, R. A.; Morrill, C. G.; Spector, S. L.; Weiser,
49 P. C. (1983) Effects of paniculate air pollution on asthmatics. Am. J. Public Health 73: 50-56.
50
51
April 1995 12-347 DRAFT-DO NOT QUOTE OR CITE
-------
1 Peto, R.; Speizer, F. E.; Cochrane, A. L.; Moore, F.; Fletcher, C. M.; Tinker, C. M.; Higgins, I. T. T.;
2 Gray, R. G.; Richards, S. M.; Gilliland, J.; Norman-Smith, B. (1983) The relevance in adults of
3 air-flow obstruction, but not of mucus hypersecretion, to mortality from chronic lung disease: results
4 from 20 years of prospective observation. Am. Rev. Respir. Dis. 128: 491-500.
5
6 Piantadosi, S.; Byar, D. P.; Green, S. B. (1988) The ecological fallacy. Am. J. Epidemiol. 127: 893-904.
7
8 Plagiannakos, T.; Parker, J. (1988) An assessment of air pollution effects on human health in Ontario. Toronto,
9 ON, Canada: Ontario Hydro, Economics and Forecast Division, Energy Economics Section; report no.
10 706.01 (#260).
11
12 Ponka, A. (1991) Asthma and low level air pollution in Helsinki. Arch. Environ. Health 46: 262-270.
13
14 Ponka, A.; Virtanen, M. (1994) Chronic bronchitis, emphysema, and low-level air pollution in Helsinki,
15 1987-1989. Environ. Res. 65: 207-217.
16
17 Pope, C. A., III. (1991) Respiratory hospital admissions associated with PM10 pollution in Utah, Salt Lake, and
18 Cache Valleys. Arch. Environ. Health 46: 90-97.
19
20 Pope, C. A., III. (1994) Particulate pollution and mortality in Utah valley. Prepared for: Critical evaluation
21 workshop on paniculate matter—mortality epidemiology studies; November; Raleigh, NC. Provo, UT:
22 Brigham Young University.
23
24 Pope, C. A., III.; Dockery, D. W. (1992) Acute health effects of PM10 pollution on symptomatic and
25 asymptomatic children. Am. Rev. Respir. Dis. 145: 1123-1128.
26
27 Pope, C. A., Ill; Kanner, R. E. (1993) Acute effects of PM10 pollution on pulmonary function of smokers with
28 mild to moderate chronic obstructive pulmonary disease. Am. Rev. Respir. Dis. 147: 1336-1340.
29
30 Pope, C. A., Ill; Xu, X. (1993) Passive cigarette smoke, coal heating, and respiratory symptoms of nonsmoking
31 women in China. Environ. Health Perspect. 101: 314-316.
32
33 Pope, C. A., Ill; Dockery, D. W.; Spengler, J. D.; Raizenne, M. E. (1991) Respiratory health and PM10
34 pollution: a daily time series analysis. Am. Rev. Respir. Dis. 144: 668-674.
35
36 Pope, C. A., Ill; Schwartz, J.; Ransom, M. R. (1992) Daily mortality and PM10 pollution in Utah valley. Arch.
37 Environ. Health 47: 211-217.
38
39 Pope, C. A., Ill; Dockery, D. W.; Schwartz, J. (1995) Review of epidemiological evidence of health effects of
40 paniculate air pollution. Inhalation Toxicol. 7: 1-18.
41
42 Pope, C. A., Ill; Thun, M. J.; Namboodiri, M. M.; Dockery, D. W.; Evans, J. S.; Speizer, F. E.; Heath, C.
43 W., Jr. (1995) Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults.
44 Am. J. Respir. Crit. Care Med. 151: 669-674.
45
46 Prentice, R. L.; Thomas, D. (1993) Methodologic research needs in environmental epidemiology: data analysis.
47 Environ. Health Perspect. 101(suppl. 4): 39-48.
48
49 Quackenboss, J. J.; Krzyzanowski, M.; Lebowitz, M. D. (1991) Exposure assessment approaches to evaluate
50 respiratory health effects of paniculate matter and nitrogen dioxide. J. Exposure Anal. Environ.
51 Epidemiol. 1: 83-107.
52
April 1995 12-348 DRAFT-DO NOT QUOTE OR CITE
-------
1 Quandt, R. E. (1958) The estimation of the parameters of a linear regression system obeying two separate
2 regimes. J. Am. Stat. Assoc. 53: 873-880.
3
4 Queiros, M.; Bonito-Vitor, A.; Costa-Pereira, A.; Costa Maia, J. (1990) Childhood asthma and outdoor air
5 pollution in Oporto area. Allergol. Immunopathol. 18: 291-295.
6
7 Raizenne, M. E.; Hargreave, F.; Sears, M.; Spengler, J.; Stern, B.; Burnett, R. (1987) Exercise and lung
8 function responses during an air pollution episode in young females with airway hyperresponsiveness to
9 methacholine. Am. Rev. Respir. Dis. 135(suppl.): A343.
10
11 Raizenne, M.; Burnett, R.; Stern, B.; Meranger, J. C. (1987) Transported air pollutants and respiratory health in
12 two Canadian communities. Chest 91: 314.
13
14 Raizenne, M. E.; Burnett, R. T.; Stern, B.; Franklin, C. A.; Spengler, J. D. (1989) Acute lung function
15 responses to ambient acid aerosol exposures in children. Environ. Health Perspect. 79: 179-185.
16
17 Raizenne, M.; Neas, L.; Damokosh, A.; Dockery, D.; Spengler, J.; Koutrakis, P.; Ware, J.; Speizer, F. (1993)
18 Health effects of acid aerosols on North American children: pulmonary function. Am. Rev. Respir. Dis.
19 147: A635.
20
21 Ramlow, J. M.; Kuller, L. H. (1990) Effects of the summer heat wave of 1988 on daily mortality in Allegheny
22 County, PA. Public Health Rep. 105: 283-289.
23
24 Rango, N. (1984) Exposure-related hypothermia mortality in the United States, 1970-79. Am. J. Public Health
25 74: 1159-1160.
26
27 Ransom, M. R.; Pope, C. A., III. (1992) Elementary school absences and PM10 pollution in Utah Valley.
28 Environ. Res. 58: 204-219.
29
30 Redline, S.; Gold, D. (1994) Challenges in interpreting gender differences in asthma. Am. J. Respir. Crit. Care
31 Med. 150: 1219-1221.
32
33 Ricci, P. F.; Wyzga, R. E. (1983) An overview of cross-sectional studies of mortality and air pollution and
34 related statistical issues. Environ. Int. 9: 177-194.
35
36 Richards, J. H.; Marriott, C. (1974) Effect of relative humidity on the rheologic properties of bronchial mucus.
37 Am. Rev. Respir. Dis. 109: 484-486.
38
39 Richardson, S.; Stucker, I.; Hemon, D. (1987) Comparison of relative risks obtained in ecological and individual
40 studies: some methodological considerations. Int. J. Epidemiol. 16: 111-120.
41
42 Richardson, S. L.; Renz, K. K.; Vogel, T. T.; Graham, J. E., Jr.; Kaufman, J. (1991) Small area analysis
43 shows differences in utilization. Qual. Assur. Util. Rev. 6: 91-94.
44
45 Robertson, J. M.; Ingalls, T. H. (1989) A case-control study of circulatory, malignant, and respiratory morbidity
46 in carbon black workers in the United States. Am. Ind. Hyg. Assoc. J. 50: 510-515.
47
48 Robinson, W. S. (1950) Ecological correlations and the behavior of individuals. Am. Soc. Rev. 15: 351-357.
49
50 Roemer, W.; Hoek, G.; Brunekreef, B. (1993) Effect of ambient winter air pollution on respiratory health of
51 children with chronic respiratory symptoms. Am. Rev. Respir. Dis. 147: 118-124.
52
April 1995 12-349 DRAFT-DO NOT QUOTE OR CITE
-------
1 Rogot, E.; Blackwelder, W. C. (1970) Associations of cardiovascular mortality with weather in Memphis,
2 Tennessee. Public Health Rep. 85: 25-39.
3
4 Rogot, E.; Padgett, S. J. (1976) Associations of coronary and stroke mortality with temperature and snowfall in
5 selected areas of the United States, 1962-1966. Am. J. Epidemiol. 103: 565-575.
6
7 Rogot, E.; Fabsitz, R.; Feinleib, M. (1976) Daily variation in USA mortality. Am. J. Epidemiol. 103: 198-211.
8
9 Rogot, E.; Sorlie, P. D.; Johnson, N. J.; Schmitt, C. (1992a) A mortality study of 1.3 million persons.
10 Bethesda, MD: National Heart, Lung, and Blood Institute; NIH publication no. 92-3297.
11
12 Rogot, E.; Sorlie, P. D.; Backlund, E. (1992b) Air-conditioning and mortality in hot weather. Am. J. Epidemiol.
13 136: 106-116.
14
15 Romieu, I.; Lugo, M. C.; Velasco, S. R.; Sanchez, S.; Meneses, F.; Hernandez, M. (1992) Air pollution and
16 school absenteeism among children in Mexico City. Am. J. Epidemiol. 136: 1524-1531.
17
18 Rosenthal, R. (1979) The "file drawer problem" and tolerance for null results. Psychol. Bull. 86: 638-641.
19
20 Rothman, K. J. (1986) Modern epidemiology. Boston, MA: Little, Brown and Co.
21
22 Rothman, K. J. (1993) Methodologic frontiers in environmental epidemiology. Environ. Health Perspect.
23 101(suppl. 4): 19-21.
24
25 Rothman, N.; Ford, D. P.; Baser, M. E.; Hansen, J. A.; O'Toole, T.; Tockman, M. S.; Strickland, P. T.
26 (1991) Pulmonary function and respiratory symptoms in wildland firefighters. J. Occup. Med. 33:
27 1163-1167.
28
29 Rotton, J. (1983) Angry, sad, happy? Blame the weather. U.S. News and World Report 95: 52-53.
30
31 Rumford, J. (1961) Mortality studies in relation to air pollution. Am. J. Public Health 51: 165-173.
32
33 Ruscone, T. G.; Crosignani, P.; Micheletti, T.; Sala, L.; Santomauro, D. (1985) Meteorological influences on
34 myocardial infarction in the metropolitan area of Milan. Int. J. Cardiol. 9: 75-80.
35
36 Sakamoto-Momiyama, M. (1977) Seasonally in human mortality. Tokyo, Japan: University of Tokyo Press.
37
38 Saldiva, P. H. N.; Bohm, G. M. (1994)
39
40 Saldiva, P. H. N.; Lichtenfels, A. J. F. C.; Paiva, P. S. O.; Barone, I. A.; Martins, M. A.; Massad, E.;
41 Pereira, J. C. R.; Xavier, V. P.; Singer, J. M.; Bohm, G. M. (1994) Association between air pollution
42 and mortality due to respiratory diseases in children in Sao Paulo, Brazil: a preliminary report. Environ.
43 Res. 65: 218-225.
44
45 Samet, J. M. (1978) A historical and epidemiologic perspective on respiratory symptoms questionnaires. Am. J.
46 Epidemiol. 108: 435-446.
47
48 Samet, J. M. (1994) Learning about air pollution and asthma. Am. J. Respir. Crit. Care Med. 149: 1398-1399.
49
50 Samet, J. M.; Speizer, F. E. (1993) Assessment of health effects in epidemiologic studies of air pollution.
51 Environ. Health Perspect. 101(suppl. 4): 149-154.
52
April 1995 12-350 DRAFT-DO NOT QUOTE OR CITE
-------
1 Samet, J. M.; Utell, M. J. (1990) The risk of nitrogen dioxide: what have we learned from epidemiological and
2 clinical studies? Toxicol. Ind. Health 6: 247-262.
3
4 Samet, J. M.; Tager, I. B.; Speizer, F. E. (1983) The relationship between respiratory illness in childhood and
5 chronic air-flow obstruction in adulthood. Am. Rev. Respir. Dis. 127: 508-523.
6
7 Sandvik, L.; Erikssen, J.; Thaulow, E.; Erikssen, G.; Mundal, R.; Rodahl, K. (1993) Physical fitness as a
8 predictor of mortality among healthy, middle-aged Norwegian men. N. Engl. J. Med. 328: 533-537.
9
10 Schafer, D. W. (1987) Covariate measurement error in generalized linear models. Biometrika 74: 385-391.
11
12 Schenker, M. (1993) Air pollution and mortality. New Engl. J. Med. 329: 1807-1808.
13
14 Schenker, M. B.; Samet, J. M.; Speizer, F. E. (1983a) Risk factors for childhood respiratory disease: the effect
15 of host factors and home environmental exposures. Am. Rev. Respir. Dis. 128: 1038-1043.
16
17 Schenker, M. B.; Speizer, F. E.; Samet, J. M.; Gruhl, J.; Batterman, S. (1983b) Health effects of air pollution
18 due to coal combustion in the Chestnut Ridge region of Pennsylvania: results of cross-sectional analysis in
19 adults. Arch. Environ. Health 38: 325-330.
20
21 Scheraga, J. S.; Sussman, F. (1995) Preliminary assessment of the benefits to the U.S. of avoiding, or adapting
22 to, climate change. U.S. Environmental Protection Agency, Climate Change Division: in press.
23
24 Schimmel, H. (1978) Evidence for possible acute health effects of ambient air pollution from time series analysis:
25 methodological questions and some new results based on New York City daily mortality, 1963-1976.
26 Bull. N. Y. Acad. Med. 54: 1052-1108.
27
28 Schneider, S. H., ed. (1995) Encyclopedia of climate and weather. New York, NY: Oxford University Press; in
29 press.
30
31 Schrenk, H. H.; Heimann, H.; Clayton, G. D.; Gafafer, W. M.; Wexler, H. (1949) Air pollution in Donora,
32 PA. Epidemiology of the unusual smog episode of October 1948: preliminary report. Washington, DC:
33 Public Health Service; Public Health Service bulletin no. 306.
34
35 Schwartz, J. (1989) Lung function and chronic exposure to air pollution: a cross-sectional analysis of NHANES
36 II. Environ. Res. 50: 309-321.
37
38 Schwartz, J. (1991a) Paniculate air pollution and daily mortality in Detroit. Environ. Res. 56: 204-213.
39
40 Schwartz, J. (1991b) The first author replies [letter re Fleisher and Nayeri (1991)]. Am. J. Epidemiol. 133:
41 632-633.
42
43 Schwartz, J. (1991/92) Paniculate air pollution and daily mortality: a synthesis. Public Health Rev. 19: 39-60.
44
45 Schwartz, J. (1992) Air pollution and the duration of acute respiratory symptoms. Arch. Environ. Health 47:
46 116-122.
47
48 Schwartz, J. (1993a) Air pollution and daily mortality in Birmingham, Alabama. Am. J. Epidemiol. 137:
49 1136-1147.
50
51 Schwartz, J. (1993b) Paniculate air pollution and chronic respiratory disease. Environ. Res. 62: 7-13.
52
April 1995 12-351 DRAFT-DO NOT QUOTE OR CITE
-------
1 Schwartz, J. (1994a) Total suspended paniculate matter and daily mortality in Cincinnati, Ohio. Environ. Health
2 Perspect. 102: 186-189.
3
4 Schwartz, J. (1994b) Air pollution and daily mortality: a review and meta analysis. Environ. Res. 64: 36-52.
5
6 Schwartz, J. (1994c) What are people dying of on high air pollution days? Environ. Res. 64: 26-35.
7
8 Schwartz, J. (1994d) Air pollution and hospital admissions for the elderly in Detroit, Michigan. Am. J. Respir.
9 Crit. Care Med. 150: 648-655.
10
11 Schwartz, J. (1994e) Air pollution and hospital admissions for the elderly in Birmingham, Alabama. Am. J.
12 Epidemiol. 139: 589-598.
13
14 Schwartz, J. (1994f) PM10, ozone, and hospital admissions for the elderly in Minneapolis, MN. Arch. Environ.
15 Health 49: 366-374.
16
17 Schwartz, J. (1994g) Nonparametric smoothing in the analysis of air pollution and respiratory illness. Can. J.
18 Stat. 22: 1-17.
19
20 Schwartz, J. (1994h) The use of generalized additive models in epidemiology. Proc. Int. Conf. Biometric Soc.
21 17: 55-80.
22
23 Schwartz, J. (1995a) Short term fluctuations in air pollution and hospital admissions of the elderly for respiratory
24 disease. Thorax: in press.
25
26 Schwartz, J. (1995b) Health effects of air pollution from traffic: ozone and particulate matter. In: Fletcher, T.,
27 ed. Health at the crossroads: transportation policy and urban health, proceedings of the fifth annual public
28 health forum of the London School of Hygiene and Tropical Medicine; April; London, United Kingdom.
29 New York, NY: John Wiley & Sons, Inc.; in preparation.
30
31 Schwartz, J.; Dockery, D. W. (1992a) Increased mortality in Philadelphia associated with daily air pollution
32 concentrations. Am. Rev. Respir. Dis. 145: 600-604.
33
34 Schwartz, J.; Dockery, D. W. (1992b) Particulate air pollution and daily mortality in Steubenville, Ohio. Am. J.
35 Epidemiol. 135: 12-19.
36
37 Schwartz, J.; Morris, R. (1995) Air pollution and cardiovascular hospital admissions. Am. J. Epidemiol.: in
38 press.
39
40 Schwartz, J.; Marcus, A. H. (1986) Statistical reanalyses of data relating mortality to air pollution during London
41 winters 1958-1972. Washington, DC: U.S. Environmental Protection Agency, Office of Policy, Planning
42 and Evaluation.
43
44 Schwartz, J.; Marcus, A. (1990) Mortality and air pollution in London: a time series analysis. Am. J. Epidemiol.
45 131: 185-194.
46
47 Schwartz, J.; Dockery, D.; Ware, J.; Spengler, J.; Wypij, D.; Koutrakis, P.; Speizer, F.; Ferris, B., Jr. (1989)
48 Acute effects of acid aerosols on respiratory symptom reporting in children. Presented at: 82nd annual
49 meeting and exhibition of the Air and Waste Management Association; June; Anaheim, CA. Pittsburgh,
50 PA: Air and Waste Management Association; paper no. 89-92.1.
51
April 1995 12-352 DRAFT-DO NOT QUOTE OR CITE
-------
1 Schwartz, J.; Gold, D.; Dockery, D. W.; Weiss, S. T.; Speizer, F. E. (1990) Predictors of asthma and
2 persistent wheeze in a national sample of children in the United States: association with social class,
3 perinatal events, and race. Am. Rev. Respir. Dis. 142: 555-562.
4
5 Schwartz, J.; Spix, C.; Wichmann, H. E.; Malin, E. (1991a) Air pollution and acute respiratory illness in five
6 German communities. Environ. Res. 56: 1-14.
7
8 Schwartz, J.; Wypij, D.; Dockery, D.; Ware, J.; Zeger, S.; Spengler, J.; Ferris, B., Jr. (1991b) Daily diaries of
9 respiratory symptoms and air pollution: methodological issues and results. Environ. Health Perspect. 90:
10 181-187.
11
12 Schwartz, J.; Slater, D.; Larson, T. V.; Pierson, W. E.; Koenig, J. Q. (1993) Paniculate air pollution and
13 hospital emergency room visits for asthma in Seattle. Am. Rev. Respir. Dis. 147: 826-831.
14
15 Schwartz, J.; Dockery, D. W.; Neas, L. M.; Wypij, D.; Ware, J. H.; Spengler, J. D.; Koutrakis, P.; Speizer,
16 F. E.; Ferris, B. G., Jr. (1994) Acute effects of summer air pollution on respiratory symptom reporting
17 in children. Am. J. Respir. Crit. Care Med. 150: 1234-1242.
18
19 Schwarz, G. (1978) Estimation of the dimension of a model. Ann. Stat. 6: 461-464.
20
21 Selvin, S.; Merrill, D.; Wong, L.; Sacks, S. T. (1984) Ecologic regression analysis and the study of the
22 influence of air quality on mortality. Environ. Health Perspect. 54: 333-340.
23
24 Serfling, R. E. (1963) Methods for current statistical analysis of excess pneumonia—influenza deaths. Public
25 Health Rep. 78: 494.
26
27 Shannon, J. D. (1981) A model of regional long-term average sulfur atmospheric pollution, surface removal, and
28 net horizontal flux. Atmos. Environ. 15: 689-701.
29
30 Shapiro, S. (1994) Meta-analysis/shmeta-analysis. Am. J. Epidemiol. 140: 771-778.
31
32 Shumway, R. H. (1988) Applied statistical time series analysis. Englewood Cliffs, NJ: Prentice Hall, Inc.
33
34 Shumway, R. H.; Tai, R. Y.; Tai, L. P.; Pawitan, Y. (1983) Statistical analysis of daily London mortality and
35 associated weather and pollution effects. Sacramento, CA: California Air Resources Board; contract no.
36 Al-154-33.
37
38 Shumway, R. H.; Azari, A. S.; Pawitan, Y. (1988) Modeling mortality fluctuations in Los Angeles as functions
39 of pollution and weather effects. Environ. Res. 45: 224-241.
40
41 Shusterman, D.; Kaplan, J. Z.; Canabarro, C. (1993) Immediate health effects of an urban wildfire. West. J.
42 Med. 158: 133-138.
43
44 Shy, C. M.; Creason, J. P.; Pearlman, M. E.; McClain, K. E.; Benson, F. B.; Young, M. M. (1970) The
45 Chattanooga school children study: effects of community exposure to nitrogen dioxide. 1. Methods,
46 description of pollutant exposure, and results of ventilatory function testing. J. Air Pollut. Control Assoc.
47 20: 539-545.
48
49 Shy, C. M.; Kleinbaum, D. G.; Morgenstern, H. (1978) The effect of misclassification of exposure status in
50 epidemiological studies of air pollution health effects. Bull. N. Y. Acad. Med. 54: 1155-1165.
51
April 1995 12-353 DRAFT-DO NOT QUOTE OR CITE
-------
1 Silverman, F.; Hosein, H. R.; Corey, P.; Holton, S.; Tarlo, S. M. (1992) Effects of paniculate matter exposure
2 and medication use on asthmatics. Arch. Environ. Health 47: 51-56.
3
4 Skobeloff, E. M.; Spivey, W. H.; St. Clair, S. S.; Schoffstall, J. M. (1992) The influence of age and sex on
5 asthma admissions. JAMA J. Am. Med. Assoc. 268: 3437-3440.
6
7 Smith, V. K. (1975) Mortality-air pollution relationships: a comment. JASA, Journal of the American Statistical
8 Association 70: 341-343.
9
10 Smith, J. B.; Tirpak, D. A., eds. (1989) The potential effects of global climate change on the United States,
11 report to Congress. Washington, DC: U.S. Environmental protection Agency, Office of Policy, Planning,
12 and Evaluation; EPA report no. EPA-230/05-89-050.
13
14 Smoyer, K. E. (1993) Socio-demographic implications in summer weather-related mortality [master's thesis].
15 Wilmington, DE: University of Delaware.
16
17 Snedecor, G. E.; Cochran, W. G. (1983) Statistical methods. 7th ed. Ames, IA: Iowa State University Press.
18
19 Sorlie, P. D.; Rogot, E. (1990) Mortality by employment status in the National Longitudinal Mortality Study.
20 Am. J. Epidemiol. 132: 983-992.
21
22 Speizer, F. E. (1989) Studies of acid aerosols in six cities and in a new multi-city investigation: design issues.
23 Environ. Health Perspect. 79: 61-67.
24
25 Speizer, F. E., convener. (1993) Environmental health agenda for the 1990s: summary of workshop, October
26 1-6, 1991, Santa Fe, New Mexico. Environ. Res. 63: 1-15.
27
28 Speizer, F. E.; Fay, M. E.; Dockery, D. W.; Ferris, B. G., Jr. (1989) Chronic obstructive pulmonary disease
29 mortality in six U.S. cities. Am. Rev. Respir. Dis. 140(suppl.): S49-S55.
30
31 Spektor, D. M.; Lippmann, M.; Lioy, P. J.; Thurston, G. D.; Citak, K.; James, D. J.; Bock, N.; Speizer, F.
32 E.; Hayes, C. (1988) Effects of ambient ozone on respiratory function in active, normal children. Am.
33 Rev. Respir. Dis. 137: 313-320.
34
35 Spektor, D. M.; Yen, B. M.; Lippmann, M. (1989) Effect of concentration and cumulative exposure of inhaled
36 sulfuric acid on tracheobronchial particle clearance in healthy humans. In: Symposium on the health
37 effects of acid aerosols; October 1987; Research Triangle Park, NC. Environ. Health Perspect. 79:
38 167-172.
39
40 Spektor, D. M.; Hofmeister, V. A.; Artaxo, P.; Brague, J. A. P.; Echelar, F.; Nogueira, D. P.; Hayes, C.;
41 Thurston, G. D.; Lippmann, M. (1991) Effects of heavy industrial pollution on respiratory function in
42 the children of Cubatao, Brazil: a preliminary report. Environ. Health Perspect. 94: 51-54.
43
44 Spengler, J. D.; Allen, G. A.; Foster, S.; Severance, P.; Ferris, B., Jr. (1986) Sulfuric acid and sulfate aerosol
45 events in two U.S. cities. In: Lee, S. D.; Schneider, T.; Grant, L. D.; Verkerk, P. J., eds. Aerosols:
46 research, risk assessment and control strategies—proceedings of the second U.S.-Dutch international
47 symposium; May 1985; Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 107-120.
48
49 Spengler, J. D.; Keeler, G. J.; Koutrakis, P.; Ryan, P. B.; Raizenne, M.; Franklin, C. A. (1989) Exposures to
50 acidic aerosols. In: Symposium on the health effects of acid aerosols; October 1987; Research Triangle
51 Park, NC. Environ. Health Perspect. 79: 43-51.
52
April 1995 12-354 DRAFT-DO NOT QUOTE OR CITE
-------
1 Spix, C.; Heinrich, J.; Dockery, D.; Schwartz, J.; Volksch, G.; Schwinkowski, K.; Collen, C.; Wichmann, H.
2 E. (1993) Air pollution and daily mortality in Erfurt, East Germany, 1980-1989. Environ. Health
3 Perspect. 101: 518-526.
4
5 Spix, C.; Heinrich, J.; Dockery, D.; Schwartz, J.; Volksch, G.; Schwinkowski, K.; Collen, C.; Wichmann, H.
6 E. (1994) Summary of the analysis and reanalysis corresponding to the publication Air pollution and
7 daily mortality in Erfurt, East Germany 1980-1989. Summary report for: Critical evaluation workshop
8 on paniculate matter—mortality epidemiology studies; November; Raleigh, NC. Wuppertal, Germany:
9 Bergische Universitat-Gesamthochschule Wuppertal.
10
11 Stalker, W. W.; Dickerson, R. C.; Kramer, G. D. (1962) Sampling station and time requirements for urban air
12 pollution surveys, part IV: 2- and 24-hour sulfur dioxide and summary of other pollutants. J. Air Pollut.
13 Control Assoc. 12: 361-375.
14
15 Stedman, R. G. (1984) A universal scale of apparent temperature. J. Climatol. Appl. Meteorol. 23: 1674-1687.
16
17 Stefanski, L. A.; Carroll, R. J. (1985) Covariate measurement error in logistic regression. Ann. Stat. 13:
18 1335-1351.
19
20 Stellman, S. D. (1987) Confounding. Prev. Med. 16: 165-182.
21
22 Sterling, T. D.; Rosenbaum, W. L.; Weinkam, J. J. (1993) Risk attribution and tobacco-related deaths. Am. J.
23 Epidemiol. 138: 128-139.
24
25 Stern, B.; Raizenne, M. E.; Burnett, R. T.; Kearney, J.; Jones, L.; Franklin, C. (1988) Regional differences in
26 respiratory symptoms and lung function in a cross-sectional survey of Canadian children. Am. Rev.
27 Respir. Dis. 137(suppl.): 254.
28
29 Stern, B.; Jones, L.; Raizenne, M.; Burnett, R.; Meranger, J. C.; Franklin, C. A. (1989) Respiratory health
30 effects associated with ambient sulfates and ozone in two rural Canadian communities. Environ. Research
31 49: 20-39.
32
33 Stern, F.; Neal, B.; Smith, J. (1993) Air conditioning saturation in selected major urban areas. Washington, DC:
34 U.S. Environmental Protection Agency, Climate Change Division.
35
36 Stern, B. R.; Raizenne, M. E.; Burnett, R. T.; Jones, L.; Kearney, J.; Franklin, C. A. (1994) Air pollution and
37 childhood respiratory health: exposure to sulfate and ozone in 10 Canadian rural communities. Environ.
38 Res. 66: 125-142.
39
40 Stokinger, H. E. (1972) Concepts of thresholds in standards setting: an analysis of the concept and its application
41 to industrial air limits (TLVs). Arch. Environ. Health 25: 153-157.
42
43 Storr, J.; Lenney, W. (1989) School holidays and admissions with asthma. Arch. Dis. Child. 64: 103-107.
44
45 Stroup, D. F.; Thacker, S. B.; Herndon, J. L. (1988) Application of multiple time series analysis to the
46 estimation of pneumonia and influenza mortality by age 1962-1983. Stat. Med. 7: 1045-1059.
47
48 Studnicka, M. J.; Frischer, T.; Meinert, R.; Studnicka-Benke, A.; Hajek, K.; Spengler, J. D.; Neumann, M. G.
49 (1995) Acidic particles and lung function in children: a summer camp study in the Austrian Alps. Am. J.
50 Respir. Crit. Care Med. 151: 423-430.
51
April 1995 12-355 DRAFT-DO NOT QUOTE OR CITE
-------
1 Styer, P.; McMillan, N.; Gao, F.; Davis, J.; Sacks, J. (1995) The effect of airborne paniculate matter on daily
2 death counts. Environ. Health Perspect.: accepted.
3
4 Suh, H. H.; Allen, G. A.; Koutrakis, P.; Burton, R. M. (1994) Spatial variation in acidic sulfate and ammonia
5 concentrations within metropolitan Philadelphia. J. Air Waste Manage. Assoc.: submitted.
6
7 Suh, H. H.; Allen, G. A.; Koutrakis, P.; Burton, R. M. (1995) Spatial variation in acidic sulfate and ammonia
8 concentrations within metropolitan Philadelphia. J. Air Wast Manage. Assoc.: submitted.
9
10 Sunyer, J.; Anto, J. M.; Murillo, C.; Saez, M. (1991) Effects of urban air pollution on emergency room
11 admissions for chronic obstructive pulmonary disease. Am. J. Epidemiol. 134: 277-286.
12
13 Sunyer, J.; Saez, M.; Murillo, C.; Castellsague, J.; Martinez, F.; Anto, J. M. (1993) Air pollution and
14 emergency room admissions for chronic obstructive pulmonary disease: a 5-year study. Am. J.
15 Epidemiol. 137: 701-705.
16
17 Tager, I. B.; Segal, M. R.; Speizer, F. E.; Weiss, S. T. (1988) The natural history of forced expiratory
18 volumes: effect of cigarette smoking and respiratory symptoms. Am. Rev. Respir. Dis. 138: 837-849.
19
20 Tager, I. B.; Hanrahan, J. P.; Tosteson, T. D.; Castile, R. G.; Brown, R. W.; Weiss, S. T.; Speizer, F. E.
21 (1993) Lung function, pre- and post-natal smoke exposure, and wheezing in the first year of life. Am.
22 Rev. Respir. Dis. 147: 811-817.
23
24 Tan, G. (1992) Socioeconomic aspects of climate change. Peking, People's Republic of China: Meteorological
25 Publication Society. (Modern Atmospheric Science Technical Series).
26
27 Tashkin, D. P.; Detels, R.; Simmons, M.; Liu, H.; Coulson, A. H.; Sayre, J.; Rokaw, S. (1994) The UCLA
28 population studies of chronic obstructive respiratory disease: XI. impact of air pollution and smoking on
29 annual change in forced expiratory volume in one second. Am. J. Respir. Crit. Care Med. 149:
30 1209-1217.
31
32 Theil, H. (1978) Introduction to econometrics. Englewood Cliffs, NJ: Prentice-Hall; p. 137.
33
34 Thibodeau, L. A.; Reed, R. B.; Bishop, Y. M. M.; Kammerman, L. A. (1980) Air pollution and human health:
35 a review and reanalysis. Environ. Health Perspect. 34: 165-183.
36
37 Thomas, K. W.; Pellizzari, E. D.; Clayton, C. A.; Whitaker, D. A.; Shores, R. C.; Spengler, J. D.; Ozkaynak,
38 H.; Wallace, L. A. (1993) Particle total exposure assessment methodology (PTEAM) study: method
39 performance and data quality for personal, indoor, and outdoor aerosol monitoring at 178 homes in
40 southern California. J. Exposure Anal. Environ. Epidemiol. 3: 203-226.
41
42 Thomson, M.; Philion, J. (1991) Children's respiratory hospitalizations and air pollution. Can. J. Public Health
43 82: 203-204.
44
45 Thurston, G. D.; Kinney, P. L. (1995) Air pollution epidemiology: considerations in time-series modeling.
46 Inhalation Toxicol. 7: 71-83.
47
48 Thurston, G. D.; Ozkaynak, H. (1992) Air pollution and mortality [letter]. Science (Washington, DC) 255:
49 382-383.
50
April 1995 12-356 DRAFT-DO NOT QUOTE OR CITE
-------
1 Thurston, G. D.; Ito, K.; Lippmann, M.; Hayes, C. (1989) Reexamination of London, England, mortality in
2 relation to exposure to acidic aerosols during 1963-1972 winters. In: Symposium on the health effects of
3 acid aerosols; October 1987; Research Triangle Park, NC. Environ. Health Perspect. 79: 73-82.
4
5 Thurston, G. D.; Ito, K.; Kinney, P. L.; Lippmann, M. (1992) A multi-year study of air pollution and
6 respiratory hospital admissions in three New York State metropolitan areas: results for 1988 and 1989
7 summers. J. Exposure Anal. Environ. Epidemiol. 2: 429-450.
8
9 Thurston, G. D.; Gorczynski, J. E., Jr.; Currie, J. H.; He, D.; Ito, K.; Hipfner, J.; Waldman, J.; Lioy, P. J.;
10 Lippmann, M. (1994a) The nature and origins of acid summer haze air pollution in metropolitan Toronto,
11 Ontario. Environ. Res. 65: 254-270.
12
13 Thurston, G. D.; Ito, K.; Hayes, C. G.; Bates, D. V.; Lippmann, M. (1994b) Respiratory hospital admissions
14 and summertime haze air pollution in Toronto, Ontario: consideration of the role of acid aerosols.
15 Environ. Res. 65: 271-290.
16
17 Touloumi, G.; Pocock, S. J.; Katsouyanni, K.; Trichopoulos, D. (1994) Short-term effects of air pollution on
18 daily mortality in Athens: a time-series analysis. Int. J. Epidemiol. 23: 1-11.
19
20 Tromp, S. W. (1980) Biometeorology. London, United Kingdom: Heyden.
21
22 Tseng, R. Y. M.; Li, C. K.; Spinks, J. A. (1992) Particulate air pollution and hospitalization for asthma. Ann.
23 Allergy 68: 425-432.
24
25 Tzonou, A.; Maragoudakis, G.; Trichopoulos, D.; Zavitsanos, X.; Dimopoulou, I.; Toupadaki, N.;
26 Kremastinou, J. (1992) Urban living, tobacco smoking, and chronic obstructive pulmonary disease: a
27 study in Athens. Epidemiology 3: 57-60.
28
29 U.S. Centers for Disease Control. (1982) Medical examiner summer mortality surveillance—United States
30 1979-1981. Morb. Mortal. Wkly Rep. 31: 336-343.
31
32 U.S. Centers for Disease Control. (1984) Heat-associated morality—New York City. Morb. Mortal. Wkly. Rep.
33 33: 430-432.
34
35 U.S. Centers for Disease Control. (1984) Illness and death due to environmental heat—Georgia and St. Louis,
36 Missouri, 1983. Morb. Mortal. Wkly. Rep. 33: 325-326.
37
38 U.S. Centers for Disease Control. (1989) Heat-related deaths—Missouri, 1979-1988. Morb. Mortal. Wkly. Rep.
39 38: 437-439.
40
41 U.S. Centers for Disease Control. (1990) Asthma—United States, 1980-1987. Morb. Mortal. Wkly. Rep. 39:
42 493-497.
43
44 U.S. Centers for Disease Control. (1992) Asthma—United States, 1980-1990. Morb. Mortal. Wkly. Rep. 41:
45 733-735.
46
47 U.S. Centers for Disease Control. (1994) Populations at risk from paniculate air pollution—United States, 1992.
48 Morb. Mortal. Wkly. Rep. 43: 290-293.
49
50 U.S. Centers for Disease Control. (1995) Asthma—United States, 1982-1992. Morb. Mortal. Wkly. Rep. 43:
51 952-955.
52
April 1995 12-357 DRAFT-DO NOT QUOTE OR CITE
-------
1 U.S. Department of Commerce. (1985) Report of the impact of the January 1985 cold wave. Climate impact
2 assessment: United States.
3
4 U.S. Department of Health and Human Services. (1994) Current estimates from the National Health Interview
5 Survey, 1992. Hyattsville, MD: Public Health Service, National Center for Health Statistics; p. 84;
6 DHHS publication no. (PHS) 94-1517. (Vital and health statistics: series 10, no. 189).
7
8 U.S. Department of Health, Education, and Welfare. (1964) Smoking and health: report of the Advisory
9 Committee to the Surgeon General of the Public Health Service. Washington, DC: Public Health Service;
10 p. 60.
11
12 U.S. Environmental Protection Agency. (1982a) Air quality criteria for paniculate matter and sulfur oxides.
13 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
14 and Assessment Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield,
15 VA; PB84-156777.
16
17 U.S. Environmental Protection Agency. (1982b) Review of the national ambient air quality standards for
18 paniculate matter: assessment of scientific and technical information. Research Triangle Park, NC: Office
19 of Air Quality Planning and Standards, Strategies and Air Standards Division; report no.
20 EPA-450/5-82-001. Available from: NTIS, Springfield, VA; PB82-177874.
21
22 U.S. Environmental Protection Agency. (1986a) Second addendum to air quality criteria for particulate matter
23 and sulfur oxides (1982): assessment of newly available health effects information. Research Triangle
24 Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
25 Office; EPA report no. EPA-600/8-86-020F. Available from: NTIS, Springfield, VA; PB87-176574.
26
27 U.S. Environmental Protection Agency. (1986b) Review of the national ambient air quality standards for
28 particulate matter: updated assessment of scientific and technical information, addendum to the 1982
29 OAQPS staff paper. Research Triangle Park, NC: Office of Air Quality Planning and Standards,
30 Strategies and Air Standards Division; report no. EPA/450/05-86/012. Available from: NTIS,
31 Springfield, VA; PB87-176871/XAB.
32
33 U.S. Environmental Protection Agency. (1986c) Air quality criteria for ozone and other photochemical oxidants.
34 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
35 and Assessment Office; EPA report nos. EPA-600/8-84-020aF-eF. 5v. Available from: NTIS,
36 Springfield, VA; PB87-142949.
37
38 U.S. Environmental Protection Agency. (1989) An acid aerosols issue paper: health effects and aerometrics.
39 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
40 and Assessment Office; EPA report no. EPA-600/8-88-005F. Available from: NTIS, Springfield, VA;
41 PB91-125864.
42
43 U.S. Environmental Protection Agency. (1992) Respiratory health effects of passive smoking: lung cancer and
44 other disorders [review draft]. Washington, DC: Office of Research and Development, Office of Health
45 and Environmental Assessment; EPA report no. EPA/600/6-90/006B. Available from: NTIS, Springfield,
46 VA; PB92-182344.
47
48 U.S. Environmental Protection Agency. (1993) Air quality criteria for oxides of nitrogen. Research Triangle
49 Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
50 Office; EPA report no. EPA/600/8-91/049aF-cF.3v. Available from: NTIS, Springfield, VA;
51 PB95-124533; PB95-124525; PB95-124517.
52
April 1995 12-358 DRAFT-DO NOT QUOTE OR CITE
-------
1 U.S. Environmental Protection Agency. (1994) Respiratory toxicology of inhaled silica [internal review draft].
2 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
3 and Assessment Office; September.
4
5 U.S. Senate. (1968) Air quality criteria staff report. Washington, DC: Committee on Public Works; 90th
6 Congress, 2nd session.
7
8 Ulm, K. (1991) A statistical method for assessing a threshold in epidemiological studies. Stat. Med. 10: 341-349.
9
10 United Kingdom Ministry of Health. (1954) Mortality and morbidity during the London fog of December 1952.
11 London, United Kingdom: Her Majesty's Stationery Office. (Reports on public health and medical
12 subjects no. 95).
13
14 Utell, M. J.; Samet, J. M. (1993) Paniculate air pollution and health: new evidence on an old problem
15 [editorial]. Am. Rev. Respir. Dis. 147: 1334-1335.
16
17 Utell, M. J.; Morrow, P. E.; Speers, D. M.; Darling, J.; Hyde, R. W. (1983) Airway responses to sulfate and
18 sulfuric acid aerosols in asthmatics: an exposure-response relationship. Am. Rev. Respir. Dis. 128:
19 444-450.
20
21 Vedal, S.; Blair, J.; Manna, B. (1991) Adverse respiratory health effects of ambient inhalable particle exposure.
22 Presented at: 84th annual meeting and exhibition of the Air and Waste Management Association; June;
23 Vancouver, BC, Canada. Pittsburgh, PA: Air and Waste Management Association; paper no. 91-180.56.
24
25 Vedal, S.; Schenker, M. B.; Samet, J. M.; Speizer, F. E. (1984) Risk factors for childhood respiratory disease:
26 analysis of pulmonary function. Am. Rev. Respir. Dis. 130: 187-192.
27
28 Vedal, S.; Schenker, M. B.; Munoz, A.; Samet, J. M.; Battennan, S.; Speizer, F. E. (1987) Daily air pollution
29 effects on children's respiratory symptoms and peak expiratory flow. Am. J. Public Health 77: 694-698.
30
31 Waldron, H. A. (1974) The blood lead threshold. Arch. Environ. Health 29: 271-273.
32
33 Walker, A. M.; Blettner, M. (1985) Comparing imperfect measures of exposure. Am. J. Epidemiol. 121:
34 783-790.
35
36 Waller, R. E. (1963) Acid droplets in town air. Int. J. Air Water Pollut. 7: 773-778.
37
38 Waller, R. E. (1971) Air pollution and community health. J. R. Coll. Physicians (London) 5: 362-368.
39
40 Waller, R. E.; Lawther, P. J. (1957) Further observations on London fog. Br. Med. J. 4: 1473-1475.
41
42 Waller, R. E.; Swan, A. V. (1992) Invited commentary: paniculate air pollution and daily mortality. Am. J.
43 Epidemiol. 135: 20-22.
44
45 Walters, S.; Griffiths, R. K.; Ayres, J. G. (1994) Temporal association between hospital admissions for asthma
46 in Birmingham and ambient levels of sulphur dioxide and smoke. Thorax 49: 133-140.
47
48 Wang, X.; Dockery, D. W.; Wypij, D.; Fay, M. E.; Ferris, B. G., Jr. (1993a) Pulmonary function between 6
49 and 18 years of age. Pediatr. Pulmonol. 15: 75-88.
50
April 1995 12-359 DRAFT-DO NOT QUOTE OR CITE
-------
1 Wang, X.; Dockery, D. W.; Wypij, D.; Gold, D. R.; Speizer, F. E.; Ware, J. H.; Ferris, B. G., Jr. (1993b)
2 Pulmonary function growth velocity in children 6 to 18 years of age. Am. Rev. Respir. Dis. 148:
3 1502-1508.
4
5 Ware, J. H.; Thibodeau, L. A.; Speizer, F. E.; Colome, S.; Ferris, B. G., Jr. (1981) Assessment of the health
6 effects of atmospheric sulfur oxides and paniculate matter: evidence from observational studies. Environ.
7 Health Perspect. 41: 255-276.
8
9 Ware, J. H.; Dockery, D. W.; Spiro, A., Ill; Speizer, F. E.; Ferris, B. G., Jr. (1984) Passive smoking, gas
10 cooking, and respiratory health of children living in six cities. Am. Rev. Respir. Dis. 129: 366-374.
11
12 Ware, J. H.; Ferris, B. G., Jr.; Dockery, D. W.; Spengler, J. D.; Stram, D. O.; Speizer, F. E. (1986) Effects
13 of ambient sulfur oxides and suspended particles on respiratory health of preadolescent children. Am.
14 Rev. Respir. Dis. 133: 834-842.
15
16 Warner, P. O.; Stevens, L. (1973) Revaluation of the "Chattanooga school children study" in the light of other
17 contemporary governmental studies: the possible impact of these findings on the present NOj air quality
18 standard. J. Air Pollut. Control Assoc. 23: 769-771.
19
20 Weiss, K. B. (1990) Seasonal trends in US asthma hospitalizations and mortality. JAMA J. Am. Med. Assoc.
21 263: 2323-2328.
22
23 Weiss, K. B.; Gergen, P. J.; Grain, E. F. (1992) Inner-city asthma: the epidemiology of an emerging US public
24 health concern. Chest 101(suppl.): 362S-367S.
25
26 Weitzman, M. (1986) School absence rates as outcome measures in studies of children with chronic illness. J.
27 Chronic Dis. 39: 799-808.
28
29 Weitzman, M.; Klerman, L. V.; Alpert, J. J.; Lamb, G. A.; Kayne, H.; Rose, L. (1986) Factors associated with
30 excessive school absence. Pediatrician 13: 74-80.
31
32 Wennberg, J. E. (1987) Population illness rates do not explain population hospitalization rates: a comment on
33 Mark Blumberg's thesis that morbidity adjusters are needed to interpret small area variations. Med. Care
34 25: 354-359.
35
36 Wennberg, J. E.; McPherson, K.; Caper, P. (1984) Will payment based on diagnosis-related groups control
37 hospital costs? N. Engl. J. Med. 311: 295-300.
38
39 Westfall, P. H.; Young, S. S. (1993) Resampling-based multiple testing: examples and methods forp-value
40 adjustment. New York, NY: John Wiley & Sons, Inc.
41
42 White, M. R.; Hertz-Picciotto, I. (1985) Human health analysis of climate related to health. In: White, M. R.,
43 ed. Characterization of information requirements for studies of CO2 effects: water resources, agriculture,
44 fisheries, forests and human health. Washington, DC: Department of Energy; pp. 172-205.
45
46 White, M. C.; Etzel, R. A.; Wilcox, W. D.; Lloyd, C. (1994) Exacerbations of childhood asthma and ozone
47 pollution in Atlanta. Environ. Res. 65: 56-68.
48
49 Whittemore, A. S.; Keller, J. B. (1988) Approximations for regression with covariate measurement error. JASA
50 J. Am. Stat. Assoc. 83: 1057-1066.
51
April 1995 12-360 DRAFT-DO NOT QUOTE OR CITE
-------
1 Whittemore, A. S.; Korn, E. L. (1980) Asthma and air pollution in the Los Angeles area. Am. J. Public Health
2 70: 687-696.
3
4 Wichmann, H.-E.; Sugiri, D.; Islam, M. S.; Haake, D.; Roscovanu, A. (1988a) Lungenfunktion und
5 Carboxyhamoglobin in der Smogsituation des Januar 1987 [Pulmonary function and carboxyhemoglobin
6 during the smog episode in January 1987]. Zentralbl. Bakteriol. Mikrobiol. Hyg. Abt. 1 Orig. B 187:
7 31-43.
8
9 Wichmann, H.-E.; Sugiri, D.; Herold, G.; Knulle, E. (1988b) Atemwiderstandsmessungen bei Gesunden im
10 Winterhalbjahr 1985/86 und im Januar/Februar 1987 [Measurement of airway resistance in healthy
11 persons during the winter of 1985/86 and in January and February 1987]. Zentralbl. Bakteriol.
12 Mikrobiol. Hyg. Ser. B 185: 509-519.
13
14 Wichmann, H. E.; Mueller, W.; Allhoff, P.; Beckmann, M.; Bocter, N.; Csicsaky, M. J.; Jung, M.; Molik, B.;
15 Schoeneberg, G. (1989) Health effects during a smog episode in West Germany in 1985. In: Symposium
16 on the health effects of acid aerosols; October 1987; Research Triangle Park, NC. Environ. Health
17 Perspect. 79: 89-99.
18
19 Williams, M. K. (1970) Sickness absence and ventilatory capacity of workers exposed to sulphuric acid mist. Br.
20 J. Ind. Med. 27: 61-66.
21
22 World Health Organization. (1977) Manual of the international statistical classification of diseases, injuries, and
23 causes of death. Geneva, Switzerland: World Health Organization.
24
25 World Health Organization. (1979) Sulfur oxides and suspended paniculate matter. Geneva, Switzerland: World
26 Health Organization. (Environmental health criteria 8).
27
28 World Health Organization. (1995) Potential health effects of climate change. Geneva, Switzerland: in press.
29
30 Wright, A. L.; Taussig, L. M.; Ray, C. G.; Harrison, H. R.; Holberg, C. J. (1989) The Tucson children's
31 respiratory study: II. lower respiratory tract illness in the first year of life. Am. J. Epidemiol. 129:
32 1232-1246.
33
34 Wyzga, R. (1978) The effect of air pollution upon mortality: a consideration of distributed lag models. JASA,
35 Journal of the American Statistical Association 73: 463-472.
36
37 Wyzga, R. E.; Lipfert, F. W. (1995) Ozone and daily mortality: the ramifications of uncertainties and
38 interactions and some initial regression results. Ozone conference: accepted.
39
40 Xu, X.; Wang, L. (1993) Association of indoor and outdoor particulate level with chronic respiratory illness.
41 Am. Rev. Respir. Dis. 148: 1516-1522.
42
43 Xu, Z.-Y.; Blot, W. J.; Xiao, H.-P.; Wu, A.; Feng, Y.-P.; Stone, B. J.; Sun, J.; Ershow, A. G.; Henderson,
44 B. E.; Fraumeni, J. F., Jr. (1989) Smoking, air pollution, and the high rates of lung cancer in Shenyang,
45 China. J. Natl. Cancer Inst. 81: 1800-1806.
46
47 Xu, X.; Dockery, D. W.; Wang, L. (1991) Effects of air pollution on adult pulmonary function. Arch. Environ.
48 Health 46: 198-206.
49
50 Xu, X.; Gao, J.; Dockery, D. W.; Chen, Y. (1994) Air pollution and daily mortality in residential areas of
51 Beijing, China. Arch. Environ. Health 49: 216-222.
52
April 1995 12-361 DRAFT-DO NOT QUOTE OR CITE
-------
1 Yano, E.; Yokoyama, Y.; Higashi, H.; Nishii, S.; Maeda, K.; Koizumi, A. (1990) Health effects of volcanic
2 ash: a repeat study. Arch. Environ. Health 45: 367-373.
3
4 Yoshimura, I. (1990) The effect of measurement error on the dose-response curve. Environ. Health Perspect. 87:
5 173-178.
6
7 Zeger, S. L.; Liang, K.-Y. (1986) Longitudinal data analysis for discrete and continuous outcomes. Biometrics
8 42: 121-130.
9
10 Zeger, S. L.; Qaqish, B. (1988) Markov regression models for time series: a quashi-liklihood approach.
11
April 1995 12-362 DRAFT-DO NOT QUOTE OR CITE
-------
APPENDIX 12A
EFFECTS OF WEATHER AND CLIMATE ON HUMAN
MORTALITY AND THEIR ROLES AS
CONFOUNDING FACTORS FOR AIR POLLUTION
April 1995 12A-1 DRAFT-DO NOT QUOTE OR CITE
-------
1 Interrelationships Between Weather, PM, and Mortality
2 A number of studies have concluded that both extreme weather and high pollution
3 adversely affect mortality. While a majority of this research has examined the independent
4 effect of these stresses on mortality, few studies have successfully separated weather-induced
5 from pollution-induced mortality. This has been especially true in the evaluation of acute
6 mortality. There have been some efforts to evaluate these differential impacts (e.g., Ramlow
7 and Kuller, 1990; Shumway et al., 1988; Schwartz and Dockery, 1992a; 1992b).
8 Some authors have conducted weather/pollution/mortality evaluations in Steubenville,
9 OH; Philadelphia, PA; London, England; Birmingham, AL; and Utah County, UT as well as
10 other locales (Table 12B-1). In all of these investigations, they have reported significant
11 associations between human mortality and PM, and in some cases, the relationship extends to
12 levels well below the current National Ambient Air Quality Standard. In several, they have
13 also alluded to a weather-mortality relationship. For Steubenville, a positive non-linear
14 relationship between both temperature and dew point temperature and mortality was detected.
15 When dummy variables were used to denote hot days, humid days, and hot/humid days, the
16 hot/humid days were a significant predictor of mortality. When seasonal variations were
17 controlled for in their Poisson regression models however, neither temperature nor dew point
18 proved to be significant predictors of mortality (Schwartz and Dockery, 1992b). In a study
19 of British Smoke in London, Schwartz and Marcus (1990) controlled for temperature and
20 humidity and improved the model results significantly over the results of a model with no
21 meteorological variables.
22 More recent studies indicate that controls for weather may probably not been adequate
23 to determine true meteorological impacts in the evaluations cited above (Kalkstein et al., in
24 press). Many PM/mortality studies utilize rank-ordered temperatures, squared temperature
25 and dewpoint values, moving averages of temperature, and mean temperatures for groupings
26 of days (refer back to Table 12-2 in main chapter text for further details), which may not
27 provide the detail to detect true weather/mortality relationships. In addition, it is probably
28 not feasible to assume that cities within a wide range of climates demonstrate similar
29 weather/PM impacts on mortality, and there are possibly some regional similarities in
30 response which have not been adequately explored. In a reanalysis of Philadelphia
31 mortality/PM relationships, Schwartz (1994) took a more direct approach to examine the
April 1995 12A-2 DRAFT-DO NOT QUOTE OR CITE
-------
1 possibility of confounding weather impacts. The reanalysis utilized Hastie and Tibshirani's
2 (1990) "Generalized Additive Model" to detect and control for nonlinearities in the
3 dependence of daily mortality on weather; nevertheless, this study uncovered findings similar
4 to the original Philadelphia study. In addition, Moolgavkar (1995) assert that the role of
5 weather was improperly evaluated within the Steubenville study, and suggest a more
6 sophisticated evaluation of meteorology in future PM/mortality analyses.
7 To further control for weather, Schwartz (1994) stated that the similar responses to air
8 pollution in the mild weather of Philadelphia (based on a mean daily temperature of 57 °F)
9 and cold weather of London reduce the confounding role of weather. In addition, Schwartz
10 (1994) notes that similarity in temperature and humidity on high and low air pollution days
11 (when different mortality response are noted), "... also would seem to eliminate weather as a
12 potential confounder." However, it is possible that these studies do not remove the total
13 confounding influence of weather, especially because of their dependence on mean
14 temperatures and other meteorological surrogate which may not truly reflect weather
15 variation.
16 There have been other studies which have attempted to assess the differential impact of
17 PM and weather on acute mortality. For example, Ostro (1993) summarized studies which
18 show strong associations between exposures to PM10 and total daily mortality for many urban
19 areas in the United States, Europe, and Canada. In addition, he notes that results are
20 remarkably consistent across regions. However, the impact of weather as a confounding
21 influence is implicitly considered rather unimportant. Ito et al. (1993) showed that daily
22 mortality in London was significantly associated with aerosol acidity levels and British
23 Smoke. Weather played a lesser role, and Ito's work confirms results obtained by others
24 who have evaluated London's mortality/PM/weather relationship (Schwartz and Marcus,
25 1990; Thurston et al., 1989; Mazumdar et al., 1982). However, it should be noted that
26 London's marine climate is rather benign when compared to many large American cities, as
27 thermal extremes are unusual.
28 Some studies for cities exhibiting higher climate variation yielded somewhat different
29 results. Wyzga (1978) used the Coefficient of Haze (COH) as a surrogate measure of
30 PM concentration, and determined that high COH values are associated with increased
31 mortality in Philadelphia. However, he recognized the potential impact of extreme weather
April 1995 12A-3 DRAFT-DO NOT QUOTE OR CITE
-------
1 as well, and noted that heat waves may also be responsible for large numbers of extra deaths.
2 In a recent study by Wyzga in which weather was treated in a more sophisticated manner
3 (Wyzga and Lipfert, in press), the impact of ozone concentrations and weather on acute
4 mortality were evaluated and results were compared to TSP. The authors conclude that a
5 determination of ozone and TSP impacts is most difficult because of the influence of
6 confounders, particularly weather. In addition, use of different explanatory models yields
7 disparate results, with pollution impacts ranging, "...from essentially no effect to response
8 similar to that associated with a 10 °F increase in ambient temperature" (Wyzga and Lipfert,
9 in press). This evaluation appeared to uncover a synergistic relationship between weather
10 and pollution, as days with maximum temperatures exceeding 85 °F contributed most to the
11 associations between TSP and mortality. Several other studies have uncovered synergistic
12 relationships, and some of these consider pollution to be of secondary importance to weather
13 in affecting acute mortality. Ramlow and Kuller (1990) found that daily mortality was most
14 closely associated with the daily average temperature of the previous day rather than any
15 pollution measure in Allegheny County, Pennsylvania. In a study which attempts to
16 determine synergistic relationships between weather and pollution on mortality in Los
17 Angeles, Shumway et al. (1988) determined that mortality is, "...an additive nonlinear
18 function of temperature and pollution, whereas there may be significant interactions present,
19 especially when low or high temperatures are combined with high pollution levels." The
20 authors found that model-predicted average mortality values increased at both temperature
21 extremes when paniculate levels were held constant. Two evaluations in the Netherlands
22 found temperature extremes in summer and winter to be primary determinants in mortality
23 variation. Kunst et al. (1993) and Mackenbach et al. (1993) determined that the relationship
24 between temperature and mortality is linear, producing a V-shaped temperature curve, with
25 minimum mortality rates observed between 10 to 15 °C. The Kunst evaluation determined
26 that summer acute mortality is not influenced by variations in air pollution concentration.
27 Although weather seems to induce mortality increases when temperatures are either
28 very warm or very cold, the impact of weather as a confounder varies seasonally. For
29 example, the impact of weather on acute mortality in winter is much more difficult to
30 evaluate, and thermal relationships are decidedly weaker (WHO/WMO/UNEP, in press).
31 Thus, although the temperature/mortality relationship might follow the V-shaped curve
April 1995 12A-4 DRAFT-DO NOT QUOTE OR CITE
-------
1 Weather/Mortality Relationship
2 Both heat and cold-related climate stresses appear to have a significant impact on
3 human mortality, but the impacts appear to vary spatially and temporally. Most research on
4 weather/mortality relationships concentrate on summer temperatures, when significant spikes
5 in daily mortality totals are apparent on certain hot days (Figures 12A-1 and 12A-2). Studies
6 of winter weather/mortality relationships have greater uncertainty, and although winter death
7 totals average about 10 to 15% higher than summer, the daily variance in mortality
8 attributable to weather is considerably smaller (National Center for Health Statistics, 1978).
9 Most summer weather/mortality analyses concentrate on temperature, which is
10 demonstrated to have the greatest impact of all meteorological variables. In fact, the notion
11 of a "temperature threshold" is quite common in a number of studies, and in many cities it is
12 apparent that elevated mortality occurs only during the 10 to 15% of all summer days which
13 exhibit temperatures above the threshold (Smith and Tirpak, 1989; Table 12A-2). The
14 magnitude of the threshold temperature appears to be related to the frequency of occurrence
15 of high temperatures among these cities. For example, the temperature exceeds 96 °F in St.
16 Louis with approximately the same frequency that it exceeds 90 °F in Detroit, values which
17 represent threshold temperatures for these cities. This strongly suggests that the notion of a
18 "heat wave" is relative on an interregional scale, and may depend upon the distribution of
19 temperatures above the threshold. Most climate/health research considers the impact of
20 temperature as absolute, and little attempt is made to consider the differential impact of heat
21 or cold on an interregional scale (Schneider, in press).
22
23 Physiological Factors That May Be Sensitive To Meteorology
24 Several studies have noted that mortality can be further exacerbated by other
25 meteorological factors such as wind and humidity (Kalkstein and Davis, 1989; Kunst et al.,
26 1993). The combination of these elements with temperature produces an "apparent
27 temperature", which is the perceived thermal load to the human body (Steadman, 1984).
28 Healthy persons have efficient heat regulatory mechanisms which cope with increases in
29 apparent temperature up to the threshold condition. When exposed to heat, the body can
30 increase radiant, convective, and evaporative heat loss by methods such as vasodilation
31 (enlargement of the blood vessels) and perspiration (Horowitz and Samueloff, 1987;
April 1995 12A-5 DRAFT-DO NOT QUOTE OR CITE
-------
70
60
50
-------
TABLE 12A-1. SUMMER THRESHOLD TEMPERATURES
FOR SELECTED U.S. CITIES
City
Atlanta
Chicago
Cincinnati
Dallas
Detroit
Kansas City
Los Angeles
Memphis
Threshold
Temperature (°F)
94
91
92
103
90
99
81
99
City
Minneapolis
New Orleans
New York
Oklahoma City
Philadelphia
St. Louis
San Francisco
Threshold
Temperature (°F)
93
a
92
a
92
96
84
aNo threshold temperature detected.
Source: Smith and Tirpak (1989).
1 Diamond, 1991). In addition, acclimatization to oppressive conditions can occur within
2 several days with continuing exposure (Kilbourne, 1992).
3 Heat shock proteins, which the body synthesizes when its internal temperature rises,
4 represent one form of physiological acclimatization to heat (Born et al., 1990). A slight
5 increase in internal temperature can induce heat shock protein production and thus,
6 "...protect...cells against subsequent, otherwise lethal sudden temperature increase..."
7 (Elliot, 1992). However, there are obvious indications that the body cannot cope with
8 oppressive conditions indefinitely. Total deaths increase substantially when apparent
9 temperatures exceed threshold values, especially for several consecutive days (Kalkstein and
10 Smoyer, 1993).
11 In some cases, hot, dry weather can also induce increases in acute mortality. The
12 desiccating impact of hot, dry conditions increases evaporation opportunity from the body,
13 creating hyperthermal conditions which may result in death (Lowry, 1988; Jendritzky, 1991).
14 High winds could potentially increase evaporation opportunity even further, but most
15 weather/mortality studies surprisingly show little evidence of an important wind effect on
16 human mortality during summer (e.g., Kunst et al., 1993). In one study where nighttime
17 winds are inversely related to mortality, it was suggested that air flow in residences is thus
April 1995
12A-7 DRAFT-DO NOT QUOTE OR CITE
-------
1 inhibited, creating oppressive conditions which pose a higher health risk (Kalkstein and
2 Davis, 1989).
3 Many causes of death appear to increase when offensive weather is present. Although
4 the most direct impact of heat stress on the human body is manifested as heat exhaustion or
5 heatstroke, these causes represent only a small proportion of total mortality increases during
6 offensive weather (Larsen, 1990a,b). For example, deaths from cardiovascular,
7 cerebrovascular, respiratory, and immune system disorders appear to increase dramatically
8 during stressful weather conditions. In southwestern Germany, cardiac infarction,
9 nonrheumatic myocardiac diseases, vascular diseases of the brain, respiratory diseases, and
10 bronchitis were isolated as the most important heat-related causes of death (Jendritzky and
11 Bucher, 1993). A recent study in the Netherlands suggested that respiratory disorders
12 account for the largest proportion of heat related deaths, followed by cardiovascular diseases
13 (Kunstetal., 1993).
14 Differential temporal responses between heat related mortality and offensive weather
15 are noted. The lag time between offensive weather and subsequent peaks in mortality is very
16 short during summer. Most studies have identified a lag period of no longer than one day,
17 and in some cases, the mortality increase occurs on the same day as the offensive weather
18 (Jones et al., 1982; White and Hertz-Picciotto, 1985; Kalkstein, 1991; Kunst et al., 1993).
19 In addition, most studies have noted that heat related mortality declines as the summer season
20 progresses (Kalkstein and Davis, 1989; WMO/WHO/UNEP, in press). There are two
21 possible explanations for this phenomenon: humans acclimatize to heat very quickly, even
22 within a single season (Rotton, 1983), or many susceptible people die during early season
23 heat waves, leaving less of a susceptible pool for later heat waves (Kalkstein, 1993). There
24 is stronger evidence to support the latter contention. It appears that a proportion of people
25 who die during heat waves would have died shortly afterward regardless of the weather.
26 If such "mortality displacement" is taking place, it is probable that mortality totals shortly
27 after a severe heat episode would be below the long term baseline (Figure 12A-3). This, in
28 fact appears to be the case, and in an examination of several heat waves in New York City
29 and St. Louis, daily mortality after the heat episode remained significantly below the baseline
30 (Kalkstein, 1993). For New York and St. Louis, it was estimated that approximately 40%
31 and 19%, respectively, of extra deaths during heat episodes represented mortality
April 1995 12A-8 DRAFT-DO NOT QUOTE OR CITE
-------
8
o
8
0)
S
DC
f
C
6-
5-
4-
.N 3-
TJ
I
I
CO
2-
1-
Legend
Dally Mortality
Mean Mortality
10 20 30 40 50
60 70 80 90
Days from May to August
100 110 120 130
Figure 12A-3. Daily summer mortality for a New York heat wave, 1966. Darker
shading represents heat related mortality; lighter shading represents
diminished mortality attributed to "mortality displacement".
Source: Kalkstein, 1993.
1 displacement (Scheraga and Sussman, in press). Similar increases may not occur at all sites,
2 however.
3 There is some evidence that the impact of weather on mortality is relative in winter as
4 well, but there is considerable uncertainty about causal mechanisms. The strongest winter
5 relationships are found in the northeastern, southeastern, and midwestern sections of the
6 country, where the requisite damp, inclement weather is most common (Kalkstein and Davis,
7 1989; Larsen, 1990a). The weakest relationships are found in the southwestern U.S. and
8 more arid sections of the country. The coldest areas do not tend to show the greatest winter
9 impacts (Frost and Auliciems, 1993). One study suggests that poor populations in the
10 Southeast are particularly at risk to cold temperatures due to relative economic disadvantages
11 which render their dwellings less suitable to handle cold weather (Larsen, 1990a). However,
April 1995
12A-9
DRAFT-DO NOT QUOTE OR CITE
-------
1 winter weather related mortality is largely attributed to indirect causes, such as infectious
2 diseases (i.e., flu, etc.)
3 Cold season mortality attributed to weather is considerably more complex to establish.
4 For example, the existence of a threshold is more difficult to discern in winter, and the
5 impact of thermal factors appears to be less than in summer (Kalkstein and Davis, 1989).
6 In addition, the causes of death related to weather are very different between the two
7 seasons. Although cardiovascular disease appears important in both seasons (Auliciems and
8 Frost, 1989), influenza, pneumonia, and increased mortality from a variety of accidents are
9 of greater significance in winter (U.S. Department of Commerce, 1985).
10 Weather related mortality in winter does not tend to increase systematically as the
11 temperatures drop. Several studies indicate that for cold weather locales such as Montreal
12 and Minneapolis, mortality increases with decreasing temperature up to a certain point, but at
13 very cold temperatures, mortality tends to decrease (Kalkstein, 1988; Auliciems and Frost,
14 1989; Frost and Auliciems, 1993). In less extreme climates, such as the Netherlands and
15 Brisbane, Australia, mortality rates rise linearly with decreasing temperature (Kunst et al.,
16 1993; Frost and Auliciems, 1993). Thus, in temperate environments, people appear
17 unaccustomed to extreme cold, and react negatively. However, in cold climates, it is
18 possible that behavioral responses, such as cold avoidance, appear to become the dominant
19 thermoregulatory processes at very cold temperatures.
20 There is evidence that stormy, rather than very cold, weather is responsible for the
21 greatest mortality in winter. Overcast and windy days with precipitation (especially snow)
22 appear to be associated with the highest mortality totals in many mid-latitude locales (Rogot
23 and Padgett, 1976; Glass and Zack, 1979; Baker Blocker, 1982; Kalkstein and Davis, 1989;
24 Auliciems and Frost, 1989). Most of these evaluations suggest that acute myocardial
25 infarction related to overexertion is the primary cause of death during inclement weather.
26 In a 1978 blizzard in Rhode Island, emergency room admissions for myocardial infarction
27 rose markedly three days after the storm, and mortality from ischemic heart disease showed a
28 large increase for a five-day period after the storm (Faiche and Rose, 1979). One study even
29 notes that relatively mild temperatures, when snow is wet, represents the most dangerous
30 situation for the shoveller (Auliciems and Frost, 1989).
April 1995 12A-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 It is also suggested that storminess in winter increases mortality by confining people
2 indoors for considerable amounts of time, permitting transmission of infectious diseases such
3 as influenza and other microbial or viral infections (Richards and Marriott, 1974;
4 WMO/WHO/UNEP, in press). Thus, lag times between offensive weather and subsequent
5 related mortality in winter are much longer than in summer, with estimates of three to over
6 eight days (Kalkstein, 1988; Larsen, 1990a; Kunst et al., 1993). Causes for this lag are
7 unclear. Interestingly, the intraseasonal decreases in mortality found in summer do not seem
8 to exist in winter (WMO/WHO/UNEP, in press).
9
10 Social/Demographic Factors
11 As might be expected, the elderly, people with pre-existing health problems, low-
12 income groups, and urban residents are most at risk because of relatively low tolerance to
13 heat and cold (Kilbourne, 1989). There is also some evidence that the very young (less than
14 one year old) represent a high-risk category (Kalkstein and Davis, 1989). Elevated heat
15 related mortality among the elderly may be due to reduced perspiration efficiency, as well as
16 an inability to adapt to rapidly changing weather conditions (Kilbourne, 1989; Ramlow and
17 Kuller, 1990). In addition, many medications taken by the elderly and people with health
18 problems can increase the risk of heatstroke and other direct weather-related causes
19 (Kilbourne etal., 1982).
20 One very important potential confounder for heat-related mortality is air conditioning.
21 There is no doubt that the number of homes with air conditioning has increased rapidly over
22 the past 30 years. In St. Louis, estimates indicated that 40% of the homes had air
23 conditioning in 1965, and this number increased to 91 % by 1992 (Stern et al., 1993).
24 The impact of weather on heat and cold related mortality shows considerable spatial
25 variation. In summer, populations in the northeastern, midwestern, and Pacific regions of
26 the U.S. show the greatest sensitivity to weather (Kalkstein and Davis, 1989; Larsen,
27 1990a,b). These areas are noted for great variability in summer temperatures, with intense
28 heat waves imbedded within periods of relatively benign summer weather. The regions with
29 the weakest summer weather/mortality responses are the southeastern, continental south, and
30 southwestern portions of the country. These areas are noted for consistently hot weather
31 throughout the summer, with lower variability in temperatures than regions to the north.
April 1995 12A-11 DRAFT-DO NOT QUOTE OR CITE
-------
1 These disparities support the viewpoint that people respond to weather in relative, rather than
2 absolute, fashion. In areas where extremely hot weather occurs irregularly, the
3 weather/mortality response is significant, but in regions where heat is consistent and
4 commonplace, the relationships are considerably weaker.
5 Other studies have uncovered similar findings. An evaluation of two Chinese cities,
6 one hi a subtropical climate (Guangzhou) and the other in a more temperate climate
7 (Shanghai), uncovered similar results (Tan, 1992). The daily death rate in Shanghai
8 exceeded that in Guangzhou by about 50% during days with temperatures above the threshold
9 maximum temperature defined in the paper. An evaluation of ten cities in Canada yielded
10 similar results (WMO/WHO/UNEP, in press). Three cities with infrequent but intense heat
11 waves in summer (Toronto, Montreal, and Ottawa) demonstrated strong weather/mortality
12 relationships. However, there appears to be a northerly extent to this impact, and Canadian
13 maritime cities, such as Vancouver and St. Johns, demonstrated no relationship, even during
14 the warmest days. It appears that temperatures in the maritime locales never become warm
15 enough to elicit a heat/mortality response.
16 Thus, in developed countries, the greatest summer responses are noted in climates with
17 high variability in temperatures, and with heat sufficient enough to create physiological
18 stress. However, this may not be the case in developing countries. An evaluation of
19 seasonal heat related mortality in Cairo, Egypt indicates a linear increase in deaths as
20 maximum temperatures increase, although the response was less than that found in cooler
21 Shanghai (Intergovernmental Panel on Climate Change [IPCC], in press). With the lack of
22 air conditioning and other amenities, very hot or tropical cities in the developing world
23 present an environment where behavioral avoidance is virtually impossible, rendering the
24 impact of heat more extreme than in developed country counterparts.
25 These results suggest that physiological and behavioral acclimatization may occur with
26 exposure to sustained periods of high temperatures. Long-term acclimatization is apparent in
27 tropical and subtropical climates subjected to continuous hot weather (Diamond, 1991).
28 Short-term acclimatization is possible in mid-latitudes as well with exposure to extended
29 periods of hot weather, leading to reduced heat related mortality toward the end of the
30 summer (Marmor, 1975; Rotton, 1983). However, mortality displacement may play a role
31 in this reduction as well.
April 1995 12A-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 Acclimatization to a perpetually hot climate may occur during childhood and may be
2 linked to activation of a greater percentage of sweat glands than would occur in an individual
3 raised in a more moderate climate (Diamond, 1991). However, visitors from mid-latitude
4 climates who are unaccustomed to extremely hot conditions can acclimatize to hot weather
5 within a few days of exposure. Studies indicate that such individuals show an increased
6 ability to sweat, and they sweat sooner than those not exposed to continual heat (Candas,
7 1987; Diamond, 1991).
8 Some research has found that low-income groups are more susceptible to heat related
9 mortality (Jones et al., 1982), and a number of socio-economic issues have been evaluated to
10 determine their possible influence on mortality. The most complete of these is a
11 determination of the impact of poverty on mortality in four U.S. cities (Atlanta, New
12 Orleans, New York, St. Louis; Smoyer, 1993). The results suggest that poverty does play a
13 role in rendering segments of the population more sensitive to heat related mortality. With
14 the exception of New York, the non-poor do not exhibit significantly higher heat related
15 mortality. Poverty appears to be an especially strong factor in St. Louis, and to a lesser
16 degree, in New Orleans. Neither poor nor non-poor groups responded negatively to weather
17 in Atlanta. A particularly interesting response was noted in New York City, where both
18 poor and non-poor populations appear highly susceptible to heat related mortality.
19 In addition, these groups appear not to be susceptible within the surrounding less urban
20 counties. This suggests that in New York the "urban heat island" impact is considerably
21 stronger than any socio-economic factor in affecting heat-related mortality.
22 The Smoyer study also detected an age-related factor; the elderly poor seem particularly
23 vulnerable to heat related mortality. In addition, those areas with the highest population
24 density, lowest percentage of detached housing, and highest percentage of housing
25 constructed before 1939 contained the most susceptible populations (Smoyer, 1993). Another
26 evaluation which attempted to correlate heat related mortality with various socio-economic
27 characteristics for 15 U.S. cities generally concurred with the Smoyer findings (Chestnut
28 et al., 1993). In addition, cities with a large percentage of labor in manufacturing, a higher
29 proportion of houses with incomplete plumbing, and with low percentages of population with
30 at least a high school education also exhibited increased sensitivity to heat related mortality.
31 It is possible that these socio-economic characteristics are not causal mechanisms, but are
April 1995 12A-13 DRAFT-DO NOT QUOTE OR CITE
-------
1 instead coincidentally associated with older eastern and midwestern cities whose climatic
2 characteristics render them more vulnerable to high heat related mortality totals. However,
3 considering some of the poor amenities associated with substandard urban housing, such as
4 reduced ventilation, increased heat load, lack of vegetation or shade, and high population
5 densities, it is likely that some of these socio-economic factors play a role in exacerbating the
6 heat/mortality problem (Lowry, 1988; Meyer, 1991; Smoyer, 1993).
7 Race appears to play a minimal direct role, although blacks appear more vulnerable to
8 weather related mortality (Kalkstein and Davis, 1989). Physiological distinctions among the
9 races seem to have little impact, but non-whites may be particularly vulnerable as they are
10 concentrated within urban areas where behavioral avoidance is less possible. There is some
11 evidence that improper or inadequate home heating or insulation during occasional cold
12 waves in the South may render non-whites more vulnerable to cold related mortality there
13 (Rango, 1984). For this reason, the Centers for Disease Control have noted that non-white
14 elderly men constitute the highest risk group, especially if adequate housing is not available
15 (Centers for Disease Control, 1982).
16 Long-term comparisons of heat related mortality as air conditioning has become more
17 prevalent have yielded conflicting results. Several studies have not found significant
18 decreases in mortality over the past several decades as use has increased (Ellis and Nelson,
19 1978; Larsen, 1990a). It has been suggested that exposure to air conditioning may decrease
20 the body's ability to acclimatize, rendering air conditioning more harmful than good (Ellis,
21 1972). However, two more recent studies have determined that air conditioning plays some
22 role in lessening heat related mortality (Rogot et al., 1992; Kalkstein, 1993). One study
23 estimated that in New York, over 3,500 deaths were avoided during the 25 years of summers
24 between 1964 and 1988 because of the mitigating impact of air conditioning. This represents
25 a saving of over 21% of heat related deaths which would have occurred attributable to
26 weather without the benefit of air conditioning (Kalkstein, 1993). In locations where air
27 conditioning is prevalent even among less affluent residents, such as in certain U.S. cities,
28 there may be reduced effects of both weather and PM.
29 Estimates of mortality attributed to weather have been attempted for a number of cities
30 throughout the world (Table 12A-2; IPCC, in press). For some cities, the number of heat
31 related deaths in an average summer exceeds 200. Under certain extreme circumstances,
April 1995 12A-14 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12A-2. ESTIMATES OF PRESENT-DAY HEAT RELATED MORTALITY
DURING AN AVERAGE SUMMER
City
Atlanta
Chicago
Cincinnati
Dallas
Detroit
Kansas City
Los Angeles
Memphis
Minneapolis
New Orleans
Mortality
18
173
42
19
118
31
84
20
46
0
City
New York
Oklahoma City
Philadelphia
St. Louis
San Francisco
CANADA - Montreal
CANADA - Toronto
CHINA - Shanghai
EGYPT - Cairo
Mortality
320
0
145
113
27
69
19
418
281
Source: IPCC (in press).
1 annual summer mortality can be much higher. For example, during a 10 year evaluation for
2 Shanghai, the number of heat related deaths estimated for the hottest summer was 1,140
3 (Tan, 1992). Most, but not all, of these deaths were among the elderly. For the 15 U.S.
4 cities listed in Table 12A-2, about 80% of the heat related deaths were individuals of 65 and
5 older (Kalkstein, 1988). Other similar evaluations confirm that the majority of deaths were
6 among elderly individuals (e.g. Ramlow and Kuller, 1990). Although estimates for winter
7 mortality have been developed, these are considerably less reliable than those for summer.
8 Most winter estimates for excess mortality attributable to weather are considerably lower
9 than summer despite the generally higher winter total mortality, and one source puts them at
10 approximately 20% of summer values (Kalkstein, 1988). However, it is difficult to assess
11 the role of weather in contributing to outbreaks of droplet-borne diseases such as influenza.
12 If mortality from these causes is included, winter values could be inflated considerably.
13
14 Controlling for Weather in PM/Mortality Analyses: The Use of Synoptic Climatological
15 Methods
16 A number of procedures have been utilized to control for weather in PM/mortality
17 studies, and although the variety has been great, they generally suffer from common
18 shortcomings. First, many depend on arbitrary decisions to remove extreme weather events
19 from the dataset. The definition of extreme weather to include, for example, days above
April 1995 12A-15 DRAFT-DO NOT QUOTE OR CITE
-------
1 90 °F may be proper for a city in the north, but not for a locale further south. Thus, these
2 arbitrary delineations consider weather as an absolute, rather than a relative, factor affecting
3 human health. It is therefore possible that some stressful weather days are not identified,
4 contaminating a PM/mortality dataset which is considered controlled for weather. Second,
5 the use of weather "dummy variables" to control for meteorology within PM/mortality
6 analyses categorizes weather within groupings which may not duplicate meteorological
7 reality. Kalkstein et al. (1991, 1994) propose that the meteorology of a locale is defined by
8 discrete, identifiable situations, which represent frequency modes for combinations of
9 weather elements. Meteorological delineation that recognize the existence of such modes can
10 be used to control for weather within this context. Third, the use of mean weather elements
11 (e.g., mean daily temperature) does not permit a proper evaluation of, or control for, daily
12 weather extremes. Finally, most all consideration of weather in PM/mortality studies are
13 thermal (temperature), and, less frequently, moisture (humidity) dependent. This creates a
14 potential weather control problem, as certain meteorological phenomena, such as stormy
15 situations associated with mid-latitude cyclones, are not associated with thermal extremes, yet
16 may be very important contributors to acute mortality (Kalkstein et al., 1994). These are
17 rarely controlled for in PM/mortality studies, as they cannot be identified on the basis of
18 temperature and humidity.
19 A completely different approach is that adjustment for weather-related variables is
20 needed only insofar as it provides a basis for removing potential confounding of excess
21 mortality with PM and other air pollutants, and that any empirical adjustment for weather is
22 adequate. One of the most completely empirical methods for adjusting daily time series data
23 for covariates is by use of nonparametric functions, such as LOESS smoothers, generalized
24 splines, or generalized additive models (GAM), as demonstrated in (Schwartz, 1994 defgh,
25 1995a,b; Schwartz and MOrris, 1995). These are empirically satisfactory and may provide a
26 better fit to data than synoptic categories, but at the loss of a basis for defining weather
27 "episodes" as a characterization of duration of exposure.
28 Application of synoptic climatological procedures to control for weather has the
29 potential to compensate for these difficulties and add further insight by defining an entire set
30 of meteorological conditions which lead to increases in mortality. Many U.S. cities possess
31 a single "offensive" summer air mass associated with unusually high mortality (e.g.,
April 1995 12A-16 DRAFT-DO NOT QUOTE OR CITE
-------
1 Philadelphia, Table 12A-3). This "moist tropical" air mass in Philadelphia, possessing the
2 highest maximum and minimum temperatures, was also associated with the greatest standard
3 deviation in mortality of all air masses evaluated. Thus, although many days within the
4 offensive air mass were associated with high mortality totals, a number of days showed little
5 mortality increase. The greatest daily mortality totals during moist tropical air mass
6 incursions occurred as part of a lengthy string of consecutive days of the air mass, and when
7 minimum temperatures were particularly high. This type of information is vital when
8 controlling for weather in PM/mortality analysis.
9 Offensive air masses which lead to mortality totals significantly higher than the long-
10 term baseline have been identified for a number of U.S. cities (Table 12A-4). In most cases
11 moist tropical air masses were deemed offensive (especially in the East), but the very
12 oppressive "dry tropical" air mass was often associated with the greatest increases in
13 mortality, especially in New York, St. Louis, Philadelphia, and in southwestern cities
14 (Scheraga and Sussman, in press). In some cases, daily mortality totals are over 50% above
15 the baseline (WHO/WMO/UNEP, in press). The air mass analyses support the notion that
16 acute mortality increases only after a meteorological threshold is exceeded. This threshold is
17 not only temperature dependent; it represents an overall meteorological situation which is
18 highly stressful. It is noteworthy that most cities demonstrate only one or two offensive air
19 masses which possesses meteorological characteristics exceeding this threshold.
20 In a PM study where stressful weather days are removed from the data base, synoptic
21 categorization provides an efficient means to remove such days with greater security that
22 very few meteorologically offensive days are contaminating the remaining dataset. In studies
23 where weather is stratified based on certain meteorological elements, synoptic categorization
24 allows for a meteorologically realistic control, and may be preferable to the use of arbitrary
25 dummy variables when identifying meteorological conditions with an elevated mortality risk.
April 1995 12A-17 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 12A-3. MEANS AND STANDARD DEVIATION FOR SUMMER AIR MASSES IN PHILADELPHIA
00
Air Mass
Category Number
1
2d
3
4
5
6
7
8
9
10
Total Mortality
Mean 3 PM
Temperature
77.0
89.0
82.4
79.0
82.6
85.0
80.6
85.5
74.7
83.6
Mean
Mortalitya
-4.11
8.89
1.63
-4.43
-2.57
3.92
0.70
2.47
-4.49
0.13
Standard
Deviation
12.87
16.14
12.82
10.19
11.14
16.83
11.82
12.49
12.53
11.80
% of Top 50
Mortality5
2.00
46.00
14.00
0.00
4.00
14.00
2.00
8.00
0.00
6.00
% Top 50
% Frequency0
0.14
3.77
1.27
0.00
0.45
1.99
0.14
1.08
0.00
1.07
Mean
Mortality
-0.91
6.72
1.59
-2.82
-1.36
3.84
0.52
2.70
-2.56
1.28
Elderly Mortality
Standard
Deviation
9.99
12.58
10.76
9.33
11.10
13.77
10.41
9.88
10.39
10.86
% of Top 50
Mortality !
0.00
46.0
14.00
2.00
4.00
10.00
4.00
10.00
2.00
4.00
% Top 50
7o Frequency
0.00
3.79
1.28
0.23
0.45
1.42
0.67
1.33
0.31
0.67
£2 aValues are evaluated against a baseline of 0.
> bRepresents the percentage of top 50 mortality days within a particular synoptic category.
cRatio of percentage of top 50 days within the synoptic category over the seasonal frequency of the category. A number greater than one indicates that a larger
proportion of days in the synoptic category are among the top 50 mortality days than might be expected based on the frequency of the category.
d"Offensive" category.
Source: Kalkstein (1993).
-------
TABLE 12A-4. DAILY EXCESSIVE MORTALITY (SUMMER SEASON) DURING
OFFENSIVE AIR MASSES
City
Birmingham
Phoenix
Los Angeles
Riverside
San Francisco
Hartford
Tampa
Atlanta
Chicago
Indianapolis
Louisville
Boston
Baltimore
Detroit
Minneapolis
Offensive Air
Mass
MT
MT
MT
DM
DT
DT
MT
MM
MT*
DP
MT*
DT
MT*
MT*
MT*
MT*
MT*
DT
MT
DT
MT*
Mortality Above
Baseline3
+2
+ 1
+9
+3
+2
+9
+3
+ 1
+3
+4
+3
+9
+ 14
+3
+2
+8
+5
+ 10
+8
+4
+6
City
Kansas City
St. Louis
Newark
Buffalo
Nassau, NY
New York
Cincinnati
Columbus
Portland
Philadelphia
Providence
Memphis
Dallas-
Ft.Worth
Houston
San Antonio
Offensive Air
Mass
DT
MT*
DT
MT*
DT
MT*
MT*
DT
MT*
DT
MT*
MT*
MT*
DT
DT
MT*
MT*
DT
MT*
DT
DT
DT
Mortality Above
Baseline3
+5
+3
+ 15
+2
+6
+4
+3
+6
+5
+49
+30
+2
+3
+5
+32
+ 10
+7
+3
+ 1
+3
+8
+ 1
aMean daily deaths above the long-term baseline.
Air Mass Abbreviations: MT = Moist Tropical; DM = Dry Temperate; DT = Dry Tropical; MM = Moist
Temperate; DP = Dry Polar. Asterisks denote a particularly offensive subset of MT.
April 1995
12A-19 DRAFT-DO NOT QUOTE OR CITE
-------
i 13. INTEGRATIVE SYNTHESIS: KEY POINTS
2 REGARDING PM EXPOSURE, DOSIMETRY,
3 AND HEALTH RISKS
4
5
6 13.1 INTRODUCTION
7 This chapter concisely characterizes the hazard identification and exposure-dose-
8 response components of the risk assessment, integrating the key information and conclusions
9 from preceding chapters into a coherent framework upon which to base interpretations
10 concerning human health risks posed by ambient or near-ambient levels of paniculate matter
11 (PM) in the United States. Certain issues of direct relevance to standard setting are not
12 explicitly addressed in this document, but are instead analyzed in the Staff Paper prepared by
13 the Office of Air Quality Planning and Standards (OAQPS) as part of its regulatory analyses.
14 Such analyses include: (1) discussion of what constitutes an "adverse effect" relative to
15 developing the primary and secondary NAAQS; (2) exposure analyses and assessment of
16 consequent risk; and (3) discussion of factors to be considered in determining an adequate
17 margin of safety. Another issue directly pertinent to standard setting is identification of
18 populations at risk, which is basically a selection by EPA of the subpopulation(s) to be
19 protected by promulgation of a given standard. This issue is addressed only partially in this
20 criteria document. For example, information is presented on factors, such as preexisting
21 disease, that may biologically predispose individuals and subpopulations to adverse effects
22 from exposures to PM. The identification of a population at risk, however, also requires
23 information above and beyond data on biological predisposition, such as information on
24 levels of exposure, activity patterns, and personal habits. Such information is included in the
25 Staff Paper.
26 While more details are presented in the earlier chapters, this chapter characterizes key
27 uncertainties involved in the data and limitations on the conclusions that can be drawn from
28 them. Additionally, a new discussion expanding and placing into perspective key information
29 and conclusions is provided. Toward this end, the chapter is organized into several sections,
30 each of which covers one or more major components of an overall exposure-dose-response
31 description: (1) chemical and size characteristics of ambient PM; (2) ambient levels of PM
April 1995 13_1 DRAFT-DO NOT QUOTE OR CITE
-------
1 and related exposure aspects; (3) dosimetry of inhaled particles in the respiratory tract; (4)
2 qualitative and quantitative characterization of key health effects of PM; (5) characterization
3 of potential health effects of chemical components of ambient PM (e.g., acid aerosols); (6)
4 hypotheses concerning the biological bases for observed health effects; (7) identification of
5 population groups potentially at enhanced risk for health effects associated with ambient PM
6 exposure; and (8) implications of quantitative exposure-effects relationships for special risk
7 groups.
8
9
10 13.2 CHEMISTRY AND PHYSICS OF ATMOSPHERIC PARTICLES
11 "Paniculate matter" is a generic term for a broad class of chemically and physically
12 diverse substances that exist as discrete particles (liquid droplets or solids) dispersed in the
13 ambient atmosphere over a wide range of sizes. The chemical and physical properties of PM
14 vary greatly with time, region of the United States, meteorology, and source category, thus
15 complicating the assessment of healths and welfare effects.
16 Atmosphere particles originate from a variety of sources (such as combustion-generated,
17 photochemically produced, wind-blown dust, and plant products). Particle diameters range
18 from < 10 run to > 100 /mi. Particles are ubiquitous in the atmosphere and can be
19 biogenic or anthropogenic in origin. A complete description of the atmospheric aerosol
20 would include an accounting of the chemical composition, morphology and size of each
21 particle, and the relative abundance of each particle type as a function of particle size.
22 However, most often the physical and chemical characteristics of particles are measured
23 separately. The physical characteristics of aerosols are described by size distributions that
24 are determined by physical measures. Chemical composition is determined by analysis of
25 collected samples.
26
27 13.2.1 Size Characteristics
28 As discussed earlier in Chapters 3 and 4, the definition of fine and coarse particles is
29 an operational one based on the observations by numerous investigators, beginning in the
30 early 1970's, that mass size distribution measurements, when appropriately plotted, usually
31 yielded bimodal distributions with a minimum between 0.7 and 3.0 /un diameter (Whitby,
April 1995 13-2 DRAFT-DO NOT QUOTE OR CITE
-------
1
2
3
4
5
6
7
8
1978; Wilson, 1995). An idealized distribution, showing the normally observed division of
ambient aerosols into fine and coarse particles, is presented in Figure 13-1. (Aerosol refers
to a suspension of solid or liquid particles in air. However, aerosol is sometimes used to
refer to only the condensed phase portion when it is desired to emphasize that the discussion
refers specifically to particles small enough to be suspended in air.) There is some overlap
of fine and coarse particles in the intermodal region between 1 and 3 /mi.
o>
"
70
60
50
40
30
20
10
Fine Particles
Course Particles
TSP
HiVol
WRAC_
0.1 0.2
0.5 1.0 1 5 10 20
Aerodynamic Particle Diameter (Da), urn
50 100
-Total Suspended Particles (TSP)-
PM10 Fraction (0-10
\ Coarse
•Fine Fraction (<2.5 |im)->/< Fraction
( (2.5-10
Figure 13-1. Sampling fractions for an idealized ambient particulate mass distribution.
1 It is also possible to define portions of the distribution by sampler design or by an
2 upper or an upper and lower 50% cut-point. For example TSP or Total Suspended
3 Particulate Matter is defined by the design of the High Volume Sampler (hivol) which
April 1995
13-3
DRAFT-DO NOT QUOTE OR CITE
-------
1 collects all of the fine particles but only a portion of the coarse particles. The upper cut-off
2 size of the hivol depends on the wind speed and direction. Heroic measures, such as were
3 undertaken with the Wide Area Aerosol Classifier (WRAC), are required to collect the entire
4 coarse mode (Lundgren and Burton, 1995).
5 In order to avoid the lack of definition of the hivol's upper size cut-off and to focus
6 regulatory concern on those particles small enough to enter the lower respiratory tract, the
7 form of the PM National Ambient Air Quality Standard (NAAQS) was changed in 1987 from
8 TSP, as measured by the hivol, to PM10 (USEPA, 1987). PM10 samplers collect all of the
9 fine particles and a portion of the coarse particles. The upper cut-point is defined as having
10 a 50 % collection efficiency at 10±0.5 />im diameter. The slope of the collection efficiency
11 curve is defined in the Federal Register, Part 53. There are also samplers with upper cut-
12 points of 2.5 and 1.0 jum. The dichotomous sampler splits the sampled particles into a
13 smaller and larger fraction.
14 In an analysis reported in 1979, EPA scientists recognized the need to measure fine and
15 coarse particles separately (Miller et al., 1979). Based on the availability of a dichotomous
16 sampler with a separation size of 2.5 pirn, they recommended 2.5 /zm as the cut-point
17 between fine and coarse particles. Except for collection in clouds, fog, or at a relative
18 humidity very close to 100%, the PM2.5 sample will contain all of the fine particles.
19 However, especially in dry areas or during dry conditions, it may also contain a small but
20 significant fraction of the coarse particles (Wilson, 1995).
21 As more experimental evidence and theoretical understanding developed, it was
22 recognized that fine particles could be best understood as being composed of a nuclei mode
23 and an accumulation mode (Whitby, 1978). The nuclei mode can be seen as a completely
24 separate mode only under special conditions such as the oxidation of SO2 to sulfuric acid in
25 otherwise particle-free air or in the vicinity of sources of hot gases in which normally
26 condensed species have been vaporized. A size distribution measurement, made in
27 automobile traffic at the General Motors Proving Grounds, which demonstrates such a
28 separation, is shown in Figure 13-2 (Wilson et al., 1977). The background air, measured
29 before and after simulating freeway traffic on the proving ground track, indicated that all of
30 the nuclei mode and a small fraction of the coarse mode were generated by the traffic, but
31 that the accumulation mode and most of the coarse mode were more regional in nature.
April 1995 13-4 DRAFT-DO NOT QUOTE OR CITE
-------
Mechanically
Generated
DP (urn)
_ Nuclei Mode ^ ^Accumulation Mode.
Coarse Mode
^ ^ ^
Fine Particles ^ ^
TSP, Total Suspended Particulate Matter
PM25
Coarse Particles
^
^ £oar§e Fraction
ofPM
10
Figure 13-2. Measured mass size distribution showing particles in nuclei and
accumulation modes of fine particles. Also shown are transformation and
growth mechanisms (e.g., nucleation, condensation, and coagulation).
Source: Wilson et al. (1977).
1 Such studies have led to a better understanding of the formation mechanisms for fine
2 and coarse particles. In simplest terms, coarse particles are formed by breaking up bigger
3 particles into smaller particles. The coarse particle mode is sometimes called the dispersion,
4 mechanically-generated or comminution mode. However, as particles become smaller and
5 smaller, more and more energy is required to break them into smaller units. This establishes
6 a lower limit of approximately 1 /xm for coarse particles (Friedlander, 1977).
7 Fine particles are usually formed from gases. The processes of nucleation and growth
8 are shown in Figure 13-2. Nucleation involves the formation of very small particles from
9 gases. Low vapor pressure substances are generated in the gas phase by high temperature
April 1995
13-5
DRAFT-DO NOT QUOTE OR CITE
-------
1 vaporization or by chemical reaction in the atmosphere. Growth of particles in the nuclei
2 mode occurs by two processes; coagulation, in which two small particles combine to form a
3 larger particle, and condensation, in which gas or vapor molecules condense onto existing
4 particles. The rate of coagulation depends on particle number and particle velocity; the rate
5 of condensation depends on surface area. These parameters decrease rapidly as the particle
6 size approaches 1 jim. As a result, particles normally do not grow by these processes to
7 diameters above 1 /mi. Since particles tend to "accumulate" in the size range from 0.1 to 1
8 fjLm, this size range is called the accumulation mode.
9 Secondary fine particles are formed by the atmospheric conversion of gases into
10 particles. In one mechanism a gas is converted into the gaseous form of a material with a
11 low vapor pressure. One example is the oxidation of SO2 (sulfur dioxide) to H2SO4 (sulfuric
12 acid) which forms fine particles by coagulation. In a second mechanism, a gas is converted
13 into a second gas which can react further to form a low vapor pressure substance. An
14 example is the oxidation of NO2 (nitrogen dioxide) to HNO3 (nitric acid) which can react
15 with NH3 (ammonia) to form fine particles of NI^NC^ (ammonium nitrate). Acid gases in
16 the atmosphere, such as SO2 and HNO3, may also react with coarse particles (such as CaCO3
17 from soil, construction or demolition or NaCl from sea spray or road salt) to form salts.
18 These would still be considered coarse particles. Metals vaporized in high temperature
19 processes such as smelting and organic material vaporized in cooking, which coagulate or
20 condense without chemical reaction, form primary fine particles.
21 In addition to anthropogenic particles, resulting from human activities, fine PM also
22 contains particles of natural origin, i.e. atmospheric reaction products of gases such as
23 terpenes emitted by plants. Thus, the organic component of fine particles contains both
24 natural and anthropogenic material and both primary and secondary material. Combustion
25 products from the burning of gasoline and diesel fuel also give rise to fine particles.
26 However, the combustion of coal and heavy fuel oil yields both fine particles, from material
27 that is vaporized during combustion, and coarse particles, i.e. flyash, from non-combustible
28 material.
29 In recent years it has been realized that during periods of fog or near 100 % relative
30 humidity certain types of particles can grow by absorption of water and reach diameters of
31 up to five times their dry size. Particle size distributions which include particles subjected to
April 1995 13-6 DRAFT-DO NOT QUOTE OR CITE
-------
1 such conditions sometimes show an accumulation mode which is split into a condensation or
2 dry mode and a droplet or wet mode. This process is discussed in Section 3.7 of Chapter 3.
3 The nuclei or ultrafine mode, with particles from a few nm to 100 nm, has become of
4 interest lately because of its suggested biological activity in the lower respiratory tract (see
5 Sections 13.4, 13.6, and 13.7). Ultrafine particles always contribute most of the atmospheric
6 aerosol by particle number (90 to 99%). Such particles, however, rapidly coagulate into the
7 accumulation mode so they make a significant contribution to mass only under conditions in
8 which they are rapidly generated, such as in vehicular traffic.
9 Available data on the measurement of atmospheric coarse particles indicate that the
10 majority of the mass is usually associated with particles of 2 to 100 pm diameters, is
11 lognormally distributed with a mass mean diameter (MMD) of 10 to 30 /-tin and a geometric
12 standard deviation near 2. Fine particles generally have the majority of their mass in the 0.1
13 to 1.0 pm diameter and a geometric standard deviation of < 2.
14 There is generally a clear separation of atmospheric aerosol into two mass fractions-a
15 fine and coarse mode with a minimum in the 1 to 3 pm range. This idealized "bimodal"
16 ambient mass size distribution consist of both dispersed (coarse) and condensed (fine)
17 particles. The total mass, mass mean diameter and the mass fraction in the two modes varies
18 from location to location and from day-to-day, while the shape of the mass/size distribution
19 remains rather constant.
20
21 13.2.2 Chemical Characterization of PM10 Particles
22 Many measurements indicate that the chemical composition of coarse and fine particles
23 are distinct, as are the processes that affect their formation and removal (Figure 13-2).
24 Coarse particles are typically generated by mechanical processes such as grinding, wind, and
25 erosion and consist of substances such as soil dust, sea spray, plant fragments, tire wear
26 particles, and emissions from rock-crushing operations. The data consistently show that
27 crustal elements such as iron, silicon, and calcium are concentrated in the coarse fraction.
28 The major sources of coarse particles are windblown dust from soil, unpaved roads, piles of
29 material containing coarse dust, etc. and dust reentrained by turbulent air generated by traffic
30 on paved or unpaved roads. Coarse particles are also generated by demolition of buildings
31 and evaporation of sea spray. Pollen, mold spores, and parts of plants and insects are also
April 1995 13.7 DRAFT-DO NOT QUOTE OR CITE
-------
1 found in the coarse particle size range. The composition of the coarse particle minerals is
2 similar to the composition of the earth's crust, with major components being calcite and
3 silicon with major ingredients of iron and aluminum (Dodd et al., 1991).
4 Fine particle chemical composition differs greatly from that of soil and ocean derived
5 coarse particles and erosion. The major components of fine particles are sulfate, nitrate,
6 organic and elemental carbon, ammonium ion and a variety of trace elements. Metals and
7 organic carbon can be found in both size fractions.
8 Fine particle mass is frequently dominated by sulfates. Sulfates and associated water
9 can represent 75 to 80% of fine mass. Organic and elemental carbon can account for 15 to
10 20% of the mass. Sulfuric acid and its neutralization products with ammonia constitute a
11 major anthropogenic contribution to fine particle aerosol.
12 Several atmospheric aerosol species, such as ammonium nitrate and certain organic
13 compounds, are semivolatile and are found in both gas and particle phases. The gas-particle
14 distributions of semivolatile organic compounds depend on composition, vapor pressure, total
15 particulate surface area, and atmospheric temperature. Diurnal temperature fluctuations can
16 cause dynamic changes in gas-particle partitioning on a time scale of a few hours. This
17 process can lead to alteration of the chemical composition of atmospheric aerosols during
18 sampling, storage, or analysis by loss of volatile components. Water is an important
19 constituent of atmospheric particles. Hygroscopic or deliquescent particles can absorb water
20 in increasing amounts as the relative humidity increases. Some but not all of this particle-
21 bound-water can be lost when the sample is exposed to drier air.
22 The carbonaceous fraction of ambient PM consist of both elemental (EC) and organic
23 carbon (OC). Wood-burning fireplaces are a major source of both EC and OC. Diesels are
24 major sources of EC, also called black carbon. In areas where wood burning is significant,
25 more particulate elemental carbon is expected in winter than summer. Most other EC is
26 attributable to diesel motor vehicle sources. The concentration of EC varies significantly,
27 depending on the location and the season. Elemental carbon concentration in rural and
28 remote areas usually vary from 0.2 to 2.0 Mg/m3 and from 1.5 to 20 jiig/m3 in urban areas,
29 with average EC concentration around 1.3 and 3.8 jug/m3 for U.S. rural and urban sites,
30 respectfully. The distribution of EC emitted by automobiles is unimodal with 85% of the
April 1995 13-8 DRAFT-DO NOT QUOTE OR CITE
-------
1 mass in particles smaller than 0.12 /mi Dae. The ambient distribution of EC is bimodal, with
2 peaks in the 0.05 to 0.12 /mi and 0.5 to 1.0 /mi size ranges.
3 The organic component of the ambient aerosol both in polluted and remote areas is a
4 complex mixture of hundreds of organic compounds. The concentration of OC is around
5 3.5 /ig/m3 in rural locations and 5 to 20 /ig/m3 in polluted atmospheres. Organic carbon can
6 represent about 24% of the PM10 mass, with 30% beirig PM2 5 mass.
7 Primary carbonaceous (OC) particles are produced by combustion, chemical,
8 geological, and natural sources. Important sources in Los Angeles, CA included meat
9 cooking (21.2%), paved road dust (15.9%), fireplaces (14%) and noncatalyst equipped
10 automobiles (11.6%). Secondary organic aerosol material is formed in the atmosphere by
11 the condensation of low vapor pressure products of the oxidation of organic gases on the
12 already existing particles. Partial oxidation products of aromatics such as toluene contribute
13 as much as 60% of the secondary organic aerosol. Organic compounds accumulate mainly in
14 the submicrometer aerosol size range.
15 Trace elements that are found predominantly in the fine particle size distribution are
16 Na, Cs, Cl, Br, Cu, Zn, As, Ag, Cd, In, Sn, W, and Pb. Greater than 75% of their mass is
17 associated with particles of diameter less than 2 /mi. Metals which are found in both fine
18 and coarse modes are V, Cr, Mn, Fe, Co, and Se, while elements found primarily within
19 large particle distributions are Ca, Al, Ti, Sc, and La. Table 13-1 shows the concentration
20 ranges of some elements found in U.S. ambient air. Potential sources of trace metals found
21 in fine airborne particles are primarily anthropogenic and include combustion of coal and oil,
22 wood burning, waste inceneration, smelters, and metal mining production.
23 Fine particles emitted from oil and coal combustion are enriched with metals.
24 Vegetative burning, which includes residential wood combustion and forest fires, is a source
25 for release of trace elements into the atmosphere. Potassium is a characteristic trace element
26 present in fine particles from wood burning. Vanadium can be related to oil burning. Over
27 90% of the mass from geological material is in the coarse particle size fraction, while the
28 combustion-related source categories contained -90% of the mass in the PM2 5 fraction.
29 Available data indicate that about 50% of aerosol nitrate and 20% of aerosol sulfate is
30 associated with the coarse fraction of PM10 particles in marine-influenced environments (John
31 et al., 1990). Wolff (1984) found coarse particle nitrate to be 18 to 90% and coarse sulfate
April 1995 13.9 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-1. CONCENTRATION RANGES OF VARIOUS ELEMENTS
ASSOCIATED WITH PARTICIPATE MATTER IN THE
UNITED STATES ATMOSPHERE (ng/m3)
Elements
As
Cd
Ni
Pb
V
Zn
Co
Cr
Cu
Fe
Hg
Mn
Se
Sb
Remote
0.007 to 1.9
0.003 to 1.1
0.01 to 60
0.007 to 64
0.001 to 14
0.03 to 460
0.001 to 0.9
0.005 to 11.2
0.029 to 12
0.62 to 4160
0.005 to 1.3
0.01 to 16.7
0.0056 to 0.19
0.0008 to 1.19
Rural
1.0 to 28
0.4 to 1000
0.6 to 78
2 to 1700
2.7 to 97
11 to 403
0.08 to 10.1
1.1 to 44
3 to 280
55 to 14530
0.05 to 160
3.7 to 99
0.01 to 3.0
0.6 to 7
Urban (USA)
2 to 2320
0.2 to 7000
1 to 328
30 to 96270
0.4 to 1460
15 to 8328
0.2 to 83
2.2 to 124
3 to 5140
130 to 13800
0.58 to 458
4 to 488
0.2 to 30
0.5 to 171
Source: Schroeder et al., 1987.
1 to be 3 to 23% of the total mass of aerosol at five different continental locations. Milford
2 and Davidson (1987), summarizing the size distribution data for sulfates and nitrates, noted
3 that the sulfate and nitrate had a major mass peak less than 2 /xm, with a signficant secondary
4 mass peak in the PM10-coarse fraction size range (2.5 to 10 fim).
5 Chow et al. (1994) simultaneously measured PM2 5 and PM10 particles at nine sites in
6 the Los Angeles Basin during the summer and fall of 1987. PM10-coarse fraction mass was
7 about one-fourth of the PM2 5 mass in the fall and one-half in the summer; the ratio was
8 similar for most sites. Sodium, aluminum, silicon, calcium, and iron were abundant only in
9 the coarse particle fraction. Nitrate was equally divided between coarse and fine particles.
10 The ratio of PM10-coarse fraction sulfate to fine particle sulfate was 5% in the summer and
11 20% in the fall.
12 Sweet and Vermette (1993) concluded that the PM10 mass is about 50% higher in urban
13 areas than rural. The average concentration of most trace elements were 3 to 10 times
14 higher in urban areas, with peaks 100 times higher. Crustal material (e.g., Ca, Si, Al, Fe)
April 1995 13-10 DRAFT-DO NOT QUOTE OR CITE
-------
1 had their major concentration in the PM10 coarse particle fraction, while anthropogenic
2 material (i.e., Cu, Zn, Pb) had bimodal distribution, with 20 to 50% of total mass in the
3 PMio coarse particle fraction in both urban and rural areas. Fugitive emission sources
4 (contaminated soil, road dust, steel and coal dust) accounted for nearly all of the
5 anthropogenic coarse particle material.
6 A secondary mass peak contained within the PM10 coarse fraction size range occurs for
7 sulfate, nitrate, and anthropogenic trace metals. The sulfate and nitrate peaks are probably
8 due to chemical reactions on the surface of coarse particles. Urban areas must be considered
9 as major sources for anthropogenic trace metals present in the PM10-coarse fraction because
10 urban concentrations of trace metals exceed rural area concentration by 5 to 10 times.
11 PM10-coarse fraction particles should not be lumped together with all coarse particles into a
12 single category because the chemical composition of particles > 10 pm is different (e.g.,
13 several types of pollens).
14 In addition to the above distinctions, there are many other differences between fine and
15 coarse mode particles, as are summarized in Table 13-2.
16
17
18 13.3 UNITED STATES PM CONCENTRATIONS AND EXPOSURE
19 CONSIDERATIONS
20 As discussed in Chapter 6, the nonurban (IMPROVE/Nescaum) and urban (AIRs)
21 networks may be used to examine different patterns for PM concentrations in various regions
22 of the U.S. and to compare urban and non-urban concentrations. Of particular interest is the
23 observation of an urban excess of PM which varies both spatially and temporally.
24 Nationally, the fine fraction at non-urban sites ranges between 0.4 and 0.8 /im. The
25 highest fine fraction is recorded east of the Mississippi River where 75 % of the non-urban
26 mass is in particles < 2.5 jtg/m3 in size. Thus fine particles dominate the non-urban aerosol
27 concentration east of the Mississippi River. The mass of the fine fraction also exceeds the
28 mass of the coarse fraction at the non-urban northwestern sites. The fine fraction is lowest
29 in the southwestern United States (< 50%) particularly in the Spring (Quarter 2). Thus the
30 non-urban southwestern PM is dominated by coarse mass.
April 1995 13_ 11 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-2. COMPARISON OF FINE VERSUS COARSE MODE PARTICLES
Fine
Coarse
Formed from:
Formed by:
Composed of:
Solubility:
Gases
Chemical reaction
Nucleation
Condensation
Coagulation
Evaporation of fog and cloud
droplets in which gases have
dissolved and reacted
Sulfate, SO4=
Nitrate, NO^
Ammonium, NH^
Hydrogen ion, H*~
Elemental carbon, C
Organic compounds
PNA
Metals, Pb, Cd, V
Ni, Cu, Zn
Particle-bound water
Biogenic organics
Largely soluble, hygroscopic
and deliquescent
Large solids/droplets
Mechanical disruption
(crushing, grinding etc.)
Evaporation of sprays
Suspension of dusts
Resuspended dusts
Soil dust, street dust
Coal and oil fly ash
Metal oxides of Si,
Al, Mg, Ti, Fe
CaCO3, NaCl, sea salt
Pollen, mold spores
Plant parts
Largely insoluble and non-
hygroscopic
Sources:
Lifetimes:
Travel
Distance:
Combustion of coal, oil,
gasoline, diesel, wood
Atmospheric transformation
products of NOX, SO2, and
organics including biogenic
organics, e.g., terpenes
High temperature processes,
smelters, steel mills, etc.
Days
100'sof km
Resuspension of soil tracked
onto roads and streets
Suspension from disturbed
soil, e.g., farming, mining
Resuspension of industrial
dusts
Construction, coal and oil
combustion, ocean spray
Hours
10's of km
April 1995
13-12 DRAFT-DO NOT QUOTE OR CITE
-------
1 In the non-urban East, the fine aerosol is dominated by sulfur aerosol types (sulfate and
2 ammonium ion, and associated water) and organics, which together constitute about 80% of
3 the fine particulate mass. Over the northwest, organics dominate the non-urban fine particle
4 mass, especially during the winter season. In the southwest, fine soil is a major component,
5 accounting for 25 to 30% of the non-urban fine mass. Non-urban sulfate, in the eastern
6 United States, exceeds the concentration over the mountainous western states by factors of
7 five or more. Fine non-urban particle nitrates are most prevalent over California, exceeding
8 4 /tg/m3 at most sites. Organic carbon non-urban concentrations are higher over California
9 and northwestern sites, as well as eastern United States sites.
10 The eastern United States is covered by a widespread contiguous PM2 5 non-urban
11 concentration that ranges from 10 /ng/m3 in winter (Quarter 1) to 17 /xg/m3 in summer
12 (Quarter 3). The lowest non-urban PM2 5 concentrations are measured over the central
13 mountainous western states. The low concentrations are about 3 jug/m3, while the summer
14 values are around 6 pig/m3. Somewhat elevated PM2 5 concentrations are observed over the
15 southwestern border adjacent to Mexico as well as in California and the Pacific northwest.
16 The spatial pattern for non-urban PM10 concentrations is similar to the PM2 5, however,
17 the PM10 concentration exceeds the PM2 5 by up to a factor of two, depending on region and
18 season. Eastern non-urban PM10 concentrations range between 12 /ng/m3 in Quarter 1 and
19 25 /ig/m3 in Quarter 3. The lowest non-urban PM10 concentrations are measured over the
20 central mountain states (5 /xg/m3 in Quarter 1). Higher PM10 non-urban concentrations
21 between 10 to 20 /-ig/m3 are measured over the southwestern United States, as well as over
22 the Pacific states from California to the northwest.
23 The highest AIRs PM2 5 concentrations are reported over eastern urban industrial
24 centers, such as Philadelphia and Pittsburgh, where concentrations around 50 jtig/m3 exceed
25 those of non-urban counterparts by a factor of 2 to 3. The excess urban PM2 5
26 concentrations are confined to the immediately vicinity of urban centers. The PM2 5
27 concentrations at remote New England, over the southeastern United States, and over the
28 upper Midwest are within about 50% for AIRs PM2 5 and Improve/Nescaum PM2 5. This
29 indicates that over the eastern United States, a regionally homogenous background of PM2 5
30 concentration exists that has smooth spatial gradients. Superimposed on the smooth residual
31 patterns are local hot-spots (within a few miles of urban industrial centers) with 2 or 3 times
April 1995 13_13 DRAFT-DO NOT QUOTE OR CITE
-------
1 excess concentrations. The eastern United States PM2 5 is composed largely of secondary
2 sulfate, nitrate, and organic aerosols that are produced several days after the emission of
3 their gaseous precursors.
4 The reported AIRs PM2 5 concentrations over the Pacific States are generally higher
5 and average 20 to 50 /tg/m3. This is 5 to 10 times higher than comparison concentrations
6 from the Improve NESCAUM PM2 5 data set. The dramatic difference is due to urban-
7 industrial-agricultural centers verses remote national parks and wilderness areas.
8 Urban PM10 data from the AIRs network are summarized in Table 13-3. The annual
9 means for all regions show a decline from 1985 to 1993. The variability within the region is
10 much greater in the west than southeast and midwest. The eastern U.S. has summer urban-
11 peaks, while the west has fall and winter-peaks. The fine fraction is more predominate in
12 the east than the southwest or upper midwest. Thus, there are distinctive differences by
13 region for PM10 size fractions by mass.
14 The highest eastern United States urban PM10 concentrations are recorded in Quarter 3.
15 The peak concentrations are over the Ohio River Valley, stretching from Pittsburgh to West
16 Virginia, southern Indiana and St. Louis. In this region, the urban PM10 concentration over
17 the industrialized midwest exceeds 40 pig/m3. Additional hot-spots with > 40 /jg/m3 are
18 recorded in Birmingham, AL, Atlanta, GA, Nashville, TN, Philadelphia, PA, and Chicago,
19 IL. The summertime PM10 urban concentrations in New England and upstate Michigan are
20 < 20 /ig/m3. The spatial variability of urban PM10 over the eastern United States is
21 derived primarily from the varying emissions density of primary aerosols and precursors of
22 secondary aerosols.
23 The eastern PMi0 seasonality is rather pronounced, with winter concentrations
24 December through March of 24 ^g/m3 and July through August peaks of 35 /ig/m3. Fine
25 and coarse urban particles have different seasonal dynamics in the East. Fine particle mass
26 is bimodal, with a major peak in July and a smaller winter peak in January. The coarse
27 particle concentration shows a single broad peak over the warm season April through
28 October. Sixty percent at PM10 is in the sub 2.5 jon size range.
29 The mountainous states, west of the Rockies, show high urban PM10 concentrations
30 (> 50 /ig/m3) at localized hot-spots during the cold season including Salt Lake City, LA
31 basin San Jonquan Valley, and site in Wyoming, Oregon and Washington. The seasonality
April 1995 13-14 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-3. CHARACTERIZATION OF URBAN PM10 DATA FROM AIRS NETWORK BY REGION FOR THE
UNITED STATES
t— '
1
H- »
O
3
6
o
o
H
0
a
0
H
W
0
*>
n
PM10 (/xg/m3)
Region 1993 1985 SE)a
Northeast 22 36 30%
Southeast 24 32 17%
Industrial Midwest 25 38 28%
Upper Midwest 25 31 19%
Southwest 26 52 45%
Northwest 25 50 45%
Southern California 32 45 40%
"Standard deviation among monitoring stations within
bSeasonal range expressed as percent.
Seasonal
Seasonality Variation %
Summer Peak July 20 %b
Summer High 37%
July, August
High June, August 37%
Low November, February
Slightly Lower Levels 19%
December/January
April-June Peak October-
November Peak
August-September Dip
Peak December 36%
Low March-May
Peak November 27%
Low March
regions.
PM25/PM10 Influences
62% Canadian and Gulf
airmasses, local sources,
long range transport
58% Flat, poor regional
ventilation
59% Winter cold Canadian
airmasses. Summer moist
Gulf Coast masses
38% Agricultural Heartland
windblown dust influence
37% low precipitation, coarse
particle dominant, dust
contribution to PM10
59% Meteorology highly
variable
50% Air flow from Pacific, dry
summer, low in remote
Basin wide elevation
-------
1 of urban PM2 5 west of the Rockies is strongly peaked in December (40 /xg/m3) — about a
2 factor of four higher than the summertime values. The coarse urban fraction shows a broad
3 peak during late summer, July through October.
4 The eastern United States has summer urban peaks, while the west has fall and winter
5 urban peaks. The fine fraction is 60% of PM10 in the east and 50% in the west. Fine
6 particles dominate during the Winter season in the Western United States.
7 The urban PM10 concentrations exhibit marked differences in the shape of their
8 distribution functions around the mean values. The day-to-day variations of urban PM10
9 concentration in Knoxville, TN are about 40% of the mean value of 35 pig/m3, while
10 Missoula, MT shows a coefficient at variation of 60% of the mean of 34 /xg/m3. During the
11 winter season, the coefficient of variation is even higher. Thus, Missoula has a significantly
12 higher short-term variation.
13 The highest logarithmic standard deviation in concentrations is recorded over the
14 northeast and northwest states, during the cold season. Regionally, the logarithmic standard
15 deviation in the north-northwest is about 2.0, with pockets of high winter variability such as
16 Salt Lake City, Utah and Missoula, MT. In southern states, the summertime logarithmic
17 standard deviation is below 1.5, indicating a more uniform summertime PM10 level, while
18 the northern states are more epidsodic.
19 Tables 13-4a and 13-4b provide representative examples of current PM10 levels
20 observed in selected U.S. SMSAs for 1993.
21 Among key issues addressed in Chapter 7 were (1) the relative contributions of ambient
22 air PM exposures to total human exposure to particles (also including those derived from
23 indoor sources) and (2) the extent to which ambient air monitoring sites might serve as
24 reasonable indices (or proxy's) of variations in personal exposures of individuals evaluated in
25 community-level epidemiologic studies of PM health effects. Suffice it to say here that
26 widely varying percentages of total human exposure to particles can be attributed to those
27 from ambient sources, depending upon a number of factors (penetration of outdoor particles
28 to indoor space, presence/absence of indoor combustion sources, presence/absence of
29 smokers in home or place of work, etc.). Probably of most importance are findings that,
30 whereas very few coarse-mode ambient particles (> 2.5 /*m) penetrate into closed indoor air
31 spaces generally, substantial percentages of fine-mode particles (< 2.5 /xg/m3) indoors are
April 1995 13-16 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-4a. PM10 LEVELS BY ANNUAL AVERAGE FOR SELECTED U.S.
SMSAs FOR 1993
Metropolitan Statistical Area
Santa Fe, NM
Amarillo, TX
Santa Rosa, CA
Springfield, MD
Casper, WY
Danbury, CT
Glens Falls, NY
Titusville Area, FL
New London Area, CT/RI
Bridgeport, CT
Fort Lauderdale, FL
Asheville, NC
Montgomery, AL
Honolulu, HI
Oakland, CA
Charleston, SC
San Francisco, CA
Dallas, TX
Louisville, KY
Baltimore, MD
Birmingham, AL
Mobile, AL
Orange County, CA
Phoenix, AZ
New York, NY
1990
Population
117,043
187,547
388,222
239,971
61,226
187,867
118,539
398,978
266,819
443,722
1,255,480
174,821
292,517
836,231
2,082,914
506,875
1,603,678
2,553,362
952,662
2,382,172
907,810
476,923
2,410,556
2,122,101
8,546,846
PM10
WTD AM1
0*g/m3)
15
16
18
18
18
19
19
19
19
21
21
22
23
24
26
26
29
30
33
35
36
38
38
44
47
PM10
2nd Max2
0*g/m3)
35
29
52
39
41
46
44
57
41
50
71
58
48
58
71
58
72
74
73
70
85
71
80
92
86
03 (ppm)3
—
-
—
—
—
0.14
—
—
0.13
0.17
—
—
—
—
0.13
—
—
0.14
0.14
0.15
0.13
—
0.17
0.13
—
Weighted Annual Mean
2Highest Second Maximum 24-hour Concentration
3Highest O3 Second Daily Maximum 1-hour Concentration
April 1995
13-17 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-4b. SELECTED U.S. PM10 LEVELS BY 2ND MAX
PM10 FOR 1993
Metropolitan Statistical Area
St. Louis, MO
Los Angeles, CA
San Diego, CA
El Paso, TX
Medford, OR
Seattle, WA
Gary, IN
Flint, MI
Bakerville, CA
Fresno, CA
Denver, CO
Chicago, IL
Eugene, OR
Salt Lake City, UT
Spokane, WA
Pittsburgh, PA
Riverside, CA
Steubenville, OH
New Haven, CT
Provo, UT
Philadelphia, PA
1990
Population
2,444,099
8,863,164
2,498,016
591,610
146,389
1,972,961
604,526
430,459
543,477
667,490
1,622,980
6,069,974
282,912
1,072,227
361,364
2,056,705
2,588,793
142,523
638,220
263,590
4,856,881
PM10
2nd Max2
G*g/m3)
101
102
105
106
106
119
122
127
128
131
142
147
151
156
166
167
172
177
178
209
531
PM10
WTD AM1
0*g/m3)
44
47
34
37
41
35
34
24
54
53
41
47
28
42
46
38
73
40
52
40
34
Note
O33 0.13
O3 - 0.25, CO4-
14
O3 0.16
O3 0.14, CO-11
—
—
—
—
O3 0.16
O3 0.14
—
—
—
—
CO 12
SO25 0.155
O3 0.23
SO2 0.244
03 0.15
CO 10
O3 0.14
'Weighted Annual Mean
2Highest Second Maximum 24-hour Concentration
3Highest O3 Second Daily Maximum 1-hour Concentration
4Highest CO Second Maximum Non-overcapping 8-hour Concentration
5Highest SO2 Second Maximum 24-hour Concentration
April 1995
13-18 DRAFT-DO NOT QUOTE OR CITE
-------
1 often found to originate from outdoor ambient air (e.g., > 60% in certain circumstances).
2 Those include, for example, high percentages of sulfates derived from outdoor air and at
3 times metals and other typical ambient air fine-mode particle constituents.
4 Chapter 7 also provides analyses of relationships between human exposures to airborne
5 particles indexed by personal monitors versus either ambient or indoor fixed monitors.
6 Several earlier studies appeared to indicate very poor associations between personal PM
7 exposure (measured by personal monitors) and either ambient or indoor PM concentrations
8 measured by fixed monitor sites, with correlations for the former (ambient) being near zero
9 and those for the latter (indoor) being somewhat better but not high. More recent studies,
10 however, have begun to demonstrate higher correlations between personal monitoring results
11 and both indoor and outdoor fixed monitoring site data. Of particular importance is the
12 finding that, although individual exposures may not be well correlated with ambient PM
13 concentrations measured at representative fixed ambient monitoring site(s), the mean personal
14 exposure of people in a community in fact tends to be correlated to the ambient
15 concentration. Basically, in terms of community air pollution, a properly sited ambient PM
16 measurement is reasonably related to the mean personal PM exposure of the community,
17 although it will not be good a indicator of any single individuals' daily PM exposure. The
18 important consideration here is that the ambient monitor(s) be properly sited. This would
19 have to be evaluated study by study, which can be difficult or impossible if data has not been
20 reported in the published epidemiologic studies. There must be limits to the acceptability to
21 using a monitor for daily PM level changes, both in terms of distance from the population
22 and terrain between population and monitoring site.
23
24
25 13.4 DOSIMETRY OF INHALED PARTICLES IN THE RESPIRATORY
26 TRACT
27 Inhaled particles are deposited in the respiratory tract by several mechanisms:
28 impaction, sedimentation, interception, diffusion, and electrostatic precipitation. Ventilation
29 rates differ for various activity patterns in humans, for different ages, and among species.
30 These ventilation differences coupled with differences in upper respiratory tract structure and
31 in size, branching pattern, and structure of the lower respiratory tract among species and
April 1995 13_19 DRAFT-DO NOT QUOTE OR CITE
-------
1 between healthy versus diseased states result in significantly different patterns of airflow that
2 in turn affect particle deposition in the respiratory tract regions. For a given aerosol, the
3 two most important parameters determining deposition are the mass mean aerodynamic
4 diameter (MMAD) and the distribution of particles about that mean (ag}. Subsequent
5 clearance of a deposited dose is dependent on the initial site of deposition, physicochemcial
6 properties of the particles (e.g., dissolution half-time), and on time since deposition.
7 An accurate description of the exposure-dose-response relationship for PM should
8 account, to the extent possible, for these mechanistic determinants of particle disposition
9 (deposition, absorption, distribution, metabolism, and elimination). Mathematical dosimetry
10 models that incorporate descriptions of disposition of chemicals have been useful in
11 describing relationships between exposure concentration and target tissue dose, particularly as
12 applied to describing these relationship for exposure-dose-response assessment. Deposited
13 dose may be an appropriate metric for acute effects (e.g., mortality) especially if the particles
14 exert their primary action on the surface contacted. An alternative to consider is deposited
15 dose rate normalized to respiratory tract surface area because insoluble particles deposit and
16 clear along the surface of the respiratory tract. Depending on the availability of
17 morphometric information, other normalizing factors such as numbers of alveoli or
18 macrophages, could be explored. Chronic effects may be better described by retained dose
19 estimates, which represent the difference between deposition of particles and clearance.
20 The deposited dose and retained dose estimates calculated for typical ambient aerosols
21 and laboratory animal data show anticipated differences due to the influence of particle
22 diameter (dae) and distribution (ag), minute ventilation, and species-specific morphometry.
23 Figure 13-3 shows the fractional deposition in the tracheobronchial (TB) region for various
24 species and "normal augmenter" and "mouth breather" adult male humans for essentially
25 monodisperse (oe = 1.3) or polydisperse (oe = 2.4) aerosols of various particle diameters.
& &
26 Species differences in inhalability versus humans are marked, with larger diameter
27 aerosols having minimal deposition in laboratory animals. This may explain why some
28 animal studies with larger particles (e.g., 2 to 3 jum) require high concentrations to elicit any
29 effects. For the smaller particles, deposition fractions in the laboratory animals species are
30 actually higher than those in humans. This emphasizes the need to explore dose metrics
31 based on particle number because concentrations of particles in this region are very small by
April 1995 13-20 DRAFT-DO NOT QUOTE OR CITE
-------
0.3
0.25
I 0.2
§ 0.15
•E
CO
I 0.1
Q
0.05
0
Nose
Mouth
Rat
Mouse
Hamster
Guinea Pig
4 6
MMAD (urn)
10
0.3
0.25
| 0.2
£
| 0.15
*±
CO
I °-1
<
0.05
Og = 2.4
4 6
MMAD (urn)
8
10
•Nose
• Mouth
•Rat
Mouse
• Hamster
• Guinea Pig
Figure 13-3. Predicted tracheobronchial deposition fractions versus MMAD of inhaled
monodispersed (ag = 1.3) and polydisperse (ag = 2.4) aerosols (top and
bottom panels, respectively).
April 1995
13-21
DRAFT-DO NOT QUOTE OR CITE
-------
1 mass but extremely high by number. Deposition in the alveolar region is also significant for
2 the smaller diameter aerosols. Figures 13-4 and 13-5 show the retained dose (/*g/g lung
3 tissue) in the alveolar region for a small particle diameter and relatively monodisperse
4 aerosol (0.5 fim MMAD, ag = 1.3) and a larger particle diameter and more polydisperse
5 aerosol (2.55 /im MMAD, ag = 2.4), assuming two different dissolution-absorption half-
6 times of 10 days and 1,000 days. These retained fractions are different than for deposited
7 alveolar dose. Further, not only is the relationship between species different between the
8 two aerosols, but within each aerosol the relationship changes depending on the assumed
9 dissolution-absorption half-time. For the smaller diameter and monodisperse aerosol, the rat
10 and hamster are consistently lower than humans in retained alveolar particle burden than are
11 humans. For the larger diameter and polydisperse aerosol, rats and hamsters are more
12 similar to humans for soluble particles (dissolution-absorption half-time of 10 days) versus
13 the more insoluble.
14 Accounting for differences in dosimetry can change the relationship of apparent effect
15 levels among species. The predictions to follow are based on several assumptions and
16 models described in Chapter 10. A major assumption is that the concentrations involved are
17 below those that cause particle overloading or alter clearance rates through chemical-specific
18 toxicity. For example, if deposited dose is normalized to regional surface area, the same
19 exposure concentration of a 0.5 /nm MMAD and ag aerosol causing a TB effect in a rat
20 versus guinea pig would be adjusted by 9.39 and 0.79, respectively, to calculate the human
21 equivalent concentration (HEC). Based on the HEC values alone and not accounting for
22 tissue sensitivity, the guinea pig would thus appear to be more sensitive than the rat when the
23 HEC is calculated on this deposited dose metric. An alveolar dose based on retained burdens
24 and assuming a soluble aerosol (dissolution-absorption halt-time of 10 days) would result in
25 an HEC for this same aerosol (0.5 /im MMAD, a% = 1.3) of 0.22 and 7.84 for the rat and
26 guinea pig, respectively. For a relatively insoluble aerosol (dissolution-absorption half-time
27 of 1,000 days) with the same MMAD and ag, the HEC would be 0.49 and 2.76 for the rat
28 and guinea pig, respectively. To illustrate, if a 0.5 /mi MMAD aerosol with a ag of 1.3
29 caused an acute effect in the tracheobronchial region of rat at an exposure concentration of
30 100 /ig/m3, it would be predicted that 939 /ig/m3 would result in a similar tracheobronchial
31 deposited dose and thereby a similar effect in humans, if species sensitivity to a given dose
April 1995 13-22 DRAFT-DO NOT QUOTE OR CITE
-------
A 0.5 urn MMAD; <$ -1.3
Guinea Pig
Mouse
Monkey
Dog
Human
Hamster
Rat
1,0003
1 1 1 1
100 200 300 400
Days of Exposure
B 2.55 urn MMAD; cj, - 2.4
500
600
700
0.1
200 300 400
Days of Exposure
500
600
700
Figure 13-4. Predicted alveolar retained dose burden for 0.5 jun MMAD monodisperse
(ag = 1.3) aerosol (top panel A) and 2.55 /im MMAD polydisperse (
-------
I
Retained Dose Burden (ng/g Lung)
Retained Dose Burden (ng/g Lung)
i i inn i i i i urn i i i i inn
-------
1 were equal. For chronic exposures when retained dose has a greater influence, an alveolar
2 effect in a rat exposed to a 0.5 /xm MM AD aerosol with a ag of 1.3 at an exposure
3 concentration of 100 jiig/m3 would be predicted to be equivalent to a human exposed to 22
4 M§/m3' assuming a dissolution-absorption half-time of 10 days and equivalent species
5 sensitivity to retained dose.
6 These results clearly emphasize the need to adjust for interspecies differences in
7 physiologic factors such as inhalability, deposition, and clearance mechanisms as well as for
8 physicochemical characteristics of the aerosol, such as particle diameter, distribution, and
9 solubility. Although not modeled in these exercises, hygroscopicity of aerosols is also
10 anticipated to affect estimates of particle deposition and subsequent clearance.
11 The effects of particle diameter, distribution, and possibly hygroscopicity were also
12 evident in the simulations of deposition for three ambient aerosols in "normal augmenter"
13 and "mouth breather" adult males. As expected from experimental studies, mouth breathing
14 alters deposition fractions in the tracheobronchial and alveolar regions compared to nasal
15 breathing. More detailed analysis presented in Chapter 10 also showed differences in
16 regional deposition fractions between activity patterns and across modes of the real-world
17 trimodal aerosols. The differences shown in these simulations for deposition between normal
18 augmenter and mouth breather adult males and among different activity patterns highlights
19 the influence of ventilation rate and morphometry. This, in turn, points to the importance
20 of characterizing the differences between the genders and the impact of age on deposition.
21 The International Commission on Radiological Protection (ICRP) model has predicted
22 differences between children of 1 year and adults across particle diameters ranging from the
23 diffusion to aerodynamic range of approximately 2.5-fold in the tracheobronchial region and
24 2-fold in the alveolar region (ICRP66, 1994). Given the complexity of the deposition
25 patterns by age, selecting any one example is an oversimplification at this tune. For
26 example, the ICRP has predicted that at about 0.2 jum MM AD, infants have greater
27 deposition than adults, but at 2 /xm MMAD, adults have greater deposition than infants.
28 Differences in ventilation and morphometry for diseased states can also be anticipated.
29 There is a mismatch of ventilation and perfusion in lung diseases such as asthma,
30 emphysema, and COPD (Bates et al., 1971, Bates, 1989). In more severe stages of these
31 diseases, a small portion of the lung volume recieves most of the tidal breathing volume. In
April 1995 !3_25 DRAFT-DO NOT QUOTE OR CITE
-------
1 emphysema, approximately one-fourth of the lung receives three-fourths of the tidal volume
2 (Bates et al., 1971). These changes in ventilation and volume will result in some portions of
3 the lower respiratory tract receiving a greatly increased particle burden compared to others.
4 For example, Miller et al. (1995) predicted that a compromised human lung would have a
5 greater number of particles (0.1 /im dae) deposited per alveolus and per macrophage. Larger
6 particles (5 /xm dae) were predicted to have greater deposition per alveolus, per macrophage,
7 per surface area, and per ventilatory unit in the compromised human lung simulation. Also,
8 Anderson et al. (1990) have shown that the deposition of ultrafine particles in patients with
9 COPD is greater than that in healthy people.
10
11
12 13.5 KEY HEALTH EFFECTS OF PARTICIPATE MATTER
13 This section concisely discusses key findings from PM10 and PM2 5 studies that
14 examine mortality and morbidity associated with ambient short- and long-term exposure.
15
16 13.5.1 Short-Term PM Exposure Mortality Studies
17 The 1982 criteria document (U.S. Environmental Protection Agency, 1982) concluded
18 that the most clearly defined effects on mortality arising from exposure to PM have been
19 sudden increases in the number of deaths occurring, on a day-to-day basis, during episodes
20 of high pollution. The most notable of these occurred in the Meuse Valley in 1930, in
21 Donora in 1948, and in London in 1952. During the December, 1952 epidsode between
22 3,000 and 4,000 excess deaths were attributable to air pollution, with the greatest increase in
23 death from chronic lung disease and heart disease (United Kingdom Ministry of Health,
24 1954). In addition, the death rate increased most dramatically in those older than 45 years of
25 age. Additional episodes with associated increases in mortality occurred in London during
26 various winters from 1948 to 1962. Collectively, studies of these and other early episodes
27 left little doubt that airborne paniculate matter contributed to mortality associated with very
28 high concentrations of urban aerosol mixes dominated by combustion products (e.g., from
29 burning coal) and/or their transformation products (e.g., H2SO4).
30 Besides evaluating mortality associated with major episodes, the 1982 criteria
31 document (U.S. Environmental Protection Agency, 1982) also focused on studies of more
April 1995 13-26 DRAFT-DO NOT QUOTE OR CITE
-------
1 moderate day-to-day variations in mortality within large cities in relation to PM pollution.
2 Various methodological problems were identified in most of the studies, precluding
3 quantitative conclusions regarding exposure-response relationships of importance for deriving
4 air quality standards. Among the main problems noted were inadequate measurement or
5 control for potentially confounding variables and inadequate quantification of exposure to
6 airborne particles and other associated pollutants (e.g., sulfates or acid aerosols). Despite
7 such problems, the U.S. Environmental Protection Agency (1982) document concluded that
8 the then available studies collectively indicated that mortality was clearly and substantially
9 increased when airborne particle 24-h concentrations exceeded 1,000 /ig/m3 (as measured by
10 the Black Smoke, or BS, method) in conjunction with elevations of sulfur dioxide (SO^
11 levels in excess of 1,000 jig/m3 (with the elderly or others with severe preexisting
12 cardiovascular or respiratory disease mainly being affected). As for evaluation of risks of
13 mortality at lower exposure levels, the U.S. Environmental Protection Agency (1982)
14 concluded that studies conducted in London by Martin and Bradley (1960) and Martin (1964)
15 yielded useful, credible bases by which to derive conclusions concerning quantitative
16 exposure-response relationships.
17 The 1986 addendum to the 1982 criteria document (U.S. Environmental Protection
18 Agency, 1986) also considered several additional acute exposure mortality studies of London,
19 England during the 1958-1959 through 1971-1972 winter periods conducted by Mazumdar
20 et al. (1982), Ostro (1984), Shumway et al. (1983), and by EPA (later published in Schwartz
21 and Marcus, 1990). After reviewing these various data analyses, and taking into account the
22 previously reviewed London results and the above noted methodological considerations, the
23 following conclusions were drawn (U.S. Environmental Protection Agency, 1986):
24
25 (1) Markedly increased mortality occurred, mainly among the elderly and chronically
26 ill, in association with BS and SO2 concentrations above 1,000 jwg/m3, especially
27 during episodes when such pollutant elevations occurred for several consecutive
28 days;
29
30 (2) During such episodes, coincident high humidity or fog was also likely important,
31 possibly by providing conditions leading to formation of sulfuric acid (H2SO4) or
32 other acidic aerosols;
33
34 (3) Increased risk of mortality is associated with exposure to BS and SC^ levels in
35 the range of 500 to 1,000 /ig/m3, for SO2 most clearly at concentrations in
36 excess of =700 /ig/m3; and
April 1995 13-27 DRAFT-DO NOT QUOTE OR CITE
-------
1 (4) Convincing evidence indicates that relatively small, but statistically significant,
2 increases in the risk of mortality exist at BS (but not SO2) levels below 500
3 Mg/m3> with no indications of any specific threshold level having been
4 demonstrated at lower concentrations of BS (e.g., at < 150 jig/m3). However,
5 precise quantitative specification of the lower PM levels associated with mortality
6 is not possible, nor can one rule out potential contributions of other possible
7 confounding variables at these low PM levels.
8
9 In setting the current U.S. PM standards, the BS levels noted above were taken as indexing
10 particles roughly in the same size range as inhalable particles reaching tracheobronchial or
11 alveolar regions of the respiratory tract; and the U.S. 24 h primary NAAQS was set as 150
12 /xg/m3 PM10.
13 The decade or so since the 1986 criteria document (U.S. Environmental Protection
14 Agency, 1986) has been an active period for the reporting of time series analyses of
15 associations between human mortality and acute exposures to PM at concentrations at or
16 below the lower end of the range indexed by the above studies of London mortality or the
17 level of the current U.S. 24-h standard. Some utilized TSP or other measures (e.g., COH,
18 BS, etc.) as an indices of PM exposure, but during the last few years, the analyses have
19 mainly focused on PM10 as a measure of PM. This is because sufficient routine PM10
20 ambient measurement data became available for such statistical analyses to be conducted in a
21 wide variety of locales.
22 With regard to the results of the new time-series analyses, numerous investigators have
23 reported very small, but statistically significant associations between increased relative risk
24 for mortality and various indices of PM (e.g., BS, COH, TSP, PM10, PM2 5, etc.) for many
25 different cities in the United States and in other countries, as well. The elderly (>65 yr
26 old), particularly those with preexisting cardiopulmonary disease, were found to have
27 distinctly higher risks than younger age groups. The small relative risk estimates for PM
28 were consistently reduced when other likely important (potentially confounding) factors were
29 also controlled for in the models, with the PM association still usually remaining statistically
30 significant, although typically accounting for much less of the variance in mortality than did
31 temperature or combinations of variables used to index contributions of weather-related
32 factors to human mortality. Thus, qualitatively, the newly emerging database appears to
33 provide indications that polluted atmospheres containing relatively low concentrations of
April 1995 13-28 DRAFT-DO NOT QUOTE OR CITE
-------
1 particles may contribute (along with other factors) to a very small increase in relative risk for
2 human mortality, especially in the elderly with preexisting cardiopulmonary diseases.
3 However, it is important to note that differences in opinion exist within the scientific
4 community with regard to: (a) how adequately other likely important confounding factors
5 (including weather and copollutants) and/or other seasonal factors were controlled for across
6 the various newly available analyses; and (b) the interpretation of the meaning or
7 implications of reported associations between increased relative risk estimates and indices of
8 ambient PM concentrations. For example, introduction of one or more other commonly-
9 present ambient air pollutants (e.g., SO2, O3, CO, NOX) into models of PM effects generally
10 reduces the estimated PM effect, often by as much as 50% and, at times, to statistically non-
11 significant levels. In a few studies, however, the size of the PM effect remained essentially
12 the same or increased slightly upon introduction of other copollutants into the model.
13 Similarly, analyses of PM-mortality effects by season (winter, spring, summer, fall), as done
14 in only a few studies, have found varying patterns of PM-mortality effects being significant
15 in one or another season(s) but not all, with specific effective seasons differing from one
16 locale to another. The copollutant and seasonality analyses results, in particular, have led to
17 considerable debate in the scientific community, typified on the one hand by (a) skepticism
18 about the size and the "realness" of reported low-level PM effects and, on the other hand, (b)
19 countervailing views asserting that the effect of PM (or any other weakly contributing factor)
20 on mortality can be made to "disappear" by overspecification of applicable models (i.e., by
21 introduction of sufficient other, possibly extraneous, variables into the models or by more
22 detailed breakdown of data, e.g., by season, that may reduce the power to detect a PM
23 effect).
24 No clear resolution of this debate or "consensus" opinion in the scientific community
25 has yet crystallized, but some agreement appears to be emerging that the results for models
26 containing only PM and no other copollutants may provide upper bound estimates for effects
27 of ambient particle-containing mixes of pollutants, whereas results derived from analyses
28 including other copollutants and extensive controls for weather, seasonality, and/or other
29 likely important contributing factors should be viewed as lower-bound estimates of PM
30 effects (which may be 50% or more lower than the upper bound or even include zero). The
31 next several sections discuss key points regarding the derivation of quantitative estimates of
April 1995 13_29 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM-related mortality and morbidity effects, taking into account the types of uncertainties and
2 scientific debate just noted.
3
4 13.5.1.1 PM10 Relative Risk Analyses
5 This section discusses key findings from studies conducted since the PM criteria
6 document addendum (U.S. Environmental Protection Agency, 1986) that have employed
7 PM10 in their analyses of the human mortality effects of acute exposures to PM, as discussed
8 in more detail in Chapter 12. Some studies considered daily mortality in the entire
9 population (i.e., all ages) and some by cause; some also considered subpopulations (e.g, the
10 elderly).
11 Two earlier published summaries of the PM literature converted all results to a
12 PM10-equivalence basis and provided quantitative intercomparisons (Ostro, 1992; Dockery
13 and Pope, 1994). Other such summaries have been conducted using TSP as the reference
14 PM metric (Schwartz, 1991; Schwartz, 1994a), and considered many of the same studies
15 included in the two PM10-equivalent summaries. The results presented in such summaries
16 suggest about a 1 percent change in acute total mortality for a 10 ^g/m3 change in PM10, but
17 the estimates range from 0.3 to 1.6% (i.e., a factor of 5). While most of the 95%
18 confidence intervals (Cis) of these estimates overlap, Cis of the highest and lowest estimates
19 do not overlap, indicating significant differences between these estimates. Note that the
20 effects indicated for a 10 /tig/m3 PM10 change cannot be reliably converted to other PM
21 increments (e.g., 50 or 100 /ig/m3 PM10), as differences in model specification (e.g., linear
22 versus log models) will cause them to differ in their conversions to other particle
23 concentration levels. The reasons for the approximately five-fold effect estimate difference
24 noted among studies are not obvious from the information provided by these references, but
25 one factor does appear to be the PM exposure averaging time, as estimates using multiple
26 day PM10 averages are all seen to be 1% or higher. This is not unexpected, given that any
27 lagged effects from prior days of PM10 exposure will be added to the effects estimate when a
28 multi-day average is employed, increasing the estimated effect on a per j*g/m3 basis.
29 It is also important to note that other air pollutants were generally not addressed in
30 deriving the coefficients reported by the above summaries. Differences among coefficients
31 are to be expected, given that the composition (and, potentially toxicity) of the PM, as well
April 1995 13-30 DRAFT-DO NOT QUOTE OR CITE
-------
1 as the demographic characteristics in each city, can be expected to differ. Moreover, the
2 conversions from other PM metrics to PM10 must necessarily introduce additional
3 uncertainty. However, though not all of these results may therefore be the most appropriate
4 available for quantifying a PM10 effect, they do consistently indicate that there is an
5 association between acute exposure to ambient air mixes containing PM and increased daily
6 mortality. Moreover, the by-cause results also reported in the summaries indicate that PM
7 effect estimates are greater for respiratory causes, which lends support to the biological
8 plausibility of the noted PM associations.
9 In an effort to more clearly quantify daily PM10-total acute mortality associations, Table
10 13-5 summarizes the total mortality relative risks (RR) of a 50 /ig/m3 increase in PM10
11 estimated from nine studies reviewed in Chapter 12, which employed PM10 data in their
12 analysis of total mortality data (or which had on-site PM10 reference data to convert other
13 PM metrics with more certainty). The studies listed were selected for this analysis mainly
14 because they can most readily be intercompared and provide direct data related to PM10
15 levels. The RR's calculated were based upon a 50 /ig/m3 increase above the mean PM10
16 24-h concentration, which is approximately the order of magnitude of the typical difference
17 between the mean and maximum in these cities. This is noted because in non-linear models
18 such as were often employed in the studies discussed in Chapter 12, the RR estimate
19 associated with a given /ig/m3 PM10 increase will vary depending upon the baseline
20 concentration to which it is added.
21 From results presented in Chapter 12, it is apparent that these studies generally have
22 yielded at least marginally significant PM10 coefficients, but the resultant excess risk
23 estimates differ by a factor of five across these studies (from 1.5% to 8.5% per 50 /xg/m3).
24 The mean and maximum PM10 concentration data are noted for each study. If the PM10
25 coefficient decreased as the mean level of PM10 decreased, then confounding as a function of
26 varying PM level would be suggested. However, the data presented indicate that the
27 variability in coefficients is not a function of PM10 level, as sites with high or low PM10
28 concentrations can report either high or low RR's. In Chapter 12, the statistical
29 methodology characteristics of each study were concisely summarized, in order to determine
30 if any factors are important to help explain the variability observed from study to study in the
31 PM10 RR estimate. As noted earlier, the RR estimate for acute mortality associated with
April 1995 13.31 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-5. COMPARISON OF RELATIVE RISK (RR) ESTIMATES FOR TOTAL MORTALITY FROM 50 jig/m3
CHANGE IN PM10, USING STUDIES WHERE PM10 WAS MEASURED OR WAS CALIBRATED FOR THE SITE
SO
so
H-
1
OJ
10
o
§>
H
6
o
o
H
O
o
Study
Utah Valley, UT
St. Louis, MO
Kingston, TN
Birmingham, AL
Athens, Greece
Toronto, ON Canada
Los Angeles, CA
Chicago, IL
Santiago, Chile
Chicago, IL
Reference
Pope et al.(1992)
Dockery et al. (1992)
Dockery et al. (1992)
Schwartz (1993)
Touloumi et al. (1994)
Ozkaynak et al. (1994)
Kinney et al. (1995)
Ito et al. (1995)
Ostro et al. (1995a)
Styer et al. (1995)
PM10
Mean
47
28
30
48
78
40
58
38
115
37
(/ig/m3)
Maximum
297
97
67
163
306
96
177
128
367
365
Other Pollutants
In Model
None
None, Winter
None, Summer
Max O3, Summer
Avg O3, Summer
None
03
None
03
None
None
SO2, CO
None
None
03, CO
03, CO
None
None
None, Poisson
SO2, Poisson
NO2, Poisson
O3, Poisson
None
Lag Times, d
< 4d
< 4d
< 4d
< 4d
< 4d
< 3d
< 3d
< 3d
< 3d
< 3d
1 d
1 d
Od
1 d
1 d
< 3d
1 d
< 4d
1 d
1 d
1 d
1 d
3d
RRper
50 /ig/m3
1.08
1.085
1.11
1.19
1.14
1.08
1.06
1.085
1.09
1.05
1.034
1.015
1.025
1.025
1.017
1.025
1.04
1.07
1.0221
1.0261
1.0431
1.0261
1.04
95 Percent
Confidence Interval
(1.05, 1.11)
(1.03, 1.14)
(0.92, 1.35)
(0.96, 1.47)
(0.92, 1.41)
(1.005, 1.15)
(0.98, 1.15)
(0.94, 1.25)
(0.94, 1.26)
(1.01, 1.10)
(1.025, 1.044)
(1.00, 1.03)
(1.015, 1.034)
(1.00, 1.055)
(0.99, 1.036)
(1.005, 1.05)
(1.005, 1.06)
(1.04, 1.10)
(1.003, 1.042)
(1.005, 1.047)
(1.020, 1.066)
(1.005, 1.047)
(1.00, 1.08)
3
O
HH
3
'Calculated on a basis of 50 /ig/m3 increase from 50 to 100 /ig/m3.
-------
1 short-term exposure to PM10 is usually larger when other pollutants have not been
2 simultaneously considered in the model. Those studies which considered PM10 both alone
3 and with other pollutants in the model often yielded smaller, and usually more marginally
4 significant, PM10 relative risks when other pollutants were also considered. This influence
5 ranges from roughly a 20 to 50 percent reduction in the estimate of excess risk associated
6 with PM10 (e.g., in Athens, Greece, the PM10 RR declines from 1.07 to 1.03 per 100 /xg/m3
7 when other pollutants are considered). Such a reduction is to be expected when co-linear
8 variables are added.
9 Older studies using BS or TSP often found a degree of correlation between S02 and the
10 PM indicator that reduced the apparent PM effect and attenuated its statistical significance
11 (Schwartz and Marcus, 1990; Schwartz and Dockery, 1992a,b; Moolgavkar et al., 1995a,b).
12 However, studies using a variety of PM indicators at cities were SO2 levels were so low as
13 to have little likelihood of SO2 being a significant confounder of a PM effect (Fairley et al.,
14 1991, for Santa Clara County; Pope et al., 1992, for Utah County) found quantitatively
15 similar significant PM effects. While there is some possibility that summertime PM effects
16 may be partially confounded with those of other pollutants (e.g., O3) derived from motor
17 vehicle fuel combustion or transformation products (Pope, 1994; Ostro et al., 1995;
18 Moolgavker et al., 1995a,b; Dockery, 1994), winter effects of PM are clearly detectable
19 when O3 levels are much lower (Moolgavkar et al., 1995b; Pope, 1994; Schwartz, 1995b).
20 If PM effects on mortality were so completely confounded with those co-pollutants so as to
21 be undetectable, then one would need to invoke many different confounders in different
22 studies of communities. While this explanation is not impossible, it appears highly unlikely,
23 but'cannot be precluded altogether since PM may derive from different sources in these
24 studies, have varying size and chemical composition from one locale to another, and
25 therefore may have different characteristics that affect health outcomes such as mortality.
26 Another factor which clearly affected the PM10 RR was the PM10 averaging period.
27 Most of the studies which utilized multi-day averages of PM10 in their regressions (i.e., Utah
28 Valley; St. Louis; eastern TN; Santiago; Chicago; and Birmingham) are among the higher
29 RR estimate studies. As discussed above, this would be expected. However, the increase
30 indicated for these studies is not proportional to the averaging time. Indeed, in sub-analyses
31 included by Pope et al (1992), the PM10 mortality risk is indicated to be roughly doubled by
April 1995 13.33 DRAFT-DO NOT QUOTE OR CITE
-------
1 using a five day average versus a single day concentration, while sub-analyses presented by
2 Ostro et al. (1995) for Santiago also indicate approximately a doubling in the PM10 RR when
3 a 3 day average is considered (i.e., from RR = 1.04 for a single day PM10 value to RR =
4 1.07 for a 3d average PM10 value). This may be due to the fact that, since autocorrelation
5 exists in the PM10 concentrations from day to day, the single day concentration is "picking
6 up" some of the effect of multi-day pollution episodes, even though they are not explicitly
7 modeled. These results suggest that a multi-day rather than a single-day average PM10
8 concentration may provide a more relevant index to gauge the effects of short-term PM
9 exposures occurring over several consecutive days.
10 Table 13-5 shows that the total acute mortality relative risk estimate associated with a
11 50 /xg/m3 increase in the one-day 24-h average PM10 can range from 1.015 to 1.085,
12 depending upon the site (i.e., the PM10 composition and population demographics) and also
13 upon whether PM10 is modeled as the sole index of air pollution or not. Relative Risk
14 estimates with PM10 as the only pollutant index in the model range from RR = 1.025 to
15 1.085, while the PM10 RR with multiple pollutants in the model range from 1.015 to 1.025.
16 As noted earlier, the former range might be viewed as approximating an upper bound of the
17 best estimate, as any mortality effects of co-varying pollutants are likely to be "picked up"
18 by the PM10 index. On the other hand, the latter multiple pollutant model range might be
19 viewed as approximating a lower bound of the best estimate, as the inclusion of highly
20 correlated covariates may weaken the PM10 estimate. Both estimates should be considered in
21 assessing the potential effects of PM10. Overall, consistently positive PM-mortality
22 associations are seen throughout these analyses, despite the use of a variety of modeling
23 approaches, and after controlling for major confounders such as season, weather, and
24 co-pollutants, with the 24-h average 50 /ig/m3 PM10 total mortality effect estimate most
25 likely falling in approximately the RR = 1.025 to 1.05 range (representing an expected 2.5
26 to 5.0% increase in risk of death over daily background mortality rates for which a 50 /ig/m3
27 increment in ambient PM10 concentration could be a contributing factor).
28 It is logical to assume that the bulk of the total mortality effects suggested by these
29 studies are among the elderly. During the historic London, 1952 pollution episode the
30 greatest increase in the mortality rate was among older citizens and those having respiratory
31 diseases. An analysis by Schwartz (1994c) of mortality in Philadelphia, PA during 1973
April 1995 13-34 DRAFT-DO NOT QUOTE OR CITE
-------
1 through 1980 comparing mortality during the 5% highest versus the 5% lowest TSP days
2 found the greatest increase in risk of death in those aged 65 to 74 and those > 74 year of age
3 (mortality risk ratios = 1.09 and 1.12, respectively, between high and low TSP days). Also,
4 in their time series analyses of Philadelphia daily mortality during this period, Schwartz and
5 Dockery (1992) found a TSP-mortality coefficient which was significantly higher (6 =
6 0.000910 + 0.000161) for persons older than 65 years of age than they found for the
7 younger population (6 = 0.000271 ± 0.000206). These coefficients indicated an effect size
8 for the elderly which was roughly three times that for the younger population (10% versus
9 3%, respectively, for a 100 /tg/m3 increase in TSP). In addition, two recent PM10 analyses
10 have directly considered the question of PM10-mortality associations among the elderly
11 population (> 65 years of age), and these provide further relevant insights into this question.
12 The first of these two analyses which directly considered the elderly when looking at
13 PM10-mortality associations was conducted by Saldiva and Bohm (1994) during May 1990
14 through April 1991 in Sao Paulo. Multiple regression models, including models with all
15 pollutants considered simultaneously, consistently attributed the association found with
16 mortality among the elderly to PM10. The PM10 relative risk reported in this study (RR =
17 1.13 for a 100 /ig/m3 increase) is higher than noted above for total mortality studies
18 addressing multiple pollutants (100 /ig/m3 RR = 1.03 to 1.05), supporting past observations
19 that the elderly represent a population especially sensitive to the health effects of air
20 pollution. A second recent study directly examined the association between PM10 and
21 mortality in the elderly population in Santiago, Chile (Ostro, 1995). In the overall
22 population, the PM10 100 jwg/m3 RR estimate was 1.08, but in the population aged 65 and
23 greater, the PM10 RR of a 100 /tg/m3 increase in PM10 rose to an estimate of RR = 1.11 in
24 the same model specification. Thus, these directly comparable estimates (i.e., using the
25 same model specification and population) suggest that the elderly experience roughly a 40
26 percent higher excess risk from exposure to PM air pollution than the overall population.
27 Overall, considering the historical pollution episode evidence and the results of recent
28 PM10-mortality analyses evaluating elderly populations, it seems evident that elderly adults
29 represent a population especially at risk for mortality implications of acute exposure to air
30 pollution, including PM.
April 1995 13_35 DRAFT-DO NOT QUOTE OR CITE
-------
1 Relatively few studies have directly examined the PM-mortality association in children.
2 It is difficult, given the limited and somewhat conflicting results available at this time, to
3 ascribe any such association to low-level PM pollution in particular. This is an area where
4 further research is clearly needed to broaden the base upon which to assess the potential for
5 PM to increase mortality among children.
6 Throughout the results and discussions presented above and in more detail in Chapter
7 12 regarding the effects of acute PM exposure on human mortality, a consistent trend was for
8 the effect estimates to be higher for the respiratory mortality category. This lends support to
9 the biological plausibility of a PM air pollution effect, as the breathing of toxic particles
10 would be expected to most directly affect the respiratory tract, and these results are
11 consistent with this expectation. For example, the respiratory mortality relative risk
12 estimates discussed in detail in Chapter 12 are all higher than the risks for the population as
13 a whole. Of particular interest is to compare the relative risk values for each study that
14 made most direct and appropriate comparisons. In the Santa Clara study (Fairley, 1990), the
15 PM-respiratory mortality RR was 4.3 times as large as for deaths as a whole (i.e., 3.5/0.8).
16 For Philadelphia, (Schwartz and Dockery, 1992a), the PM (TSP)-respiratory mortality RR
17 was 2.7 times as large as for total mortality (i.e., 3.3/1.2). In the Utah Valley study (Pope
18 et al., 1992), the PM10-respiratory mortality RR was 2.5 times as large as for deaths as a
19 whole (i.e., 3.7/1.5). In the Birmingham, AL study (Schwartz, 1994), the respiratory
20 mortality RR of PM10 was 1.5 times as large as for deaths as a whole (i.e., 1.5/1.0). More
21 recently, the Santiago, Chile PM10 study by Ostro et al. (1995) reported excess respiratory
22 mortality RR of PM10 to be 1.8 times as large as for deaths as a whole (comparing 1.15/1.08
23 RR per 100 /ig/m3). Thus, in these studies, the PM RR for respiratory diseases is indicated
24 to range from 50 to over 400% higher for respiratory disease categories than for all causes of
25 death, indicating that increases in respiratory deaths are a major contributor to the overall
26 PM-mortality associations noted previously. Moreover, since evidence suggests that an acute
27 pollution episode is most likely be inducing its primary effects by stressing already
28 compromised individuals (rather than, for example, inducing chronic respiratory disease from
29 a single air pollution exposure episode), the above results indicate that persons with
30 pre-existing respiratory disease represent a population especially at risk to the mortality
31 implications of acute exposures to air pollution, including PM.
April 1995 13-36 DRAFT-DO NOT QUOTE OR CITE
-------
1 In overall summary, the time-series mortality studies reviewed in this and past PM
2 criteria documents provide reasonably strong evidence that increases in daily human mortality
3 are associated with short-term exposures to air pollution mixes containing elevated PM
4 levels. Recent studies provide indications that small increases in such risk occur in
5 association with air pollution indexed by moderate increases of PM10 (-50 ^g/m3) above
6 routine ambient levels averaging around 50 to 100 /*g/m3. Overall, the PM10 relative risk
7 estimates derived from the most recent PM10 total mortality studies suggest an acute
8 exposure effect on the order of RR = 1.025 to 1.05 in the general population for increases
9 in ambient air pollution indexed by a 24-h average 50 /xg/m3 PM10 increment, with higher
10 (30-40%) relative risks indicated for the elderly sub-population and for those with
11 pre-existing respiratory conditions.
12
13 13.5.1.2 Fine Particles/Acid Aerosols Relative Risks
14 As noted earlier in this document (Chapters 11 and 12) some epidemiologic and
15 experimental toxicology data point toward fine particles as a class or certain constituents
16 (e.g., acidic aerosols) as possibly being key contributors to observed PM-mortality and/or
17 PM-morbidity associations. Only a few epidemiologic studies provide direct comparisons
18 between various PM indices, including fine particle and acidity measurements.
19 For example, Dockery et al. (1992) investigated the relationship between multiple air
20 pollutants to include PM2 5, and total daily mortality during a one year period in St. Louis,
21 MO and Kingston/Harrirman, TN and surrounding counties, as discussed in Chapter 12.
22 The distribution of mortality, and air pollution between September 1, 1985 and August 31,
23 1986 are presented for St. Louis and Eastern Tennessee in Table 13-6.
24 In Poisson regressions controlling for weather and season (Table 13-7), previous day's
25 PM10 was the only significant predictor of daily mortality (6 = 0.00175 + 0.00067), but the
26 association dropped off at 3 days (B = 0.00042 ± 0.00063). The size-fractionated PM data
27 were examined to determine whether this association could be attributed to either the fine
28 (PM2 5, aerodynamic diameter da < 2.5 pirn) or the coarse (2.5 /im < da < 10 /xm)
29 component of the PM10 mass. The fine fraction (PM2.5) was positively associated with
30 mortality (B = 0.00171 + 0.00096, P = 0.075). Coarse particles were also positively
April 1995 13_37 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-6. MEAN VALUES OF VARIABLES RELATED TO DAILY MORTALITY IN ST. LOUIS AND
EASTERN TENNESSEE
H- >
VO
VO
I— '
to
oo
o
Variable
Mortality
Deaths/day
Paniculate pollution
PM10 (fig/m3)
PM2 5 (/xg/m3)
SO4 (jug/m )
H+ (nmole/m3)
aNumber of days of measurements.
bStandard deviation.
cMinimum-maximum.
TABLE 13-7.
Na
365
311
312
311
220
COMPARISON
St. Louis
Mean ± SDb Range0
56.0 ± 8.2 31-81
27.6 ± 14.9 1-97
17.7 ± 10.0 1-75
8.0 ± 5.1 0-38
9.7 ± 12.0 0-88
OF UNIVARIATE REGRESSIONS WITH
^ CONTROLLING FOR WEATHER AND SEASON
6
0
1
o
§
o
n
H
w
Pollutant
PM10 (Mg/m3)
PM2 5 (/tg/m )
SO^2 (/*g/m3)
H+ (nmole/m3)
ft
0.00150
0.00171
0.00608
0.00086
St. Louis
(SE) t
(0.00069) 2.17
(0.00096) 1.78
(0.00577) 1.05
(0.00118) 0.73
N
365
330
331
330
232
POLLUTION ON
VARIABLES
ft
0.00160
0.00228
0.008
0.00017
Eastern Tennessee
Mean ± SD
15.5 ± 4.2
30.0 + 12.1
21.0 ± 9.4
8.7 ± 5.0
36.1 ± 39.6
PREVIOUS DAY
Eastern Tennessee
(SE)
(0.00149)
(0.00186)
(0.012)
(0.00055)
Range
5-29
4-67
4-58
1-27
0-90
t
1.07
1.23
0.67
0.31
-------
1 associated (6 = 0.00247 ± 0.00129, P = 0.056). Neither fine nor coarse particles showed
2 a stronger association than the other when considered simultaneously.
3 Both daily SO4 and H+ concentrations were significantly correlated with PM10 (Pearson
4 correlations 0.52 and 0.76, respectively). Sulfate (SO42~) as measured by the sulfur fraction
5 of PM10 (/3 = 0.00608 ± 0.00577) and H+ (B = 0.00086 + 0.00118) were positively, but
6 not significantly, associated with daily mortality. Among the other paniculate elemental
7 concentrations, those correlated with PM10 concentrations were also associated with
8 mortality. In particular, aluminum, calcium, chromium, iron, and silica all had correlations
9 with PM10 of 0.5 or higher and had positive associations with mortality. Neither S02 nor
10 NO2 nor O3 was significantly associated (P > 0.30) with total mortality.
11
12 13.5.2 Long-Term PM10/PM2 5 Exposure Mortality Studies
13 13.5.2.1 Population-Based Cross Sectional Mortality Studies
14 Ecological cross-sectional studies employing averages across various geopolitical units
15 (cities, SMS As, etc.) present data that examines the relation between mortality and PM
16 levels. These community-based studies seeks to define the (average) community
17 characteristics that are associated with its overall average health status, in this case annual
18 mortality rate. Ozkaynak and Thurston (1987) analyzed 1980 total mortality in 98 SMS As,
19 using data on PM15 and PM2 5 from the EPA inhalable particle (IP) monitoring network for
20 38 of these locations. Ozkaynak and Thurston ranked the importance of the various
21 pollutants mainly by relative statistical significance in separate regressions. They concluded
22 that the results were "suggestive" of an effect of particles on mortality decreasing with
23 particle size; although in the basic model only SO42' was significant. In some of the other
24 models, PM2 5 was also significant and PM15, nearly so. However, if the effects are judged
25 by elasticities rather than significance levels, SO42-> PM2 5, and PM15 would be judged as
26 equivalent, with TSP ranking somewhat lower. Ozkaynak and Thurston (1987) also used
27 source apportionment techniques to estimate that particles from coal combustion and from the
28 metals industry appeared to be the most important. The coefficients and significance levels
29 obtained for TSP by Ozkaynak and Thurston may be the result of the TSP data they used,
30 which were based on a single monitoring station in each SMSA and thus are unlikely to be
31 fully representative of population exposures. Thus, alternative interpretations of these
April 1995 13.39 DRAFT-DO NOT QUOTE OR CITE
-------
1 findings are certainly possible. In addition, because smoking, diet, and other socioeconomic
2 or lifestyle variables were not considered in the regression model, the pollution coefficients
3 may have been biased. Finally, this study did not specifically address the question of acute
4 vs. chronic responses by exploring lagged pollution variables.
5 Data from up to 149 metropolitan areas (mostly SMSAs) were analyzed in a study of
6 the relationships between community air pollution and "excess" mortality due to various
7 causes for the year 1980 (Lipfert, 1993). Several socioeconomic models were used in cross-
8 section multiple regression analyses to account for non- pollution effects. Two different
9 sources of (measured) air quality data were utilized: data from the EPA AIRS database (TSP,
10 SC>42, Mn, and ozone) and data from the inhalable paniculate (IP; PM15) network; the latter
11 data (PM15,PM2.5 and SO4= from the IP filters) were only available for 63 locations. All
12 paniculate data were averaged across all the monitoring stations available for each SMSA;
13 the TSP data were restricted to the year 1980 and were based on an average of about 10 sites
14 per SMSA. Using these models, statistically significant associations were found between
15 TSP and mortality due to non-external causes with the log-linear models evaluated, but not
16 with a linear model. Sulfates, manganese, inhalable particles (PM15), and fine particles
17 (PM2 5) were not significantly (P < 0.05) associated with mortality with any of the
18 parsimonious models, although PM2 5 and manganese were close with linear models
19 (p=0.07) and significance may have been affected by the use of smaller data sets. This
20 study showed that PM2 5 was the "strongest" paniculate variable with linear models, but that
21 TSP performed better in log-linear models. This study supported the previous findings of
22 associations between TSP and premature mortality and also the hypothesis that improving the
23 accuracy of pollutant exposure data tends to increase statistical significance.
24
25 13.5.2.2 Prospective Mortality Studies
26 Prospective studies consider data on the relative survival rates of individuals, as
27 affected by age, sex, race, smoking habits, and certain other individual risk factors. This
28 type of analysis has a substantial advantage over the population-based studies, because the
29 identification of the actual decedents allows stratification according to important risk factors
30 such as smoking. However, since none of the prospective cohort studies had data on
31 personal exposures to air pollution, these studies are also "ecological."
April 1995 13-40 DRAFT-DO NOT QUOTE OR CITE
-------
1 The newer prospective studies (Abbey et al., 1991; Dockery et al., 1993; and Pope et
2 al., 1995), are of most interest. Abbey et al. (1991a) describes a prospective study of about
3 6,000 white, non-hispanic, nonsmoking, long-term California residents who were followed
4 for 6 to 10 years, beginning in 1976. The study was designed to test the use of cumulative
5 exposure data as an explanatory factor for disease incidence and chronic effects. Pollutant
6 species were limited to TSP and ozone in this paper; oxidant concentrations were used in the
7 early part of the monitoring record. In a follow-up analysis, Abbey et al. (1995) considered
8 exposures to SO42", PM10 (estimated from site-specific regressions on TSP), PM2 5 (estimated
9 from visibility), and visibility per se (extinction coefficient). No significant associations with
10 nonexternal mortality were reported, and only high levels of TSP or PM10 were associated
11 with symptoms of asthma, chronic bronchitis or emphysema. The finding of Abbey et al.
12 (199la) of no association between long-term cumulative exposure to TSP or O3 and all
13 natural-cause mortality may be interpreted as showing the absence of chronic responses after
14 10 years but not necessarily the absence of (integrated) acute responses, since coincident air
15 pollution exposures were not considered.
16 Dockery et al. (1993) analyzed survival probabilities among 8,111 adults who were first
17 recruited in the mid-1970s in six cities in the eastern portion of the United States. The cities
18 are: Portage, WI, a small town north of Madison; Topeka, KS; a geographically-defined
19 section of St. Louis, MO; Steubenville, OH, an industrial community near the West Virginia-
20 Pennsylvania border; Watertown, MA, a western suburb of Boston; and Kingston-Harriman,
21 TN, two small towns southwest of Knoxville. This selection of locations thus comprises a
22 transect across the Northeastern and Northcentral United States, from suburban Boston,
23 through Appalachia, and into the upper Midwest. The adults were white and aged 25 to 74
24 at enrollment. The final cohorts numbered 1,400 to 1,800 persons in each city. Follow-up
25 periods ranged from 14 to 16 years, during which from 13 to 22% of the enrollees died. Of
26 the 1,430 death certificates, 98% were located, including those for persons who had moved
27 away and died elsewhere. The bulk of the analysis was based on all-cause mortality.
28 Individual characteristics of the members (and thus of the decedents) considered included
29 smoking habits, an index of occupational exposure, body mass index, and completion of a
30 high school education. The effects of air pollution were evaluated in two ways: by
31 evaluating the relative risks of residence in each city relative to Portage (the city with the
April 1995 13_41 DRAFT-DO NOT QUOTE OR CITE
-------
1 lowest pollution levels for most indices), and by including the community-average air quality
2 levels directly in the models.
3 Based on statewide mortality data, substantial differences in survival rates would be
4 expected across this transect of the Northeastern U.S. and were in fact observed. The long-
5 term average mortality rate in Steubenville was 16.2 deaths per 1,000 person-years; in
6 Topeka, it was 9.7, yielding a 67% variation in the range of annual average (crude) relative
7 risk across the six cities. After individual adjustment for age, smoking status, education, and
8 body-mass index, the range in average relative risk was reduced to 26%.
9 Dockery et al. (1993) report that "mortality was more strongly associated with the
10 levels of fine, inhalable, and sulfate particles" than with the other pollutants, which they
11 attributed primarily to factors of particle size. They provided relative risk estimates and
12 confidence limits based on the differences between air quality in Steubenville and in Portage
13 for these three pollutants.
14 Table 13-8 shows only small differences among many pollutants, including SO2 and
15 NO2, owing in part to the strong collinearity present. Note that TSP and the coarse particle
16 variables created by subtracting PM15 from TSP and PM2 5 from PM15 were not significant,
17 suggesting that particles larger than about 15 urn in aerodynamic diameter may be less
18 important; this outcome may reflect in part greater spatial variability within the communities
19 for these measures. The non sulfate portion of PM2 5 had the tightest confidence limits
20 (SO42" was multiplied by 1.2 before subtraction, assuming an average composition of
21 NH4HSO4). Note also that the estimated 1970 TSP variable performed better than the TSP
22 data used by Dockery et al. (ca. 1982). However, all of the differences in relative risks and
23 their confidence limits could have occurred due to chance, given the availability of only
24 6 observations. No relationship was found for aerosol acidity (H+), but only limited data
25 were available.
26 The authors of this study appear to have made the most of the available individual data
27 on some of the most important mortality risk factors. They were quite cautious in their
28 conclusions, stating only that the results suggest that fine-particulate air pollution "contributes
29 to excess mortality in certain U.S. cities." There are several other important outcomes:
30
31 • None of the population subgroups examined appeared to be significantly more
32 sensitive to air pollution than any other.
April 1995 13-42 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-8. ESTIMATED RELATIVE RISKS IN SIX U.S. CITIES
ASSOCIATED WITH A RANGE OF AIR POLLUTANTS
Species
PM15
PM2.5
so42-
TSP
TSP-PM15
PM15-PM25
PM25-SO42-
PM15-SO42-
S02
NO2
1970 TSP
Regr.
Coeff.
0.0085
0.0127
0.0297
0.0037
0.0042
0.0178
0.0255
0.0121
0.0093
0.0126
0.0014
Standard
Error
(0.0026)
(0.0034)
(0.0081)
(0.0014)
(0.0032)
(0.0098)
(0.0029)
(0.0034)
(0.0032)
(0.0046)
(0.00044)
Range
28.3
18.6
8.5
55.8
27.5
9.7
8.4
18.1
19.8
15.8
154.0
Rel.
Risk
1.27
1.27
1.29
1.22
1.12
1.19
1.24
1.24
1.20
1.22
1.25
95% CIs
(n=6)
(1.04-1.56)
(1.06-1.51)
(1.06-1.56)
(0.99-1.53)
(0.88-1.43)
(0.91-1.55)
(1.16-1.32)
(1.05-1.48)
(1.01-1.43)
(1.00-1.49)
(1.03-1.50)
Source: Dockery et al. (1993)
1 • The implied regression coefficients are much larger (about an order of magnitude)
2 than those found in either type of population-based study. This could be interpreted
3 as evidence that the chronic effects of air pollution far exceed the acute effects, or
4 that not all of the spatial confounding has been controlled. Use of linear models for
5 non-linear effects (body-mass index) and failure to control for alcohol consumption,
6 diet, exercise and migration may have contributed to the relatively large effects
7 indicated for air pollution.
8
9 • If the responses to air pollution truly are chronic in nature, it is logical to expect
10 that cumulative exposure would be the preferred metric (Abbey et al., 1991).
11 Pollution levels 10 years before this study began were much higher in Steubenville
12 and St. Louis, as indexed by TSP from routine monitoring networks. Estimates of
13 previous levels of fine particles are more difficult, but atmospheric visibility data
14 suggest that previous levels may have been higher in winter, but not necessarily in
15 summer. These uncertainties make it difficult to accept quantitative regression
16 results based solely on coincident monitoring data. For example, annual average
17 TSP in 1965 in Steubenville was about three times the value used by Dockery et al.
18 (1993).
19
April 1995 13.43 DRAFT-DO NOT QUOTE OR CITE
-------
1 Because it seems unlikely that any of the perceived shortcomings of this study could have
2 resulted in bias sufficient to reduce the risk estimates to levels less than those found in acute
3 mortality studies, the study of Dockery et al. (1993) appears to provide support for the
4 hypothesis that the results of long-term air pollution studies must also reflect the presence of
5 acute effects on mortality as integrated over the long term, as suggested by Evans et al.
6 (1984a). It may also be concluded that support has been shown for the existence of chronic
7 effects; these two possibilities are not mutually exclusive. However, these conclusions must
8 be qualified by the realization that not all of the relevant socioeconomic factors may have
9 been properly controlled in this study.
10 Pope et al. (1995) analyzed 7-year survival data (1982 to 1989) for about 550,000 adult
11 volunteers obtained by the American Cancer Society (ACS). The Cox proportional hazards
12 model was used to define individual risk factors for age, sex, race, smoking (including
13 passive smoke exposure), occupational exposure, alcohol consumption, education, and body-
14 mass index. The deaths, about 39,000 in all, were assigned to geographic locations using the
15 3-digit zip codes listed at enrollment into the ACS study in 1982. Relative risks were then
16 computed for 151 metropolitan areas defined by these zip codes and were compared to the
17 corresponding air quality data, ca. 1980. The sources of air quality data used were the EPA
18 AIRS system for sulfates, as obtained from high-volume sampler filters for 1980, and the
19 Inhalable Paniculate Network for fine particles (PM2 5). The latter data were obtained from
20 dichotomous samplers during 1979-81. Causes of death considered included all causes,
21 cardiopulmonary causes (ICD-9 401-440, 460-519), lung cancer (ICD-9 162), and all other
22 causes.
23 This study took great care to control for those potential confounding factors for which
24 data were available. Several different measures of active smoking were considered, as was
25 the time exposed to passive smoke. The occupational exposure variable was specific to any
26 of: asbestos, chemicals/solvents, coal or stone dusts, coal tar/pitch/asphalt, diesel exhaust,
27 or formaldehyde. The education variable was an indicator for having less than a high-school
28 education. Risk factors not considered by Pope et al. include income, employment status,
29 dietary factors, drinking water hardness and physical activity levels, all of which have been
30 shown to affect longevity (Sorlie and Rogot, 1990; Belloc, 1973; Pocock et al., 1980). In
31 addition, they did not discuss the possible influences of other air pollutants.
April 1995 13-44 DRAFT-DO NOT QUOTE OR CITE
-------
1 The ACS cohort is by no means a random sample of the U.S. population; it is 94%
2 white and better educated than the general public, with a lower percentage of smokers than in
3 the Six City Study. The (crude) death rate during the 7.25 years of follow-up was just under
4 1% per year, which is about 20% lower than expected for the white population of the
5 U.S. in 1985, at the average age reported by Pope et al. In contrast, the corresponding rates
6 for the Six- Cities study (Dockery et al., 1993) discussed above tended to be higher than the
7 U.S. average.
8 The adjusted total mortality risk ratios for the ACS study (computed for the range of
9 the pollution variables) were 1.15 (95% CL = 1.09 to 1.22) for sulfates and 1.17 (95% CL
10 = 1.09 to 1.26) for PM2 5. When expressed as log-linear regression coefficients, these
11 values were quite similar for both pollution measures: 0.0070 (0.0014) per /ig/m3 for SO42"
12 and 0.0064 (0.0015) for PM2 5, suggesting that particle chemistry may be relatively
13 unimportant as an independent risk factor (it is possible that the SO42" results have been
14 biased high by the presence of filter artifacts). Pope et al. found that the pollution
15 coefficients were reduced by 10 to 15% when variables for climate extremes were added to
16 the model.
17 The results of the long-term prospective cohort studies are shown in Table 13-9. The
18 results of the American Cancer Society prospective study were qualitatively consistent with
19 those of the Six City study with regard to their findings for sulfates and fine particles; but
20 relative standard errors were smaller, as expected because of the substantially larger
21 database. However, no other pollutants were investigated in the ACS analysis, so that no
22 further progress was made in attempting to identify the "responsible" pollutants. In addition,
23 the ACS regression coefficients were about 1/4 to 1/2 of the corresponding Six City values
24 and were much closer to the corresponding values obtained in various acute mortality studies.
25 Thus it is not clear to what extent chronic effects (as opposed to integrated acute effects) are
26 indicated by these results and to what extent the limited air quality data base used was
27 responsible for this outcome.
28 The California and Six-City studies suffer from small sample sizes and inadequate
29 degrees of freedom, which partially offset the specificity gained by considering individuals
30 instead of population groups. All of them may have neglected some important risk factors.
31 The studies of California nonsmokers by Abbey et al. (1991, 1994) that had the best
April 1995 13_45 DRAFT-DO NOT QUOTE OR CITE
-------
i
o
u
0,
Cfl
i
PH
S
a
.-aa a
•SJ"
•3 3 « "i
4> .S2 * I— «*i
g "
"
. B •-
-
<5 S
2
a-,
o
op
.
— Q S
•fi w
- '§
O Q 5
als
< -a
VO -•
U O.
Soo
I
> .S '3
H
S
o
vS
i
i
s
•1.
at.
§
o"
S
3
g
o
o x .H o «i S 'J3
o « <3 o o« w o
5
S'
:Ic?:l
U CO U
B
* I
4> w O
'333
C
.5
o •a o.
O
ff
.s
•581?
:1
Illl
§-
April 1995
13-46
DRAFT-DO NOT QUOTE OR CITE
-------
1 cumulative exposure estimates found no significant mortality effects of previous air pollution
2 exposure. The Six Cities and ACS studies agree in their findings of strong associations
3 between fine particles and excess mortality. At this time, the long-term studies provide
4 support for the existance of short-term PM-related mortality increases which are not
5 subsequently offset by decreases below normal rates. However, they do not exclude the
6 possible existence of additional chronic exposure effects, nor do they provide convincing
7 evidence as to the specific pollutant(s) involved, and they do not rule out the possible
8 existence of pollutant thresholds.
9
10 13.5.3 Morbidity Outcomes Associated With PM Exposure
11 Dockery and Pope (1994) reviewed the effects of PM on respiratory mortality and
12 morbidity. The authors considered five primary health endpoints: mortality, hospital usage,
13 asthma attacks, respiratory symptoms and lung funtion. In order to include as many studies
14 as possible, they converted both British smoke and TSP measurements to PM10. Results
15 from each study were converted to an estimated percent change in the health endpoint per
16 10 pig/m3 PM10. These converted results were then combined across studies of similar
17 endpoints using the standard inverse variance weighted method (fixed effects model). The
18 authors concluded that there was a coherence of effects across the endpoints, with most
19 endpoints showing a one to three percent change per 10 jig/m3 PM10. Pulmonary function
20 showed a smaller change of 0.15 percent for FEV and 0.08 percent for PEFR. These
21 smaller percent changes are to be expected because there is much less variation in pulmonary
22 function measurements than in the other measures. The limitations of the methodological
23 considerations as they pertain to quantitative assessment of the individual studies were
24 discussed earlier in Chapter 12. Dockery and Pope (1994) also noted such limitations in
25 their review.
26 The primary difficulties in combining studies can be summarized as follows. Most
27 studies used several endpoints and it is not clear that results for all of the different endpoints
28 were reported. Most studies used different lag times or moving averages for the pollutants,
29 and in some cases reported only those which gave positive results. For those studies which
30 did report results for similar endpoints, many were analyzed with different statistical models.
31 The short-term studies must take into account serial correlation, and this was done in a
April 1995 13.47 DRAFT-DO NOT QUOTE OR CITE
-------
1 variety of ways in those studies which did adjust for it. For these reasons, key findings from
2 most of the studies will be described rather than combined formally.
3
4 13.5.3.1 Short-Term PM Exposure Hospital Admission Studies
5 Hospitalization data can provide a measure of the morbidity status of a community
6 during a specified time frame. Hospitalization data specific for respiratory illness diagnosis
7 or more specifically for COPD and pneumonia give a measure of the respiratory status.
8 Such studies provide an outcome measure that relates to mortality studies for total and
9 specified respiratory measures. Tables 13-10 through 13-13 summarize these studies. These
10 studies associate hospitalization data with various measures of PM. Some of the same factors
11 and concerns related to the mortality studies are an issue for these studies also.
12 Both COPD and pneumonia hospitalization studies show moderate but statistically
13 significant relative risks in the range of 1.06 to 1.25 resulting from an increase of 50 pig/m3
14 in PM10 or its equivalent. There is a possible suggestion of a relationship to heart disease,
15 but the evidence is very inconclusive. The admission studies of respiratory disease show a
16 similar effect. The hospitalization studies in general use very similar analytic methodologies
17 and the majority of the papers are written by a single author. Overall, these hospitalization
18 studies are indicative of health outcomes related to PM. They are also supportive of the
19 mortality studies, especially with the more specific diagnosis relationships.
20
21 13.5.3.2 Short and Long-Term Exposure Respiratory Disease Studies
22 Respiratory illness and the factors determining its occurrence and severity are
23 important public health concerns. This effect is of public health importance because of the
24 widespread potential for exposure to PM and because the occurrence of respiratory illness is
25 common (Samet et al., 1983; Samet and Utell, 1990). Of added importance is the fact that
26 recurrent childhood respiratory illness may be a risk factor for later susceptibility to lung
27 damage (Glezen, 1989; Samet et al., 1983; Gold et al., 1989). The occurrence of lower
28 respiratory morbidity in early childhood may be associated with impaired lung function and
29 growth that appears to persist through adolescence. Denny and Clyde (1986) stated that
30 infections, reactive airways, and inhaled pollutants (mostly cigarette smoke) are the most
31 important risk factors in the development of chronic lung disease. Thus, factors such as the
April 1995 13-48 DRAFT-DO NOT QUOTE OR CITE
-------
u>
O
O
1
O
c;
o
?o
n
Study
Burnett et al. (1994)
All ages in Ontario,
Canada, 1983-1988
Thurston et al. (1994)
All ages in Ontario,
Canada, July and August,
1986-1988
Thurston et al. (1992)
All ages in Buffalo,
Albany, New York City,
July and August, 1988-1989
Schwartz (in press)
Elderly in New Haven,
1988-1990
Schwartz (in press)
Elderly in Tacoma, 1988-
1990
PM Type &
No. Sites
9 monitoring
stations
measuring
sulfate
3 monitoring
stations
measuring
sulfate, TSP,
and PM10
3 monitoring
stations (one
per city)
measuring
sulfate, H+
PM10
monitoring
stations
averaged, no.
of stations not
given
PM10
monitoring
stations
averaged, no.
of stations not
given
PM Mean Ave. Count
& Range per Day
sulfate means 108
ranged from 3.1
to 8.2 /ig/m3
mean sulfate 14.4
ranged 38 to 124
(nmole/m3), PM10
30 to 39 pg/m3,
TSP 62 to 87
/*g/m3
(values not given) Buffalo, 24
Albany, 12,
New York,
137
mean = 41, 8.1
10% tile = 19,
90% tile = 67
mean = 37, 4.2
10% tile = 14,
90% tile = 67
Model Type
&Lag
Structure
Lin. regress.
on filtered
data, 1-d lag
best
Linear
regression on
filtered data,
0-d lag best
Linear
regression on
filtered data
Poisson log-
linear
regression,
19 day mov.
ave. filter,
0-d lag best
Poisson log-
lin. regress.
19 day mov.
ave. filter,
0-d lag best
Other pollutants
measured
Ozone
Ozone, H+, SO2,
N02
Ozone, H +
Weather &
Other Factors
Temperature
Temperature
Temperature
Result*
Pollutants (Confidence
in model Interval)
none 1 .03
(1.02,
none PM10
1.09
(0.96,
ozone PM10
1.01
(0.87,
1.04)
1.22)
1.15)
ozone (not given for
PM measures)
Ozone (ppb): mean
= 29; 10% tile =
16; 90% tile = 45;
SO2 (ppb): mean =
30; 10% tile = 9;
90% tile = 61
Ozone (ppb): mean
= 25; 10% tile =
13; 90% tile = 36;
SO2 (ppb): mean =
17; 10% tile = 6;
90% tile = 28
Temperature
and dew point
adjusted for in
the moving
average
Temperature
and dew point
adjusted for in
the moving
average
none 1 .06
(1.00,
SO2(2day 1.07
lag) (1.01,
none 1.10
(1.03,
SO2(2day 1.11
lag) (1.02,
1.13)
1.14)
1.17)
1.20)
* Relative risk calculated from parameters given by author assuming a 50 jtg/m3 increase in PM10 on 100 jug/m3 increase in TSP.
-------
TABLE 13-11. HOSPITAL ADMISSIONS STUDIES FOR COPD
T3
S:
»— '
Ul
£
tin
0
O
£>
2
O1
o
2
o
H
O
O
H
W
O
Q
H
W
Study
Sunyer et al. (1993)
Adults in Barcelona,
1985-1989
Schwartz (1994f)
Elderly in Minneapolis,
1986-1989
Schwartz (1994e)
Elderly in Birmingham,
1986-1989
Schwartz (1994d)
Elderly in Detroit
1986-1989
* Relative risk calculated
PM Type &
No. Sites
15 monitoring
stations measuring
black smoke
6 monitoring
stations measuring
PM10
1 to 3 monitoring
stations measuring
PM10
2 to 11 PM10
monitoring stations,
data available for
82% of possible
days
from parameters given
PM Mean
& Range
winter 33% tile
= 49, 67% tile
= 77, summer
33% tile = 36,
67% tile = 55
mean = 36, 10%
tile = 18, 90%
tile = 58
mean = 45,
10% tile = 19.
90% tile = 77
mean = 48,
10% tile = 22,
90% tile = 82
Ave. Model Type
Count & Lag
per Day Structure
12 Autoregressive
linear
regression
analysis, 0-d
lag best
2.2 Autoregressive
Poisson model,
1-d lag best
2.2 Autoregressive
Poisson model,
0-d lag best
5.8 Poisson auto-
regressive
model using
GEE, 0-d lag
best
by author assuming a 50 /jg/m3 increase in PM
Other
pollutants
measured
Sulfur dioxide,
winter 33% tile
= 49Mg/m3,
67%tile = 77,
summer
33%tile = 36,
67%tile = 55
Ozone, mean
= 26 ppb,
10%tile =11,
90%tile = 41
Ozone, mean
= 25 ppb,
10%tile = 14,
90%tile = 37
Ozone, mean
= 21 ppb,
10%tile = 7,
90%tile = 36
10 or 100 jtg/m3
Weather &
Other Pollutants
Factors in model
min temp, none
dummies for day
of week and
year
SO2
8 categories of none
temp. & dew
pt., month,
year, lin. &
quad, time trend
7 categories of none
temp. & dew
pt., month,
year, lin. &
quad, time trend
Dummy vars. ozone
for temp,
month, lin. &
quad, time trend
increase in TSP.
Result*
(Confidence
Interval)
winter: 1.15
(1.09,1.21)
summer: 1.05
(0.98,1.12)
winter: 1.05
(1.01,1.09)
summer:
1.01
(0.97,1.05)
1.25
(1.10,1.44)
1.13
(1.04,1.22)
1.11 (1.04,
1.17)
-------
3.
TABLE 13-12. HOSPITAL ADMISSIONS STUDIES FOR PNEUMONIA
Study
Schwartz (1994f)
Elderly in Minneapolis,
1986-1989
Schwartz (1994e)
Elderly in Birmingham,
1986-1989
Schwartz (I994d)
Elderly in Detroit
1986-1989
PM Type & PM Mean
No. Sites & Range
6 monitoring stations mean = 36,
measuring PM10 10%tile = 18,
90% tile = 58
1 to 3 monitoring mean = 45,
stations measuring 10% tile = 19.
PM10 90% tile = 77
2 to 11 PM10 mon. mean = 48,
stations, data for 10% tile = 22,
82% of possible days 90% tile = 82
Ave.
Count
per Day
6.0
5.9
15.7
Model Type
& Lag Other pollutants
Structure measured
Autoregressive Ozone: mean 26
Poisson mod., ppb; 10% tile 11;
1-d lag best 90% tile 41
Autoregressive Ozone: mean 25
Poisson mod., ppb; 10% tile 14;
0-d lag best 90% tile 37
Poisson auto- Ozone: mean 21
regress, mod. ppb; 10% tile 7;
using GEE, 90% tile 36
0-d lag best
TABLE 13-13. HOSPITAL ADMISSIONS STUDIES FOR HEART
Study
Schwartz and Morris
(in press)
Elderly in Detroit
1986-1989
Ischemic Heart Disease
Burnett et al. (in press)
All ages in Ontario,
Canada, 1983-1988
Cardiac disease admission
PM Type & PM Mean
No. Sites & Range
2 to 11 PM10 mean = 48,
monitoring 10% tile = 22,
stations, data 90% tile = 82
available for
82% of possible
days
22 sulfate station means
monitoring ranged from 3.0 to
stations 7.7 in the summer
and 2.0 and 4.7 in
the winter
Ave.
Count
per Day
44.1
14.4
Model Type
& Lag Other pollutants
Structure measured
Weather &
Other Pollutants
Factors in model
8 categories of temp. &none
dew pt., month, year,
lin. & quad, time trend
7 cat. of temp. & dew none
pt., month, year, lin.
& quad, time trend
Dummy variables for ozone
temp, month, lin. &
quad, time trend
DISEASE
Weather &
Other Pollutants
Factors in model
Poisson auto- SO2, mean = 25 Dummy vars. for none
regressive ppb, 10% tile =
model using 11, 90% tile =
GEE, 0-d lag CO, mean 2.4
best ppm, 10% tile 1
90% tile = 3.8
Linear Ozone averaged
regression on a ppb
19 day linear
filter, 1-d lag
best
: temp, month, lin.
44 & quad, time trend
ozone,
2, CO, SO2
36 Temperature none
included in
separate analyses
by summer and ozone
winter
Result*
(Confidence
Interval)
1.08
(1.01,1.15)
1.09
(1.03, 1.15)
1.06
(1.02, 1.10)
Result*
(Confidence
Interval)
1.06
(1.02, 1.10)
1.06
(1.02, 1.10)
1.04
(1.03, 1.06)
1.04
(1.03, 1.05)
* Relative risk calculated from parameters given by author assuming a 50 /ig/m3 increase in PM10 on 100 /*g/m3 increase in TSP.
-------
1 presence of PM (which increases the risk for respiratory symptoms and related respiratory
2 morbidity) are important because of associated public health concern with regard to both the
3 immediate symptoms produced and the longer term potential for increases in the development
4 of chronic lung disease.
5
6 Acute Respiratory Disease Studies
7 Acute respiratory disease studies include several different endpoints, but the majority of
8 authors reported results on at least two of: (1) upper respiratory illness; (2) lower
9 respiratory illness; or (3) cough (See Table 13-14). These relative risks are all estimated for
10 an increase of 50 /xg/m3 in PM10 or its equivalent. The results for upper respiratory illness
11 are very inconsistent: two studies estimate a relative risk near 1.00 whereas four others
12 obtain estimates between 1.14 and 1.55. The relative risks for lower respiratory illness are
13 spread between 1.01 and 2.03, but all are positive. The relative risks for cough include two
14 below 1.0 and go as high as 1.51. All of these are suggestive of an effect. Whereas the
15 hospital admission studies were all done in a similar manner and resulted in very similar
16 results, these studies are done with very different designs and give very inconsistent results.
17
18 Chronic Respiratory Disease Studies
19 The three studies are based on a similar type of questionnaire but were done by two
20 different groups of researchers. All three studies suggest a chronic effect of paniculate
21 matter on respiratory disease, but the studies suffer from the usual difficulty of cross
22 sectional studies. The effect of paniculate matter is based on variations in exposure which
23 are determined by the different number of locations. In the first two studies there were six
24 locations and in the second there were four. The results seen were consistent with a
25 paniculate matter gradient, but it is impossible to separate out the effect of paniculate matter
26 and any other factors or pollutants which have the same gradient (See Table 13-15).
27
28 13.5.3.3 Short and Long-Term Exposure Pulmonary Function Studies
29 Pulmonary function studies are part of a comprehensive investigation of the possible
30 effects of any air pollutant. Measurements can be made in the field, they are noninvasive,
31 and their reproductibility has been well documented. Guidelines for reference values and
April 1995 13-52 DRAFT-DO NOT QUOTE OR CITE
-------
&
H^
8 Study
Schwartz et al.
elementary sch<
in Six-Cities in
1984-1988
TABLE 13-14. ACUTE RESPIRATORY DISEASE STUDIES
(1994) 300
)ol children
U.S.,
PM Type &
No. Sites
PM10
monitoring in
each city
PMMean
& Range
median 30 /xg/m3
10th percentile =
13, 90th
percentile = 53
Ave.
Rate
per Day
3.1
Model Type
& Lag Structure
Autoregressive
logistic
regression using
GEE
Other
pollutants
measured
Ozone, NO2,
S02
Weather &
Other
Factors
Temperature
Other
pollutants
in model
none
SO2
ozone
Result*
(Confidence
Interval)
1.51 (1.12,2
1.39(0.98, 1
1.49(1.10,2
.05)
.96)
.01)
Pope et al. (1991),
students in the Utah
Valley, winter 1989-1990
PM1?
monitoring
stations at 3
sites
mean = 46
jig/m3,
range = 11 to
195
Pope et al. (1991), PM10 mean = 46
asthmatic children in the monitoring /ig/m3,
Utah Valley, winter 1989- stations at 3 range = 11 to
1990 sites 195
(not given) Fixed effects Limited Variables for none
logistic monitoring of temperature and
regression NO2, SO2, and time trend
ozone. Values
were well
below the
standard
(not given) Fixed effects Limited Variables for none
logistic monitoring of low
regression NO2, SO2, and temperature and
ozone. Values time trend
were well
below the
Upper resp.
1.20(1.03, 1.39)
Lower resp.
1.28(1.06, 1.56)
Upper resp.
0.99(0.81, 1.22)
Lower resp.
1.01 (0.81, 1.27)
o
o
2
o
o
c:
o
H
W
O
Q
H
W
Pope and Dockery (1992), PM10 mean = 76 (not given)
symptomatic children in monitoring /ng/m3.
the Utah Valley, winter stations at 2 range = 7 to 251
1990-1991 sites
sioiiuam
Autoregressive none
logistic
regression using
GEE
Variable for none Upper resp.
low 1.20(1.03,
temperature Lower resp.
1.27(1.08,
Cough
1.29(1.12,
1.39)
1.49)
1.48)
-------
TABLE 13-14 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
g.
VO
&
H- '
u>
"^
O
>
H
6
0
^
g
o
H
W
O
Study
Pope and Dockery (1992),
asymptomatic children hi
the Utah Valley, whiter
1990-1991
Hoek and Brunekreef
(1993), respiratory disease
in school children aged 7
to 12 in Wageningen,
Netherlands, whiter
1990-1991
Schwartz et al. (1991)
Study of acute respiratory
illness in children in
5 German communities,
1983-1985
PM Type &
No. Sites
PM10
monitoring
stations at
2 sites
Two to 4
monitoring
stations
measured
PM10
Two to 4
monitoring
stations in
each area
measured
TSP
Ave.
PM Mean Rate
& Range per Day
mean = (not given)
76 jtg/m3,
range = 7 to
251
max = (not given)
110/*g/ni3
medians ranged 0.5 to 2.9
from 17 to
56 /*g/m3,
10% tiles from
5 to 34, 90%
tiles from 41 to
118
Model Type
&Lag
Structure
Autoregressive
logistic regression
using GEE
Autoregressive
logistic regression
using GEE
Autoregressive
Poisson regression
using GEE
Other
pollutants
measured
none
Max SO2 = 105
jig/m3, max NO2
= 127 /ig/m3
median SO2
levels ranged
from 9 to
48 jig/m3,
median NO2
levels ranged
from 14 to
5 ftg/m3
Weather &
Other
Factors
Variable for
low temperature
Variable for
ambient
temperature and
day of study
Most
significant
terms of day
of week, time
trend, and
weather
(terms not
listed)
Other Result*
pollutants (Confidence
hi model Interval)
none Upper resp.
0.99
(0.78, 1.26)
Lower resp.
1.13
(0.91, 1.39)
Cough
1.18
(1.00, 1.40)
none Upper resp.
1.14
(1.00, 1.29)
Lower resp.
1.06
(0.86, 1.32)
Cough
0.98
(0.86, 1.11)
none (TSP 1.26
was not (1.12,1.42)
significant
when NO2
added to model)
n
H-I
H
W
-------
TABLE 13-14 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
3.
vo
S
t— »
u>
t!/«
LH
!>
H
i
O
2
s
o
r^^
O
H
W
0
90
o
H
Study
Schwartz et al. (1994)
Study of respiratory
symptoms in 6 U.S.
cities, 1984-1988
Braun-Fahrlander et al.
(1992)
Study of preschool
children in four areas
of Switzerland
Roemer et al. (1993)
Study of children with
chronic respiratory
symptoms in
Wageningen, The
Netherlands
Dusseldorf et al.
(1994)
Study of adults near a
steel mill in The
Netherlands
PM Type &
No. Sites
Daily
monitoring
of PM10,
PM2 5 at
each city
Daily
monitoring of
TSP
Daily
monitoring of
PM10
Daily
monitoring of
PM10, iron,
sodium,
silicon, and
manganese
Ave.
PM Mean Rate
& Range per Day
median PMi0 (not given)
= 30 /tg/rn ,
10% tile = 13,
90% tile = 53
median PM2 5
= 18 ng/m ,
10% tile = 7,
90% tile = 37
(not given) 4.4
6 days above .094
110 /ig/m3 incidence
rate
mean PM10 (not given)
= 54 /tg/m3,
range =
4 to 137)
Model Type
&Lag
Structure
Autoregressive
logistic
regression
using GEE
Logistic
regression
Autoregressive
logistic
regression
Logistic
regression
Other
pollutants
measured
SO2, median = 4
ppb, 10% tile =
1, 90% tile = 18
NO2, median =
13 ppb, 10% tile
5, 90% tile = 24,
ozone
SO2, NO2, and
ozone levels not
given
SO2 and NO2
means not given
Geometric mean
iron = 501
ng/m3, manganese
= 17 ng/m3,
silicon =
208 ng/m3
Weather &
Other
Factors
temperature,
day of week,
city or
residence
city, risk
strata, season,
temperature
(not given)
(not given)
Other Result*
pollutants (Confidence
in model Interval)
all two Cough
pollutant (PM10 lag 1)
models were 1.51
fitted with (1.12,2.05)
minimal effect Upper resp.
on PM (PMIO lag 2)
1.39
(0.97, 2.01)
Lower resp.
(PM10 lag 1)
2.03
(1.36,3.04)
none Upper resp.
1.55
(1.10,2.24)
none Cough
(not given,
probably less than
one)
none Cough
1.14
(0.98, 1.33)
-------
TABLE 13-14 (cont'd). ACUTE RESPIRATORY DISEASE STUDIES
a.
Ul
Study
PM Type &
No. Sites
PMMean
& Range
Ave.
Rate
per Day
Model Type
&Lag
Structure
Other
pollutants
measured
Weather &
Other
Factors
Other
pollutants
in model
Result*
(Confidence
Interval)
Ostro et al. (1991)
Study of adult asthmatics
in Denver, Colorado
November 1987 to
February 1988
Ostro et al. (1993)
Study of non-smoking
adults in Southern
California
Two monitors
provided daily
measurements of
PM2.5
22 jig/m3, range 15 (out of 108)
= 0.5 to 73
Autoregressive nitric acid,
logistic sulfates,
regression nitrates, SO2,
andH+
day of survey, none
day of week, gas
stove, minimum
temperature
Apparently one mean sulfate = 84.2/person for
site (Azusa). /*g/m3, range = lower,
PM measurements 2 to 37 mean 10.2/person,
included sulfate COHS = 12 per upper
fraction and 100 ft, range
COHS = 4 to 26
Logistic ozone, mean temperature, rain none
regression = 7 pphm, humidity
range = 1 to 28
Cough
1.09
(0.57, 2.10)
Sulfates:
Upper resp.
0.91
(0.73, 1.15)
Lower resp.
1.48
(1.14, 1.91)
£^ * Relative risk calculated from parameters given by author assuming a 50 /tg/m3 increase in PM10 on 100 /tg/m3 increase in TSP.
O\
-------
TABLE 13-15. CHRONIC RESPIRATORY DISEASE STUDIES
3;
h— »
\o
VO
H-
1
^J
O
5
Jfl
H
6
25
O
H
O
§
W
O
50
n
Study
Ware et al. (1984)
Study of respiratory
symptoms in children in 6
cities in the U.S. Survey
done 1974-1977
Dockery et al. (1989)
Study of respiratory
symptoms in children in 6
cities in the U.S. Survey
done 1980-1981
Chapman et al. (1985)
Study of persistent cough
and phlegm (bronchitis) in
adults in four
communities in Utah.
Survey done in 1976
Neas et al. (1994)
Study of children aged
7 to 1 1 from six cites in
U.S. Survey done
1983-1986.
* Estimates calculated from
PM Type &
No. Sites
Daily monitoring
of TSP, SO2,
NO2, and ozone at
each city
Daily monitoring
of PM15, sulfate
fraction at each
city
Daily monitoring
of TSP, and
sulfate fraction at
each city
PM2.5
PMMean
& Range
City TSP means
ranged from 39
to 114 /ig/m3
City PM15
means ranged
from 20 to 59
/ig/m3
Previous 5 year
TSP ranged from
11 to 115 /ig/m3
Not given
Overall
Symptom
Rate
Cough, .08,
Bronchitis
.08,
Lower resp.
.19
Cough, .02 to
.09, Bronchitis
.04 to .10,
Lower resp.
.07to .16
.02 to .05 by
city
Not given
data tables assuming a 50 /tg/m3 increase in PM10 on
Model
Type
&Lag
Structure
Logistic
regression
Logistic
regression
Logistic
regression
Logistic
regression
100 /ig/m3
Other
Other Other pollutants
pollutants Covariates in model
measured
SO2, NO2, age, gender, none
and ozone parental
education,
maternal
smoking
SO2, NO2, age, gender, none
and ozone maternal
smoking
SO2, NO2 smoking none
NO2 household none
smoking, gas
stove, age,
gender none
none
increase in TSP.
Result*
(Confidence
Interval)
Cough
2.75 (1.92,
Bronchitis
2.80(1.17,
Lower resp.
2.14(1.06,
Cough
5.39(1.00,
Bronchitis
3.26(1.13,
Lower resp.
2.93 (0.75,
Mothers
1.75(1.21,
Fathers
1.94(1.16,
Cough
1.08(0.76,
Bronchitis
1.32(0.98,
Lower resp.
1.23(0.98,
3.94)
7.03)
4.31)
28.6)
10.28)
11.60)
2.54)
3.25)
1.53)
1.79)
1.55)
-------
1 interpretative strategies of lung function tests have been prepared (American Thoracic
2 Society, 1991). Various factors are important determinants of lung functions.
3 Lung function in children has been related to genetic factors that exert their greatest
4 influence through general stature as measured by height and age. Growth patterns in
5 children differ by gender. Lung function declines with age among adults
6 (Dockery et al., 1988). The study of the growth of pulmonary function and generalized
7 growth models consider factors of how growth is statistically dependent on the initial
8 measures of function, and how it is related to respiratory illness in childhood. The effects of
9 active smoking and passive smoking (Lebowitz and Holberg et al., 1987) are also considered.
10 Epidemiological studies relating particulate matter to decrements in pulmonary function
11 represent a potentially important health effect.
12 The acute pulmonary function studies are suggestive of a short term effect resulting
13 from particulate pollution. Peak flow rates show decreases in the range of 30 to 40 ml/sec
14 resulting from an increase of 50 /ig/m3 in PM10 or its equivalent. The results appear to be
15 larger in symptomatic groups such as asthmatics. The effects are seen across a variety of
16 study designs, authors, and analysis methodologies. Effects using FEVj or FVC as
17 endpoints are less consistent (See Table 13-16).
18 The chronic pulmonary function studies are less numerous than the acute studies
19 (Table 13-16). The one study with good monitoring showed no effect from particulate
20 pollution. Cross sectional studies require very large sample sizes to detect differences
21 because the studies cannot eliminate person to person variation which is much larger than the
22 within person variation. Thus the lack of statistical significance cannot be taken as proof of
23 no effect.
24
25 13.5.3.4 Comparison of the Effect of PM10 to PM2 5 on Respiratory Disease and
26 Pulmonary Function
27 The most direct comparison of the effect of PM10 to PM2 5 results when studies include
28 both exposure measures in their analyses. This occurred in the Six City study (Schwartz
29 et al. 1994), the Tuscon study (Quackenboss et al., 1991) and the Uniontown study (Neas
30 et al., 1995). None of these studies could directly show that one of these measures was a
31 significantly better predictor than the other. The Schwartz et al. (1994) study suggested that
32 PM10 was a better predictor of respiratory disease. The Quackenboss et al. (1991) study
April 1995 13-58 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-16. ACUTE AND CHRONIC PULMONARY FUNCTION CHANGES
3.
VO
l/i
i— »
LA
vo
0
$>
i-j
O1
O
as
0
*j
c
o
,_j
a
o
Study
Dockery et al. (1982)
School age children in
Steubenville, OH,
measured at three times
between 1978 and 1980
Dassen et al. (1986)
School age children in
The Netherlands,
measured in November,
1984 and January, 1985
Quackenboss et al.
(1991)
Asthmatic children aged
6 to 15 years in Tuscon,
AZ, measured in May
and November, 1988
Pope et al. (1991)
Study of asthmatic
children in the Utah
Valley
Pope and Dockery
(1992)
Study of non-asthmatic
symptomatic and
asymptomatic children
in the Utah Valley
PM Type & PM Mean
No. Sites & Range
Single station up to 455 /ig/m3
measuring TSP
Six station network TSP and RSP both
measuring TSP, exceeded 200
RSP (PM10) /ig/m3
Individual
monitoring of
homes of PM2 5,
PM10
PM10 monitors in PM10 ranged from
Orem and Lindon, 11 to 195 /ig/m3
Utah
PMjo monitors in PM10 ranged from
Orem and Lindon, 11 to 195 /ig/m3
Utah
Model Type Other
& Lag pollutants
Structure measured
individual SO2
regression analyses
for each child,
coefficients pooled
across time
multiple linear SO2
regression
Random effects NO2
linear model
Weighted least SO2, NO2, ozone
squares regression
Weighted least SO2, NO2, ozone
squares regression
Weather &
Other Pollutants
Factors in model
average TSP
temperature
technician, RSP
appliance,
presence of
colds
temperature, PM2 s
wind speed,
dew point
low PM10
temperature
low PM10
temperature
Decrease*
(Confidence
Interval)
FVC: 8.1 ml
FEV075: 1.8ml
Note: decreases
were statistically
significant
slopes not given but
FVC, FEVj, and
PEFR were
significantly reduced
during episodes
PEFR: 375 ml/s
Note: these are
diurnal rather than
daily changes
PEFR: 55 ml/s
(24, 86)
Symptomatic
PEFR
30 ml/s
(10, 50)
Asympto-
matic PEFR
21 ml/s
(4, 38)
-------
TABLE 13-16 (cont'd). ACUTE AND CHRONIC PULMONARY FUNCTION CHANGES
w
h- t-
h-*
s
H- k
o\
o
0
!>
T1
i
o
2
o
H
O
H
W
o
n
H
w
Study
Koenig et al. (1993)
Study of asthmatic and
non-asthmatic elementary
school children in Seattle,
WA in 1989 and 1990
Hoek and Brunekreef
(1993)
Study of children aged 7 to
12 in Wageningen,
Netherlands
Roemer et al. (1993)
Study of children with
chronic respiratory
symptoms in The
Netherlands
Pope and Kanner (1993)
Study of adults in the Utah
Valley from 1987 to 1989
Neas et al. (1994)
Study of lung function in
children in 6 cities in the
U.S. Data collected from
1983-1988.
* Decreases in lung function
PM Type &
No. Sites
PM2 5 calibrated
from light
scattering
Single site
measure black
smoke. PM10 was
measured during
episodes
Single site
measure black
smoke. PM10 was
measured using an
Anderson dichot
PM10 was collected
daily from the north
Salt Lake site
Daily monitoring of
PM2 5, sulfate
fraction at each city
PM Mean
& Range
PM2 5 ranged
from 5 to
45 /ig/m3
range of PM10
was 30 to
144 /ig/m3
range of PM10
was 30 to
144 jtg/m3
PM10 daily
mean =
55 ^g/m ,
ranged from 1
to 181 /tg/m3
not given
calculated from parameters given by
Model Type
&Lag
Structure
Random effects
linear regression
SAS procedure
AUTOREG
multiple linear
regression
analysis
Linear regression
on difference in
PFT as a function
of PM10
Linear regression
using logarithm of
PFT value
author assuming a 50
Other Weather &
pollutants Other
measured Factors
none height,
temperature
SO2, NO2 day of study
SO2, NO2 none
Limited low
monitoring of temperature
SO2, NO2, and
ozone
SO2, NO2, and city, gender
ozone parental
education,
history of
asthma, age,
height, weight
jig/m3 increase in PM10 on 100
Decrease*
Pollutants (Confidence
in model Interval)
PM2 5 Asthmatics
FEV! 42 ml
(12, 73)
FVC 45 ml
(20, 70)
Non-asthmatics
FEV, 4 ml
(-7, 15)
FVC -8 ml
(-20, 3)
PM10 PEFR
41 ml/s (-8, 90)
PM10 PEFR
34 ml/s (9, 59)
PM2 5 FEVj
29 ml (7, 51)
FVC 15 ml (-
15, 45)
PM2 5 FVC and FEVj not
changed. Values
could not be
converted to mis.
fig/m3 increase in TSP.
-------
1 suggested that PM2 5 was a better predictor of lung function change. The Neas et al. (1995)
2 study used only the PM2 5 values in their analysis, but this may have been due to the fact that
3 the PM10 values were not available as 12 h averages whereas the other pollutants were.
4 There were at least three other studies that used PM2 5 as a measure of paniculate
5 exposure. The Ostro et al. (1991) study of respiratory disease in Denver found an effect that
6 was in the middle of the range of effects found by the PM10 studies. The Koenig et al.
7 (1993) study of lung function found a slightly larger effect for asthmatics and slightly smaller
8 effect for non-asthmatics when compared with the PM10 studies. Finally, the Neas et al.
9 (1994) Six City study of lung function did not give parameter estimates which could be
10 compared directly with the other studies.
11 Based on the above information, there is currently no obvious way by which to clearly
12 distinguish morbidity effects of PM10 versus PM2 5. Even the suggestive evidence leaves the
13 scales in a balanced position.
14
15 13.5.4 Coherence of Epidemiologic Findings
16 Factors involved in evaluating both the data and the entire group of epidemiological
17 studies, include the strength of association, the consistence of the association, as evidenced
18 by its repeated observation by different persons, in different places, circumstances and time,
19 and the consistency with other known facts (Bates, 1992). One can look for
20 interrelationships between different health indices to provide a stronger and more consistent
21 synthesis of available information. The various findings that support a picture of coherence
22 would provide a stronger case with quantitative studies as opposed to qualitative studies.
23 Other studies may be inappropriate to use in such a discussion, the quality of the study
24 should be considered. Bates (1992) states that the difficulty with discussing any index of
25 internal coherence is that this requires a series of judgements on the reliability of the
26 individual findings and observations. The outcome of a coherence discussion then is a
27 qualitative presentation in the end, not quantitative. Thus, coherence cannot be formally
28 measured.
29 Bates (1992) also noted that the strength of different health indexes are important as are
30 difficulties in assessing exposure. Bates (1992) also suggests three areas to look for
April 1995 13-61 DRAFT-DO NOT QUOTE OR CITE
-------
1 coherence: (1) within epidemiological data, (2) between epidemiological and animal
2 toxicological data, and (3) between epidemiological, controlled human and animal data.
3 Coherence by its nature considers biological relationships of exposure to health
4 outcome. The biologic mechanism underlying an acute pulmonary function test reduction in
5 children is most likely not part of the acute basis for a change in the mortality rate of a
6 population exposed in an older group of individuals. In looking for coherence one should
7 compare outcomes that look at similar time frames—daily hospitalizations compared to daily
8 mortality rather than monthly hospitalizations. Overall the data indicates that PM has a
9 relationship with a continuum of health outcomes, but the studies may not establish a
10 coherence between them. The underlying mechanisms may be different.
11 The principal health outcome for which coherence is desirable is mortality, the death
12 rate in a population. This can be considered within the endpoint and/or in other endpoints.
13 Of the various morbidity outcomes studied and discussed in the earlier part of the chapter,
14 hospitalization studies reviewed in the chapter support this notion. The mortality studies
15 suggest that these specific causes provide stronger relationships (i.e., larger RR estimates)
16 than total mortality. The outcome potentially most related is hospital admission for
17 respiratory or cardiovascular causes in the older age group (i.e., > 65 years old). In a
18 qualitative sense, the increased mortality found in that age group should also be paralled by
19 increased hospital admissions.
20 Partial coherence is established by those studies in which increased incidence of
21 different health outcomes associated with PM are found in the same population, as is the case
22 for the following examples, based on currently published studies:
23 •Detroit: Mortality mainly in elderly populations, hospital admissions for respiratory
24 causes and for cardiovascular causes in the elderly;
25 •Birmingham: Mortality mainly in the elderly, hospital admissions for the elderly;
26 •Philadelphia: Mortality and hospital admissions for pneumonia in the elderly;
27 »Utah Valley: Mortality and hospital admissions for respiratory causes in adults.
28
29 Also, pulmonary function, respiratory symptoms, and medication use in asthmatic subjects of
30 all ages; hospital admissions for respiratory symptoms, pulmonary function, respiratory
31 symptoms, and medication use in healthy school children, pulmonary function in
April 1995 13_62 DRAFT-DO NOT QUOTE OR CITE
-------
1 symptomatic and asymptomatic children; and elementary school absences in children were
2 found to be associatied with PM exposures in Utah Valley. A similar study found a PM
3 effect on pulmonary function in smokers with COPD in Salt Lake Valley. The Utah Valley
4 population was largely non-smoking, so smoking was not likely to be a source of
5 confounding.
6 While these multiple outcomes did not occur in strictly identical subgroups of each
7 population, there was probably a sufficient degree of overlap to indicate that PM was a
8 significant predictor of a wide range of health outcomes within a specific community. The
9 symptoms serious enough to warrant hospitalization and the major part of the excess
10 mortality occurred in the elderly sub-group of the population. However, a significant
11 decrement in pulmonary function and increased incidence of symptoms associated with daily
12 increases in PM occurred in children in Utah Valley, along with a "quality of life" effect
13 measured by lost school days. Thus, there is evidence for increased risk of health effects
14 related to PM exposure ranging in seriousness from asymptomatic pulmonary function
15 decrements, to respiratory symptoms and cardiopulmonary symptoms sufficiently serious to
16 warrant hospitalization, and to excess mortality from respiratory and cardiovascular causes,
17 especially in those older than 65 years of age.
18 Children may also be at increased risk of pulmonary function changes and increased
19 incidence of symptoms associated with PM exposure. While we have arrayed these health
20 outcomes in order of increasing severity, there is as yet little indication that there is a
21 progression of effects in any single individual associated with increasing exposure to PM.
22 The "exposure-response" relationship that is derived in most studies must be understood as
23 characterizing population risk from population exposure. Additional studies are needed to
24 define the relationship(s) among individual exposure to PM and other stress factors,
25 individual risk, and individual progression among disease states. Differences in PM
26 dosimetry in the developing, aged, or diseased respiratory tract may also contribute to
27 increased susceptability.
28 The coherence of the various health effects in humans could be established more
29 conclusively from epidemiology studies if there were better evidence. We cannot prove that
30 the people that suffered respiratory symptoms in response to PM exposure were among the
31 same people who suffered pulmonary function decrements from PM exposure in the past,
April 1995 13.63 DRAFT-DO NOT QUOTE OR CITE
-------
1 that those who were admitted to hospital for respiratory or cardiopulmonary causes in
2 response to PM exposure were among those who had suffered respiratory symptoms or
3 pulmonary function decrements from earlier PM exposures, nor that those who died from
4 PM exposure were among those who had earlier shown other health endpoints associated
5 with PM exposures. Such information could, in principle, be extracted from longitudinal
6 data bases such as those collected by health care providers; however, although some such
7 efforts are now being considered, the preferred design for such a study is a prospective
8 design rather than a retrospective design. If and when these studies are completed, they
9 could be useful in future PM health assessments.
10
11
12 13.6 HEALTH EFFECTS OF AMBIENT PM CONSTITUENTS
13 There are literally hundreds of ambient PM consituents, having varying degrees of
14 health effects information. The key constituents (defined as those individual compounds
15 likely to constitute > 1 ng/m3 in ambient U.S. urban air) include several heavy metals (e.g.,
16 cadmium, manganese, chromium, aluminum, lead, etc.), asbestos, silica, carbon particles,
17 nitrates, sulfates, and acid aerosols. Other constituents of interest include transition metals
18 (e.g., iron) and ultrafine particles. Of all of these materials, only a few have a direct
19 relevance to PM as a criteria pollutant. For example, acid aerosols are a generic subclass of
20 PM and are summarized here. The others are of indirect interest insofar as they can
21 elucidate and extend the epidemiologic findings. For example, because heavy metals are
22 regulated as hazardous air pollutants (i.e., they are not criteria pollutants, except for lead),
23 their data base is mainly of value here to help illustrate the great importance of particulate
24 chemistry to health outcome. Transition metals and ultrafines may be important mechanisms
25 for lung morbidity and mortality observed in epidemiologic studies. These issues are
26 discussed in more depth later, in an attempt to present hypotheses for the biological
27 plausibility of the epidemiologic findings for PM exposure.
28 The carcinogencity of diesel (and carbon) particles and particle-bound organics was
29 summarized in Chapter 11. Diesel emissions are regulated separately (mobile source
30 provisions of the Clean Air Act), and many of the organics of concern on PM are regulated
31 as hazardous air pollutants. Therefore, they do not have much direct relevance per se for
April 1995 13-64 DRAFT-DO NOT QUOTE OR CITE
-------
1 deriving criteria for PM standards. Nevertheless, they do have an indirect relevance. The
2 diesel and carbon studies assist in interpreting other animal toxicity studies by showing that:
3 (1) high concentrations of particles can cause unrealistic phenomena (particle overloading)
4 that can produce health outcomes not likely to occur at typical non-episodic ambient PM
5 concentrations; and that (2) the carbon core of the diesel particle may be significantly more
6 responsible than adsorbed organics for its carcinogenic properties. Data on particle-bound
7 organics illustrate that PM can act as a physical carrier of other pollutants, including
8 carcinogenic compounds. However, quantitative relationships between (a) results of animal
9 studies of organic extracts of particles concentrated from ambient air and (b) potential human
10 carcinogencity associated with ambient or near-ambient exposures to such particles remain to
11 be defined.
12 Because of the above factors, this section focuses mainly on the health effects of acid
13 aerosols as the one specific PM constituent class having the most extensive data most clearly
14 pointing toward likely contributions to reported ambient PM-mortality/morbidity effects. The
15 bulk of the recent data base of controlled exposure studies on PM involves sulfur oxide
16 particles, primarily H2SO4, and the available evidence indicates that the observed responses
17 to these are likely due to H+ rather than to SO^2. Acidic sulfates exert their action
18 throughout the respiratory tract, with the response and location of effect dependent upon
19 particle size and mass and number concentration. The other constituents discussed in
20 Chapter 11 are alluded to later in this integrated synthesis chapter as they pertain to the
21 biological plausibility of the epidemiologic results.
22
23 13.6.1 Mortality Effects of Acid Aerosols
24 Few epidmiological studies have examined mortality data for an association with
25 ambient particulate strong acid aerosol (H+) exposures. The scarcity of the analyses is due
26 to the absence of adequate ambient acid measurement techniques in the past, and to the lack
27 of routine acid aerosol monitoring in more recent years. Some studies now exist which
28 suggest that human health effects may be associated with expsoures to ambient acid aerosols,
29 both: (1) as derived from reexamination of older, historically important data on air pollution
30 episode events in North America and Europe and (2) as can be deduced from limited recent
31 epidemiology studies carried out in the U.S., Canada, and Europe.
April 1995 13.55 DRAFT-DO NOT QUOTE OR CITE
-------
1 Historical and present-day evidence suggests that strongly acidic PM can be associated
2 with both acute and chronic human health effects. Evidence from historical pollution for
3 episodes, notably the London Fog episodes of the 1950's and early 1960's, indicates that
4 extremely elevated daily acid aerosol concentrations (on the order of 400 /xg/rn3 as H2SO4, or
5 roughly 8,000 nmoles/m3 H+) may be associated with excess acute human mortality when
6 present as a copollutant with elevated concentrations of PM and SO2. In addition, Thurston
7 et al. (1989) and Ito et al. (1993) both found significant associations between acid aerosols
8 and mortality in London during non-episode pollution levels (< 30 ^g/m3 as H2SO4, or
9 < approximately 600 nmoles/m3 H+), although these associations could not be separated
10 from those for BS or SO2. The only attempts to date to associate present-day levels of acidic
11 aerosols with acute and chronic mortality (Dockery et al., 1992; Dockery et al., 1993b,
12 respectively) were unable to do so, but weaknesses in these analyses (in particular, too
13 limited H+ data for the analysis) may have made associations undetectable. At very high
14 concentrations that do not occur in the ambient air, mortality in laboratory animals can occur
15 following acute exposure, due primarily to laryngeal or bronchoconstriction; larger particles
16 are more effective in this regard than are smaller ones.
17
18 13.6.2 Respiratory Illness Effects of Acid Aerosols
19 Historical and present-day evidence suggests that there can be both acute and chronic
20 effects of strongly acidic PM on human health. Increased hospital admissions for respiratory
21 causes were documented during the London Fog episode of 1952, and this association has
22 now been observed under present-day conditions, as well. Thurston et al. (1992) and
23 Thurston et al. (1994) have noted associations between ambient acidic aerosols and
24 summertime respiratory hospital admissions in both New York State and Toronto, Canada,
25 respectively, even after controlling for potentially confounding temperature effects. In the
26 latter of these studies, significant independent H+ effects remained even after simultaneously
27 considering the other major copollutant, O3, in the regression model. In these studies, H +
28 effects were estimated to be the largest during acid aerosol episodes (H+ > 10 /xg/m3 as
29 H2SO4, or «200 nmoles/m3 H+), which occur roughly 2 to 3 times per year in eastern
30 North America. These studies provide evidence that present-day strongly acidic aerosols
April 1995 13-66 DRAFT-DO NOT QUOTE OR CITE
-------
1 may represent a portion of PM which is particularly associated with significant acute
2 respiratory disease health effects in the general public.
3 Results from recent acute symptoms studies of healthy children indicate the potential for
4 acute acidic PM effects in this population. While the 6-City study of diaries kept by parents
5 of children's respiratory and other illness did not demonstrate H+ associations with lower
6 respiratory symptoms except at H+ above 110 nmoles/m3 (Dockery et al., 1994), upper
7 respiratory symptoms in two of the cities were found to be most strongly associated with
8 daily measurements of H2SO4 (Schwartz, et al. 1991).
9 Studies of the effects of chronic H+ exposures on children's respiratory health and lung
10 function are generally consistent with effects as a result of chronic H+ exposure.
11 Preliminary analyses of bronchitis prevalence rates as reported across the 6-City study locales
12 were found to be more closely associated with average H+ concentrations than with PM in
13 general (Speizer, 1989). A follow-up analysis of these cities and a seventh locality which
14 controlled the analysis for maternal smoking and education and for race, suggested
15 associations between summertime average H+ and chronic bronchitic and related symptoms
16 (Damokosh et al., 1993). The relative odds of bronchitic symptoms with the highest acid
17 concentration (58 nmoles/m3 H+) versus the lowest concentration (16 nmoles/m3) was 2.4
18 (95% CI: 1.9 to 3.2). Furthermore, in a follow-up study of children in 24 U.S. and
19 Canadian communities (Dockery et al., 1993a) in which the analysis was adjusted for the
20 effects of gender, age, parental asthma, parental education, and parental allergies, bronchitic
21 symptoms were confirmed to be significantly associated with strongly acidic PM (relative
22 odds = 1.7, 95% CI: 1.1 to 2.4). It was also found in the 24-Cities study that mean FVC
23 and PEX^ 0 were lower in locales having high particle strong acidity (Raizenne et al., 1993).
24 Thus, chronic exposures to strongly acidic PM may have effects on measures of respiratory
25 health in children.
26 The respiratory tract has an array of defense mechanisms to kill, detoxify, and
27 physically remove inhaled material, and these defenses may be altered by exposure to H2SO4
28 at levels < 1,000 /ig/m3. Acid aerosols alter mucociliary clearance in human and laboratory
29 animals, with effects dependent on exposure concentration and the region of the lung being
30 studied. For example 1- to 2-h resting exposures of humans to 100 /*g/m3 accelerate
31 clearance in large bronchi, but slows clearance in smaller more peripheral airways.
April 1995 13.57 DRAFT-DO NOT QUOTE OR CITE
-------
1 Clearance in asthmatics is also affected, but the results are not clearly interpretable. Long-
2 term exposure also affects mucociliary clearance in animals. For example, in rabbits
3 exposed intermittently for 125 pig/m3 H2SO4 for 1 year, clearance was accelerated during
4 exposure but was depressed 6 months after exposure ceased. These responses are complex
5 and are accompanied by histological and chemical changes in mucus and epithelial secretory
6 cells. Defenses, such as resistance to bacterial infection, may be altered by acute exposure
7 to concentrations of H2SO4 around 1,000 /*g/m3.
8 Severe morphologic alterations in the respiratory tracts of animals occur at high acid
9 levels. At low levels and with chronic exposure, the main response seems to be hypertrophy
10 and/or hyperplasia of mucus secretory cells in the epithelium; these alterations may extend to
11 the small bronchi and bronchioles, where secretory cells are normally rare or absent.
12 Limited data also suggest that exposure to acid aerosols may affect the phagocytic
13 functioning of alveolar macrophages; the lowest level examined to date is 500 /ig/m3 H2SO4.
14 Alveolar region particle clearance is accelerated by repeated H2SO4 exposures to as low as
15 250 /xg/m3; higher levels retard clearance. Acute exposure of rabbits to lower concentrations
16 (e.g., 75 /xg/m3 H2SO4) can affect other alveolar macrophage functions.
17
18 13.6.3 Pulmonary Function Effects of Acid Aerosols
19 Both acute and chronic exposure of laboratory animals to H2SO4 at levels well below
20 lethal ones will produce functional changes in the respiratory tract. The pathological
21 significance of some of these are greater than for others. Acute exposure will alter
22 pulmonary function, largely due to bronchoconstrictive action. However, attempts to
23 produce changes in airway resistance in healthy animals at levels below 1 mg/m3 have been
24 largely unsuccessful, except when the guinea pig has been used. The lowest effective level
25 of H2SO4 producing bronchoconstriction to date in the guinea pig is 100 />tg/m3 (1-h
26 exposure). In general, smaller size droplets are more effective in altering pulmonary
27 function, especially at low concentrations. Yet even in the guinea pig, there are
28 inconsistencies in the type of response exhibited towards acid aerosols. Chronic exposure to
29 H2S04 is also associated with alterations in pulmonary function (e.g., changes in the
30 distribution of ventilation and in respiratory rate in monkeys). But, in these cases, effective
31 concentrations are >500 /ug/m3. Hyperresponsive airways have been induced with repeated
April 1995 13-68 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposures to 250 /xg/m3 H2SO4 in rabbits, and have been suggested to occur following single
2 exposures at 75 pig/m3.
3 Ten human clinical studies since 1988 have confirmed previous findings that healthy
4 subjects do not experience decrements in lung function following single exposures to H2SO4
5 at levels up to 2,000 /zg/rn3 for 1 h, even with exercise and use of acidic gargles to minimize
6 neutralization by oral ammonia. Mild lower respiratory symptoms occur at exposure
7 concentrations in the mg/m3 range, particularly with larger particle sizes.
8 There is no clearly established exposure-response relationship across studies. Asthmatic
9 subjects appear to be more sensitive than healthy subjects to the effects of acid aerosols on
10 lung function, but reported effective concentrations differ widely among studies. Adolescent
11 asthmatics may be more sensitive than adult asthmatics, and may experience small
12 decrements in lung function in response to H2SO4 at exposure levels only slightly above peak
13 ambient levels (e.g., less than 100 /ig/m3). Although the reasons for the inconsistency
14 among studies remain largely unclear, individual variability in sensitivity and subject
15 selection may be an important factors. Even in studies reporting an overall absence of
16 effects on lung function, occasional asthmatic subjects appear to demonstrate clinically
17 important effects. Two studies from different laboratories have suggested that responsiveness
18 to acid aerosols may correlate with the degree of baseline airway hyperresponsiveness.
19 However, based on very limited studies, the elderly and individuals with chronic obstructive
20 pulmonary disease do not appear to be particularly more susceptible to the effects of acid
21 aerosols on lung function than healthy adults.
22 Two recent studies have examined the effects of exposure to both H2SO4 and ozone on
23 lung function in healthy and asthmatic subjects. Both studies found evidence that 100 /zg/m3
24 H2SO4 may potentiate the response to ozone, in contrast with previous studies. Recent
25 summer camp (and schoolchildren) studies of lung function have also indicated a significant
26 association between acute exposures to acidic PM and decreases in the lung function of
27 children independent of those associated with O3 (Studnicka et al., 1995; Neas et al, 1995).
28 In view of uncertainties about differences between high acid concentrations needed to
29 produce effects in animal studies and low concentrations found in the human environment,
30 the epidemiologic evidence does not establish a clear role for acid aerosols as a primary
31 agent contributing to ambient PM exposure effects on pulmonary function.
April 1995 13.69 DRAFT-DO NOT QUOTE OR CITE
-------
1 13.7 BIOLOGICAL PLAUSIBILITY: POTENTIAL MECHANISMS OF
2 ACTION
3 13.7.1 Introduction
4 Epidemiologic studies have suggested that ambient paniculate exposure may be
5 associated with increased mortality and morbidity at PM concentrations below those
6 previously thought to affect human health (Chapter 12). However, the biological plausibility
7 of a causal relationship between low concentrations of PM and daily mortality and morbidity
8 rates is neither intuitively obvious nor expected based on experimental studies of the toxicity
9 of inhaled particles. As indicated in Chapter 11, chronic toxicity from poorly soluble
10 particles has been observed based on the slow accumulation of large lung burdens of
11 particles, not due to small daily fluctuations of one or another of the specific PM constituents
12 discussed in that chapter. Two possible exceptions can be noted. Acute toxicity from
13 inhaled particles has been demonstrated with acidic particles, but only at much higher particle
14 concentrations than those observed in the recent epidemiology studies reporting an association
15 between low-level PM concentrations and morbidity/mortality. Acute toxicity resulting in
16 death has also been reported in rats inhaling singlet ultrafine particles (<0.05 /mi) formed in
17 the pyrolysis of perfluorinated compounds at concentrations of 60 to 200 /*g/m3' (Oberdorster
18 et al., 1995; Warheit et al., 1990), but the significance of these findings for ambient human
19 exposures is yet to be determined.
20 To approach the difficult problem of determining if the association between low-level
21 PM concentrations and daily morbidity and mortality is biologically plausible, one must
22 consider: the chemical and physical characteristics of the particles in the inhaled
23 atmospheres; the characteristics of the morbidity/mortality observed and the affected
24 population; as well as potential mechanisms that might link the two. Several salient
25 considerations related to the evaluation of biological plausibility of the epidemiology findings
26 are discussed below.
27
28 13.7.2 Characteristics of Observed Morbidity and Mortality
29 If daily mortality rates are increased in association with elevated ambient paniculate
30 concentrations, what are the people dying of? Schwartz (1994) addressed this question by
31 comparing causes of death in Philadelphia on high pollution days (average = 141 /ig/m3)
April 1995 13-70 DRAFT-DO NOT QUOTE OR CITE
-------
1 with causes of deaths on lower pollution days ( average = 47 /ig/m3). On the high pollution
2 days there was a higher relative increase in deaths due to chronic obstructive pulmonary
3 disease (COPD) (RR = 1.25), pneumonia (RR = 1.13), cardiovascular disease (RR = 1.09)
4 and stroke (RR = 1.15). There was also an increase in reports of respiratory factors being
5 contributing causes in the deaths and a higher relative age of those dying. The patterns of
6 causes of death and age of those dying were found to be similar to the patterns observed in
7 the London smog deaths of 1952.
8 Other studies on associations of morbidity with paniculate pollution noted small
9 decreases (2 to 2.5%) in pulmonary function (FVC or FEVj) in smokers on high pollution
10 days (100 ^g/m3; Salt Lake City; Pope and Kanner, 1993) and in nonsmokers (>60 Mg/m3;
11 NHANES I data, Chestnut et al., 1991). An increased number of asthma attacks among
12 working age adults was correlated with increases in paniculate pollution over a 3-year period
13 (average particle level = 76 ^g/m3) in Helsinki (Ponka, 1991). Thus, the characteristics of
14 health effects on high particle pollution days are mainly cardiopulmonary in nature and are
15 the types of effects that can be considered plausibly related to airborne toxicants.
16 It is also of interest to consider the health status of the people affected. People with
17 previously existing health conditions (such as COPD, asthma, or other chronic debilitating
18 conditions) are logically likely to be more susceptible to effects from exposure to paniculate
19 pollutants than would be healthy persons. Such a situation might result in an increased daily
20 mortality rate on days with higher PM10, followed by a decreased daily mortality rate so that
21 the average mortality rate over a longer time period would not be affected.
22 Data on the relative effect of particle exposures in persons with pre-existing pulmonary
23 disease compared to healthy persons do not yield a clear picture. Pope and Kanner (1993)
24 reported an approximate 2 % decline in FEVj in smokers with mild to moderate COPD
25 during an increased concentration in PM10 of 100 /zg/m3 in Salt Lake City. However,
26 persons with severe COPD (average FEVj equal to 50% of predicted) had no further
27 reduction in pulmonary function upon acute (2 h) exposure to 90 /*g/m3 H2SO4 in clinical
28 studies (Morrow et al., 1994). Exercising asthmatics experienced mild bronchoconstriction
29 following the same exposures. In a separate study, exercising adolescent asthmatics exposed
30 to 68 jiig/m3 H2SO4 experienced reduced pulmonary function (average of 6% decrease in
31 FEVj) (Koenig et al., 1989), but in another study, exercising asthmatics did not respond to
April 1995 13_71 DRAFT-DO NOT QUOTE OR CITE
-------
1 exposures to as high as 130 Mg/m3 H2S04 (Avol et al., 1990). Using an elastase-induced rat
2 model of emphysema, Mauderly et al. (1990) found that exposure to diesel exhaust, which
3 contains aggregates of ultrafine soot particles, resulted in less particle deposition in the lungs
4 of emphysematous rats than in normal rats, thus sparing the emphysematous rats the health
5 effects induced by the soot particles in normal animals.
6
7 13.7.3 Influence of Particle Size, Chemical Composition, and Respiratory
8 Tract Deposition/Clearance
9 The PM10 standard is the only U.S. national ambient air quality standard that is not
10 chemical-specific. The chemical composition of a particle will greatly affect its toxicity and,
11 if possible, should be considered in determining if the observed associations between
12 atmospheric PM concentrations and increases in morbidity/mortality are causal. For
13 example, alpha-quartz particles are more toxic than TiO2 particles (Driscoll and Maurer,
14 1991); and acid sulfate aerosols are more likely to cause acute health effects than are neutral
15 sulfate aerosols (Fine et al., 1987).
16 Size is also important in defining the toxicity of particles. Recent studies indicate that
17 ultrafine particles (<20 nm) are much more toxic than larger inhalable particles (Oberdorster
18 et al., 1992; Driscoll and Maurer, 1991). The ultrafine particles have a greater number and
19 surface area per unit mass than fine or coarse particles, which may account, in part, for their
20 greater toxicity. Fine particles tend to have a different chemical composition than larger
21 particles, because their source is often combustion processes. A study of the chemical
22 composition of PM2 5 particles versus PM10 particles in Los Angeles indicated that nitrates,
23 sulfates, ammonium and organic and elemental carbon were the most abundant species in the
24 PM2.5 fraction, while the coarser particles contained soil-related species, such as aluminum,
25 silicon, calcium, and iron (Chow et al., 1994). Chemical composition of PM10 is discussed
26 in Chapter 3 and summarized earlier in this chapter (Section 13.2).
27 In a few epidemiology studies, the investigators attempted to determine what size
28 and/or chemical form of particles had the strongest association with health effects. For
29 example, in the Harvard 6-cities study (Dockery et al., 1993), the excess chronic mortality
30 was most strongly associated with the ambient fine particles, including sulfates. However, in
31 a study of daily air pollution in St. Louis and eastern Tennessee by Dockery et al. (1992),
April 1995 13-72 DRAFT-DO NOT QUOTE OR CITE
-------
1 the strongest association of participate pollution with daily mortality rates was PM10, with
2 progressively weaker associations with PM2 5, sulfate, and aerosol acidity. This is the
3 opposite of what one would expect if aerosol acidity were the main cause of increased
4 mortality, as has been suggested (Lippmann, 1989). The Six Cities study investigators state,
5 however, that the low daily death counts, the short study period, and the large geographic
6 areas considered in the St. Louis/Eastern Tennessee study limited the statistical power of the
7 study, and they could not conclude that the acidity of the aerosol was not associated with
8 mortality.
9 If the chemical and physical forms of the PM are important in determining the health
10 effects induced by PM, one would expect different concentration-response curves to be
11 observed in different epidemiology studies, depending on the type of aerosol present in the
12 atmosphere. Spurny (1993) in his analysis of studies conducted in the south-western part of
13 Germany found differences in concentrations, composition, and cell-toxic effects among
14 urban, residential, and remote areas. The different slopes of the concentration-response
15 curves for the different cities could be due to several factors, including differences in
16 physicochemical properties and resultant potency of the PM in the different cities.
17 It is also worth noting that considerations of dosimetry could potentially provide insight
18 on plausible mechanisms or alter the exposure-dose-response relationships evaluated. To
19 date, most analyses have used the exposure concentration (/xg/m3) of particles. Because
20 deposition of particles in the respiratory tract are determined by particle diameter and
21 distribution, calculation of the RR estimates based on various internal dose metrics (e.g.,
22 deposited dose (mass) rate per tracheobronchial or alveolar surface area or deposited particle
23 number rate per surface area), could alter some of these relationships. Different dose
24 metrics may be more appropriate to characterize acute effects (e.g., mortality) versus chronic
25 effects (e.g., morbidity). Certainly dosimetry can provide insight on the variability of
26 inhaled dose due to differences in airway morphometry and ventilation rates among species,
27 age, genders, and disease status of the respiratory tract. For example, it has been shown that
28 patients with COPD have increased deposited particle burdens when compared to healthy
29 subjects (Anderson et al., 1992). To the extent that particle composition alters the particle
30 diameter and distribution of a given aerosol, dosimetry will also be effected. Solubility of an
31 aerosol influences clearance rates and subsequent retained dose estimates. The potential for
April 1995 13.73 DRAFT-DO NOT QUOTE OR CITE
-------
1 dosimetry to influence the exposure-dose-relationship should be considered to the extent that
2 mathematical modeling andd norphometric data allow.
3
4 13.7.4 Potential Mechanisms of Causality Between Low Levels of
5 Participate Pollution and Health Effects
6 Pathophysiologic mechanisms by which various specific PM constituents can cause
7 health effects are discussed in detail in Chapter 11. Here, the focus is on mechanisms by
8 which airborne particles are known to cause health effects and the extent to which such
9 mechanisms provide plausible evidence or explanations for the reported epidemiologic
10 findings of increases in morbidity and daily mortality rates at low PM concentrations. For
11 purposes of this discussion, health effects of particle inhalation are discussed below in terms
12 of: clinical considerations, acute lung injury, chronic pulmonary toxicity from accumulation
13 of particles in the lung, effects on pulmonary function, effects on pulmonary defense
14 mechanisms, and pathophysicologic mechanisms. Also considered in this section is the
15 potential for interactive mechanisms among air pollutants that might influence the health
16 effects induced by airborne particles. This is an area in which there is little information;
17 most studies have been directed toward determining the toxicity of single compounds.
18
19 Clinical Considerations
20 Potential mechanisms which might help explain the phenomenon of particle related
21 mortality have been considered by Frampton and Utell (1995). These mechanisms include:
22 (1) "premature" death, that is the hastening of death for individuals already near death (i.e.,
23 hastening of an already certain death by hours or days); (2) increased susceptibility to
24 infectious disease; and (3) exacerbation of chronic underlying cardiac or pulmonary disease.
25 Particulate pollution could contribute to daily mortality rates by affecting those at
26 greatest risk of dying, i.e., those individuals for whom death is already very imminent.
27 Acute exposure to moderately elevated concentrations which might only be a minor irritant to
28 healthy people could be the "last straw" that tips over the precariously balanced physiology
29 of a dying patient. Other studies suggest that the full effect of particles on mortality cannot
30 be explained solely by acute PM exposure death-bed effects (Frampton and Utell, 1995), i.e.,
April 1995 13-74 DRAFT-DO NOT QUOTE OR CITE
-------
1 some studies also indicate an effect on annual mortality rates which cannot be explained
2 simply by the hastening of death for individuals already near death.
3 Particle exposure could also increase susceptibility to respiratory infection with bacteria
4 or viruses, leading to an increased incidence of (and death from) pneumonia in susceptable
5 members of the population. However, pneumonia rarely results in death within 24 h of
6 onset; serious infections of the lower respiratory tract generally develop and evolve over
7 weeks, and would not explain effects on daily mortality. If pollutant exposure increased
8 susceptibility to infectious disease, it should be possible to detect differences in the incidence
9 of such diseases in communities with low vs. high particle concentrations. Emergency room
10 visits and hospitalizations for pneumonia caused by the relevant agent should also be
11 measurably higher on days with elevated ambient particle concentrations. However, no such
12 relationship has been observed, and laboratory animal data to support such a mechanism are
13 weak.
14 Paniculate air pollution might also aggravate the severity of underlying chronic lung
15 disease. This mechanism could explain increases in daily mortality (through effects on those
16 near death from their disease) and longitudinal increases in mortality (if individuals with
17 chronic airways disease experienced more frequent or severe exacerbation of their disease, or
18 more rapid loss of function as a result of particulate exposure).
19 What chronic disease processes are most likely to be affected by inhaled particulate
20 matter? To explain the daily mortality statistics, there must be common conditions that
21 contribute significantly to overall mortality from respiratory causes. The most likely
22 candidates are the chronic airways diseases, particularly chronic obstructive pulmonary
23 disease (COPD). This group of diseases encompasses both emphysema and chronic
24 bronchitis, however, information on death certificates does not allow differentiation between
25 these diagnoses. The pathophysiology includes chronic inflammation of the distal airways as
26 well as destruction of the lung parenchyma. There is loss of supportive elastic tissue, so that
27 the airways collapse more easily during expiration, obstructing outflow of air. Processes that
28 enhance airway inflammation or edema, increase smooth muscle contraction in the
29 conducting airways, or slow mucociliary clearance could adversely affect gas exchange and
30 host defenses. Moreover, the uneven ventilation-perfusion matching characteristics of this
31 disease, with dependence on fewer functioning airways and alveoli for gas exchange, means
April 1995 13.75 DRAFT-DO NOT QUOTE OR CITE
-------
1 inhaled particles may be directed to the few remaining functioning lung units in higher
2 concentration than in normal lungs (Bates, 1992)
3 Paniculate pollutants have been associated with increases in cardiovascular mortality
4 both in the major air pollution episodes and in the more recent time-series analyses. Bates
5 (1992) has postulated three ways in which pollutants could affect cardiovascular mortality
6 statistics. These include: (1) acute airways disease misdiagnosed as pulmonary edema;
7 (2) increased lung permeability, leading to pulmonary edema in people with underlying heart
8 disease and increased left atrial pressure and (3) acute bronchiolitis or pneumonia induced by
9 air pollutants precipitating congestive heart failure in those with pre-existing heart disease.
10 Moreover, the pathophysiology of many lung diseases is closely intertwined with cardiac
11 function. For example, one postulated cause of the increasing mortality rate in asthma is
12 overuse of adrenergic agonist medications leading to fatal cardiac arrhythmias. Many
13 individuals with COPD also have cardiovascular disease caused by smoking, aging, or
14 pulmonary hypertension accompanying COPD. Terminal events in patients with end-stage
15 COPD are often cardiac complications, and may therefore be misclassified as cardiovascular
16 deaths. Hypoxemia associated with abnormal gas exchange can precipitate cardiac
17 arrhythmias and sudden death.
18
19 Acute Lung Injury
20 The acute toxicity of particles in the respiratory tract has been the topic of numerous
21 studies to determine the potential pulmonary toxicity of dusts, particularly those of concern in
22 industrial processes. Toxic particles that deposit in the lung can induce an inflammatory
23 response that, if it persists, may lead to pulmonary fibrosis and impaired pulmonary function.
24 The response of the respiratory tract to such particles includes the release of numerous
25 cytokines from alveolar macrophages and epithelial lining cells that promote healing and
26 repair or, if healing does not occur because of the persistence of toxic particles, may
27 promote development of fibrosis. Although such acute responses are well known, they
28 typically only occur after several days or weeks of exposure to airborne particle
29 concentrations many fold higher than those that have been shown to be associated with
30 increased mortality and morbidity in epidemiology studies. Recently, however, it has been
31 observed in experimental animal studies that certain types of particles are acutely toxic to the
April 1995 13-76 DRAFT-DO NOT QUOTE OR CITE
-------
1 lung at low exposure concentrations. A half-hour exposure of rats to freshly generated
2 ultrafine polytetrafluoroethylene particles at a concentration of 64 /zg/m3 resulted in severe
3 pulmonary inflammation and death (Oberdorster et al, 1995). Warheit et al. (1990) also
4 found that fresh ultrafine aerosols resulted in mortality in rats by causing severe lung injury.
5 The significance for environmental exposures of the highly toxic fresh aerosols formed from
6 pyrolysis of perfluorinated materials is unknown at this time, because of the rapid loss of
7 toxicity of the aerosols with time and the lack of information on the concentration of those
8 specific aerosols in the ambient atmosphere. Although it is known that combustion processes
9 emit ultrafine aerosols into the environment (Cantrell and Whitby, 1978), it is not clear how
10 much ultrafine particulate matter is present as the product of pyrolysis of perfluorinated
11 compounds. Nor is there much information on typical ambient concentrations of other
12 ultrafine particles (e.g., metals from high temperature smelting) or their persistance as
13 ultrafmes in urban aerosol mixes.
14
15 Toxicity Resulting from Accumulation of Particles in the Lung
16 The accumulation of large lung burdens of poorly soluble particles can lead to
17 decreased clearance of subsequently inhaled particles and an enhanced rate of accumulation
18 of particles in the lung (Morrow, 1992). Large lung burdens of particles of even relatively
19 low inherent toxicity have been shown to induce lung cancer in animal models such as the rat
20 (Mauderly et al., 1994). But how much prior exposure to particles is required to accumulate
21 enough particles to impair clearance of subsequently inhaled particles? Rats exposed to 350
22 fj-g/m* diesel soot (aggregated ultrafine carbon particles) for 24 months did not accumulate
23 enough particles to induce pulmonary inflammation (as measured by both histopathology and
24 analysis of lung lavage fluid) or to impair particle clearance, but rats exposed to 3500 /xg/m3
25 for the same length of time did. Rats that inhaled carbon black particles at an 8-h
26 time-weighted concentration of 10,000 /ig/m3 5 days a week for 12 weeks also accumulated
27 enough particles to induce an inflammatory response by 6 weeks (Henderson et al., 1992).
28 In general, the toxicity resulting from accumulation of large burdens of particles in the
29 lung does not likely provide a plausible biological basis for reported associations between
30 acute exposures to low level PM concentrations (ca, 30 to 200 /zg/m3) of inhalable particles
31 (PM10) and daily mortality/morbidity rates. One possible exception that stands out as a
April 1995 13_77 DRAFT-DO NOT QUOTE OR CITE
-------
1 relatively sizeable segment of the general population would be smokers or former smokers
2 among the elderly. In such individuals, particle overload from 40 to 60 years of directly
3 inhaled tobacco smoke particles could make them more vulnerable to the impacts of
4 relatively small additional acute increments in their lung particle burdens, as would the
5 preexisting chronic cardiorespiratory diseases caused by smoking.
6 A second possible exception might be elderly persons who experienced notable past
7 exposures over many years to very high ambient or workplace PM concentrations, as would
8 be the case for individuals who resided or worked in heavily industrialized cities before
9 effective occupational and air pollution control measures were introduced in the 1950s to
10 1970s to reduce such exposures. For example, in the Harvard 6-cities study, an association
11 was found between daily mortality rates and PM levels across a few rural communities,
12 lightly industrialized cities, and some heavily industrialized cities. Because the ranking of
13 the cities in terms of air-pollution levels did not change during the study period, it is not
14 possible to distinguish completely between effects due to past historical exposures and those
15 due to recent exposures. Therefore, the elevation in daily mortality rates in industrialized
16 cities such as Steubenville compared to less industrialized cities (such as Topeka or Portage)
17 may be in part based on accumulated past exposures to higher particle levels and
18 consequently larger lung particle burdens in the former.
19
20 Impaired Respiratory Function
21 Very few of the specific PM constituents discussed in Chapter 11 have acute exposure
22 effects on respiratory function, except possibly at very high concentrations (in the /xg/m3
23 range). One possible exception is acid aerosols, which appear to have acute effects on
24 pulmonary function among some sensitive individuals at levels below 1,000 jwg/m3.
25 Exposures to acid particles are known to induce hyperreactive airways and in some cases,
26 bronchoconstriction, but at concentrations in the mg/m3 range, well above peak U.S. ambient
27 acidity levels of 50 to 75 /-tg/m3. In healthy humans, inhalation of 1,000 jug/m3 H2SO4
28 aerosol for 3 h did not cause any influx of inflammatory cells into the lung based on analysis
29 of lung lavage fluid obtained 18 h after the exposures (Frampton et al., 1992). However,
30 mild bronchoconstriction has been reported after brief exposures to as low as 68 /ig/m3
31 H2SO4 in exercising adolescent asthmatics and 90 /ig/m3 in excersing adult asthmatics
April 1995 13-78 DRAFT-DO NOT QUOTE OR CITE
-------
1 (Morrow et al., 1994); (Koenig et al., 1989), although this has not always been observed
2 (Avol, et al., 1990). Also of interest is the finding that hyperresponsive airways developed
3 after exposure of healthy rabbits to as little as 75 jig/m3 H2SO4 for 3 h (El-Fawal and
4 Schlesinger, 1994). Additional studies have also found that acid-coated particles were more
5 potent than the acid or particles alone. Therefore, under some circumstances, one possible
6 mechanism for increased mortality among some elderly persons with a debilitating disease
7 (asthma) on days with moderately high PM pollution might be that acid aerosols place a
8 stress on their cardiopulmonary system, leading to death.
9
10 Impaired Pulmonary Defense Mechanisms
11 The ability of particulate exposures to reduce pulmonary defense mechanisms has been
12 documented for aerosols of H2SO4 and trace metals. Trace metals have been shown to be
13 cytotoxic to alveolar macrophages (AMs) and immunosuppressive, but only at much higher
14 concentrations than encountered in ambient atmospheres (Zelikoff et al., 1993). Sulfuric
15 acid aerosols have also been shown to alter resistance to bacterial infection in mice after
16 acute exposures to 1,000 /ig/m3; repeated exposures to 100 /*g/m3 reduced mucociliary
17 transport rates in animals. Even these levels of H2SO4 are much higher than have been
18 reported in atmospheres of cities evaluated in the recent epidemiology studies. Also, one
19 would expect effects from impaired pulmonary defense mechanisms to develop over an
20 extended period of continuing exposure, not within a few days.
21
22 Synergistic Effects
23 An area for which there is little information is the potential interactive effects of
24 mixtures of air pollutants and/or with other factors (e.g., aging). The potential significance
25 of mixtures is illustrated by the studies of Amdur and Chen (1989), in which a repeated daily
26 3-h exposure for 5 days of guinea pigs to 20 ^g/m3 of H2SO4 coated on metal particles
27 resulted in decrements in lung volume and pulmonary diffusing capacity and elevations of
28 lung weight/body weight ratio, protein, and number of neutrophils in pulmonary lavage fluid.
29 For example, A 1-h exposure to 20 /ig/m3 H2SO4 coated on metallic particles increased
30 bronchial reactivity in guinea pigs; a 10-fold higher concentration of H2SO4 alone was
31 required to produce the same response (Chen et al., 1992b). However, such synergistic
April 1995 13.79 DRAFT-DO NOT QUOTE OR CITE
-------
1 effects were not observed by Anderson et al. (1992), who studied the effects on 15 healthy
2 and 15 asthmatic volunteers of 1-hr exposures to 100 jwg/m3 H2SO4 (0.5 ^m) or 250 /^g/m3
3 carbon black (0.5 /xm) separately or with the H2SO4 coated on the particles. The exposures
4 did not result in changes in symptoms or pulmonary function, except for an equivocal
5 response in one person.
6 The population segment most susceptible to elevations in ambient PM are the elderly
7 (> 65 years old) with preexisting respiratory disease. Aging, in the absence of pathology, is
8 an extremely complex biological phenomenon and is described as being a multifactorial
9 process composed of both genetic and environmental components (Cristofalo et al., 1994).
10 While the physiological characteristics of the healthy older population is an area of active
11 research, significant decrements in key physiological parameters including lung volumes,
12 FEVj, flow velocity/volume curves, resting cardiac output, and cardiac output reserve with
13 age have been reported (Kenney, 1989). However, there is controversy concerning
14 decrements in physiological function associated with the aging process alone as well as with
15 accompanying disease processes or with other environmental stressors. Moreover, there is
16 little information on the extent to which an older population might be more susceptible to the
17 effects of ambient paniculate pollution (Cooper et al., 1991). It is possible that the elderly
18 are more susceptible to ambient particles because of numerous changes in the body's
19 protective mechanisms and protracted exposures to particles over a life time. This could
20 allow time for latent effects from earlier life tune exposures to manifest themselves, and for
21 potential cumulative effects to emerge. Virtually nothing is known of the possibilities for
22 interaction among toxicants over a long life time or the possibilities for interaction between
23 medications and ambient pollutants.
24
25 Pathophysiologic mechanisms
26 The respiratory system may be compromised and become less efficient in older people
27 or as a result of disease, and inhaled particles could, conceivably, further compromise their
28 respiratory function. Because small increases in environmental particle concentrations would
29 not be lethal to most people, the effect must result from initiating or promoting a lethal
30 failing of a critical function, such as ventilation, gas exchange, pulmonary circulation, or
April 1995 13-80 DRAFT-DO NOT QUOTE OR CITE
-------
1 cardiorespiratory control in subjects brought to the limits of tolerance by preexisting
2 conditions (Mauderly, 1995).
3 Inhaled particles or their pathophysiological reaction products could further impair
4 ventilation in the chronically ill individual by further reducing airway caliber. For example,
5 particles may activate airway smooth muscle, constricting airways, or may influence various
6 airway secretions which could add to and thicken the mucous blanket. Inhaled particles or
7 their pathophysiological reaction products could decrease the diffusing capacity of the lungs
8 by decreasing the area of the respiratory membrane available for diffusion, by increasing
9 diffusion distances across the respiratory membrane, and/or by causing abnormal ventilation-
10 perfusion ratios in some parts of the lung. Particles or their products could also act at the
11 level of the pulmonary vascularure to elicit changes in pulmonary vasculature resistance,
12 which could further alter ventilation-perfusion abnormalities in people with respiratory
13 disease. Furthermore, particles could conceivably alter respiratory and cardiovascular
14 control by affecting local control mechanisms located in the endothelial cells or other sites.
15 This could produce changes in peripheral and central control mechanisms and directly affect
16 the respiratory and cardiovascular control centers. Little evidence is currently available that
17 directly addresses the above speculative possibilities.
18
19 13.7.6 Biological Plausibility Conclusions
20 Having considered the characteristics of the particulate exposure atmospheres and the
21 types of morbidity and mortality associated with the polluted atmospheres, what can be
22 concluded about the biological plausibility of the epidemiological results? The types of
23 morbidity and mortality reported to be associated with increased ambient particle
24 concentrations are consistent with the types of health effects that one might expect from
25 exposures to high levels of PM. Therefore, the type of response seems plausible, if one
26 accepts the temporal relationships modeled in the epidemiological studies. The analyses
27 found associations with 1-day or multi-day (usually 3 to 5 day) lags. The concentrations of
28 particulate matter reported to be associated with such health responses, however, are much
29 lower than would be expected based on animal and human clinical studies of responses to
30 single particulate pollutants. This is true even when one considers that there is evidence that
31 the people who make up the excess mortality population may be susceptible subpopulations.
April 1995 13_81 DRAFT-DO NOT QUOTE OR CITE
-------
1 Moreover, it is not clear what portion of the inhalable paniculate matter constitutes the
2 delivered dose that is associated with the observed morbidity or mortality. There are
3 suggestions from both animal toxicology data and epidemiology data that ultrafine acid
4 aerosols may be of greater health significance than the rest of the particulate mass. Finally,
5 the potential for interactive effects between PM of different types and PM and other air
6 pollutants is not known.
7 Thus, although there are several hypotheses as outlined above, little clear or convincing
8 evidence is available at this time to support the biological plausibility of a causal relationship
9 for the reported epidemiologic associations between low ambient concentrations of PM and
10 daily mortality and morbidity rates.
11
12
13 13.8 IDENTIFICATION OF POPULATION GROUPS POTENTIALLY
14 SUSCEPTIBLE TO HEALTH EFFECTS FROM PM EXPOSURE
15 Certain groups within the population may be more susceptible to the effects of PM
16 exposure, including persons with preexisting respiratory disease, children, and the elderly.
17 The reasons for paying special attention to these groups is that (1) they may be affected by
18 lower levels of PM than other subpopulations and (2) the impact of an effect of given
19 magnitude may be greater. Some potential causes of heightened susceptibility are better
20 understood than others. Subpopulations that already have reduced ventilatory reserves (e.g.,
21 the elderly and persons with asthma, emphysema, and chronic bronchitis) would be expected
22 to be more impacted than other groups by a given decrement in pulmonary function. For
23 example, a healthy young person may not even notice a small percentage change in
24 pulmonary function, but a person whose activities are already limited by reduced lung
25 function may not have the reserve to compensate for the same percentage change.
26 Based on Chapter 12 discussions, it is clear that the bulk of the total mortality effects
27 suggested by the epidemiology studies discussed earlier are among the elderly. During the
28 historic London, 1952 pollution episode the greatest increase in the mortality rate was among
29 older citizens and those having respiratory diseases. An analysis by Schwartz (1994c) of
30 mortality in Philadelphia, PA found the greatest increase in risk of death in those aged 65 to
31 74 and those >74 year of age (mortality risk ratios = 1.09 and 1.12, respectively, between
April 1995 13-82 DRAFT-DO NOT QUOTE OR CITE
-------
1 high and low TSP days). Other studies also suggest that the elderly experience a higher
2 excess risk from exposure to PM air pollution than the population overall.
3 Other potentially susceptible groups include patients with COPD, such as emphysema
4 and chronic bronchitis. Some of these patients have airway hyperresponsiveness to physical
5 and chemical stimuli. A major concern with COPD patients is the absence of an adequate
6 ventilatory reserve, a susceptibility factor described above. In addition, altered distribution
7 of respiratory tract ventilation in COPD may lead to a greater delivery of PM to the segment
8 of the lung that is well ventilated, thus resulting in a greater regional tissue dose. Also, PM
9 exposure may alter already impaired defense mechanisms, making this population potentially
10 more susceptible to respiratory infection. It is estimated (U.S. Department of Health and
11 Human Services, 1990; Collins, 1988) that 14 million persons ( = 6%) suffer from COPD in
12 the United States. Bronchial mucous transport clearance may be impaired in people with
13 chronic bronchitis, asthma, and in association with various acute infections. Rates of
14 alveolar region clearance appear to be reduced in humans with chronic obstructive lung
15 disease.
16 Throughout the results and discussions presented above and in Chapter 12 regarding the
17 effects of acute PM exposure on human mortality, a consistent trend was for the effect
18 estimates to be higher for the respiratory mortality category. This lends support to the
19 biological plausibility of a PM air pollution effect, as the breathing of toxic particles would
20 be expected to most directly affect the respiratory tract, and these results are consistent with
21 this expectation. For example, the estimates of relative risk for PM-induced mortality due to
22 respiratory causes discussed in Chapter 12 are all higher than the risks for the population as
23 a whole and for other causes. More specifically, the PM RR for respiratory diseases ranged
24 from 50 to more than 400% higher for respiratory disease categories than for all causes of
25 death, indicating that increases in respiratory deaths are a key major contributor to the
26 overall PM-mortality associations noted previously. PM relative risk estimates for
27 cardiovascular causes were also notably elevated. Moreover, since evidence suggests that an
28 acute pollution episode is most likely be inducing its primary effects by stressing already
29 compromised individuals (rather than, for example, inducing chronic respiratory disease from
30 a single air pollution exposure episode), the above results indicate that persons (especially the
April 1995 13_83 DRAFT-DO NOT QUOTE OR CITE
-------
1 elderly) with pre-existing cardiovascular or respiratory disease constitute a population
2 segment especially at risk for mortality implications of acute ambient PM exposures.
3 Apropos to the identification of individuals with pre-existing respiratory and
4 cardiovascular diseases as being at special risk for ambient PM exposure effects, it is
5 important to highlight smoking as a key etiological agent for such diseases. The U.S.
6 Environmental Protection Agency (1992) report on environmental tobacco smoke indicates
7 that smoking is the major cause of chronic obstructive pulmonary disease (COPD), which
8 includes emphysema, and is thought to be responsible for approximately 61,000 COPD
9 deaths yearly, i.e., about 82% of U.S. COPD deaths (U.S. DHHS, 1989). Tobacco use is
10 also a major risk factor for cardiovascular diseases, the leading cause of death in the United
11 States. It is estimated that each year 156,000 heart disease deaths and 26,000 deaths from
12 stroke are attributable to smoking (CDC, 1991). Smoking is also a risk factor for various
13 respiratory infections, such as influenza, bronchitis, and pneumonia. An estimated 20,000
14 influenza and pneumonia deaths per year are attributable to smoking (CDC, 1991).
15 The U.S. Environmental Protection Agency report also indicates that in children, ETS
16 exposure is causally associated with an increased risk of lower respiratory tract infections
17 such as bronchitis and pneumonia. It is estimated that 150,000 to 300,000 cases annually in
18 infants and young children up to 18 months of age are attributable to ETS. ETS exposure is
19 also causally associated with additional episodes and increased severity of symptoms in
20 children with asthma. It is estimated that 200,000 to 1,000,000 asthmatic children have their
21 condition worsened by exposure to ETS. ETS is also a risk factor for new cases of asthma
22 in children who have not previously displayed symptoms (U.S. Environmental Protection
23 Agency).
24 Lastly, the EPA report also indicates that environmental tobacco smoke (ETS) is a
25 human lung carcinogen, responsible for approximately 3,000 lung cancer deaths annually in
26 U.S. nonsmokers (U.S. Environmental Protection Agency, 1992).
27 Overall, then, the most susceptible population segment that can be most clearly
28 identified as being at likely increased risk for low-level ambient PM exposure-induced
29 mortality or morbidity are elderly individuals with pre-existing cardiovascular respiratory
30 diseases, the majority of which are likely either current or former smokers. Smoking may
31 also be a key ancillary contributor to any low-level PM exposure-induced exacerbation of
April 1995 13-84 DRAFT-DO NOT QUOTE OR CITE
-------
1 respiratory infections among other adults and children and to any increased cancer mortality
2 attributable to chronic ambient PM exposures.
3 Asthmatic subjects appear to be more sensitive than healthy subjects to the effects of
4 acid aerosols on lung function, but the effective concentration differs widely among studies.
5 Adolescent asthmatics may be more sensitive than adults, and may experience small
6 decrements in lung function in response to H2SO4 at exposure levels only slightly above peak
7 ambient levels. Although the reasons for the inconsistency among studies remain largely
8 unclear, subject selection may be an important factor. Even in studies reporting an overall
9 absence of effects on lung function, occasional asthmatic subjects appear to demonstrate
10 clinically important effects. Studies from different laboratories suggest that responsiveness to
11 acid aerosols may correlate with degree of baseline airway hyperresponsiveness. On the
12 other hand, based on very limited studies, elderly and individuals with chronic obstructive
13 pulmonary disease do not appear to be particularly susceptible to the effects of acid aerosols
14 on lung function.
15 Alveolar deposition at different flow rates was lower (26% versus 48% thoracic
16 deposition) in subjects after induced bronchoconstriction. In asthmatics, thoracic deposition
17 of particles was higher than healthy subjects (83% versus 73% of total deposition).
18 Trachial/bronchial deposition was also found to be higher in asthmatics. The results are
19 similar to those found in subjects with obstructive lung disease. The buffering capacity of
20 mucus may be altered in persons with compromised lungs. For example, sputum from
21 asthmatics had a lower pH than that from normals and a reduced buffering capacity, and so
22 may represent a population segment especially sensitive to inhaled acidic particles.
23 The National Institutes of Health (1991) estimates that approximately 10 million persons
24 in the United States have asthma. In the general population, asthma prevalence rates
25 increased by 29% from 1980 to 1987. For those under 20 years old, asthma rates increased
26 from approximately 35 to 50 per 1,000 persons, a 45% increase. The airways of asthmatics
27 may be hyperresponsive to a variety of inhaled materials, including pollens, cold-dry air,
28 allergens, and air pollutants. The potential addition of an PM-induced increase in airway
29 response to the already heightened responsiveness to other substances raises the possibility of
30 exacerbation of this pulmonary disease by PM.
31
April 1995 13-85 DRAFT-DO NOT QUOTE OR CITE
-------
1 13.9 IMPLICATIONS OF RELATIVE RISK ESTIMATES
2 Preceding sections of this Chapter concluded that the newly emerging epidemiologic
3 data base on PM-mortality/morbidity effects provides reasonably consistent results indicative
4 of increased risk of mortality and morbidity effects being associated with exposures of the
5 general population to ambient air pollutant mixes containing PM concentrations currently
6 found in many U.S. urban areas. This includes effects associated with ambient air exposures
7 to pollutant mixes having 24-h PM10 concentrations falling in the range of 30 ^g/m3 to 200
8 /ig/m3, including evidence suggestive of effects below 150 /ig/m3 (the level of the current 24-
9 h U.S. PM10 NAAQS).
10 It was also noted in Chapter 12 that the relative risk (RR) estimates for both the
11 mortality and morbidity effects associated with short-term (ca. 24-h or a few days) exposures
12 to ambient PM are very small compared to RR values typically viewed in epidemiologic
13 literature as providing strong evidence for a likely causative association. Section 13.5.4
14 further noted the relatively limited evidence directly demonstrating coherence between the
15 mortality and morbidity effects findings from epidemiologic studies, with the most
16 compelling evidence for coherence now being findings of both increased hospital admissions
17 (for cardiopulmonary endpoints) and increased mortality in relation to increments in 24-h PM
18 concentrations in the same population group (the elderly) within several U.S. urban areas
19 (Detroit, Birmingham, Philadelphia) and the Utah Valley. However, only very limited
20 evidence for the biological plausibility of acute low-level PM exposure effects at the above-
21 stated PM10 concentration range now appears to exist to support several hypotheses discussed
22 in Section 13.7 with regard to possible mechanisms of action. A key point emerging from
23 the plausibility discussion (Section 13.7) and the ensuing Section (13.8) was the identification
24 of elderly individuals (65 yr. old) with preexisting chronic cardiovascular and respiratory
25 disease conditions (the majority of whom are likely current or former smokers) as being the
26 most susceptible general population segment most clearly at special risk for mortality and
27 morbidity effects associated with exposures to ambient air mixes containing moderately
28 elevated PM concentrations.
29 The meaning or interpretation of quantitative estimates of PM-related effects (i.e.,
30 relative risk estimates) discussed earlier as having been generated by the newly available PM
31 epidemiology studies remains a subject of controversy, with divided opinions still existing in
April 1995 13-86 DRAFT-DO NOT QUOTE OR CITE
-------
1 the scientific community as noted earlier in this chapter. Thus, in attempting to interpret
2 such risk estimates, several caveats should be kept in mind. First, caveats analogous to those
3 made in point (4) at the top of page 13-29 for key conclusions drawn from the last previous
4 PM criteria review still apply. That is, although new evidence has emerged which points
5 toward very small, but statistically significant increases in risk of human mortality and
6 morbidity effects being associated with exposures to ambient air mixes containing moderately
7 elevated PM (with no evident thresholds being identified in the studied range of PM
8 concentrations), precise quantitative specification of relative contributions of such low-level
9 concentrations of ambient PM to reported mortality and morbidity effects is not possible at
10 this time. Nor can one now separate out with confidence potential relative contributions to
11 the reported PM effects of several other likely important confounding or interacting
12 variables.
13 With regard to the latter, it is as of yet very difficult, for example, to sort out with
14 confidence relative contributions of weather versus PM per se. It is clear that temperature
15 extremes (very hot or very cold days in relation to typical ranges of temperature for any
16 given locale) have notable effects on variations in daily mortality, with temperature or other
17 combinations of variables indexing weather impacts usually being found to be significant
18 predictors of daily human mortality in modeling of PM effects and to account for distinctly
19 larger proportions of the variance in daily mortality than do indices of PM pollution. On the
20 other hand, in most of the newer PM studies, small elevations in relative risk attributable to
21 PM still remained even after control for temperature extremes and/or other weather indices;
22 and PM effects were found to be significant in several analyses (e.g., for London) restricted
23 to days not involving wide variations in temperature that would constitute geographic-specific
24 extremes. It is also not yet clear to what extent any given relative risk estimate derived from
25 any of the newer analyses represent actual risk due to an increase in ambient PM or to what
26 extent the elevations in risk attributed to modeled PM indices more broadly represent
27 increased mortality or morbidity risks due to human exposure to the overall pollutant mix in
28 the particular airshed evaluated (including not only the ambient PM aerosols present but
29 other copollutants, such as SO2, CO, O3, NOX or non-particulate organic air toxics).
30 Other caveats bear on the issue of how generalizable the reported PM relative risk
31 estimates are. It is not yet possible to determine the extent to which the risk estimates for
April 1995 13-87 DRAFT-DO NOT QUOTE OR CITE
-------
1 PM mortality or morbidity effects are generalizable to other geographic areas or are highly
2 site-specific, i.e., narrowly applicable to the specific cities from which they were derived or,
3 at least, most credibly confined for use in projecting any estimates of likely PM risk to other
4 airsheds with fairly similar ambient aerosol mixes in terms of particle size distribution and
5 chemical composition. Thus, it is not clear, for example, how credible the use of PM-
6 related relative risk estimates derived from Philadelphia, St. Louis, or other midwestern or
7 eastern U.S. conurbanations (or foreign cities such as Sao Paulo, Santiago, or Athens) with
8 high percentages of particles from combustion processes might be in attempting to estimate
9 PM-related risks for other cities, e.g., in the western U.S., with much greater proportions of
10 crustal materials in the ambient air pollutant mix. Use of presently available PM-related risk
11 estimates to attempt to quantify potential PM-related risks across various seasons in locales
12 where widely varying seasonal mixes of particles of different sizes/chemical composition may
13 also be of dubious scientific credibility at this time.
14 Another issue of much interest and debate has been that of "threshold" for the estimated
15 PM effects derived from the newly available analyses. As noted above and discussed in
16 Chapter 12, no evident thresholds have yet been demonstrated for reported PM-related
17 mortality or morbidity effects, based on the presently available published analyses. On the
18 other hand, as also discussed in Chapter 12, only very limited efforts have been made to date
19 to undertake statistical analyses by which to more definitively address the issue; and serious
20 doubt exists as to whether any thresholds, if they do exist, even in the range of the observed
21 data (i.e., roughly from 30 to 200 /xg/m3 PM10) can be demonstrated, given notable
22 statistical power limitations associated with necessary breaking down of data into more
23 refined intervals as part of any threshold "search". Nor is there now any scientifically
24 credible basis by which to make a "no-threshold" argument in support of extrapolating
25 currently available PM relative risk estimates to ambient PM concentrations below the range
26 of observed data used in the reported analyses. This is especially true in view of the lack of
27 any well demonstrated evidence for one or another hypothesized potential mechanisms of
28 action that might plausibly explain the elevated risk of mortality or morbidity at the very low
29 PM concentrations implied by the results of the newly available epidemiology studies.
30 It is also clear from the available analyses that the occurrence of any increased risk of
31 mortality or morbidity due to short-term moderate elevations in PM (either alone or in
April 1995 13_gg DRAFT-DO NOT QUOTE OR CITE
-------
1 concert with other copollutants) likely represents the outcome of a combination of risk factors
2 culminating in relatively rare health events (as clarified further by the ensuing quantitative
3 discussion below). By far the greatest risk is posed for the elderly over 65 years old and
4 especially those with preexisting cardiopulmonary diseases, with very distinctively lower risk
5 estimates having been derived for younger individuals and those without chronic respiratory
6 or cardiovascular diseases. Thus, in order for notable health effects to occur in association
7 with short-term exposures to ambient PM (and/or copollutants), it appears that other
8 predisposing conditions and/or contributing risk factors must be present, as well. That is,
9 low-level ambient PM exposures alone do not typically appear to be sufficient per se to
10 induce increased morbidity or mortality, but may contribute to such health outcomes under
11 conditions when one or more other contributing risk factors also converge. Thus, for
12 example, short-term low-level exposures to ambient PM at concentrations in the ranges
13 evaluated in most of the newer epidemiology studies are extremely unlikely alone to cause
14 lung function decrements or respiratory symptoms of much note (except possibly for some
15 highly sensitive asthmatic patients), based on currently available epidemiologic and controlled
16 human exposure study results. Nor are such exposures likely to markedly reduce or impair
17 respiratory tract defenses (e.g., alveolar macrophage numbers or function, retrociliary
18 clearance mechanisms, lung immune response, etc.) sufficiently so as to cause increased
19 susceptibility to respiratory infections, based on available experimental toxicology findings.
20 On the other hand, once a respiratory infection were to occur due to other causes, then it is
21 conceivable that added stress due to low-level PM exposure in terms of small further
22 decrements in pulmonary function or exacerbation of respiratory symptoms could lead to
23 worsening of the acute illness and, possibly, the need for medical attention and/or hospital
24 admission in some cases.
25 Still additional converging risk factors appear to be necessary for exposures to ambient
26 air pollution mixes containing low concentrations of typical outdoor urban aerosols to
27 contribute to increased mortality. By far the most important are the cooccurrence of
28 advanced age (> 65 yr old) and already compromised cardiopulmonary function. In older
29 individuals with preexisting COPD, emphysema, chronic heart disease, etc. resulting from
30 other predisposing risk factors (e.g., long-term earlier high-level particle exposures from
31 smoking or past occupational or ambient PM exposures before effective control measures
April 1995 13_89 DRAFT-DO NOT QUOTE OR CITE
-------
1 were introduced), it appears conceivable that additional stress from low-level ambient PM
2 exposures might cause further complications that might lead to terminal consequences in
3 some cases. Several possibilities were discussed earlier as having been hypothesized, e.g.,
4 increased air flow to and consequent greater particle deposition/retention in remaining
5 functioning areas of the compromised lung, possible tipping over by small additional particle
6 burdens of already saturated lung defenses due to particle overloads from past long-term high
7 level particle exposures, and/or the induction of cascading inflammatory or other immune
8 responses-(due to particularly toxic specific PM consultants, e.g., possibly certain transition
9 metals) that overwhelm remaining lung reserves and oxygen exchange mechanisms.
10 However, at this time, no clearly convincing scientific evidence has yet been reported by
11 which to either compellingly substantiate or refute such hypothesized possibilities. Thus,
12 considerable uncertainties still exist with regard to what the relative risk estimates from the
13 newly available epidemiologic studies may imply.
14 In evaluating the potential public health significance of the relative risk increases for
15 mortality or morbidity effects reported in the newer PM epidemiology studies, much recent
16 interest has focussed on use of such relative risk estimates to generate projections of numbers
17 of excess deaths or morbidity events likely to be associated with ambient PM exposures at
18 concentrations currently found in the United States or other countries. Given the above-
19 noted caveats and uncertainties pertaining both to the derivation of the relative risk estimates
20 and their interpretation, there are substantial reasons to caution against attempting such
21 calculations at this time and to have major reservations about accepting any such projections
22 as credible quantitative estimates of additional deaths or morbidity events likely to actually
23 occur with current or future exposures in the United States or elsewhere. At best, such
24 projections might be associated with exposures to PM-containing ambient air mixes in cities
25 with closely similar particle size/chemical composition characteristics and population
26 demographics to those cities from which the relative risk estimates were derived. It is
27 currently questionable as to whether widely generalizable, broadly applicable projections can
28 be made based on some single "best estimate" of PM-related relative risk and, also, whether
29 such projections can be credibly aggregated across PM exposure variations during different
30 seasons and/or across geographic locales with widely disparate mixes of PM aerosols and/or
31 other copollutants.
April 1995 13-90 DRAFT-DO NOT QUOTE OR CITE
-------
1 Despite the above caveats and reservations, however it may be useful to provide
2 illustrative examples of possible quantitative implications of relative risk estimates of the type
3 generated by the recent PM epidemiology analyses. Table 13-5 earlier showed that total
4 acute mortality relative risk estimates associated with exposure to ambient air pollution
5 having a 50 jug/m3 increase in one-day 24-h average PM10 can range from 1.015 to 1.085,
6 depending upon the site (i.e., the PM10 composition and population demographics) and also
7 upon whether PM10 is modeled as the sole index of air pollution or not. Relative Risk
8 estimates with PM10 as the only pollutant index in the model range from RR = 1.025 to
9 1.085, while the PM10 RR with multiple pollutants in the model range from 1.015 to 0.025.
10 The former range, as noted earlier, might be viewed as approximating an upper bound of the
11 best estimate, as any mortality effects of co-varying pollutants may be "picked up" by the
12 PM10 index, whereas the latter multiple pollutant model range might be viewed as
13 approximating a lower bound of the best estimate, as the inclusion of highly correlated
14 covariates might weaken the PM10 estimate. Thus, "typical" total mortality effect estimates
15 (per 50 jiig/m3 PM10 increase) most likely fall within an approximate RR = 1.025 to 1.06
16 range, based on the various coefficients reported in the published studies. The formal EPA
17 meta-analyses results discussed in Chapter 12 yielded a best estimate of 1.031 with 95%
18 confidence intervals (CI) of 1.025 to 1.038 for PM10 studies using models that include 0-1
19 day lags but no copollutants. For those analyses with longer lag times (3-5 days) and no
20 copollutants in the models, the EPA meta-analyses yielded a best estimate of 1.064 (CI =
21 1.047 to 1.082). Thus, the very small increased risks of about 3.1 to 6.4% over baseline
22 mortality levels (per 50 /-ig/m3 increase in 24-h PM10 concentration in the 30 to 200 /ig/m3
23 range) derived from the EPA meta-analyses probably represent currently best available upper
24 bound estimates for reported PM10-related total mortality effects. Lower bound estimates,
25 from analyses that included other copollutants in the models for acute PM-mortality effects,
26 could be as much as 50% lower than the above upper bound estimates or, possibly, even
27 include zero (i.e., represent no increased risk) especially during some seasons in different
28 locales.
29 To help place these findings into a context by which to better understand the potential
30 implications of such relative risk estimates, Table 13-17 summarizes important information
31 by which to project potential increases in excess mortality in a city of one million people that
April 1995 13.91 DRAFT-DO NOT QUOTE OR CITE
-------
1 might be associated with exposure to ambient air mixes in which an increment in 24-h
2 exposure of 50 pig/m3 may be a contributing factor. First it is important to note that the
3 typical general population baseline or background risk of death on any given day in the U.S.
4 is only about 23.6 in a million (23.6 x 10~6) or 23.6 deaths per day in a city of 1 million
5 people. If the 24-h PM10 concentration increased by 50 ^g/m3 on a given day (e.g., from a
6 usual level of about 50 jig/m3 to around 100 Mg/m3) then risk for mortality in the total
7 general population would be expected to increase by about 3.0 to 6.0% over baseline, i.e., to
8 increase from 23.6 in a million to about 24 or 25 in a million as an upper bound estimate.
9 In other words, exposure to the ambient mix of pollutants indexed by the 50 /ig/m3 increase
10 in 24-h PM10 levels might contribute to as much as an additional 0.7 to 1.5 deaths per
11 million people exposed, as shown in the Table 13-17 far right column.
12 Of the 23.6 baseline deaths per day expected in a city of 1 million, about 17 would be
13 attributable to elderly individuals (aged 65 or over), who only constitute about 12.6% of the
14 1991 U.S. population but for whom the background risk of dying on a given day is much
15 higher than for the total general population. For such individuals, the upper bound estimate
16 for increased numbers of excess deaths possibly contributed to by the 50 pig/m3 PM10
17 increase would be projected to be approximately 1.0 (more than half of the higher estimate
18 for total mortality among the entire population), assuming that the specific city has a typical
19 demographic distribution of percentages of people in different age brackets. In other cities
20 or locations with notably higher elderly populations (e.g., some retirement communities or
21 cities left with higher percentages of the elderly possibly due to outmigration of younger
22 people), then the overall risk and expected deaths per day would be higher. Conversely, in
23 other locations with much younger than average populations and lower percentages of elderly
24 residents, the risk and expected numbers of PM-related excess deaths would be lower.
25 If the increment in PM10 concentration continued to average about 50 /ig/m above
26 routine ambient levels for 3-5 days in the given city of 1.0 million people, then relative risk
27 estimates derived from PM10-mortality models using 3-5 d lags might more appropriately
28 apply. Then, the expected number of deaths to which the 3-5 day 50 pig/m3 PM elevation
29 might be projected to contribute could range up to about 1.5 deaths per day among the total
30 general population per 1 million people exposed; or up to about 4.5 to 7.5 deaths during the
April 1995 13-92 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-17. ESTIMATED EXCESS MORTALITY PER DAY IN A POPULATION
OF ONE MILLION FOR WHICH AN INCREASE OF 50 jig/m3 PM10 (24-h)
COULD BE A CONTRIBUTING FACTOR
Health
Outcome
Total
Mortality
Total
Mortality
Respiratory
Mortality
Cardiovascular
Mortality
Age
Group
All
65 +
All
All
All
Population
Baseline
Annual
Mortality
8,603J
6,2013
8,603
6761
3,635
Population
Baseline
Daily
Mortality
23.6
17.0
23.6
1.85
10.0
PM10
Lag
Time
< 2d
2d
3-5d
3-5d
3-5d
Upper Bound
Excess Risk
Per PMjr,
50 ng/m*
0.032
0.064
0.062
0.195
0.095
Possible
Number of
PM-Related
Deaths/Day
0.7
1.0
1.5
0.3
0.9
Monthly Vital Statistics Report for 1991 (U.S. CD 1993).
2From EPA meta-analyses, Table 12-25; all models without co-pollutants
3Elderly as 12.6% of 1991 U.S. population
4From Saldiva and Bohn (1994) and Ostro et al. (1995), variance-weighted average (TWA); Section 12.3.1.3
5From Pope, et al. (1991), Schwartz (1993) for Utah Valley and Birmingham TWA, Table 12-4
TABLE 13-18. ESTIMATED NUMBER OF DEATHS PER DAY IN CITIES OF
10,000 to 10 MILLION1 FOR WHICH AN INCREASE OF 50 jig/m3 PM10
COULD BE A CONTRIBUTING FACTOR
Expected Number of PM-Related Excess Deaths Per Day
Population
of City
10 Million
5 Million
1 Million
500,000
100,000
50,000
10,000
Whole Pop.
All Causes
< 2d Lag
-4 -7
-2 -4
-0.4 -0.7
-0.2 -0.4
-0.05
-0.03
-0.005
65+ Pop.
All Causes
< 2d Lag
-5 - 10
-2.5 - 5
-0.5 - 1
-0.3 - 0.5
-0.07
-0.04
-0.01
Whole Pop.
All Causes
3-5 Day Lag
-7 - 15
-4 - 8
-0.8 - 1.5
-0.4 -0.8
-0.01
-0.05
-0.01
Whole Pop.
Respiratory
3-5 Day Lag
-2 - 3
-1 - 2
-0.2 - 0.3
-0.1 - 0.15
-0.03
-0.02
-0.002
Whole Pop.
Cardiovascular
3-5 Day Lag
-5 -9
-2 -5
-0.5- 1.0
-0.2-0.5
-0.07
-0.04
-0.008
!Upper end of range for each city size calculated from upper bound estimates in Table 13-17 for population of 1
million. Lower end of range derived as lower bound estimate roughly 50% less than the upper bound, as per
text.
April 1995
13-93
DRAFT-DO NOT QUOTE OR CITE
-------
1 full 3-5 days of elevated PM10. Of the 1.50 excess death per day attributed to the 3-5 day 50
2 jng/m3-increment in PM10 24-h concentrations, an estimated 0.34 would likely be due to
3 respiratory causes and about 0.91 to cardiovascular causes. Obviously, both the increased
4 deaths due to respiratory and cardiovascular causes would mainly occur in elderly persons
5 having preexisting chronic respiratory or cardiovascular disease conditions. Note that small
6 numerical inconsistencies in Table 13-17 and in succeeding tables on morbidity arise from
7 the fact that the excess risk estimates are based on different studies in a number of different
8 populations, with different baseline death or hospital admissions rates.
9 Table 13-18 simply takes the information from the far right column of Table 13-17 on
10 upper bound estimates of the number of possible PM-contributory deaths per day (for the
11 total population, for the elderly 65 + , and for respiratory and cardiovascular causes), and
12 depicts ranges of lower and upper bound estimates for comparable numbers of estimated
13 possible deaths per day contributed to by exposure to ambient air pollution mixes containing
14 50 fig/m3 increments in PM10 concentrations in cities ranging from 10 thousand to 10 million
15 in size. Table 13-18 is extremely informative in showing that no appreciable risk for
16 mortality is expected to occur with exposure to such ambient air mixes for cities less than 1
17 million population, even if the PM10 elevation lasts for 3-5 days or occurs several times a
18 year; nor is there much appreciable risk for the elderly in smaller population cities, unless
19 perhaps a particular city with less than 1 million population has an extraordinarily high
20 percentage of elderly residents. This applies even for days when 100 jng/m3-increments in
21 PM10 might occur for 3-5 days in a row. Even for cities of 1 million population, the
22 projected upper bound risk may be of dubious public health significance unless 50-100 /xg/m3
23 PM10 elevations were to occur numerous times per year, especially in view of such tiny
24 increased risk likely mainly being posed for elderly individuals with preexisting
25 cardiopulmonary disease conditions that predominantly arise from voluntary smoking. Any
26 risk of excess mortality associated with short-term, acute exposures to ambient air pollutant
27 mixes having 50-100ptg/m3 increments in 24-h PM10 levels would most likely be projected as
28 possibly causing meaningful numbers of excess deaths mainly as such exposures occur for
29 large segments of the elderly population (age 65 + yrs) with preexisting cardiopulmonary
30 diseases in rather large cities exceeding 1-2 million population. The level of public concern,
31 however, even in such cases may be tempered by the likelihood that a majority of those at
April 1995 13.94 DRAFT-DO NOT QUOTE OR CITE
-------
1 special risk are most probably current or former smokers, given the predominant role
2 attributed (U.S. EPA, 1992) to smoking in the etiology of preexisting cardiopulmonary
3 diseases that distinguish those identified as being at greater risk.
4 There is some limited direct evidence for an interaction between smoking status and
5 excess mortality attributable to PM exposure. Based on the Six Cities Study, Dockery, et.
6 al. (1993) reported an increased RR for PM 2 5 between the least polluted city (Portage) and
7 the most polluted city (Steubenville) that is substantially (albeit not significantly) higher in
8 individuals who are current or former smokers, compared to never-smokers. This is shown
9 in Table 13-19. Prospective studies that have individual data on smoking status probably
10 offer the best opportunity for detecting differential effects of smoking status on PM-related
11 mortality and morbidity for use in future criteria assessments.
12
13
TABLE 13-19. ASSOCIATION BETWEEN CIGARETTE SMOKING STATUS AND
EXCESS MORTALITY RISK FROM AIR POLLUTION IN THE SIX CITIES STUDY
Relative Risk for Worst PM2.5 City (Steubenville)
Versus Lowest Best PM2 5 City (Portage)
Smoking Status
Never Smoker
Former Smoker
Current Smoker
M+F
1.19
1.35
1.32
(95% CI)
(0.90,1.57)
(1.02,1.77)
(1.04,1.68)
M
1.29
1.31
1.42
(95% CI)
(0.80,2.09)
(0.96,1.80)
(1.05,1.92)
F
1.15
1.48
1.23
(95% CI)
(0.82,1.62)
(0.82,2.66)
(0.83,1.83)
'Based on Table 3 from Dockery, et. al (1993)
1 The prematurity of the excess deaths is also a matter of considerable importance, but
2 there is as yet little firm evidence from acute mortality studies by which to judge whether
3 PM-related excess deaths generally represent highly comprised elderly individuals dying a
4 few days or weeks sooner than they would have otherwise versus several months or years of
5 prematurity of death for some.
6 Morbidity effects demonstrated as likely being associated with short-term exposures to
7 ambient U.S. PM exposures include increased hospital admissions for respiratory and
8 cardiovascular disease conditions, increased respiratory symptoms (including exacerbation of
9 asthma), and small pulmonary function decrements (e.g., 2-3% decreases in FEVj or FVC).
April 1995 13.95 DRAFT-DO NOT QUOTE OR CITE
-------
1 Probably of most immediate public health concern are the hospital admissions, which are also
2 more readily quantifiable and understandable as an index of obviously serious health impacts.
3 Table 13-20 summarizes key types of information by which one might attempt to project
4 increments might contribute per of 1 million people exposed. Table 13-20 can be interpreted
5 in an analogous fashion to Table 13-17. Note from Table 13-20 that the typical number of
6 PM10-related hospital admissions for cardiovascular causes would be projected to be only
7 about 2.5 times as high as the number of potential deaths during the same event, and the
8 number of respiratory admissions about 6 times as high as the possible number of deaths
9 from respiratory causes shown in Table 13-17. However, many deaths from cardiovascular
10 or respiratory causes occur without a prior hospital admission. There is, nonetheless, a
11 reasonable numeric consistency between the rough estimates of potential hospital admissions
12 or discharges and possible total deaths contributed to by exposure in a community to PM-
13 containing ambient-air mixes.
14 Table 13-21 then scales expected daily hospital admissions potentially associated with
15 exposures to ambient air mixes having a 50 /xg/m3 increase in PM10 (24-h) for towns and
16 cities with populations of 10 thousand to 10 million (analogous to what was done earlier in
17 Table 13-18, based on Table 13-17 calculations). However, in this case, both tables 13-20
18 and 13-21 provide only upper bound estimates for hospital admissions based on available
19 analyses, which did not include copollutants in the models. Essentially the same types of
20 statements as made with regard to the very small increases in excess risk depicted in Tables
21 13-18 and 13-19 for mortality also generally apply here for hospital admissions, except to
22 note somewhat larger projected numbers for possible hospital admission cases for which the
23 ambient PM exposure might be a contributing factor.
24 Overall, based on the foregoing discussions, there appears to exits credible evidence for
25 a likely very small, but real PM effect on human health in some susceptible subpopulations
26 (including contributing along with other risk factors to premature deaths among the elderly
27 with preexisting cardiopulmonary diseases) at PM10 24-h concentrations in the range of 30 to
28 200 )wg/m3. However, the biological mechanisms by which such effects occur are as yet not
29 well understood and remain to be delineated, as is the case for clearer characterization and
30 interpretation of relative risk estimates for PM-related effects and their appropriate use in
31 projecting potential public health impacts.
April 1995 13-96 DRAFT-DO NOT QUOTE OR CITE
-------
TABLE 13-20. ESTIMATED HOSPITAL ADMISSIONS PER DAY IN A POPULATION OF
ONE MILLION FOR WHICH AN INCREASE OF 50 uglm3 (24-h) PM10 COULD BE A
CONTRIBUTING FACTOR
First
Listed
Diagnosis
All conditions
Respiratory
Conditions (all)
Pneumonia
COPD
Heart
Disease
Age
Group
All
65 +
All
65 +
All
65 +
All
65 +
All
65 +
Population
Baseline
Annual
Discharges
124, HO1
42,8452
12.1801
5,101*
4.3401
2,3352
3,3377
2,5607
12,310
13,502
Population
Baseline
Daily Hospital
Discharges
340.0
117.4
33.4
14.0
11.9
6.4
9.1
7.0
58.4
37.0
Excess Risk
per PMjn
50 /ig/m3
(Lag <.! d)
0.063
0.083
0.084
0.155
0.163
0.046
0.066
Possible Number
of PM-Related
Hosp. Admissions
Per Day
—
2.0
1.1
0.5
1.4
1.1
2.3
2.2
'From Table 12-7
3From Table 12-9, average
5From Table 12-10, average
7From 1992 detailed Tables; excludes asthma (ICD9 493-493.9)
2From Table 12-7, assuming 12.6% age 65 +
4From Table 12-11, average
6From Table 12-12
TABLE 13-21. ESTIMATED NUMBERS OF HOSPITAL ADMISSIONS FOR RESPIRATORY
AND CARDIOVASCULAR CAUSES PER DAY IN CITIES OF 10,000 to 10 MILLION FOR
WHICH AN INCREASE OF 50 jig/m3 PM10 (24-h) COULD BE A CONTRIBUTING FACTOR
All Respiratory
Conditions
Population
of City
10 Million
5 Million
1 Million
500,000
100,000
50,000
10,000
Whole
Pop.
20.0
10.0
2.0
1.0
0.2
0.1
0.02
65 +
Pop.
11.0
5.5
1.1
0.55
0.11
0.05
0.01
Pneumonia
Whole 65 +
Pop. Pop.
5.0
2.5
0.5
0.25
0.05
0.02
0.01
COPD
Whole
Pop.
14.0
7.0
1.4
0.7
0.14
0.07
0.02
65 +
Pop.
11.0
5.5
1.1
0.55
0.11
0.05
0.01
Heart Disease
Whole
Pop.
23.0
11.5
2.3
1.15
0.23
0.12
0.02
65 +
Pop.
22.0
11.0
2.2
1.1
0.22
0.11
0.02
April 1995
13-97
DRAFT-DO NOT QUOTE OR CITE
-------
1 REFERENCES
2
3 Abbey, D. E.; Mills, P. K.; Petersen, F. F.; Beeson, W. L. (1991) Long-term ambient concentrations of total
4 suspended particulates and oxidants as related to incidence of chronic disease in California Seventh-Day
5 Adventists. Environ. Health Perspect. 94: 43-50.
6
7 Abbey, D. E.; Lebowitz, M. D.; Mills, P. K.; Petersen, F. F.; Beeson, W. L.; Burchette, R. J. (1995) Long-
8 term ambient concentrations of particulates and oxidants and development of chronic disease in a cohort
9 of nonsmoking California residents. Inhalation Toxicol. 7: 19-34.
10
11 Amdur, M. O. (1989) Sulfuric acid: the animals tried to tell us. Appl. Ind. Hyg. 4: 189-197.
12
13 Amdur, M. O.; Chen, L. C. (1989) Furnace-generated acid aerosols: speciation and pulmonary effects. In:
14 Symposium on the health effects of acid aerosols; October 1987; Research Triangle Park, NC. Environ.
15 Health Perspect. 79: 147-150.
16
17 American Thoracic Society. (1991) Lung function testing: selection of reference values and interpretative
18 strategies Am. Rev. Respir. Dis. 144: 1202-1218.
19
20 Anderson, K. R.; Avol, E. L.; Edwards, S. A.; Shamoo, D. A.; Peng, R.-C.; Linn, W. S.; Hackney, J. D.
21 (1992) Controlled exposures of volunteers to respirable carbon and sulfuric acid aerosols. J. Air Waste
22 Manage. Assoc. 42: 770-776.
23
24 Aris, R.; Christian, D.; Sheppard, D.; Balmes, J. R. (1990) Acid fog-induced bronchoconstriction: the role of
25 hydroxymethanesulfonic acid. Am. Rev. Respir. Dis. 141: 546-551.
26
27 Aris, R.; Christian, D.; Sheppard, D.; Balmes, J. R. (1991a) The effects of sequential exposure to acidic fog and
28 ozone on pulmonary function in exercising subjects. Am. Rev. Respir. Dis. 143: 85-91.
29
30 Aris, R.; Christian, D.; Sheppard, D.; Balmes, J. R. (1991b) Lack of bronchoconstrictor response to sulfuric
31 acid aerosols and fogs. Am. Rev. Respir. Dis. 143: 744-750.
32
33 Avol, E. L.; Linn, W. S.; Whynot, J. D.; Anderson, K. R.; Shamoo, D. A.; Valencia, L. M.; Little, D. E.;
34 Hackney, J. D. (1988a) Respiratory dose-response study of normal and asthmatic volunteers exposed to
35 sulfuric acid aerosol in the sub-micrometer size range. Toxicol. Ind. Health 4: 173-184.
36
37 Avol, E. L.; Linn, W. S.; Wightman, L. H.; Whynot, J. D.; Anderson, K. R.; Hackney, J. D. (1988b) Short-
38 term respiratory effects of sulfuric acid in fog: a laboratory study of healthy and asthmatic volunteers.
39 JAPCA 38: 258-263.
40
41 Avol, E. L.; Linn, W. S.; Shamoo, D. A.; Anderson, K. R.; Peng, R.-C.; Hackney, J. D. (1990) Respiratory
42 responses of young asthmatic volunteers in controlled exposures to sulfuric acid aerosol. Am. Rev.
43 Respir. Dis. 142: 343-348.
44
45 Balmes, J. R.; Fine, J. M.; Christian, D.; Gordon, T.; Sheppard, D. (1988) Acidity potentiates
46 bronchoconstriction induced by hypoosmolar aerosols. Am. Rev. Respir. Dis. 138: 35-39
47
48 Bates, D. V. (1992) Health indices of the adverse effects of air pollution: the question of coherence. Environ.
49 Res. 59: 336-349.
50
51 Belloc, N. D. (1973) Relationship of health practices and mortality. Prev. Med. 2: 67-81.
52
April 1995 13_98 DRAFT-DO NOT QUOTE OR CITE
-------
1 Burnett, R. T.; Dales, R. E.; Raizenne, M. E.; Krewski, D.; Summers, P. W.; Roberts, G. R.; Raad-Young,
2 M.; Dann, T.; Brook, J. (1994) Effects of low ambient levels of ozone and sulfates on the frequency of
3 respiratory admissions to Ontario hospitals. Environ. Res. 65: 172-194.
4
5 Cantrell, B.; Whitby, R. T. (1978) Aerosol size distributions and aerosol volume formation for a coal-fired
6 power plant plume. Atmos. Environ. 12: 324.
7
8 Centers for Disease Control. (1991) Smoking-attributable mortality and years of potential life lost—United States,
9 1988. Morb. Mortal. Wkly. Rep. 40: 62-71.
10
11 Centers for Disease Control. (1993) Advance report of final mortality statistics, 1991. Hyattsville, MD: U.S.
12 Department of Health and Human Services, Public Health Service, National Center for Health Statistics.
13 (Monthly vital statistics report: v. 42, no. 2, suppl.).
14
15 Chapman, R. S.; Calafiore, D. C.; Hasselblad, V. (1985) Prevalence of persistent cough and phlegm in young
16 adults in relation to long-term ambient sulfur oxide exposure. Am. Rev. Respir. Dis. 132: 261-267.
17
18 Chestnut, L. G.; Schwartz, J.; Savitz, D. A.; Burchfiel, C. M. (1991) Pulmonary function and ambient
19 paniculate matter: epidemiological evidence from NHANES I. Arch. Environ. Health 46: 135-144.
20
21 Chow, J. C.; Watson, J. G.; Fujita, E. M.; Lu, Z.; Lawson, D. R.; Ashbaugh, L. L. (1994) Temporal and
22 spatial variations of PM2 5 and PM10 aerosol in the Southern California Air Quality Study. Atmos.
23 Environ. 28: 2061-2080.
24
25 Chow, J. C.; Watson, J. G.; Houck, J. E.; Pritchett, L. C.; Rogers, C. F.; Frazier, C. A.; Egami, R. T.; Ball,
26 B. M. (1994) A laboratory resuspension chamber to measure fugitive dust size distributions and chemical
27 compositions. Atmos. Environ. 28: 3463-3481.
28
29 Collins, J. G. (1988) Prevalence of selected chronic conditions, United States, 1983-85. Hyattsville, MD: U.S.
30 Department of Health and Human Services, National Center for Health Statistics; DHHS publication no.
31 (PHS) 88-1250. (Advance data from vital and health statistics no. 155).
32
33 Commins, B. T.; Waller, R. E. (1967) Observations from a ten-year-study of pollution at a site in the city of
34 London. Atmos. Environ. 1: 49-68.
35
36 Cooper, R. L.; Goldman, J. M.; Harbin, T. J., eds. (1991) Aging & environmental toxicology: biological &
37 behavioral perspectives. Baltimore, MD: Johns Hopkins University Press. (Series in environmental
38 toxicology).
39
40 Cristofalo, V. J.; Gerhard, G. S.; Pignolo, R. J. (1994) Molecular biology of aging. Surg. Clin. North Am.
41 74: 1-21.
42
43 Damokosh, A. I.; Spengler, J. D.; Dockery, D. W.; Ware, J. H.; Speizer, F. E. (1993) Effects of acidic
44 particles on respiratory symptoms in 7 US communities. Am. Rev. Respir. Dis. 147: A632.
45
46 Dassen, W.; Brunekreef, B.; Hoek, G.; Hofschreuder, P.; Staatsen, B.; De Groot, H.; Schouten, E.; Biersteker,
47 K. (1986) Decline in children's pulmonary function during an air pollution episode. J. Air Pollut. Control
48 Assoc. 36: 1223-1227.
49
50 Denny, F. W.; Clyde, W. A. (1986) Acute lower respiratory tract infections in nonhospitalized children. J.
51 Pediatr. (St. Louis) 108: 635-646.
52
53 Dice, J. F. (1993) Cellular and molecular mechanisms of aging. Physiol. Rev. 73: 149-159.
54
April 1995 13.99 DRAFT-DO NOT QUOTE OR CITE
-------
1 Dockery, D. W.; Pope, C. A., III. (1994) Air pollution and mortality: the authors reply [letter]. N. Engl. J.
2 Med. 330: 1238.
3
4 Dockery, D. W.; Pope, C. A., III. (1994) Acute respiratory effects of paniculate air pollution. Annu. Rev.
5 Public Health 15: 107-132.
6
7 Dockery, D. W.; Ware, J. H.; Ferris, B. G., Jr.; Speizer, F. E.; Cook, N. R.; Herman, S. M. (1982) Change
8 in pulmonary function in children associated with air pollution episodes. J. Air Pollut. Control Assoc.
9 32: 937-942.
10
11 Dockery, D. W.; Speizer, F. E.; Ferris, B. G., Jr.; Ware, J. H.; Louis, T. A.; Spiro, A., III. (1988)
12 Cumulative and reversible effects of lifetime smoking on simple tests of lung function in adults. Am.
13 Rev. Respir. Dis. 137: 286-292.
14
15 Dockery, D. W.; Speizer, F. E.; Stram, D. 0.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr. (1989) Effects
16 of inhalable particles on respiratory health of children. Am. Rev. Respir. Dis. 139: 587-594.
17
18 Dockery, D. W.; Schwartz, J.; Spengler, J. D. (1992) Air pollution and daily mortality: associations with
19 particulates and acid aerosols. Environ. Res. 59: 362-373.
20
21 Dockery, D. W.; Damokosh, A. I.; Neas, L. M.; Raizenne, M.; Spengler, J. D.; Koutrakis, P.; Ware, J. H.;
22 Speizer, F. E. (1993) Health effects of acid aerosols on North American children: respiratory symptoms
23 and illness. Am. Rev. Respir. Dis. 147: A633.
24
25 Dockery, D. W.; Pope, C. A., Ill; Xu, X.; Spengler, J. D.; Ware, J. H.; Fay, M. E.; Ferris, B. G., Jr.;
26 Speizer, F. E. (1993) An association between air pollution and mortality in six U.S. cities. N. Engl. J.
27 Med. 329: 1753-1759.
28
29 Dodd, J. A.; Ondov, J. M.; Tuncel, G.; Dzubay, T. G.; Stevens, R. K. (1991) Multimodal size spectra of
30 submicrometer particles bearing various elements in rural air. Environ. Sci. Technol. 25: 890-903.
31
32 Driscoll, K. E.; Maurer, J. K. (1991) Cytokine and growth factor release by alveolar macrophages: potential
33 biomarkers of pulmonary toxicity. Toxicol. Pathol. 19: 398-405.
34
35 Dusseldorf, A.; Kruize, H.; Brunekreef, B.; Hofschreuder, P.; de Meer, G.; van Oudvorst, A. B. (1994) Acute
36 effects of PM10 and airborne iron on respiratory health: a panel study among adults living near a steel
37 industry in the Netherlands. Am. J. Respir. Crit. Care Med.: submitted.
38
39 El-Fawal, H. A. N.; Schlesinger, R. B. (1994) Nonspecific airway hyperresponsiveness induced by inhalation
40 exposure to sulfuric acid aerosol: an in vitro assessment. Toxicol. Appl. Pharmacol. 125: 70-76.
41
42 Evans, J. S.; Tosteson, T.; Kinney, P. L. (1984a) Cross-sectional mortality studies and air pollution risk
43 assessment. Environ. Int. 10: 55-83.
44
45 Evans, J. S.; Kinney, P. L.; Koehler, J. L.; Cooper, D. W. (1984b) The relationship between cross-sectional
46 and time series studies. J. Air Pollut. Control Assoc. 34: 551-553.
47
48 Fairley, D. (1990) The relationship of daily mortality to suspended particulates in Santa Clara county, 1980-86.
49 Environ. Health Perspect. 89: 159-168.
50
51 Fine, J. M.; Gordon, T.; Sheppard, D. (1987a) The roles of pH and ionic species in sulfur dioxide- and sulfite-
52 induced bronchoconstriction. Am. Rev. Respir. Dis. 136: 1122-1126.
53
April 1995 13-100 DRAFT-DO NOT QUOTE OR CITE
-------
1 Fine, J. M.; Gordon, T.; Thompson, J. E.; Sheppard, D. (1987b) The role of titratable acidity in acid aerosol-
2 induced bronchoconstriction. Am. Rev. Respir. Dis. 135: 826-830.
3
4 Folkow, B.; Svanborg, A. (1993) Physiology of cardiovascular aging. Physiol. Rev. 725-764.
5
6 Frampton, M. W.; Voter, K. Z.; Morrow, P. E.; Roberts, N. J., Jr.; Culp, D. J.; Cox, C.; Utell, M. J. (1992)
7 Sulfuric acid aerosol exposure in humans assessed by bronchoalveolar lavage. Am. Rev. Respir. Dis.
8 146: 626-632.
9
10 Frampton, M. W.; Morrow, P. E.; Cox, C.; Levy, P. C.; Condemi, J. J.; Speers, D.; Gibb, F. R.; Utell, M. J.
11 (1995) Sulfuric acid aerosol followed by ozone exposure in healthy and asthmatic subjects. Environ. Res.:
12 in press.
13
14 Gearhart, J. M.; Schlesinger, R. B. (1986) Sulfuric acid-induced airway hyperresponsiveness. Fundam. Appl.
15 Toxicol. 7: 681-689.
16
17 Gearhart, J. M.; Schlesinger, R. B. (1988) Response of the tracheobronchial mucociliary clearance system to
18 repeated irritant exposure: effect of sulfuric acid mist on function and structure. Exp. Lung Res. 14: 587-
19 605.
20
21 Gergen, P. J.; Weiss, K. B. (1992) The increasing problem of asthma in the United States. Am. Rev. Respir.
22 Dis. 146: 823-824.
23
24 Glezen, W. P. (1989) Antecedents of chronic and recurrent lung disease: childhood respiratory trouble. Am.
25 Rev. Respir. Dis. 140: 873-874.
26
27 Gold, D. R.; Tager, I. B.; Weiss, S. T.; Tosteson, T. D.; Speizer, F. E. (1989) Acute lower respiratory illness
28 in childhood as a predictor of lung function and chronic respiratory symptoms. Am. Rev. Respir. Dis.
29 140: 877-884.
30
31 Henderson, R. F.; Barr, E. B.; Cheng, Y. S.; Griffith, W. C.; Hahn, F. F. (1992) The effect of exposure
32 pattern on the accumulation of particles and the response of the lung to inhaled particles. Fundam. Appl.
33 Toxicol. 19: 367-374.
34
35 Heyder, J.; Gebhart, J.; Rudolf, G.; Schiller, C. F.; Stahlhofen, W. (1986) Deposition of particles in the human
36 respiratory tract in the size range 0.005-15 ^m. J. Aerosol Sci. 17: 811-825.
37
38 Hoek, G.; Brunekreef, B. (1993) Acute effects of a winter air pollution episode on pulmonary function and
39 respiratory symptoms of children. Arch. Environ. Health 48: 328-335.
40
41 Ito, K.; Thurston, G. D.; Hayes, C.; Lippmann, M. (1993) Associations of London, England, daily mortality
42 with paniculate matter, sulfur dioxide, and acidic aerosol pollution. Arch. Environ. Health 48: 213-220.
43
44 Ito, K.; Kinney, P.; Thurston, G. D. (1995) Variations in PM10 concentrations within two metropolitan areas and
45 their implications to health effects analyses. Inhalation Toxicol.: accepted.
46
47 John, W.; Wall, S. M.; Ondo, J. L.; Winklmayr, W. (1990) Modes in the size distributions of atmospheric
48 inorganic aerosol. Atmos. Environ. Part A 24: 2349-2359.
49
50 Kenney, R. A., ed. (1989) Physiology of aging. St. Louis, MO: Mosby-Year Book, Inc.
51
52 Kinney, P. L.; Ito, K.; Thurston, G. D. (1995) A sensitivity analysis of mortality/PM10 associations in Los
53 Angeles. Inhalation Toxicol. 7: 59-69.
54
April 1995 13-101 DRAFT-DO NOT QUOTE OR CITE
-------
1 Koenig, J. Q.; Pierson, W. E.; Horike, M. (1983) The effects of inhaled sulfuric acid on pulmonary function in
2 adolescent asthmatics. Am. Rev. Respir. Dis. 128: 221-225.
3
4 Koenig, J. Q.; Pierson, W. E.; Horike, M.; Frank, R. (1983) A comparison of the pulmonary effects of 0.5 ppm
5 versus 1.0 ppm sulfur dioxide plus sodium chloride droplets in asthmatic adolescents. J. Toxicol.
6 Environ. Health 11: 129-139.
7
8 Koenig, J. Q.; Covert, D. S.; Pierson, W. E. (1989) Effects of inhalation of acidic compounds on pulmonary
9 function in allergic adolescent subjects. In: Symposium on the health effects of acid aerosols; October
10 1987; Research Triangle Park, NC. Environ. Health Perspect. 79: 173-178.
11
12 Koenig, J. Q.; Covert, D. S.; Larson, T. V.; Pierson, W. E. (1992) The effect of duration of exposure on
13 sulfuric acid-induced pulmonary function changes in asthmatic adolescent subjects: a dose-response study.
14 Toxicol. Ind. Health 8: 285-296.
15
16 Koenig, J. Q.; Dumler, K.; Rebolledo, V.; Williams, P. V.; Pierson, W. E. (1993) Respiratory effects of
17 inhaled sulfuric acid on senior asthmatics and nonasthmatics. Arch. Environ. Health 48: 171-175.
18
19 Koenig, J. Q.; Larson, T. V.; Hanley, Q. S.; Rebolledo, V.; Dumler, K.; Checkoway, H.; Wang, S.-Z.; Lin,
20 D.; Pierson, W. E. (1993) Pulmonary function changes in children associated with fine paniculate
21 matter. Environ. Res. 63: 26-38.
22
23 Kunst, A. E.; Looman, C. W. N.; Mackenbach, J. P. (1993) Outdoor air temperature and mortality in the
24 Netherlands: a time-series analysis. Am. J. Epidemiol. 137: 331-341.
25
26 Lakatta, E. G. (1993) Cardiovascular regulatory mechanisms in advanced age. Physiol. Rev. 73: 413-467.
27
28 Laube, B. L.; Bowes, S. M., Ill; Links, J. M.; Thomas, K. K.; Frank, R. (1993) Acute exposure to acid fog:
29 effects on mucociliary clearance. Am. Rev. Respir. Dis. 147: 1105-1111.
30
31 Lebowitz, M. D.; Holberg, C. J. (1987) Physiological changes in pulmonary development related to passive
32 smoking and its interaction with active smoking. Toxicol. Lett. 35: 101-105.
33
34 Li, Y.; Roth, H. D. (1995) Daily mortality analysis by using different regression models in Philadelphia County,
35 1973-1990. Inhalation Toxicol. 7: 45-58.
36
37 Linn, W. S.; Avol, E. L.; Anderson, K. R.; Shamoo, D. A.; Peng, R.-C.; Hackney, J. D. (1989) Effect of
38 droplet size on respiratory responses to inhaled sulfuric acid in normal and asthmatic volunteers. Am.
39 Rev. Respir. Dis. 140: 161-166.
40
41 Linn, W. S.; Shamoo, D. A.; Anderson, K. R.; Peng, R.-C.; Avol, E. L.; Hackney, J. D. (1994) Effects of
42 prolonged, repeated exposure to ozone, sulfuric acid, and their combination in healthy and asthmatic
43 volunteers. Am. J. Respir. Crit. Care Med. 150: 431-440.
44
45 Lipfert, F. W. (1993a) Community air pollution and mortality: analysis of 1980 data from US metropolitan areas.
46 I. Particulate air pollution. Upton, NY: U.S. Department of Energy, Brookhaven National Laboratory;
47 report no. BNL 48446-R.
48
49 Lipfert, F. W. (1993b) A critical review of studies of the association between demands for hospital services and
50 air pollution. Environ. Health Perspect. 101(suppl. 2): 229-268.
51
52 Lippmann, M. (1989) Progress, prospects, and research needs on the health effects of acid aerosols. In:
53 Symposium on the health effects of acid aerosols; October 1987; Research Triangle Park, NC. Environ.
54 Health Perspect. 79: 203-205.
April 1995 13-102 DRAFT-DO NOT QUOTE OR CITE
-------
1 Logan, W. P. D. (1953) Mortality in the London fog incident, 1952. Lancet (Feb. 14): 336-339.
2
3 Martin, A. E. (1964) Mortality and morbidity statistics and air pollution. Proc. R. Soc. Med. 57: 969-975.
4
5 Martin, A. E.; Bradley, W. H. (1960) Mortality, fog and atmospheric pollution: an investigation during the
6 winter of 1958-59. Mon. Bull. Minist. Health Public Health Lab. Serv. (GB) 19: 56-73.
7
8 Mauderly, J. L.; Bice, D. E.; Cheng, Y. S.; Gillett, N. A.; Griffith, W. C.; Henderson, R. F.; Pickrell, J. A.;
9 Wolff, R. K. (1990) Influence of preexisting pulmonary emphysema on susceptibility of rats to diesel
10 exhaust. Am. Rev. Respir. Dis. 141: 1333-1341.
11
12 Mazumdar, S.; Schimmel, H.; Higgins, I. T. T. (1982) Relation of daily mortality to air pollution: an analysis of
13 14 London winters, 1958/59-1971/72. Arch. Environ. Health 37: 213-220.
14
15 Milford, J. B.; Davidson, C. I. (1987) The sizes of paniculate sulfate and nitrate in the atmosphere—a review.
16 JAPCA 37: 125-134.
17
18 Morrow, P. E. (1992) Dust overloading of the lungs: update and appraisal. Toxicol. Appl. Pharmacol.
19 113: 1-12.
20
21 Morrow, P. E.; Utell, M. J.; Bauer, M. A.; Speers, D. M.; Gibb, F. R. (1994) Effects of near ambient levels
22 of sulphuric acid aerosol on lung function in exercising subjects with asthma and chronic obstructive
23 pulmonary disease. Ann. Occup. Hyg. 38(suppl. 1): 933-938..
24
25 National Institutes of Health. (1991) Guidelines for the diagnosis and management of asthma. Bethesda, MD:
26 U.S. Department of Health and Human Services, National Heart, Lung, and Blood Institute, National
27 Asthma Education Program; publication no. 91-3042.
28
29 Neas, L. M.; Dockery, D. W.; Ware, J. H.; Spengler, J. D.; Ferris, B. G., Jr.; Speizer, F. E. (1994)
30 Concentration of indoor paniculate matter as a determinant of respiratory health in children. Am. J.
31 Epidemiol. 139: 1088-1099.
32
33 Neas, L. M.; Dockery, D. W.; Koutrakis, P.; Tollerud, D. J.; Speizer, F. E. (1995) The association of ambient
34 air pollution with twice daily peak expiratory flow rate measurements in children. Am. J. Epidemiol.
35 141: 111-122.
36
37 Oberdorster et al. (1995) (Abstract). Am. J. Respir. Crit. Care Med.: in press.
38
39 Oberdorster et al. (1995) (Abstract). In: Advances in combustion toxicology: proceedings of an international
40 colloquium; Oklahoma City, OK.
41
42 Oberdorster, G.; Ferin, J.; Gelein, R.; Soderholm, S. C.; Finkelstein, J. (1992a) Role of the alveolar
43 macrophage in lung injury: studies with ultrafine particles. Environ. Health Perspect. 97: 193-199.
44
45 Oberdorster, G.; Ferin, J.; Morrow, P. E. (1992b) Volumetric loading of alveolar macrophages (AM): a possible
46 basis for diminished AM-mediated particle clearance. Exp. Lung Res. 18: 87-104.
47
48 Oberdorster, G.; Gelein, R.; Ferin, J.; Weiss, B. (1995) Association of paniculate air pollution and acute
49 mortality: involvement of ultrafine particles? In: Colloquium on paniculate air pollution and human
50 mortality and morbidity; January 1994; Irvine, CA. Inhalation Toxicol. 7: 111-124.
51
52 Ostro, B. (1984) A search for a threshold in the relationship of air pollution to mortality: a reanalysis of data on
53 London winters. Environ. Health Perspect. 58: 397-399.
54
April 1995 13_103 DRAFT-DO NOT QUOTE OR CITE
-------
1 Ostro, B. D.; Lipsett, M. J.; Wiener, M. B.; Seiner, J. C. (1991) Asthmatic responses to airborne acid aerosols.
2 Am. J. Public Health 81: 694-702.
3
4 Ostro, B. D.; Lipsett, M. J.; Mann, J. K.; Krupnick, A.; Harrington, W. (1993) Air pollution and respiratory
5 morbidity among adults in Southern California. Am. J. Epidemiol. 137: 691-700.
6
7 Ostro, B.; Sanchez, J. M.; Aranda, C.; Eskeland, G. S. (1995) Air pollution and mortality: results of a study of
8 Santiago, Chile. J. Exposure Anal. Environ. Epidemiol.: submitted.
9
10 Ozkaynak, H.; Thurston, G. D. (1987) Associations between 1980 U.S. mortality rates and alternative measures
11 of airborne particle concentration. Risk Anal. 7: 449-461.
12
13 Ozkaynak, H.; Xue, J.; Severance, P.; Burnett, R.; Raizenne, M. (1994) Associations between daily mortality,
14 ozone, and paniculate air pollution in Toronto, Canada. Presented at: Colloquim on particulate air
15 pollution and human mortality and morbidity; January; Irvine, CA. Irvine, CA: University of California
16 at Irvine, Air Pollution Health Effects Laboratory; report no. 94-02.
17
18 Pepelko, W. E.; Mattox, J. K.; Cohen, A. L. (1980) Toxicology of ammonium sulfate in the lung. Bull.
19 Environ. Contam. Toxicol. 24: 156-160.
20
21 Ponka', A. (1991) Asthma and low level air pollution in Helsinki. Arch. Environ. Health 46: 262-270.
22
23 Pope, C. A., III.; Dockery, D. W. (1992) Acute health effects of PMi0 pollution on symptomatic and
24 asymptomatic children. Am. Rev. Respir. Dis. 145: 1123-1128.
25
26 Pope, C. A., Ill; Kanner, R. E. (1993) Acute effects of PM10 pollution on pulmonary function of smokers with
27 mild to moderate chronic obstructive pulmonary disease. Am. Rev. Respir. Dis. 147: 1336-1340.
28
29 Pope, C. A., Ill; Dockery, D. W.; Spengler, J. D.; Raizenne, M. E. (1991) Respiratory health and PM,0
30 pollution: a daily time series analysis. Am. Rev. Respir. Dis. 144: 668-674.
31
32 Pope, C. A., Ill; Schwartz, J.; Ransom, M. R. (1992) Daily mortality and PM10 pollution in Utah valley. Arch.
33 Environ. Health 47: 211-217.
34
35 Pope, C. A., Ill; Dockery, D. W.; Schwartz, J. (1995) Review of epidemiological evidence of health effects of
36 particulate air pollution. Inhalation Toxicol. 7: 1-18.
37
38 Pope, C. A., Ill; Thun, M. J.; Namboodiri, M. M.; Dockery, D. W.; Evans, J. S.; Speizer, F. E.; Heath, C.
39 W., Jr. (1995) Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults.
40 Am. J. Respir. Crit. Care Med. 151: 669-674.
41
42 Quackenboss, J. J.; Krzyzanowski, M.; Lebowitz, M. D. (1991) Exposure assessment approaches to evaluate
43 respiratory health effects of particulate matter and nitrogen dioxide. J. Exposure Anal. Environ.
44 Epidemiol. 1: 83-107.
45
46 Raizenne, M.; Neas, L.; Damokosh, A.; Dockery, D.; Spengler, J.; Koutrakis, P.; Ware, J.; Speizer, F. (1993)
47 Health effects of acid aerosols on North American children: pulmonary function. Am. Rev. Respir. Dis.
48 147: A635.
49
50 Roemer, W.; Hoek, G.; Brunekreef, B. (1993) Effect of ambient winter air pollution on respiratory health of
51 children with chronic respiratory symptoms. Am. Rev. Respir. Dis. 147: 118-124.
52
53 Saldiva, P. H. N.; Bohm, G. M. (1994)
54
April 1995 13-104 DRAFT-DO NOT QUOTE OR CITE
-------
1 Samet, J. M.; Utell, M. J. (1990) The risk of nitrogen dioxide: what have we learned from epidemiological and
2 clinical studies? Toxicol. Ind. Health 6: 247-262.
3
4 Samet, J. M.; Tager, I. B.; Speizer, F. E. (1983) The relationship between respiratory illness in childhood and
5 chronic air-flow obstruction in adulthood. Am. Rev. Respir. Dis. 127: 508-523.
6
7 Schiller, C. F.; Gebhart, J.; Heyder, J.; Rudolf, G.; Stahlhofen, W. (1988) Deposition of monodisperse
8 insoluble aerosol particles in the 0.095 to 0.2 pm size range within the human respiratory tract. Ann.
9 Occup. Hyg. 32(suppl.): 41-49.
10
11 Schlesinger, R. B. (1988) Biological disposition of airborne particles: basic principles and application to vehicular
12 emissions. In: Air pollution, the automobile, and public health. Washington, DC: National Academy
13 Press; pp. 239-298.
14
15 Schlesinger, R. B.; Naumann, B. D.; Chen, L. C. (1983) Physiological and histological alterations in the
16 bronchial mucociliary clearance system of rabbits following intermittent oral or nasal inhalation of
17 sulfuric acid mist. J. Toxicol. Environ. Health 12: 441-465.
18
19 Schlesinger, R. B.; Gorczynski, J. E.; Dennison, J.; Richards, L.; Kinney, P. L.; Bosland, M. C. (1992) Long-
20 term intermittent exposure to sulfuric acid aerosol, ozone and their combination: alterations in
21 tracheobronchial mucociliary clearance and epithelial secretory cells. Exp. Lung. Res. 18: 505-534.
22
23 Schroeder, W. H.; Dobson, M.; Kane, D. M.; Johnson, N. D. (1987) Toxic trace elements associated with
24 airborne paniculate matter: a review. JAPCA 37: 1267-1285.
25
26 Schwartz, J. (1991) Paniculate air pollution and daily mortality in Detroit. Environ. Res. 56: 204-213.
27
28 Schwartz, J. (1991) The first author replies [letter re Fleisher and Nayeri (1991)]. Am. J. Epidemiol.
29 133: 632-633.
30
31 Schwartz, J. (1993) Paniculate air pollution and chronic respiratory disease. Environ. Res. 62: 7-13.
32
33 Schwartz, J. (1993) Air pollution and daily mortality in Birmingham, Alabama. Am. J. Epidemiol.
34 137: 1136-1147.
35
36 Schwartz, J. (1994a) Total suspended paniculate matter and daily mortality in Cincinnati, Ohio. Environ. Health
37 Perspect. 102: 186-189.
38
39 Schwartz, J. (1994b) Air pollution and daily mortality: a review and meta analysis. Environ. Res. 64: 36-52.
40
41 Schwartz, J. (1994c) What are people dying of on high air pollution days? Environ. Res. 64: 26-35.
42
43 Schwartz, J. (1994d) Air pollution and hospital admissions for the elderly in Detroit, Michigan. Am. J. Respir.
44 Crit. Care Med. 150: 648-655.
45
46 Schwartz, J. (1994e) Air pollution and hospital admissions for the elderly in Birmingham, Alabama. Am. J.
47 Epidemiol. 139: 589-598.
48
49 Schwartz, J. (1994f) PM10, ozone, and hospital admissions for the elderly in Minneapolis, MN. Arch. Environ.
50 Health 49: 366-374.
51
52 Schwartz, J. (1995) Short term fluctuations in air pollution and hospital admissions of the elderly for respiratory
53 disease. Thorax: in press.
54
April 1995 13-105 DRAFT-DO NOT QUOTE OR CITE
-------
1 Schwartz, J.; Dockery, D. W. (1992a) Increased mortality in Philadelphia associated with daily air pollution
2 concentrations. Am. Rev. Respir. Dis. 145: 600-604.
3
4 Schwartz, J.; Dockery, D. W. (1992b) Particulate air pollution and daily mortality in Steubenville, Ohio. Am. J.
5 Epidemiol. 135: 12-19.
6
7 Schwartz, J.; Marcus, A. (1990) Mortality and air pollution in London: a time series analysis. Am. J. Epidemiol.
8 131: 185-194.
9
10 Schwartz, J.; Morris, R. (1995) Air pollution and cardiovascular hospital admissions. Am. J. Epidemiol.: in
11 press.
12
13 Schwartz, J.; Spix, C.; Wichmann, H. E.; Malin, E. (1991) Air pollution and acute respiratory illness in five
14 German communities. Environ. Res. 56: 1-14.
15
16 Schwartz, J.; Wypij, D.; Dockery, D.; Ware, J.; Zeger, S.; Spengler, J.; Ferris, B., Jr. (1991) Daily diaries of
17 respiratory symptoms and air pollution: methodological issues and results. Environ. Health Perspect.
18 90: 181-187.
19
20 Schwartz, J.; Dockery, D. W.; Neas, L. M.; Wypij, D.; Ware, J. H.; Spengler, J. D.; Koutrakis, P.; Speizer,
21 F. E.; Ferris, B. G., Jr. (1994) Acute effects of summer air pollution on respiratory symptom reporting
22 in children. Am. J. Respir. Crit. Care Med. 150: 1234-1242.
23
24 Shumway, R. H.; Tai, R. Y.; Tai, L. P.; Pawitan, Y. (1983) Statistical analysis of daily London mortality and
25 associated weather and pollution effects. Sacramento, CA: California Air Resources Board; contract no.
26 Al-154-33.
27
28 Sorlie, P. D.; Rogot, E. (1990) Mortality by employment status in the National Longitudinal Mortality Study.
29 Am. J. Epidemiol. 132: 983-992.
30
31 Speizer, F. E. (1989) Studies of acid aerosols in six cities and in a new multi-city investigation: design issues.
32 Environ. Health Perspect. 79: 61-67.
33
34 Spengler, J. D.; Keeler, G. J.; Koutrakis, P.; Ryan, P. B.; Raizenne, M.; Franklin, C. A. (1989) Exposures to
35 acidic aerosols. In: Symposium on the health effects of acid aerosols; October 1987; Research Triangle
36 Park, NC. Environ. Health Perspect. 79: 43-51.
37
38 Spurny, K. R. (1993) Atmospharisches, Anthropogenes Aerosol und seine toxischen und krebserregenden
39 Komponenten [Atmospheric, anthropogenic aerosol and its toxic and carcinogenic components]. Wiss.
40 Umwelt(2): 139-151.
41
42 Stahlhofen, W.; Rudolf, G.; James, A. C. (1989) Intercomparison of experimental regional aerosol deposition
43 data. J. Aerosol Med. 2: 285-308.
44
45 Studnicka, M. J.; Frischer, T.; Meinert, R.; Studnicka-Benke, A.; Hajek, K.; Spengler, J. D.; Neumann, M. G.
46 (1995) Acidic particles and lung function in children: a summer camp study in the Austrian Alps. Am. J.
47 Respir. Crit. Care Med. 151: 423-430.
48
49 Styer, P.; McMillan, N.; Gao, F.; Davis, J.; Sacks, J. (1995) The effect of airborne paniculate matter on daily
50 death counts. Environ. Health Perspect.: accepted.
51
52 Sunyer, J.; Saez, M.; Murillo, C.; Castellsague, J.; Martinez, F.; Anto, J. M. (1993) Air pollution and
53 emergency room admissions for chronic obstructive pulmonary disease: a 5-year study. Am. J.
54 Epidemiol. 137: 701-705.
April 1995 13-106 DRAFT-DO NOT QUOTE OR CITE
-------
1 Sweet, C.; Vermette, S. (1993) Sources of toxic trace elements in urban air in Illinois. Environ. Sci. Technol.
2 27: 2502-2510.
3
4 Thurston, G. D.; Ito, K.; Lippmann, M.; Hayes, C. (1989) Mortality in relation to exposure to acidic aerosols
5 during 1963-1972 winters. Environ. Health Perspect. 79: 73-82.
6
7 Thurston, G. D.; Ito, K.; Lippmann, M.; Hayes, C. (1989) Reexamination of London, England, mortality in
8 relation to exposure to acidic aerosols during 1963-1972 winters. In: Symposium on the health effects of
9 acid aerosols; October 1987; Research Triangle Park, NC. Environ. Health Perspect. 79: 73-82.
10
11 Thurston, G. D.; Ito, K.; Kinney, P. L.; Lippmann, M. (1992) A multi-year study of air pollution and
12 respiratory hospital admissions in three New York State metropolitan areas: results for 1988 and 1989
13 summers. J. Exposure Anal. Environ. Epidemiol. 2: 429-450.
14
15 Thurston, G. D.; Gorczynski, J. E., Jr.; Currie, J. H.; He, D.; Ito, K.; Hipfner, J.; Waldman, J.; Lioy, P. J.;
16 Lippmann, M. (1994) The nature and origins of acid summer haze air pollution in metropolitan Toronto,
17 Ontario. Environ. Res. 65: 254-270.
18
19 Thurston, G. D.; Ito, K.; Hayes, C. G.; Bates, D. V.; Lippmann, M. (1994) Respiratory hospital admissions and
20 summertime haze air pollution in Toronto, Ontario: consideration of the role of acid aerosols. Environ.
21 Res. 65: 271-290.
22
23 Touloumi, G.; Pocock, S. J.; Katsouyanni, K.; Trichopoulos, D. (1994) Short-term effects of air pollution on
24 daily mortality in Athens: a time-series analysis. Int. J. Epidemiol. 23: 1-11.
25
26 U.S. Department of Health and Human Services. (1989) Reducing the health consequences of smoking: 25 years
27 of progress. A report of the Surgeon General. Rockville, MD: U.S. Department of Health and Human
28 Services, Centers for Disease Control; DHHS publication no. (CDC) 89-8411.
29
30 U.S. Department of Health and Human Services. (1990) Vital and health statistics: current estimates from the
31 National Health Interview Survey, 1989. Hyattsville, MD: Public Health Service, National Center for
32 Health Statistics; DHHS publication no. (PHS) 90-1504. (Series 10: data from the National Health Survey
33 no. 176).
34
35 U.S. Department of Health, Education, and Welfare. (1969) Air quality criteria for sulfur oxides. Washington,
36 DC: National Air Pollution Control Administration; report no. AP-50. Available from: NTIS,
37 Springfield, VA; PB-190252.
38
39 U.S. Environmental Protection Agency. (1982) Air quality criteria for paniculate matter and sulfur oxides.
40 Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
41 and Assessment Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield,
42 VA; PB84-156777.
43
44 U.S. Environmental Protection Agency. (1986) Second addendum to air quality criteria for paniculate matter and
45 sulfur oxides (1982): assessment of newly available health effects information. Research Triangle Park,
46 NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office;
47 EPA report no. EPA-600/8-86-020F. Available from: NTIS, Springfield, VA; PB87-176574.
48
49 U.S. Environmental Protection Agency. (1992) Respiratory health effects of passive smoking: lung cancer and
50 other disorders. Washington, DC: Office of Research and Development, Office of Health and
51 Environmental Assessment; EPA report no. EPA/600/6-90/006F. Available from: NTIS, Springfield,
52 VA; PB93-134419/XAB.
53
April 1995 13_107 DRAFT-DO NOT QUOTE OR CITE
-------
1 United Kingdom Ministry of Health. (1954) Mortality and morbidity during the London fog of December 1952.
2 London, United Kingdom: Her Majesty's Stationery Office. (Reports on public health and medical
3 subjects no. 95).
4
5 Utell, M. J.; Morrow, P. E.; Hyde, R. W. (1982) Comparison of normal and asthmatic subjects' responses to
6 sulphate pollutant aerosols. In: Walton, W. H., ed. Inhaled particles V: proceedings of an international
7 symposium organized by the British Occupational Hygiene Society; September 1980; Cardiff, Wales.
8 Ann. Occup. Hyg. 26: 691-697.
9
10 Ware, J. H.; Dockery, D. W.; Spiro, A., Ill; Speizer, F. E.; Ferris, B. G., Jr. (1984) Passive smoking, gas
11 cooking, and respiratory health of children living in six cities. Am. Rev. Respir. Dis. 129: 366-374.
12
13 Warheit, D. B.; Seidel, W. C.; Carakostas, M. C.; Hartsky, M. A. (1990) Attenuation of perfluoropolymer
14 fume pulmonary toxicity: effect of filters, combustion method, and aerosol age. Exp. Mol. Pathol.
15 52: 309-329.
16
17 Watson, J. G.; Chow, J. C.; Lowenthal, L. C.; Pritchett, C. A.; Frazier, C. A.; Neuroth, G. R.; Robbins, R.
18 (1994) Differences in the carbon compostion of source profiles for diesel- and gasoline-powered vehicles.
19 Atmos. Environ. 28: 2493-2505.
20
21 Watson, J. G.; Chow, J. C.; Lu, Z.; Fujita, E. M.; Lowenthal, D. H.; Lawson, D. R.; Ashbaugh, L. L. (1994)
22 Chemical mass balance source apportionment of PM10 during the Southern California Air Quality Study.
23 Aerosol Sci. Technol. 21: 1-36.
24
25 Watson, J. G.; Chow, J. C.; Lurmann, F. W.; Musarra, S. P. (1994) Ammonium nitrate, nitric acid, and
26 ammonia equilibrium in wintertime Phoenix, Arizona. Air Waste 44: 405-412.
27
28 Watson, J. G.; Chow, J. G.; Roth, P. M. (1994) Clear sky visibility as a challenge for society. Annu. Rev.
29 Energy Environ. 19: 241-266.
30
31 Wolff, G. T. (1984) On the nature of nitrate in coarse continental aerosols. Atmos. Environ. 18: 977-981.
32
33 Zelikoff, J. T.; Parsons, E.; Schlesinger, R. B. (1993) Inhalation of paniculate lead oxide disrupts pulmonary
34 macrophage-mediated functions important for host defense and tumor surveillance in the lung. Environ.
35 Res. 62: 207-222.
36
£ U.S. GOVERNMENT PRINTING OFFICE: 1995-652-452
April 1995 13-108 DRAFT-DO NOT QUOTE OR CITE
-------
|