Policy Assessment for the Review of the
Particulate Matter National Ambient Air
Quality Standards
Preliminary Draft
September 2009
-------
DISCLAIMER
This preliminary draft document has been prepared by staff from the Ambient Standards
Group, Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency.
Any opinions, findings, conclusions, or recommendations are those of the authors and do not
necessarily reflect the views of the EPA. This document is being circulated for informational
purposes and to facilitate discussion with the Clean Air Scientific Advisory Committee
(CAS AC) on the overall structure, areas of focus, and level of detail to be included in an external
review draft Policy Assessment, which EPA plans to release for CASAC review and public
comment later this year. Questions related to this preliminary draft document should be
addressed to Beth Hassett-Sipple, U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards, C504-06, Research Triangle Park, North Carolina 27711 (email: hassett-
sipple.beth@epa.gov).
-------
EPA 450/P-09-007
September 2009
Policy Assessment for the Review of the Paniculate
Matter National Ambient Air Quality Standards
Preliminary Draft
U.S. Environmental Protection Agency
Office of Air and Radiation
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Ambient Standards Group
Research Triangle Park, North Carolina 27711
-------
This page left intentionally blank.
-------
TABLE OF CONTENTS
1 INTRODUCTION 1-1
1.1 PURPOSE 1-1
1.2 BACKGROUND 1-3
1.2.1 Legislative Requirements 1-3
1.2.2 Overview of theNAAQS Review Process 1-5
1.2.3 History of PMNAAQS Reviews 1-6
1.2.4 Litigation Related to the 2006 PM Standards 1-14
1.2.4.1 Fine Particles 1-14
1.2.4.2 Thoracic Coarse Particles 1-16
1.2.5 Current PM NAAQS Review 1-17
1.3 GENERAL APPROACH AND ORGANIZATION OF THE DOCUMENT 1-19
1.4 REFERENCES 1-21
2 CHARACTERIZATION OF AMBIENT PM 2-1
2.1 INTRODUCTION 2-1
2.2 PROPERTIES OF AMBIENT PM 2-1
2.2.1 Particle Size Distribution 2-2
2.2.1.1 Modes 2-2
2.2.1.2 Sampler Cutpoints 2-4
2.2.2 Sources and Formation Processes 2-7
2.2.3 Chemical Composition 2-11
2.2.4 Fate and Transport 2-12
2.2.5 Optical Properties of Particles 2-13
2.2.6 Other Radiative Properties of Particles 2-14
2.3 AMBIENT PM MEASUREMENT METHODS 2-14
2.3.1 Particle Mass Measurement Methods 2-15
2.3.2 Indirect Optical Methods 2-18
2.3.3 Size-Differentiated Particle Number Concentration Measurement Methods 2-19
2.3.4 Chemical Composition Measurement Methods 2-19
2.3.5 Measurement Issues 2-20
2.4 AMBIENT PM MONITORING NETWORKS 2-21
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
-------
2.4.1 PM2.5 2-22
2.4.1.1 PM2.5FRM Network 2-23
2.4.1.2 PM2.5 Continuous Monitor Network 2-23
2.4.1.3 PM2.5 Chemical Speciation Network (CSN) 2-27
2.4.2 PMio 2-30
2.4.3 PMio-2.5 2-32
2.4.4 National Core (NCore) 2-32
2.5 PM CONCENTRATIONS, TRENDS, AND SPATIAL PATTERNS 2-35
2.5.1 PM2.5 2-35
2.5.2 PMio 2-40
2.5.3 PMio-2.5 2-43
2.5.4 Ultrafme Particles 2-44
2.5.5 PM Components 2-44
2.5.5.1 PM2.5 2-44
2.5.5.2 PMio-2.5 2-55
2.5.5.3 Ultrafme Particles 2-56
2.5.6 Relationships among PM2.5, PMio, and PMio-2.s 2-57
2.6 PM TEMPORAL PATTERNS 2-58
2.7 PM BACKGROUND LEVELS 2-66
2.8 RELATIONSHIP BETWEEN AMBIENT PM MEASUREMENTS AND HUMAN
EXPOSURE 2-72
2.8.1 Definitions 2-73
2.8.2 Centrally Monitored PM Concentration as a Surrogate for Particle Exposure 2-74
2.9 RELATIONSHIP BETWEEN AMBIENT PM AND VISIBILITY 2-78
2.9.1 Particle Mass and Light Extinction 2-80
2.9.2 Other Measures of Visibility 2-81
2.9.3 Visibility at PM Background Conditions 2-82
2.10 REFERENCES 2-85
3 POLICY-RELEVANT ASSESSMENT OF HEALTH EFFECTS EVIDENCE 3-1
3.1 INTRODUCTION 3-1
3.2 MECHANISMS 3-5
3.2.1 Dosimetry 3-6
3.2.2 Possible Pathways/Modes of Action 3-12
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
-------
3.3 NATURE OF EFFECTS 3-18
3.3.1 Causal Determinations 3-19
3.3.2 Fine Particles 3-23
3.3.2.1 Effects Associated with Short-term PM2.5 Exposures 3-23
3.3.2.2 Effects Associated with Long-term PM25 Exposures 3-40
3.3.2.3 PM2.5 Components and Sources 3-54
3.3.2.4 Ultrafme Particles 3-61
3.3.3 Thoracic Coarse Particles 3-64
3.3.3.1 Effects Associated with Short-term PMi0-2.5 Exposures 3-65
3.3.3.2 Effects Associated with Long-term PMio-2.5 Exposures 3-71
3.3.3.3 PMio-2.5 Components 3-72
3.4 ISSUES RELATED TO INTERPRETINGEPIDEMIOLOGIC EVIDENCE 3-73
3.4.1 Issues Related to Alternative Exposure Periods in Epidemiological Studies 3-73
3.4.1.1 Lag Structure in Short-term Exposure Studies 3-73
3.4.1.2 Seasonal and Regional Differences in Time-Series Epidemiological Results 3-76
3.4.1.3 Health Effects Related to Different Short-term Exposure Time Periods 3-77
3.4.1.4 Exposure Periods Used in Prospective Cohort Studies 3-78
3.4.2 Concentration-Response Relationships and Potential Thresholds 3-81
3.4.3 Air Quality Data in Epidemiologic Studies 3-82
3.4.4 Measurement Error and Exposure Error 3-83
3.4.5 Alternative Model Specifications 3-86
3.4.5.1 Time-series epidemiologic studies 3-86
3.4.5.2 Prospective cohort epidemiologic studies 3-87
3.4.6 Co-pollutant Confounding and Effect Modification 3-88
3.4.6.1 Co-pollutant Confounding 3-89
3.4.6.2 Effect Modification 3-91
3.5 PM-RELATED IMPACTS ON PUBLIC HEALTH 3-92
3.5.1 Potentially At-Risk Populations 3-93
3.5.1.1 Older Adults 3-94
3.5.1.2 Children 3-95
3.5.1.3 Pregnancy and Developmental Effects 3-96
3.5.1.4 Gender 3-96
3.5.1.5 Race/Ethnicity 3-97
3.5.1.6 Gene-Environment Interaction 3-98
3.5.1.7 Pre-Existing Disease 3-99
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite iii
-------
3.5.1.8 Socioeconomic Status 3-105
3.5.2 Potential Public Health Impacts 3-106
3.6 SUMMARY AND CONCLUSIONS 3-109
3.7 REFERENCES 3-110
CHARACTERIZATION OF HEALTH RISKS 4-1
4.1 INTRODUCTION 4-1
4.2 SCOPE 4-1
4.2.1 Key design elements 4-1
4.2.2 Alternative sets of standards included in the risk assessment 4-1
4.3 METHODS USED IN THE URBAN CASE STUDY ANALYSIS 4-1
4.3.1 General approach 4-1
4.3.2 Air quality inputs 4-2
4.3.3 Selection of model inputs 4-2
4.3.4 Addressing uncertainty and variability 4-2
4.4 METHODS USED IN THE REPRESENTATIVENESS ANALYSIS 4-2
4.5 METHODS USED IN THE NATIONAL-SCALE PM-RELATED MORTALITY
ANALYSIS 4-2
4.6 RESULTS 4-2
4.6.1 Core risk estimates 4-2
4.6.1.1 Recent conditions 4-2
4.6.1.2 Current set of standards 4-2
4.6.1.3 Alternative sets of standards 4-2
4.6.2 Sensitivity analysis results 4-3
4.6.2.1 Key sources of uncertainty 4-3
4.6.2.2 Additional set of reasonable risk estimates 4-3
4.6.3 Representativeness analysis 4-3
4.6.4 National-scale PM2.5-related mortality analysis 4-3
STAFF CONCLUSIONS ON PRIMARY PM NAAQS 5-1
5.1 INTRODUCTION 5-1
5.2 APPROACH 5-1
5.3 FINE PARTICLE STANDARDS 5-1
5.3.1 Adequacy of Current PM25 Standards 5-1
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
-------
5.3.1.1 Evidence-based Considerations 5-1
5.3.1.2 Risk-based Considerations 5-1
5.3.1.3 Summary 5-1
5.3.2 Indicators 5-1
5.3.3 Averaging Times 5-1
5.3.4 Alternative PM2.5 Standards to Address Health Effects Related to Long-term Exposure
5-1
5.3.4.1 Evidence-based Considerations 5-1
5.3.4.2 Risk-based Considerations 5-1
5.3.4.3 Summary 5-1
5.3.5 Alternative PM2 5 Standards to Address Health Effects Related to Short-term Exposure
5-1
5.3.5.1 Evidence-based Considerations 5-1
5.3.5.2 Risk-based Considerations 5-1
5.3.5.3 Summary 5-1
5.3.6 Alternative Forms for Annual and 24-hour PM2.5 Standards 5-1
5.3.6.1 Form of Annual Standard 5-1
5.3.6.2 Form of 24-Hour Standard 5-1
5.3.7 Summary of Staff Conclusions on Primary PM2.5NAAQS 5-1
5.4 THORACIC COARSE PARTICLE STANDARD 5-2
5.4.1 Adequacy of Current PMio Standards 5-2
5.4.2 Indicators 5-2
5.4.3 Averaging Time 5-2
5.4.4 Alternative Standards to Address Health Effects Related to Short-term Exposure 5-2
5.4.5 Summary of Staff Conclusions 5-2
5.5 SUMMARY OF KEY UNCERTAINTIES AND RESEARCH RECOMMENDATIONS
RELATED TO SETTING PRIMARY PM STANDARDS 5-2
5.6 REFERENCES 5-2
6 POLICY-RELEVANT ASSESSMENT OF WELFARE EFFECTS EVIDENCE 6-1
6.1 INTRODUCTION 6-1
6.2 EFFECTS ON VISIBILITY 6-2
6.2.1 2006 PM NAAQS Review 6-2
6.2.2 Scope of Current Review 6-3
6.2.3 Nature of Urban Focused PM-related Visibility Effects 6-5
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite v
-------
6.2.4 Public Welfare Impacts of Urban Visibility Impairment 6-6
6.2.5 Reanalysis of Public Perception Studies 6-7
6.2.6 Considerations in Selecting Alternative Light Extinction Scenarios for Assessment6-12
6.2.7 Characterization of Current Urban Visibility Conditions 6-13
6.2.8 Projected Light Extinction When Meeting Alternative NAAQS Scenarios 6-16
6.2.9 Summary 6-23
6.3 EFFECTS ON CLIMATE 6-25
6.3.1 Aerosol Direct Effects 6-26
6.3.2 Aerosol Indirect Effects 6-27
6.3.3 Summary and Conclusions 6-27
6.4 EFFECTS ON VEGETATION AND ECOSYSTEMS 6-29
6.4.1 Major Ecosystem Stressors inPM 6-30
6.4.2 Ecosystem Services 6-31
6.4.3 Effects of Deposited PM on Ecosystem Receptors 6-33
6.4.3.1 Plants 6-33
6.4.3.2 Soil and Nutrient Cycling 6-35
6.4.3.3 Wildlife 6-36
6.4.3.4 Water 6-37
6.4.3.5 Ecosystem and Regional Responses 6-37
6.4.4 Summary and Conclusions 6-38
6.5 EFFECTS ON MATERIALS 6-39
6.5.1 Materials Damage Effects 6-40
6.5.2 Soiling Effects 6-41
6.5.3 Summary and Conclusions 6-42
6.6 REFERENCES 6-43
7 STAFF CONCLUSIONS ON SECONDARY PM NAAQS 7-1
7.1 INTRODUCTION 7-1
7.2 APPROACH 7-1
7.3 STANDARDS TO ADDRESS VISIBILITY IMPAIRMENT 7-1
7.3.1 Adequacy of Current PM2.s Standards 7-1
7.3.2 Indicators 7-1
7.3.3 Averaging Times 7-1
7.3.4 Alternative Standards to Address Visibility Impairment 7-1
7.3.5 Summary of Staff Conclusions 7-1
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite vi
-------
7.4 STANDARDS TO ADDRESS OTHER PM-RELATED WELFARE EFFECTS 7-1
7.4.1 Climate Change 7-1
7.4.2 Vegetation and Ecosystems 7-1
7.4.3 Materials 7-1
7.4.4 Summary of Staff Conclusions 7-1
7.5 SUMMARY OF KEY UNCERTAINTIES AND RESEARCH RECOMMENDATIONS
RELATED TO SETTING SECONDARY STANDARDS 7-1
7.6 REFERENCES 7-1
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite vii
-------
1 LIST OF TABLES
2 Table 1-1. Summary of National Ambient Air Quality Standards Promulgated for Particulate
3 Matter 1971-2006 1-8
4 Table 2-1. Particle Size Fraction Terminology 2-5
5 Table 2-2. Characteristics of ambient fine (ultrafme plus accumulation-mode) and coarse
6 particles. (US EPA, 2009a, Table 3-1) 2-6
7 Table 2-3. Constituents of atmospheric particles and their major sources 2-10
8 Table 2-4. 2005-2007 Average PRB Daylight PM Light Extinction 2-84
9 Table 3-1. Summary of Causality Determinations by PM Size Fraction 3-22
10 Table 3-2. Percent of the U.S. population with respiratory diseases, cardiovascular diseases, and
11 diabetes 3-108
12 Table 6-1. PM2.5 design values and percent of days with maximum light extinction exceeding
13 three candidate protection levels (CPLs) for 15 Urban study areas 6-13
14 Table 6-2. Light extinction design values for "just meeting" secondary NAAQS scenarios based
15 on PM2.5 mass 6-19
16 Table 6-3. Percentage reductions required in the non-PRB light extinction or PM2.5 mass to "just
17 meet" the NAAQS scenarios 6-20
18 Table 6-4. Percentage of days with a maximum 1-hour daylight light extinction above the three
19 CPL for each of the eight NAAQS scenario cases 6-22
20
21 LIST OF FIGURES
22 Figure 2-1. Particle size distributions by number and volume 2-3
23 Figure 2-2. PM2.5 FRMs Reporting to AQS 2-25
24 Figure 2-3. PM2.5 Continuous Monitors Reporting to AQS 2-26
25 Figure 2-4. IMPROVE and Chemical Speciation Networks 2-29
26 Figure 2-5. PMio Monitoring Network 2-31
27 Figure 2-6. Proposed NCore Monitoring Network 2-34
28 Figure 2-7. Annual average and 24-hour (98th percentile 24-hour concentrations) PM2 5
29 concentrations in ug/m3, 2005, 2006, and 2007 2-37
30 Figure 2-8. Ambient annual PM2.5 concentrations in the U.S., 1999-2007, 2-38
31 Figure 2-9. Ambient 24-hour PM2.5 concentrations in the U.S., 1999-2007, 2-39
32 Figure 2-10. 24-hour PMio concentrations in ug/m3, 2005- 2007 2-41
33 Figure 2-11. Ambient 24-hour PMio concentrations in the U.S., 1988-2007, 2-42
34 Figure 2-12. Three-year average 24-hour PM2.5 OC concentrations measured at CSN sites across
35 the U.S.,2005-2007 2-47
36 Figure 2-13 Three-year average 24-hour PM2 5 EC concentrations measured at CSN sites across
37 the U.S.,2005-2007 2-48
38 Figure 2-14. Three-year average 24-hour PM2.5 SC>42concentrations measured at CSN sites
39 across the U.S., 2005-2007 2-49
40 Figure 2-15 Three-year average 24-hour PM2 5 NOs concentrations measured at CSN sites
41 across the U.S., 2005-2007 2-50
42 Figure 2-16 Three-year average 24-hour PM2.5 NH4+ concentrations measured at CSN sites
43 across the U.S., 2005-2007 2-51
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite viii
-------
1 Figure 2-17. Three-year average PM2.5 speciation estimates for 2005-2007 derived using the
2 SANDWICH method 2-53
3 Figure 2-18. Seasonally-stratified three-year average PM2.5 speciation estimates for
4 2005-2007 derived using the SANDWICH method 2-54
5 Figure 2-19. Diel plot generated from hourly FRM-like PM2.5 data 2-59
6 Figure 2-20. Diel plots generated from hourly FEM PMio data 2-60
7 Figure 2-21. Relationship of annual average and 98th percentile 24-hour PM2.5 concentrations at
8 the site and year level, all sites with complete enough data for calculation of valid
9 2005-2007 design values 2-63
10 Figure 2-22 Relationship of annual average and 98th percentile 24-hour PM2.5 3-year design
11 values, only for sites with the highest annual design value in their CBS A 2-64
12 Figure 2-23. Three-year history of 24-hour PM2.s concentrations at highest design values site in
13 four example CBS As 2-65
14 Figure 2-24. Graphical representation of the relationships between the three commonly used
15 metrics of VAQ 2-82
16 Figure 6-1. Summary of results of urban visibility studies in four North American cities,
17 showing the identified range of the 50% acceptance criteria 6-10
18 Figure 6-2. Distributions of estimated daylight 1-hour light extinction (top) and maximum daily
19 daylight 1-hour light extinction (bottom) in each study area 6-15
20 Figure 6-3. Light extinction box and whisker plots by time of day for each of the 15 urban areas.
21 6-16
22 Figure 6-4. Distributions of daily maximum daylight 1-hour total light extinction under two
23 "just meeting" secondary NAAQS scenarios 6-18
24 Figure 6-5. Millennium ecosystem assessment categorization of ecosystem 6-32
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite ix
-------
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
A/C
AAC
ACS
ABI
AERMOD
AHR
AHRQ
AHS
AHSMOG
Al
ANS
APEX
APHEA
AQCD
AQI
AQS
AR4
ARIES
As
ASOS
BAD
BAM
BASE
BC
BC
BenMAP
BMI
BNP
BP
BS
LIST OF ACRONYMS/ABBREVIATIONS
Air conditioning
Abdominal Aortic Calcium
American Cancer Society
Ankle-brachial index
American Meteorological Society/EPA Regulatory Model
Improvement Committee (AERMIC) air quality dispersion
modeling system
Airway Hyperesponsiveness
Agency for Healthcare Research and Quality
American Housing Survey
California Seventh Day Adventist Study
Aluminum
Autonomic nervous system
EPA's Air Pollutants Exposure model, version 4
Air Pollution and Health: a European Approach
Air Quality Criteria Document
Air Quality Index
EPA's Air Quality System
Fourth Assessment Report of the Intergovernmental Panel on Climate
Change
Aerosol Research and Inhalation Epidemiological Study
Arsenic
Automated Surface Observing System
Bronchial Artery Diameter
Beta Attenuation Mass Monitor
Building Assessment Survey and Evaluation
Black carbon
British Columbia
Benefits Mapping Analysis Program
Body Mass Index
B-type natriuretic peptide/Brain natriuretic peptide
Blood pressure
Black smoke
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
-------
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
C
Ca
CAA
CAC
CAIR
CAMx
CAP
CARB
CASAC
CBS A
CBVD
CCN
CCSP
Cd
CDPHE
CEM
CHAD
CHD
CHF
CHS
CIMT
CMAQ
CNS
CO
CONUS
COPD
COV
CPL
Cr
C-R
CRA
CRP
CSA
CSN
Carbon
Calcium
Clean Air Act
Coronary Artery Calcium
Clean Air Interstate Rule
Comprehensive Air quality Model
Concentrated Ambient Particles
California Air Resources Board
Clean Air Scientific Advisory Committee
Consolidated Business Statistical Area
Cerebrovascular disease
Cloud Condensation Nuclei
US Climate Change Science Program
Cadmium
Colorado Department of Public Health and Environment
Continuous Emission Monitoring
EPA's Consolidated Human Activity Database
Coronary Heart disease
Congestive heart failure
Childrens Health Study
Carotid intima-media thickness
Community Multiscale Air Quality
Central Nervous System
Carbon Monoxide
CMAQ simulations covering continental US
Chronic obstructive pulmonary disease
Coefficient of Variation
Candidate Protection Level
Chromium
Concentration-response relationship
Charles River Associates
C-reactive protein
Consolidated Statistical Area
Chemical Speciation Network
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
XI
-------
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
CTM
CTS
Cu
cv
CVD
DBF
DE
DNA
ORE
dv
DVT
EC
ECG
ED
EPA
ETV
eVNA
FEM
FEVi
FIP
FMD
FRM
GAMs
GEOS
GHG
GI
GLMs
GSH
GST
HA
HCUP
HEAPSS
HEI
Chemical transport models
CMAQ model run
Copper
Cardiovascular
Cardiovascular disease
Diastolic Blood Pressure
Diesel Exhaust
Deoxyribonucleic Acid
Direct Radiative Effects
deciview
Deep Vein Thrombosis
Elemental Carbon
Electrocardiogram
Emergency department
Environmental Protection Agency
Environmental Technology Verification program
enhanced Voronoi Neighbor Averaging
Federal Equivalent Method
Change in forced expiratory volume in one second
Federal Implementation Plan
Flow-mediated vasodilation
Federal Reference Method
Generalized additive models
Global Scale Air Circulation Model
Greenhouse Gas
Group Interviews
Generalized linear models
Glutathione
Glutathione-S-transferase
Hospital admissions
Healthcare Cost and Utilization Project
Health Effects of Air Pollution Among Susceptible Subpopulations
study
Health Effects Institute
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
XII
-------
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
HF
Hg
HR
HRV
ICAM-1
ICD
ICD
ICR
IEC
IFG
IHD
IMPROVE
IPCC
IRP
ISA
ISCST
IT
Km
Lag
LagO
Lag 0-x
LCD
LF
LOESS
LAV
MAP
MCAPS
ME
MEA
MENTOR
MI
Mm
High Frequency
Mercury
Heart rate
Heart rate variability
Intercellular adhesion molecule-1
International Classification of Disease
Implantable/Implanted Cardioverter defibrillator
Information Collection Request
Industrial Economics, Incorporated
Investigative Focus Groups
Ischemic heart disease
Interagency Monitoring of Protected Visual Environment
Intergovernmental Panel on Climate Change
Integrated Review Plan
Integrated Science Assessment
EPA's Industrial Source Complex Short-Term model
Intratracheal
Kilometer
Time between one event and another
Same day as the death, test, hospital, ED, clinic, physician visit;
that occurs on the same day as the exposure to the pollutant(s)
All the deaths test, hospital, ED, clinic, physician visit; that occurs
on the same day as the exposure to the pollutant(s) and the x days
following the day of exposure
Liquid Crystal Display
Low Frequency
Locally weighted Scatter Plot Smoothing
Ratio of lumen to wall
Mean Arterial Pressure
Medicare Air Pollution Study
Microenvironment
Millennium Ecosystem Assessment
Modeling ENvironment for TOtal Risk
Myocardial infarction
Megameter
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
Xlll
-------
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
MOA
MSA
N
NAAQS
NAPS
NARSTO
NCDC
NCEA
NCore
NEI
NEM
NERL
NHANES
Ni
NLCS
NMMAPS
NN intervals
NOAA
NO
NO2
NO3"
NOx
NFS
NRC
NWS
03
OAQPS
OAR
OC
OMB
OR
ORD
OS
Mode(s) or mechanism(s) of action
Metropolitan Statistical Area
Nitrogen
National Ambient Air Quality Standards
National Air Pollution Surveillance
North American Research Strategy for Tropospheric Ozone
National Climatic Data Center
National Center for Environmental Assessment
National Core Monitoring Network
National Emissions Inventory
NAAQS Exposure Model
National Exposure Research Laboratory
National Health and Nutrition Examination Survey
Nickel
Netherlands Cohort Study
National Morbidity, Mortality, and Air Pollution Study
Normal-to-normal (NN or RR, sinus) time interval between each
QRS complex in the EKG
National Oceanic and Atmospheric Administration
Nitric Oxide
Nitrogen dioxide
Nitrate
Nitrogen oxides
National Park Service
National Research Council
National Weather Service
Ozone
Office of Air Quality Planning and Standards
Office of Air and Radiation
Organic Carbon
Office of Management and Budget
Odds Ratio
Office of Research and Development
Observational Study
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
xiv
-------
1 PA Policy Assessment
2 PA Pulmonary arterial
3 PAH Polyaromatic Hydrocarbon
4 Pb Lead
5 PBDE Polybromiated diphenyl ether
6 PCA Principal Component Analysis
7 PEF Peak Expiratory Flow L/min
8 PM Particulate matter
9 PMio Particles with a 50% upper cut-point of 10± 0.5 um aerodynamic
10 diameter and a penetration curve as specified in the Code of
11 Federal Regulations.
12 PMio-2.5 Particles with a 50% upper cut-point of 10 um aerodynamic
13 diameter and a lower 50% cut-point of 2.5 um aerodynamic
14 diameter.
15
16 PM2.5 Particles with a 50% upper cut-point of 2.5 um aerodynamic
17 diameter and a penetration curve as specified in the Code of
18 Federal Regulations.
19
20 PMx The legal definition for PMx, as defined in the Code of Federal
21 Regulations, includes both a 50% cut-point and a penetration
22 curve. A 50% cut-point of X um diameter means that 50% of
23 particles with aerodynamic diameter of X are removed by the inlet
24 and 50% pass through the inlet and are collected on the filter.
25 Depending on the specific penetration curve specified, particles
26 larger than X um aerodynamic diameter are collected with an
27 efficiently than decreases rapidly for particles larger than X while
28 the collection efficiency for particles smaller than X increases
29 rapidly with decreasing size until 100 % efficiency is reached.
30 PMF Positive Matrix Factorization
31 POP Persistant Organic Pollutant
32 PRB Policy-Relevant Background
33 PSAS The French National Program on Air Pollution Health Effects
34 PT Prothrombin Time
35 PTT Partial Thomboplastin Time
36 QA Quality assurance
37 QT Time for depolarization and repolarization of the ventricles
38 REA Risk and Exposure Assessment
39 RF Radiative forcing
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite xv
-------
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
RH
RNS
rMSSD
ROFA
ROS
RR
RV
SAB
SANDWICH
SAP
SBP
SCAB
SD
SDNN
SEARCH
SEDD
SES
SH
SHEDS-PM
Si
SID
SMOKE
S
SO2
SO42"
sox
SOPHIA
STP
TB
TEARS
TEACH
TEOM
TNF-a
Relative humidity
Reactive Nitrogen Species
Square root of the mean squared successive NN interval
differences in the EKG
Residential Oil Fly Ash
Reactive Oxygen Species
Relative risk
Right Ventricular
Science Advisory Board
Sulfate, Adjusted Nitrate, Derived Water, Inferred Carbonaceous
mass approach
Synthesis and Assessment Product
Systolic Blood Pressure
South Coast Air Basin
Standard deviation
standard deviation normal-to-normal (NN or RR) time interval
between each QRS complex in the EKG
Southeastern Aerosol Research and Characterization Study
State Emergency Department Databases
Socioeconomic Status
Spontaneously Hypertensive
Stochastic Human Exposure and Dose Simulation model for PM
Silicon
State Inpatient Database
Sparse Matrix Operator Kernal Emissions
Sulfur
Sulfur Dioxide
Sulfate
Sulfur Oxides
Study of Particulates and Health in Atlanta
Standard Temperature and Pressure
Tracheobronchial
Thiobarbituric acid reactive substances
Toxicity and Exposure Assessment for Children's Health
Tapered Element Oscillating Microbalance
Tumor Necrosis Factor-Alpha
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
xvi
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
TSP
UACR
UBC
UFPs
UFVA
V
VAQ
VCAM-1
VOC
WACAP
WBC
WHI
Zn
Total suspended particulate
Urinary Albumin/Creatine Ratio
University of British Columbia
Ultrafine particles
Urban-Focused Visibility Impact Assessment
Vanadium
Visual Air Quality
Vascular cell adhesion molecule-1
Volatile organic compounds
Western Airborne Contaminants Assessment Project
White Blood Cell
Women's Health Initiative
Zinc
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
xvii
-------
1
2 This page intentionally left blank.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite xviii
-------
1 1 INTRODUCTION
2 1.1 PURPOSE
3 The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
4 the participate matter (PM) national ambient air quality standards (NAAQS). The EPA's overall
5 plan and schedule for this PM NAAQS review were presented in the Integrated Review Plan for
6 the National Ambient Air Quality Standards for Paniculate Matter (US EPA, 2008a). The
7 Integrated Review Plan (IRP) outlined the Clean Air Act (CAA) requirements related to the
8 establishment and reviews of the NAAQS, the process and schedule for conducting the current
9 PM NAAQS review, and the key components in the NAAQS review process: an Integrated
10 Science Assessment (ISA), Risk and Exposure Assessments (REAs, referred to as assessment
11 documents), and policy assessment/rulemaking. It also presented the key policy-relevant issues
12 to be addressed in this review as a series of questions that frames our approach to determining
13 whether the current primary (health-based) and secondary (welfare-based) NAAQS for PM
14 should be retained or revised.
15 At this time, three draft documents prepared as part of this PM NAAQS review have
16 recently been released for review by the public and EPA's Clean Air Scientific Advisory
17 Committee (CASAC) PM NAAQS Review Panel at a public meeting to be held on October 5-6,
18 2009. These documents include the Integrated Science Assessment for P articulate Matter:
19 Second External Review Draft (ISA, July 2009), prepared by EPA's National Center for
20 Environmental Assessment, Research Triangle Park, NC (NCEA-RTP), and two draft
21 assessment documents, prepared by EPA's Office of Air Quality Planning and Standards
22 (OAQPS), Risk Assessment to Support the Review of the PM Primary National Ambient Air
23 Quality Standards — External Review Draft (September 2009) and P articulate Matter Urban-
24 Focused Visibility Assessment - External Review Draft (September 2009).
25 In addition, OAQPS staff are preparing a Policy Assessment (PA) to help "bridge the
26 gap" between the relevant scientific information and assessments and the judgments required of
27 the EPA Administrator in determining whether it is appropriate to retain or revise the NAAQS
28 for PM. Preparation of a PA by OAQPS staff reflects Administrator Jackson's decision to
29 modify the NAAQS review process as presented in the IRP by reinstating the use of a policy
30 assessment document (referred to as a Staff Paper in past reviews) in lieu of issuing a policy
31 assessment in the form of an advance notice of proposed rulemaking (see
32 http://www.epa.gov/ttn/naaqs/review.html).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1 -1
-------
1 When complete, the PA will present factors relevant to EPA's review of the primary
2 (health-based) and secondary (welfare-based) PM NAAQS. It will consider both evidence- and
3 risk-based approaches in evaluating the adequacy of the current PM standards and in identifying
4 potential alternative standards for the Administrator's consideration. In preparing the PA,
5 OAQPS staff considers the available scientific evidence presented in the ISA and the quantitative
6 assessments presented in the REAs, together with related limitations and uncertainties, and
7 focuses on information that is most pertinent to evaluating the basic elements of air quality
8 standards: indicator1, averaging time, form2, and level. These elements, which together serve to
9 define each standard, must be considered collectively in evaluating the health and welfare
10 protection afforded by the PM standards.
11 We are releasing a preliminary draft PA at this time, prior to the upcoming CASAC
12 meeting, for informational purposes and to facilitate discussion with the CASAC PM Panel on
13 the overall structure, areas of focus, and level of detail to be included in an external review draft
14 PA. This preliminary draft document, which is clearly a work in progress, draws from the
15 information and conclusions presented in the second draft ISA (U.S. EPA, 2009a) and the two
16 draft REAs (U.S. EPA, 2009b,c).
17 Following this introductory chapter, this preliminary draft PA includes early draft
18 chapters addressing the following: a characterization of ambient PM (Chapter 2), an overview of
19 the policy-relevant health effects evidence (Chapter 3), and an overview of the policy-relevant
20 evidence of PM-related welfare effects and the assessment of PM-related visibility impairment
21 (Chapter 6). Chapter 4, which will present an overview of the health risk assessment, is
22 previewed by way of an annotated outline. Chapters 5 and 7, which will present staff
23 conclusions on a range of policy options for the Administrator to consider concerning whether,
24 and if so how, to revise the primary and secondary PM NAAQS, are previewed only by way of
25 the organization shown in the draft table of contents. These two chapters will discuss both
26 evidence- and risk-based approaches to informing consideration of the basic elements of each of
27 the PM NAAQS: indicator, averaging time, form, and level.
28 As noted above, this preliminary draft PA is a work in progress. For example, we are still
29 working to identify and potentially adapt figures from the ISA to sharpen our focus and
30 discussion of the health effects evidence presented in Chapter 3. Additionally, we are continuing
1 The "indicator" of a standard defines the chemical species or mixture that is to be measured in determining
whether an area attains the standard.
2 The "form" of a standard defines the air quality statistic that is to be compared to the level of the standard in
determining whether an area attains the standard.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-2
-------
1 our efforts to obtain relevant air quality data from epidemiologic study authors to aid in our
2 evidence-based approach to evaluating the current and potential alternative primary standards.
3 We are also continuing to consider how to focus our presentation of the scientific evidence and
4 quantitative analyses in a manner that will most directly help to address the key policy-relevant
5 questions that frame this review. Staff will consider CAS AC review and public comments on the
6 second draft ISA and the two draft assessment documents, as well as CAS AC discussion of this
7 preliminary draft PA, in subsequently preparing a complete external review draft PA. We plan
8 to release a complete external review draft PA later this year for review by the CASAC PM
9 Panel and the public in early 2010.
10 While this preliminary draft PA should be of use to all parties interested in the PM
11 NAAQS review, it is written with an expectation that the reader has some familiarity with the
12 technical discussions contained in the second draft ISA (US EPA, 2009a) and two draft
13 assessment documents (US EPA, 2009b,c).
14 1.2 BACKGROUND
15 1.2.1 Legislative Requirements
16 Two sections of the Clean Air Act (Act) govern the establishment and revision of the
17 NAAQS. Section 108 (42 U.S.C. 7408) directs the Administrator to identify and list air
18 pollutants that "in his judgment, cause or contribute to air pollution which may reasonably be
19 anticipated to endanger public health and welfare" and whose "presence ... in the ambient air
20 results from numerous or diverse mobile or stationary sources" and to issue air quality criteria
21 for those that are listed. Air quality criteria are to "accurately reflect the latest scientific
22 knowledge useful in indicating the kind and extent of all identifiable effects on public health or
23 welfare which may be expected from the presence of [a] pollutant in ambient air . . ." 42 U.S.C.
24 § 7408(b).
25 Section 109 (42 U.S.C. 7409) directs the Administrator to propose and promulgate
26 "primary" and "secondary" NAAQS for pollutants for which air quality criteria are issued.
27 Section 109(b)(l) defines a primary standard as one "the attainment and maintenance of which in
28 the judgment of the Administrator, based on [air quality] criteria and allowing an adequate
29 margin of safety, are requisite to protect the public health."3 A secondary standard, as defined in
3 The legislative history of section 109 indicates that a primary standard is to be set at "the maximum permissible
ambient air level.. . which will protect the health of any [sensitive] group of the population," and that for this
purpose "reference should be made to a representative sample of persons comprising the sensitive group rather than
to a single person in such a group." S. Rep. No. 91-1196, 91st Cong., 2d Sess. 10 (1970).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-3
-------
1 Section 109(b)(2), must "specify a level of air quality the attainment and maintenance of which,
2 in the judgment of the Administrator, based on such [air quality] criteria, is requisite to protect
3 the public welfare from any known or anticipated adverse effects associated with the presence of
4 [the] pollutant in the ambient air."4
5 The requirement that primary standards include an adequate margin of safety was
6 intended to address uncertainties associated with inconclusive scientific and technical
7 information available at the time of standard setting. It was also intended to provide a reasonable
8 degree of protection against hazards that research has not yet identified. Lead Industries
9 Association v. EPA, 647F.2d 1130, 1154 (D.C. Cir 1980), cert, denied, 449 U.S. 1042 (1980);
10 American Petroleum Institute v. Costle, 665 F.2d 1176, 1186 (D.C. Cir. 1981), cert, denied, 455
11 U.S. 1034 (1982); American Farm Bureau Federation v. EPA, 559 F. 3d 512, 533 (D.C. Cir.
12 2009). Both kinds of uncertainties are components of the risk associated with pollution at levels
13 below those at which human health effects can be said to occur with reasonable scientific
14 certainty. Thus, in selecting primary standards that include an adequate margin of safety, the
15 Administrator is seeking not only to prevent pollution levels that have been demonstrated to be
16 harmful but also to prevent lower pollutant levels that may pose an unacceptable risk of harm,
17 even if the risk is not precisely identified as to nature or degree.
18 In selecting a margin of safety, EPA considers such factors as the nature and severity of
19 the health effects involved, the size of the sensitive population(s) at risk, and the kind and degree
20 of the uncertainties that must be addressed. The selection of any particular approach to
21 providing an adequate margin of safety is a policy choice left specifically to the Administrator's
22 judgment. Lead Industries Association v. EPA, supra, 647 F.2d at 1161-62.
23 In setting standards that are "requisite" to protect public health and welfare, as provided
24 in section 109(b), EPA's task is to establish standards that are neither more nor less stringent
25 than necessary for these purposes. In so doing, EPA may not consider the costs of implementing
26 the standards. See generally Whitman v. American Trucking Associations, 531 U.S. 457, 471,
27 475-76 (2001).
28 Section 109(d) (1) of the Act requires that "not later than December 31, 1980, and at 5-
29 year intervals thereafter, the Administrator shall complete a thorough review of the criteria
30 published under section 108 and the national ambient air quality standards . . . and shall make
4 Welfare effects as defined in section 302(h) (42 U.S.C. 7602(h)) include, but are not limited to, "effects on soils,
water, crops, vegetation, man-made materials, animals, wildlife, weather, visibility and climate, damage to and
deterioration of property, and hazards to transportation, as well as effects on economic values and on personal
comfort and well-being."
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-4
-------
1 such revisions in such criteria and standards and promulgate such new standards as may be
2 appropriate . ..." 42 U.S.C. § 7409(d)(l). Section 109(d)(2) requires that an independent
3 scientific review committee "shall complete a review of the criteria . . . and the national primary
4 and secondary ambient air quality standards . . . and shall recommend to the Administrator any
5 new . . . standards and revisions of existing criteria and standards as may be appropriate . . ." 42
6 U.S.C. § 7409(d)(2). Since the early 1980's, this independent review function has been
7 performed by CAS AC.
8 1.2.2 Overview of the NAAQS Review Process
9 Since completion of the last PM NAAQS review, the Agency has made a number of
10 changes to the process for reviewing the NAAQS.5 In making these changes, the Agency
11 consulted with CASAC, which provides advice to the Administrator on key elements of NAAQS
12 reviews, and the public. This revised process contains four major components: planning,
13 science assessment, risk/exposure assessment, and policy assessment/rulemaking. Each of these
14 components is described in this section.
15 The planning phase of the review process begins with a "kick-off workshop early in the
16 planning phase to get input from CASAC, internal and external experts, and the public regarding
17 policy-relevant science issues that have emerged since the last review. The workshop
18 discussions help inform the preparation of an IRP jointly by NCEA and OAQPS staff. A draft
19 IRP is presented for consultation with CASAC and for public comment. A final IRP reflects
20 CASAC and public comments together with early guidance from Agency management. The IRP
21 includes the science-policy questions that will frame the review, an outline of the process and
22 schedule that the entire review will follow, and more complete descriptions of the purpose,
23 contents, and approach for developing each of the key documents in the review.
24 The science assessment phase involves the preparation of an ISA by NCEA-RTP staff.
25 The ISA provides a concise evaluation and integration of the policy-relevant science, including
26 key science judgments that are important to inform the design and scope of the risk and exposure
27 assessments. The ISA and its supporting annexes provide a comprehensive assessment of the
28 current scientific literature pertaining to known and anticipated effects on public health and
29 welfare associated with the presence of the pollutant in the ambient air, emphasizing information
30 that has become available since the last review. The process generally includes production of a
31 first and second draft ISA, both of which undergo CASAC and public review prior to completion
32 of the final IS A.
5 See http://www.epa.gov/ttn/naaqs/ for more information.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-5
-------
1 In the risk/exposure assessment phase, OAQPS staff draws upon information and
2 conclusions presented in the ISA to develop quantitative estimates of the risks/exposures for
3 health and/or welfare effects associated with current ambient levels of PM, with levels that just
4 meet the current standards, and with levels that just meet potential alternative standards. The
5 REAs provide a concise presentation of methods, key results, observations, and related
6 uncertainties. These assessments begin with the preparation of a planning document that
7 discusses the scope and methods planned for use in conducting the quantitative assessments.
8 Such Scope and Methods Plans are generally prepared in conjunction with the first draft ISA and
9 presented for consultation with CASAC and for public comment. Comments received on the
10 Scope and Methods Plan are considered in preparing a draft REA that undergoes CASAC and
11 public review, generally in conjunction with review of the second draft ISA, prior to completion
12 of a final RE A.
13 The review process ends with a policy assessment/rulemaking phase. Under recent
14 revisions to NAAQS review process, the EPA Administrator has reinstated the use of a Policy
15 Assessment (PA), which is, like the previous Staff Paper, a document that provides a transparent
16 staff analysis of the scientific basis for alternative policy options for consideration by the
17 Administrator prior to the issuance of proposed and final rules (Jackson, 2009). The PA
18 integrates and interprets the information from the ISA and REA to frame policy options for
19 consideration by the Administrator. A draft PA is released for CASAC review and public
20 comment prior to completion of the final PA. It is intended to facilitate CASAC's advice and
21 recommendations to the Administrator on any new standards or revisions to existing standards as
22 may be appropriate, as provided for in the CAA. Following issuance of the final PA, the Agency
23 publishes a proposed rule, followed by a public comment period. Taking into account comments
24 received on the proposed rule, the Agency issues a final rule to complete the rulemaking process.
25 1.2.3 History of PM NAAQS Reviews
26 Particulate matter is the generic term for a broad class of chemically and physically
27 diverse substances that exist as discrete particles (liquid droplets or solids) over a wide range of
28 sizes. Particles originate from a variety of anthropogenic stationary and mobile sources as well
29 as from natural sources. Particles may be emitted directly or formed in the atmosphere by
30 transformations of gaseous emissions such as sulfur oxides (SOX), nitrogen oxides (NOX), and
31 volatile organic compounds (VOC). The chemical and physical properties of PM vary greatly
32 with time, region, meteorology, and source category, thus complicating the assessment of health
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-6
-------
1 and welfare effects. Table 1-1 summarizes the NAAQS that have been promulgated for PM to
2 date. These reviews are briefly described below.
3 The EPA first established NAAQS for PM in 1971, based on the original air quality
4 criteria document (DHEW, 1969). The reference method specified for determining attainment of
5 the original standards was the high-volume sampler, which collects PM up to a nominal size of
6 25 to 45 micrometers (|im) (referred to as total suspended particles or TSP). The primary
7 standards (measured by the indicator TSP) were 260 |ig/m3, 24-hour average, not to be exceeded
8 more than once per year, and 75 |ig/m3, annual geometric mean. The secondary standard was
9 150 |ig/m3, 24-hour average, not to be exceeded more than once per year.
10 In October 1979 (44 FR 56731), EPA announced the first periodic review of the criteria
11 and NAAQS for PM, and significant revisions to the original standards were promulgated in
12 1987 (52 FR 24634, July 1, 1987). In that decision, EPA changed the indicator for PM from TSP
13 to PMio, the latter including particles with a mean aerodynamic diameter6 less than or equal to
14 10 jim, which delineates that subset of inhalable particles small enough to penetrate to the
15 thoracic region (including the tracheobronchial and alveolar regions) of the respiratory tract
16 (referred to as thoracic particles). The EPA also revised the level and form of the primary
17 standards by: (1) replacing the 24-hour TSP standard with a 24-hour PMio standard of 150
18 |ig/m3 with no more than one expected exceedance per year; and (2) replacing the annual TSP
19 standard with a PMio standard of 50 |ig/m3, annual arithmetic mean. The secondary standard
20 was revised by replacing it with 24-hour and annual standards identical in all respects to the
21 primary standards. The revisions also included a new reference method for the measurement of
22 PMio in the ambient air and rules for determining attainment of the new standards. On judicial
23 review, the revised standards were upheld in all respects. Natural Resources Defense Council v.
24 Administrator, 902 F. 2d 962 (D.C. Cir. 1990), cert, denied, 498 U.S. 1082 (1991).
6 The more precise term is 50 percent cutpoint or 50 percent diameter (D50). This is the aerodynamic particle
diameter for which the efficiency of particle collection is 50 percent. Larger particles are not excluded altogether,
but are collected with substantially decreasing efficiency and smaller particles are collected with increasing (up to
100 percent) efficiency.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-7
-------
1 Table 1-1. Summary of National Ambient Air Quality Standards Promulgated for
2 Particulate Matter 1971-20067
Final Rule
1971
(36 FR 8186)
1987
(52 FR 24634)
1997
(62 FR 38652)
2006
(71 FR 61144)
Indicator
TSP - Total
Suspended
Particles
( < 25-45 urn)
PM10
PM25
PM10
PM25
PM10
Ave. Time
24-hour
Annual
24-hour
Annual
24-hour
Annual
24-hour
Annual
24-hour
Annual
24-hour
Level
260 ug/m3
(primary)
150 ug/m3
(secondary)
75 ug/m3
(primary)
150 ug/m3
50 ug/m3
65 ug/m3
15 ug/m3
150 ug/m3
50 ug/m3
35 ug/m
15 ug/m3
150 ug/m3
Form
Not to be exceeded more than once
per year
Annual average
Mot to be exceeded more than once
per year on average over a 3-year
period
Annual arithmetic mean, averaged
over 3 years
98th percentile, averaged over 3
years
Annual arithmetic mean, averaged
over 3 years8
Initially promulgated 99th
percentile, averaged over 3 years;
when 1 997 standards were vacated,
the form of 1987 standards
remained in place (not to be
exceeded more than once per year
on average over a 3-year period)
Annual arithmetic mean, averaged
over 3 years
98th percentile, averaged over 3
years
Annual arithmetic mean, averaged
over 3 years9
Mot to be exceeded more than once
per year on average over a 3-year
period
7 When not specified, primary and secondary standards are identical.
8 The level of the 1997 annual PM25 standard was to be compared to measurements made at the community-oriented
monitoring site recording the highest level, or, if specific constraints were met, measurements from multiple
community-oriented monitoring sites could be averaged ("spatial averaging"). These criteria and constraints were
intended to ensure that spatial averaging would not result in inequities in the level of protection afforded by the
PM2 5 standards. Community-oriented monitoring sites were specified to be consistent with the intent that a spatially
averaged annual standard provide protection for persons living in smaller communities, as well as those in larger
population centers.
9 In the revisions to the PM NAAQS finalized in 2006, EPA tighten the constraints on the spatial averaging criteria
by further limiting the conditions under which some areas may average measurements from multiple community-
oriented monitors to determine compliance (see 71 FR 61165-61167, October 17, 2006).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
1-8
-------
1 In April 1994, EPA announced its plans for the second periodic review of the criteria and
2 NAAQS for PM, and promulgated significant revisions to the NAAQS in 1997 (62 FR 38652,
3 July 18, 1997). In that decision, EPA revised the PM NAAQS in several respects. Most
4 significantly, EPA determined that the fine and coarse fractions of PMio should be considered
5 separately. The Administrator's decision to modify the standards was based on evidence that
6 serious health effects were associated with short- and long-term exposure to fine particles in
7 areas that met the existing PMio standards. The EPA accordingly added new standards, using
8 PM2.5, as the indicator for fine particles, and retained PMio as the indicator for regulating coarse
9 particles (referred to as thoracic coarse particles or coarse-fraction particles; generally including
10 particles with a nominal mean aerodynamic diameter greater than 2.5 jim and less than or equal
11 to 10 jim, or PMio-2.s). The EPA established two new PM2.5 standards: an annual standard of 15
12 Hg/m3, based on the 3-year average of annual arithmetic mean PM2.5 concentrations from single
13 or multiple community-oriented monitors; and a 24-hour standard of 65 |ig/m3, based on the 3-
14 year average of the 98th percentile of 24-hour PM2.5 concentrations at each population-oriented
15 monitor within an area. Also, EPA established a new reference method for the measurement of
16 PM2.5 in the ambient air and adopted protocols for determining attainment of the new standards.
17 To continue to address thoracic coarse particles, the annual PMio standard was retained, while
18 the 24-hour PMio standard was revised to be based on the 99th percentile of 24-hour PMio
19 concentrations at each monitor in an area. The EPA revised the secondary standards by making
20 them identical in all respects to the primary standards.
21 Following promulgation of the revised PM NAAQS in 1997, petitions for review were
22 filed by a large number of parties, addressing a broad range of issues. In May 1998, a three-
23 judge panel of the U.S. Court of Appeals for the District of Columbia Circuit issued an initial
24 decision that upheld EPA's decision to establish fine particle standards, holding that "the
25 growing empirical evidence demonstrating a relationship between fine particle pollution and
26 adverse health effects amply justifies establishment of new fine particle standards." American
27 Trucking Associations v. EPA , 175 F. 3d 1027, 1055-56 (D.C. Cir. 1999) (rehearing granted in
28 part and denied in part, 195 F. 3d 4 (D.C. Cir. 1999), affirmed in part and reversed in part,
29 Whitman v. American Trucking Associations., 531 U.S. 457 (2001). The panel also found "ample
30 support" for EPA's decision to regulate coarse particle pollution, but vacated the 1997 PMio
31 standards, concluding in part that PMio is a "poorly matched indicator for coarse particulate
32 pollution" because it includes fine particles. Id. at 1053-55. Pursuant to the court's decision,
33 EPA removed the vacated 1997 PMio standards from the Code of Federal Regulations (CFR) (69
34 FR 45592, July 30, 2004) and deleted the regulatory provision [at 40 CFR section 50.6(d)] that
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-9
-------
1 controlled the transition from the pre-existing 1987 PMio standards to the 1997 PMio standards.
2 The pre-existing 1987 PMio standards remained in place (65 FR 80776, December 22, 2000).
3 The Court also upheld EPA's determination not to establish more stringent secondary standards
4 for fine particles to address effects on visibility (175 F. 3d at 1027).
5 More generally, the panel held (with one dissenting opinion) that EPA's approach to
6 establishing the level of the standards in 1997, both for PM and for the ozone (O3) NAAQS
7 promulgated on the same day, effected "an unconstitutional delegation of legislative authority."
8 Id. at 1034-40. Although the panel stated that "the factors EPA uses in determining the degree of
9 public health concern associated with different levels of ozone and PM are reasonable," it
10 remanded the rule to EPA, stating that when EPA considers these factors for potential non-
11 threshold pollutants "what EPA lacks is any determinate criterion for drawing lines" to
12 determine where the standards should be set. Consistent with EPA's long-standing interpretation
13 and D.C. Circuit precedent, the panel also reaffirmed its prior holdings that in setting NAAQS
14 EPA is "not permitted to consider the cost of implementing those standards" Id. at 1040-41.
15 On EPA's petition for rehearing, the panel adhered to its position on these points.
16 American Trucking Associations v. EPA, 195 F. 3d 4 (D.C. Cir. 1999). The full Court of
17 Appeals denied EPA's request for rehearing en bane, with five judges dissenting. Id. at 13.
18 Both sides filed cross appeals on these issues to the United States Supreme Court, and the
19 Court granted certiorari. In February 2001, the Supreme Court issued a unanimous decision
20 upholding EPA's position on both the constitutional and cost issues. Whitman v. American
21 Trucking Associations, 531 U.S. 457, 464, 475-76. On the constitutional issue, the Court held
22 that the statutory requirement that NAAQS be "requisite" to protect public health with an
23 adequate margin of safety sufficiently guided EPA's discretion, affirming EPA's approach of
24 setting standards that are neither more nor less stringent than necessary. The Supreme Court
25 remanded the case to the Court of Appeals for resolution of any remaining issues that had not
26 been addressed in that court's earlier rulings. Id. at 475-76. In March 2002, the Court of
27 Appeals rejected all remaining challenges to the standards, holding under the traditional standard
28 of review that EPA's PM2.5 standards were reasonably supported by the administrative record
29 and were not "arbitrary and capricious." American Trucking Associations v. EPA, 283 F. 3d 355,
30 369-72 (D.C. Cir. 2002).
31 In October 1997, EPA published its plans for the third periodic review of the air quality
32 criteria and NAAQS for PM (62 FR 55201, October 23, 1997), including the 1997 PM2.5
33 standards and the 1987 PMio standards. After CAS AC and public review of several drafts, EPA's
34 NCEA finalized the Air Quality Criteria Document for Particulate Matter (henceforth, the
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-10
-------
1 "Criteria Document") in October 2004 (U.S. EPA, 2004) and EPA's OAQPS finalized a
2 technical support document, Particulate Matter Health Risk Assessment for Selected Urban
3 Areas (Abt, 2005), and a "Staff Paper," Review of the National Ambient Air Quality Standards
4 for Particulate Matter: Policy Assessment of Scientific and Technical Information, in December
5 2005 (U.S. EPA, 2005).
6 For the primary fine particle standards, most CASAC PM Panel members favored the
7 option of revising the level of the 24-hour PM2.5 standard in the range of 35 to 30 |ig/m3 with a
8 98th percentile form, in concert with revising the level of the annual PM2.5 standard in the range
9 of 14 to 13 |ig/m3 (Henderson, 2005a). Most of the members of the CASAC PM Panel also
10 strongly supported establishing a new, secondary PM2.s standard to protect urban visibility and
11 recommended establishing a sub-daily (4- to 8-hour averaging time) PM2.5 standard within the
12 range of 20 to 30 |ig/m3 with a form within the range of the 92nd to 98th percentile (Henderson,
13 2005a). For thoracic coarse particles, there was general concurrence among CASAC PM Panel
14 members to revise the PMio standards by establishing a primary standard specifically targeted to
15 address particles in the size range of 2.5 to 10 jim. The CASAC PM Panel was also in general
16 agreement "that coarse particles in urban or industrial areas are likely to be enriched by
17 anthropogenic pollutants that tend to be inherently more toxic than the windblown crustal
18 material which typically dominates coarse particle mass in arid rural areas." Based on its review
19 of the Staff Paper, there was general agreement among the CASAC PM Panel members that a
20 24-hour PMio-2.5 standard with a level in the range of 50 to 70 |ig/m3, with a 98th percentile form,
21 was reasonably justified and that a PMi0-2.5 standard with an annual averaging time was not
22 warranted (Henderson, 2005b).
23 On January 17, 2006, EPA proposed to revise the primary and secondary NAAQS for PM
24 (71 FR 2620) and solicited comment on a broader range of options. Proposed revisions included:
25 revising the level of the 24-hour PM2.s primary standard to 35 |ig/m3; revising the form, but not
26 the level, of the annual PM2.s primary standard by tightening the constraints on the use of spatial
27 averaging; replacing the 24-hour PMio primary standard with a 24-hour standard defined in
28 terms of a new indicator, PMio-2.5 10 set at a level of 70 |ig/m3; revoking the annual PMio primary
29 standard; and revising the secondary standards by making them identical in all respects to the
30 proposed suite of primary standards for fine and coarse particles. Subsequent to the proposal,
31 CASAC provided additional advice to EPA in a letter to the Administrator requesting
10 This proposed indicator was qualified so as to include any ambient mix of PM10.25 that is dominated by particles
generated by high-density traffic on paved roads, industrial sources, and construction sources, and to exclude any
ambient mix of particles dominated by rural windblown dust and soils and agricultural and mining sources.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-11
-------
1 reconsideration of CASAC's recommendations for both the primary and secondary PM2.5
2 standards as well as the standards for thoracic coarse particles (Henderson, 2006a).
3 On September 21, 2006, EPA announced its final decisions to revise the primary and
4 secondary NAAQS for PM to provide increased protection of public health and welfare,
5 respectively (71 FR 61144, October 17, 2006). With regard to the primary and secondary
6 standards for fine particles, EPA revised the level of the 24-hour PM2 5 standard to 35 |ig/m3,
7 retained the level of the annual PM2.5 annual standard at 15 |ig/m3, and revised the form of the
8 annual PM2 5 standard by narrowing the constraints on the optional use of spatial averaging. The
9 EPA revised the secondary standards for fine particles by making them identical in all respects to
10 the primary standards. With regard to the primary and secondary standards for thoracic coarse
11 particles, EPA retained PMio as the indicator for coarse particles, retained the level and form of
12 the 24-h PMio standard (so the standard remained at a level of 150 |ig/m3 with a one expected
13 exceedance form), and revoked the annual standard. The Agency concluded that dosimetric,
14 lexicological, occupational and epidemiologic evidence supported retention of a primary
15 standard that included all thoracic coarse particles (i.e. both urban and non-urban), consistent
16 with the Act's requirement that primary NAAQS provide an adequate margin of safety. EPA
17 further concluded that retention of PMio as an indicator for thoracic coarse particles was
18 reasonable, notwithstanding that PMio measures both fine and coarse particles. The Agency
19 reasoned that because fine particle levels are generally higher in urban areas, a PMio standard set
20 at a single unvarying level will generally result in less coarse particles being allowed in urban
21 areas than in non-urban areas. This was considered to be an appropriate targeting of protection
22 given that the strongest evidence of effects associated with thoracic coarse particles came from
23 epidemiologic studies conducted in urban areas. The Agency concluded that epidemiologic
24 evidence for both urban and non-urban coarse particles supported retention of the existing level
25 of the 24-h standard, 150 |ig/m3.
26 In the same rulemaking revising the PM NAAQS, EPA also included a new Federal
27 Reference Method (FRM) for the measurement of PMio-2.s in the ambient air (71 FR 61212 to
28 61213, October 17, 2006). Although the standards for thoracic coarse particles do not use a
29 PMio-2.5 indicator, the new FRM for PMi0.2.5 was established to provide a basis for approving
30 Federal Equivalent Methods (FEMs) and to promote the gathering of scientific data to support
31 future reviews of the PM NAAQS. With an FRM, researchers will likely include PMio-2.s
32 measurements of thoracic coarse particles in health studies either by directly using the FRM or
33 by utilizing approved FEMs.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-12
-------
1 In making final decisions for the PM2.5 NAAQS in 2006, the Administrator relied
2 primarily on evidence-based considerations to inform his conclusions on the levels for the 24-
3 hour and annual standards. The Administrator believed, at that time, that the estimates of risks
4 likely to remain upon attainment of the 1997 suite of PM2.5 standards were indicative of risks that
5 could be reasonably judged important from a public health perspective, and, thus, supported
6 revision of the standards. The quantitative risk assessment provided information supporting the
7 need to revise the PM2.5 standards. However, the Administrator judged that this assessment had
8 important limitations and did not provide an appropriate basis for selecting either the level of the
9 24-hour or annual PM2.5 standard. The Administrator more heavily weighed the implications of
10 the uncertainties associated with the quantitative risk assessment than CAS AC apparently did in
11 their comments on the proposed rulemaking. He also disagreed with CASAC and many public
12 commenters that the risk assessment results could appropriately serve as a primary basis for a
13 decision for the level of either the 24-hour or the annual PM2.5 standards.n
14 Following issuance of the final rule, CASAC articulated its concern that "EPA's final
15 rule on the NAAQS for PM does not reflect several important aspects of the CASAC's advice"
16 (Henderson et al, 2006b). With respect to the 24-hour primary PM2.5 standard, CASAC was
17 pleased with the Agency's decision to revise the level to 35 |ig/m3, which was within the range
18 recommended by most members of the CASAC PM Panel. With respect to the PM2.5 annual
19 standard, CASAC expressed serious concerns regarding the decision to retain the level of the
20 standard at 15 |ig/m3. Specifically, CASAC stated:
21
22 It is the CASAC's consensus scientific opinion that the decision to retain without change
23 the annual PM2.5 standard does not provide an "adequate margin of safety... requisite to
24 protect the public health" (as required by the Clean Air Act), leaving parts of the
25 population of this country at significant risk of adverse health effects from exposure to
26 fine PM... .to our knowledge there is no science, medical or public health group that
27 disagrees with this very important aspect of the CASAC's recommendations (Henderson
28 et. al, 2006b, p.2).
29
30 With regard to EPA's final decision to retain PMio as the indicator for coarse particles
31 and to retain the 24-hour standard at a level of 150 |ig/m3, CASAC acknowledged concerns
32 associated with retaining this standard while recognizing the need to have a standard in place to
33 protect against effects associated with short-term exposures to thoracic coarse particles.
34 Specifically, CASAC stated:
11 See discussion in Section II.F of the preamble to the final rule, 71 FR 61167-61177, October 17, 2006.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-13
-------
1
2 Furthermore, the CASAC was completely surprised at the decision in the final PM
3 NAAQS to revert to the use of PMio as the indicator for coarse particles. In our
4 September 15, 2005 letter, the CASAC recommended a new indicator of PMio-2.s, which
5 EPA put forward in its proposed rule for the PM NAAQS. The option of retaining the
6 existing daily PMio standard of 150 |ig/m3 was not discussed during the advisory process,
7 and in fact the CASAC views this as highly-problematic since PMio includes both fine
8 and coarse particulate matter. The Committee acknowledges the need for the Agency to
9 increase its understanding of the health risks of coarse particles and is concerned that
10 ongoing dependence on PMio sampling as an imprecise measure of coarse particulate
11 matter will provide inadequate information on coarse PM concentrations, compositions
12 and exposures in both urban and rural areas. However, the CASAC agrees that having a
13 standard for PMio is better than no standard at all for coarse particles, and was pleased
14 with the Agency's decision against offering exemptions to specific industries (i.e.,
15 agricultural, mining) in its regulation of coarse particles (Henderson et. al, 2006b, p.2).
16
17 With regard to EPA's final decision to revise the secondary PM2.s standards to be
18 identical in all respects to the revised primary PM2.5 standards, CASAC expressed concerns that
19 CASAC's advice to establish a distinct secondary standard for fine particles to address visibility
20 impairment was not followed. Specifically, CASAC stated:
21
22 In our June 6, 2005 letter, the CASAC affirmed the recommendation of Agency staff
23 regarding a separate secondary fine particle standard to protect visibility. This sub-daily
24 secondary PM2.5 standard is a better indicator of visibility impairment than the 24-hour
25 primary standard. The CASAC wishes to emphasize that continuing to rely on primary
26 standards to protect against all PM-related adverse environmental and welfare effects
27 assures neglect, and will allow substantial continued degradation, of visual air quality
28 over large areas of the country (Henderson et. al, 2006b, p.2).
29 1.2.4 Litigation Related to the 2006 PM Standards
30 Several parties filed petitions for review following promulgation of the revised PM
31 NAAQS in 2006. These petitions addressed the following issues: (1) selecting the level of the
32 annual primary PM2.s standard, (2) setting the secondary PM2.s standards identical to the primary
33 standards; (3) retaining PMio as the indicator for coarse particles and retaining the level and form
34 of the 24-hour PMio standard, and (4) revoking the PMio annual standard.
35 1.2.4.1 Fine Particles
36 On judicial review, the D.C. Circuit remanded the primary annual PM2.5 NAAQS to EPA
37 because the Agency failed to adequately explain why the standard provided the requisite
38 protection from both short- and long-term exposures to fine particles including protection for at
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-14
-------
1 risk populations. The court also remanded the secondary PM2.5 NAAQS to EPA because the
2 Agency failed to adequately explain why the standards provided the required protection from
3 visibility impairment. American Farm Bureau Federation v. EPA, 559F.3d512, (D.C. Cir.
4 2009).
5 With respect to human health protection from short-term PM2.5 exposures, the court
6 considered the different approaches used by EPA in the 1997 and 2006 PM NAAQS decisions.
7 In the!997 decision, EPA determined that the primary annual PM2.5 standard was the generally
8 controlling standard for lowering both short- and long-term PM2.5 ambient concentrations and the
9 24-hour standard was set to "provide an adequate margin of safety against infrequent or isolated
10 peak concentrations that could occur in areas that attain the annual standard" (62 FR 38676-77,
11 July 18, 1997). In the 2006 decision, the Administrator considered it appropriate to use a
12 different evidence-based approach from that used in 1997 to set the level of the 24-hour and
13 annual PM2.5 standards. In the 2006 decision, the Administrator relied upon evidence from the
14 short-term exposure studies as the principal basis for selecting the level of the 24-hour PM2.5
15 standard and relied upon evidence from the long-term exposure studies as the principal basis for
16 selecting the level of the annual PM2.5 standard. The court found EPA failed to adequately
17 explain this change in approach in light of CAS AC and staffs recommendations to do otherwise.
18 The court also found that EPA failed to adequately explain why a 24-hour PM2.5 standard by
19 itself would provide the protection needed from short-term exposures and remanded the annual
20 PM2.s standard to EPA "for further consideration of whether it is set at a level requisite to protect
21 the public health while providing an adequate margin of safety from the risk of short-term
22 exposures to PM2.5." American Farm Bureau Federation, 559 F. 3d at 520-24. .
23 With respect to protection from long-term exposure to fine particles, the court found that
24 EPA failed to adequately explain how the current primary annual PM2.5 standard provided an
25 adequate margin of safety in children and other vulnerable subpopulations. Specifically, the
26 court found that EPA did not provide a reasonable explanation of why certain studies, including
27 a study of children in Southern California showing lung damage associated with long-term PM2.5
28 exposure (Gauderman et.al, 2000) and a multi-city study (24-Cities Study) evaluating decreased
29 lung function in children associated with long-term PM2.5 exposures (Raizenne et al., 1996), did
30 not call for a more stringent annual PM2.5 standard. Id. at 522-23. Second, the court found that
31
32 "EPA was unreasonably confident that, even though it relied solely upon long-term
33 mortality studies, the revised standard would provide an adequate margin of safety with
34 respect to morbidity among children. Notably absent from the final rule, moreover, is
35 any indication of how the standard will adequately reduce risk to the elderly or to those
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-15
-------
1 with certain heart or lung diseases despite (a) the EPA's determination in its proposed
2 rule that those subpopulations are at greater risk from exposure to fine particles and (b)
3 the evidence in the record supporting that determination." American Farm Bureau
4 Federation. 559 F. 3d at 525-26.
5
6 The court also remanded the secondary standards for fine particles, based on EPA's
7 failure to adequately explain why setting the secondary PM NAAQS equivalent to the primary
8 standards provided the required protection for public welfare including protection from visibility
9 impairment. The court found that EPA failed to identify a target level of visibility impairment
10 that would be requisite to protect public welfare. The court determined that this was contrary to
11 the statute and resulted in a lack of a reasoned basis for the final decision. In addition, the court
12 found that EPA's near exclusive reliance on a comparison of numbers of counties that would be
13 in nonattainment under alternative standards scenarios was an inadequate basis for making a
14 decision. The court concluded that EPA did not take into account the relative visibility
15 protection of the alternative standards nor did the Agency consider the failure of a 24-hour PM2.5
16 standard to address regional differences in humidity and its effect on visibility. Id. at 528-31.
17 1.2.4.2 Thoracic Coarse Particles
18 The court upheld EPA's decision to retain the primary 24-hour PMio NAAQS to provide
19 protection for coarse particle exposures and to revoke the annual PMio standard. The court
20 found that EPA reasonably included all coarse PM within the standard, both of urban and non-
21 urban origin, to protect all of the country from exposure to coarse PM. The court rejected
22 arguments that the evidence showed there are no risks from exposure to non-urban coarse PM.
23 Id. at 531-33. The court further found that EPA had a reasonable basis to not set separate
24 standards for urban and non-urban coarse PM, namely the inability to reasonably define what
25 ambient mixes would be included under either 'urban' or 'non-urban' and that the evidence in
26 the record supported EPA's cautious decision to provide "some protection from exposure to
27 thoracic coarse particles... in all areas." Specifically, the court stated,
28
29 Although the evidence of danger from coarse PM is, as EPA recognizes, "inconclusive,"
30 (71 FR 61193, October 17, 2006), the agency need not wait for conclusive findings
31 before regulating a pollutant it reasonably believes may pose a significant risk to public
32 health. The evidence in the record supports the EPA's cautious decision that "some
33 protection from exposure to thoracic coarse particles is warranted in all areas." Id. As the
34 court has consistently reaffirmed, the CAA permits the Administrator to "err on the side
35 of caution" in setting NAAQS.
36
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-16
-------
1 559 F. 3d at 533 (block citation of cases omitted)
2
3 The court also upheld EPA's decision to use PMio as the indicator for coarse particles,
4 and to retain the level of the standard at 150 |ig/m3. The EPA's final rule acknowledged that
5 evidence of harm from urban-type coarse PM is stronger than for other types of coarse particles,
6 and targeted protection at areas where urban-type coarse PM is most likely present (71 FR
7 61185-61203, October 17, 2006). The targeting is done by using the indicator PMio for coarse
8 particles. PMio includes both coarse and fine particles. Urban and industrial areas tend to have
9 higher levels of fine particles than rural areas, so that in those areas less coarse particles are
10 allowed - the desired targeting. Conversely, fine particle levels tend to be lower in rural areas,
11 so more coarse particles are allowed in those areas - again the desired targeting. Likewise, the
12 court concluded that EPA's choice of the level for the PMio standard was reasonable, for many
13 of the same reasons. American Farm Bureau Federation at 533-36. The court also upheld
14 EPA's decision to revoke the annual PMio standard. Id. at 537-38.
15 1.2.5 Current PM NAAQS Review
16 The EPA initiated the current review of the air quality criteria for PM in June 2007 with a
17 general call for information (72 FR 35462, June 28, 2007). EPA invited a wide range of external
18 experts, as well as EPA staff, representing a variety of areas of expertise (e.g., epidemiology,
19 human and animal toxicology, statistics, risk/exposure analysis, atmospheric science, ecology,
20 biological sciences) to participate in two workshops: (1) Workshop to Discuss Policy-Relevant
21 Science to Inform EPA's Integrated Plan for the Review of the Primary PM NAAQS (conducted
22 July 11-13, 2007 in Research Triangle Park, NC) and (2) Workshop to Discuss Policy-Relevant
23 Science to Inform EPA's Integrated Plan for the Review of the Secondary PM NAAQS
24 (conducted July 16, 2007 in Chapel Hill, NC) (72 FR 34003 and 34005, June 20, 2007).12 These
25 workshops provided an opportunity for the participants to broadly discuss the key policy-relevant
26 issues around which EPA would structure the PM NAAQS review and to discuss the most
27 meaningful new science that would be available to inform our understanding of these issues.
28 Based in part on the workshop discussions, EPA developed a draft IRP outlining the
29 schedule, the process, and the key policy-relevant science issues that would guide the evaluation
30 of the air quality criteria for PM and the review of the primary and secondary PM NAAQS. On
31 November 30, 2007, EPA held a consultation with CAS AC on the draft IRP (72 FR 63177,
12 See workshop materials http://www.regulations.gov/search/Regs/home.htmMhome Docket ID numbers EPA-HQ-
OAR-2007-0492-008; EPA-HQ-OAR-2007-0492-009; EPA-HQ-OAR-2007-0492-010; and EPA-HQ-OAR-2007-
0492-012.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-17
-------
1 November 8, 2007). Public comments were also requested on the draft plan and presented at that
2 CASAC teleconference. The final IRP incorporated comments received from CASAC and the
3 general public on the draft plan as well as input from senior Agency managers (US EPA, 2008a).
4 As part of the process of preparing the PM ISA, NCEA hosted a peer review workshop in
5 June 2008 on preliminary drafts of key ISA chapters (73 FR 30391, May 27, 2008). The first
6 external review draft ISA (US EPA, 2008b) was reviewed by CASAC at a meeting held in April
7 2009 (74 FR 2688, February 19, 2009). Based on CASAC and public comments, NCEA revised
8 the draft ISA and released a second draft ISA (US EPA, 2009a) in July 2009 for CASAC review
9 and public comment at an upcoming public meeting to be held on October 5-6, 2009.
10 In preparing the REA documents that build on the scientific evidence presented in the
11 ISA, OAQPS released two planning documents, Paniculate Matter National Ambient Air
12 Quality Standards: Scope and Methods Plan for Health Risk and Exposure Assessment and
13 Particulate Matter National Ambient Air Quality Standards: Scope and Methods Plan for Urban
14 Visibility Impact Assessment (henceforth, Scope and Methods Plans)13 outlining the scope,
15 approaches, and key issues that staff planned to use in conducting the quantitative assessments
16 (US EPA, 2009d,e). We considered CASAC comments on the Scope and Methods Plans made
17 during an April 2009 consultation as well as public comments (74 FR 11580, March 18, 2009) in
18 designing and conducting the assessments. OAQPS released two draft assessment documents,
19 Risk Assessment to Support the Review of the PM2.s Primary National Ambient Air Quality
20 Standards: External Review Draft - September 2009 (US EPA 2009b) and Paniculate Matter
21 Urban-Focused Visibility Assessment - External Review Draft - September 2009 (US EPA,
22 2009c) for CASAC review and public comment at the upcoming public meeting on October 5-6,
23 2009.
24 This document, a preliminary draft PA, builds upon the information presented in the
25 second draft ISA and the two draft assessment documents. As noted above, this preliminary
26 draft PA is being released at this time, prior to the upcoming CASAC meeting, for informational
27 purposes and to facilitate discussion with the CASAC PM Panel at the upcoming meeting on the
28 overall structure, areas of focus, and level of detail to be included in an external review draft PA.
29 We will consider CASAC review and public comments on the second draft ISA and the two draft
30 assessment documents, as well as CASAC discussion of this preliminary draft PA, in
31 subsequently preparing a complete external review draft PA. We plan to release a complete
13 EPA-452/P-09-001 and -002; February 2009 ; Available:
http://www.epa.gov/ttn/naaqs/standards/pm/sjm 2007jd.html
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-18
-------
1 external review draft PA later this year for review by the CASAC PM Panel and the public in
2 early 2010.
3 1.3 GENERAL APPROACH AND ORGANIZATION OF THE DOCUMENT
4 The PA will include staffs evaluation of the policy implications of the scientific
5 assessment of the evidence presented in the ISA and results of quantitative assessments based on
6 that evidence, taking into consideration CASAC advice and public comments. Taken together,
7 this information will inform various conclusions and the identification of policy options for the
8 Administrator to consider in addressing public health and welfare effects associated with
9 exposure to ambient PM.
10 Partly as a consequence of EPA's decision in the last review to continue to consider fine
11 particles and thoracic coarse particles separately, much new information is now available on PM
12 air quality and human health effects directly in terms of PM2.5 and, to a much more limited
13 degree, PMi0.2.5. Since the purpose of this review is to evaluate the adequacy of the current
14 standards, which separately address fine and thoracic coarse particles, staff is focusing this
15 policy assessment and associated quantitative analyses primarily on the evidence related directly
16 to PM2.s and PMi0-2.5. In so doing, we are considering PMi0-related evidence primarily to help
17 inform our understanding of key issues and to help interpret and provide context for
18 understanding the public health and welfare impacts of ambient fine and coarse particles.
19 Following this introductory chapter, this preliminary draft PA is organized into three
20 main parts: the characterization of ambient PM (chapter 2), PM-related health effects and
21 primary PM NAAQS (chapters 3, 4 and 5), and PM-related welfare effects and secondary PM
22 NAAQS (chapters 6 and7). The characterization of ambient PM is presented in chapter 2, which
23 focuses on properties of ambient PM, measurement methods, spatial and temporal patterns in
24 ambient PM concentrations, PM background levels, and ambient PM relationships with human
25 exposure and with visibility impairment. Thus, chapter 2 provides information relevant to both
26 the health and welfare assessments in the other subsequent chapters of this document.
27 Chapters 3 through 5 comprise the second main part of this policy assessment document
28 dealing with human health and primary standards. Chapter 3 presents a policy-relevant
29 assessment of PM health effects evidence, including an overview of the evidence, key human
30 health-related conclusions from the second draft ISA, and an examination of issues related to the
31 quantitative assessment of health risks based on the epidemiologic health evidence. Chapter 4, to
32 be developed for inclusion in the external review draft, will present key results and observations
33 from a quantitative assessment of PM2.s-related health risks, including risk estimates for current
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-19
-------
1 air quality levels as well as those associated with just meeting the current PM2.5 NAAQS and
2 alternative standards under consideration. In this preliminary draft, chapter 4 is previewed by
3 way of an annotated outline. Chapter 5, to be developed for inclusion in the external review
4 draft, will present staff observations and conclusions related to the current primary standards for
5 fine and thoracic coarse particles. As outlined in the table of contents of this preliminary draft,
6 chapter 5 is planned to begin with a discussion of the broad approach used by staff in this review,
7 generally reflecting consideration of both evidence-based and quantitative risk-based
8 considerations. The discussion is then planned to focus on consideration of the adequacy of the
9 current PM standards, staff conclusions as to alternative indicators, averaging times, levels and
10 forms, and staff conclusions with regard to potential alternative primary standards for
11 consideration by the Administrator.
12 Chapters 6 and 7 comprise the third main part of this Staff Paper dealing with welfare
13 effects and secondary standards. Chapter 6 presents a policy-relevant assessment of PM welfare
14 effects evidence, including evidence related to visibility impairment as well as to effects on
15 vegetation and ecosystems, climate change processes, and man-made materials. The emphasis of
16 this chapter is on visibility impairment, reflecting the availability of a significant amount of
17 policy-relevant information and related assessments, which serve as the basis for staff
18 consideration of a distinct secondary standard specifically for protection of visual air quality.
19 Chapter 7, to be developed for inclusion in the external review draft, will present staff
20 observations and conclusions related to the current secondary standards. As outlined in the table
21 of contents of this preliminary draft, chapter 7 is planned to begin with a discussion of the
22 approach used by staff in this review. The discussion is planned to focus on consideration of the
23 adequacy of the current standards, staff conclusions as to alternative indicators, averaging times,
24 levels and forms, and staff conclusions with regard to potential standards for consideration by the
25 Administrator.
26 As noted above, we recognize that this preliminary draft PA is incomplete and that much
27 work needs to be done to prepare an external review draft PA. Nonetheless, we decided to
28 release this preliminary draft at this time both for informational purposes, to illustrate the most
29 recent change made to the NAAQS review process with regard to the policy assessment, and to
30 have the benefit of early discussion with the CASAC PM Panel on the overall structure, areas of
31 focus, and level of detail to be included in an external review draft PA.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-20
-------
1 1.4 REFERENCES
2 Abt Associates (2005). Paniculate Matter Health Risk Assessment for Selected Urban Areas. Final Report.
3 Bethesda, MD. Prepared for the Office of Air Quality Planning and Standards, U.S. Environmental
4 Protection Agency, Contract No. 68-D-03-002. EPA 452/R-05-007A. Available:
5 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_td.html.
6 Gauderman, W. I; McConnell, R.; Gilliland, F.; London, S.; Thomas, D.; Avol, E.; Vora, H.; Berhane, K.;
7 Rappaport, E. B.; Lurmann, F.; Margolis, H. G.; Peters, J. (2000). Association between air pollution and
8 lung function growth in southern California children. Am. J. Respir. Crit. CareMed. 162: 1383-1390.
9 Henderson, R. (2005a). EPA's Review of the National Ambient Air Quality Standards for Paniculate Matter
10 (Second Draft PM Staff Paper, January 2005): A review by the Paniculate Matter Review Panel of the EPA
11 Clean Air Scientific Advisory Committee. June 6, 2005. Available: http://www.epa.gov/sab/pdf/casac-05-
12 007.pdf.
13 Henderson, R. (2005b). Clean Air Scientific Advisory Committee (CASAC) Review of the EPA Staff
14 Recommendations Concerning a Potential Thoracic Coarse PM Standard in the Review of the National
15 Ambient Air Quality Standards for Particulate Matter: Policy Assessment of Scientific and Technical
16 Information (Final PM OAQPS Staff Paper, EPA-452/R-05-005). September 15, 2005. Available:
17 http://www.epa.gov/sab/panels/casacpmpanel.html.
18 Henderson, R. (2006a). Letter from Dr. Rogene Henderson, Chair, Clean Air Scientific Advisory Committee to the
19 Honorable Stephen L. Johnson, Administrator, US EPA. Clean Air Scientific Advisory Committee
20 Recommendations Concerning the Proposed National Ambient Air Quality Standards for Particulate
21 Matter. March 21, 2006. Available: http://www.epa.gov/sab/pdf/casac-ltr-06-002.pdf
22 Henderson, R.; Cowling, E.; Crapo, J.D..; Miller, F. J.; Poirot, R.L.; Speizer, F.; Zielinska, B. (2006b). Letter from
23 Clean Air Scientific Advisory Committee to the Honorable Stephen L. Johnson, Administrator, US EPA.
24 Clean Air Scientific Advisory Committee Recommendations Concerning the Final National Ambient Air
25 Quality Standards for Particulate Matter. September 29, 2006. Available:
26 http://www.epa.gov/sab/pdf/casac-ltr-06-002.pdf
27 Jackson (2009). Memo from Administrator Lisa P. Jackson to Elizabeth Craig, Acting Assistant Administrator for
28 OAR and Lek Kadeli, Acting Assistant Administrator for ORD. Process for Reviewing the National
29 Ambient Air Quality Standards. May 21, 2009. Available:
30 http://www.epa.gov/ttn/naaqs/pdfs/NAAQSReviewProcessMemo52109.pdf.
31 Raizenne, M.; Neas, L. M.; Damokosh, A. L; Dockery, D. W.; Spengler, J. D.; Koutrakis, P.; Ware, J. H.; Speizer, F.
32 E. (1996). Health effects of acid aerosols on North American children: pulmonary function. Environ.
33 HealthPerspect. 104: 506-514.
34 US Department of Health, Education and Welfare (DHEW). (1969). Air Quality Criteria for Particulate Matter.
35 U.S. Government Printing Office, Washington DC, AP-49.
36 US EPA (2004). Air Quality Criteria for Particulate Matter. National Center for Environmental Assessment, Office
37 of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711;
3 8 report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October 2004. Available:
39 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr cd.html
40 US EPA (2005). Review of the National Ambient Air Quality Standards for Particulate Matter: Policy Assessment
41 of Scientific and Technical Information, OAQPS Staff Paper. Research Triangle Park, NC 27711: Office
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-21
-------
1 of Air Quality Planning and Standards; report no. EPA EPA-452/R-05-005a. December 2005.
2 Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr sp.html
3 US EPA (2008a). Integrated Review Plan for the National Ambient Air Quality Standards fro Paniculate Matter.
4 National Center for Environmental Assessment and Office of Air Quality Planning and Standards, U.S.
5 Environmental Protection Agency, Research Triangle Park, NC. Report No. EPA 452/R-08-004. March
6 2008. Available at:
7 http://www.epa. gov/ttn/naaqs/standards/pm/data/2008_03_final_inte grated_review_plan.pdf
8 U.S. EPA (2008b). Integrated Science Assessment for Paniculate Matter: First External Review Draft. National
9 Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
10 Research Triangle Park, NC. EPA/600/R-08/139 and 139A. December 2008. Available:
11 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_isa.html.
12 US EPA (2009a). Integrated Science Assessment for Paniculate Matter: Second External Review Draft. National
13 Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
14 Research Triangle Park, NC. EPA/600/R-08/139B. July 2009. Available:
15 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007 isa.html.
16 US EPA (2009b). Risk Assessment to Support the Review of the PM Primary National Ambient Air Quality
17 Standards - External Review Draft. Office of Air Quality Planning and Standards, U.S. Environmental
18 Protection Agency, Research Triangle Park, NC. EPA-452/P-09/XXX. September 2009. Available:
19 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_index.html.
20 US EPA (2009c). Particulate Matter Urban-Focused Visibility Assessment - External Review Draft. Office of Air
21 Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC.
22 EPA-452/P-09/XXX. September 2009. Available:
23 http://www.epa.gov/ttn/naaqs/standards/prn/sjm index.html
24 US EPA (2009d). Particulate Matter National Ambient Air Quality Standards: Scope and Methods Plan for Health
25 Risk and Exposure Assessment. Office of Air Quality Planning and Standards, U.S. Environmental
26 Protection Agency, Research Triangle Park, NC. EPA-452/P-09-002. February 2009. Available:
27 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_pd.html.
28 US EPA (2009e). Particulate Matter National Ambient Air Quality Standards: Scope and Methods Plan for Urban
29 Visibility Impact Assessment. Office of Air Quality Planning and Standards, U.S. Environmental
30 Protection Agency, Research Triangle Park, NC. EPA-452/P-09-001. February 2009. Available:
31 http://www.epa.gOv/ttn/naaqs/standards/pm/s pm 2007 pd.html.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 1-22
-------
1 2 CHARACTERIZATION OF AMBIENT PM
2 2.1 INTRODUCTION
3 This chapter generally characterizes various classes of ambient PM in terms of physical
4 and chemical properties, measurement methods, recent concentrations and trends, and
5 relationships with human exposure and visibility impairment. This information is useful for
6 interpreting the available health and welfare effects information and for making
7 recommendations on appropriate indicators for primary and secondary PM standards. The
8 information presented in this chapter was drawn from the much more extensive treatment of
9 these subjects in the second draft Integrated Science Assessment (ISA, US EPA, 2009a), the
10 2004 Air Quality Criteria Document (AQCD, US EPA, 2004) and US EPA, 2005 (US EPA,
11 2005) from the review completed in 2006, and other sources. (To assist reviewers of this
12 preliminary draft, some paragraphs that have been copied or condensed from one of these
13 documents, but would not otherwise be easily identified as such, are denoted by a parenthetical
14 comments in italics at the end of the paragraph. In some cases, these comments alert the
15 reviewer that there has been augmentation of the original text for clarity or accuracy or to bring
16 the material up to date.)
17 Section 2.2 presents information on the basic physical and chemical properties of classes
18 of PM. Section 2.3 presents information on the methods used to measure ambient PM and some
19 important considerations in the design of these methods. Section 2.4 describes the design and
20 locations of ambient PM monitoring networks. Section 2.5 presents data on PM concentrations,
21 trends, and spatial patterns in the U.S. Section 2.6 provides information on the temporal
22 variability of PM. Much of the information in Sections 2.5 and 2.6 is derived from analyses of
23 data collected by the nationwide networks of PM2.5 and PMio monitors in 2005 through 2007.
24 Section 2.7 defines and discusses background levels of ambient PM. Section 2.8 addresses the
25 relationships between ambient PM levels and human exposure to PM. Section 2.9 addresses the
26 relationship between ambient PM2 5 levels and visibility impairment.
27 2.2 PROPERTIES OF AMBIENT PM
28 Particulate matter represents a broad class of chemically and physically diverse
29 substances that exist as discrete particles in the condensed (liquid or solid) phase. Particles can
30 be characterized by size, formation mechanism, origin, chemical composition, and atmospheric
31 behavior. Classes of particles have historically been characterized largely in terms of particle
32 size. Fine particles and coarse particles, which are defined below, are relatively distinct entities
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-1
-------
1 with fundamentally different sources and formation processes, chemical composition,
2 atmospheric residence times and behaviors, transport distances, and optical and radiative
3 properties. (USEPA, 2005, section 2.2)
4 2.2.1 Particle Size Distribution
5 2.2.1.1 Modes
6 The diameters of atmospheric particles span 5 orders of magnitude, ranging from
7 0.001 micrometers to 100 micrometers (|im).14 The size and associated composition of particles
8 determine their behavior in the respiratory system, including how far the particles are able to
9 penetrate, where they deposit, and how effective the body's clearance mechanisms are in
10 removing them. Furthermore, particle size is one of the most important parameters in
11 determining the residence time and spatial distribution of particles in ambient air, key
12 considerations in assessing exposure. Particle size is also a major determinant of visibility
13 impairment, a welfare effect linked to ambient particles. Particle surface area, number, chemical
14 composition, and water solubility all vary with particle size and are also influenced by the
15 formation processes and emissions sources. (US EPA, 2005, section 2.2.1)
16 Atmospheric chemical and microphysical processing of direct emissions of PM and its
17 precursors together with mechanical generation of particles tend to produce distinct lognormal
18 modes (Whitby, 1978 as shown in Figure 2-1). Discussions in this and subsequent sections will
19 focus on particles in specific size ranges (i.e., PM2.5, PMi0.2.5, and PMi0). The subscripts after
20 PM refer to the aerodynamic diameter in micrometers (|im) of 50% cut points of sampling
21 devices, see section 2.2.1.2. For example, EPA defines PMio as particles collected by a sampler
22 with an upper 50% cut point of 10 |im aerodynamic diameter and a specific, fairly sharp
23 penetration curve, as defined in the Code of Federal Regulations (40 CFR Part 58). PM2 5 is
24 defined in an analogous way. Ultrafme particles, defined here as particles with a diameter less
25 than or equal to 0.1 jim (typically based on physical size, thermal diffusivity, or electrical
26 mobility), will also be discussed. (USEPA, 2009a, section 3.2)
27 In present usage, the term "fine particles" is most often associated with the PM2.5
28 fraction, which includes the nucleation, Aitken, and accumulation modes and some particles
14 In this preliminary draft Policy Assessment, particle size or diameter refers to a normalized measure called
aerodynamic diameter unless otherwise noted. Most ambient particles are irregularly shaped rather than spherical.
The aerodynamic diameter of any irregular shaped particle is defined as the diameter of a spherical particle with a
material density of 1 g/cm3 and the same settling velocity as the irregular shaped particle. Particles with the same
physical size and shape but different densities will have different aerodynamic diameters (US EPA 2009a, section
3.2).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-2
-------
1
2
3
4
5
6
7
8
9
10
12
13
14
15
16
17
18
19
20
21
22
from the lower-size tail of the coarse particle mode between about 1 and 2.5 jim aerodynamic
diameter. "Thoracic coarse particles" are frequently used in reference to PMi0.2.5, which does not
include the low-end tail of the coarse particle mode. With high relative humidity, larger
hygroscopic particles in the accumulation mode could also extend into the 1 to 3 jim size range.
These relationships can be seen in Figure 2-1, which shows the number distribution for ultrafine
particles and the volume distribution (or mass distribution if particle density is constant across
the size range) for fine and (thoracic) coarse particles. The figure is arranged this way because
particle number is most highly concentrated in the ultrafine size range, but volume (or mass) is
most concentrated in the larger size ranges. {US EPA, 2009a, section 3.2}
Figure 2-1. Particle size distributions by number and volume.
Fine Particles
Coarse Particles
0.01
0.1 1
Diameter (micrometers)
Figure 2-1 Notes: Dashed lines refer to values in individual modes and solid lines to their sum. Note that ultrafine
particles are a subset of fine particles. (US EPA, 2009a, Figure 3-1). Source: Pandis (2004).
Characterizing particle size is important because different size particles penetrate to
different regions of the human respiratory tract (see further discussion in section 3.2). Most
commonly, PM2.5 is used as an indicator of fine particles, PMio-2.5 is used as an indicator of the
thoracic component of coarse particles that is sometimes referred to as thoracic coarse (noting
that it excludes some coarse particles below 2.5 jim and above 10 jim), and, PMiois the sum of
fine and thoracic coarse particles.. In the current NAAQS for PM, PMi0 is used as the indicator
for thoracic coarse particles. (USEPA, 2009a, section 3.2)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-3
-------
1 As can be seen from Table 2-1, particles in individual size modes are characterized by
2 rather distinct sources, composition, chemical properties, lifetimes in the atmosphere (T), and
3 distances over which they can travel. Whereas particles in the smaller size modes are formed
4 mainly by combustion processes and by nucleation and condensation of gases, coarse particles
5 are generated mainly by mechanical activity, such as the action of wind on either the ground or
6 the sea surface or construction or resuspension by traffic. Among the mechanical activities that
7 generate coarse particles are various types of material handling and resuspension of dust at
8 industrial facilities such as metallurgical operations, cement kilns, mines, quarries, etc. Particles
9 in the ultrafme size range are either emitted directly to the atmosphere or are formed by
10 nucleation of gaseous constituents in the atmosphere as shown in Table 2-2. (US EPA, 2005,
11 Table 2-1; US EPA 2009a, Table 3-1 and section 3.2; ISA text augmented)
12 2.2.1.2 Sampler Cutpoints
13 Particle samplers typically use size-selective inlets that are defined by their 50 percent
14 cutpoint, which is the particle aerodynamic diameter at which 50 percent of particles of a
15 specified diameter are captured by the inlet, and their penetration efficiency as a function of
16 particle size. The usual notation for these classifications is "PMx", where x refers to
17 measurements with a 50 percent cut point of x jim aerodynamic diameter. Because of the
18 overlap in the size distributions of fine and coarse-mode ambient particles, and the fact that inlets
19 do not have perfectly sharp cut points, no single sampler can completely separate them. Given a
20 specific size cut, the smaller the particles the greater the percentage of particles that are captured,
21 although for particles on the order of 1-10 nm, many may diffuse to sampler surfaces before
22 being captured on the filter. The objective of size-selective sampling is usually to measure
23 particle size fractions that provide a relationship to human health impacts, visibility impairment,
24 deposition, or emissions sources. (US EPA, 2005, section 2.2.1.2, modified regarding particles
25 1-10 nm).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-4
-------
1 Table 2-1. Particle Size Fraction Terminology
Term
Description
Size Distribution Modes
Coarse Particles
Thoracic Coarse Particles
Fine Particles
Accumulation-Mode
Particles
Ultrafine Particles
Aitken-Mode Particles
Nucleation-Mode
Particles
The distribution or particles that are mostly larger than the intermodal minimum
in volume or mass distributions; also referred to as coarse-mode particles. This
intermodal minimum generally occurs between 1 and 3 um.
A subset of coarse particles that includes particles that can be inhaled and
penetrate to the thoracic region (i.e., the tracheobronchial and the gas-exchange
regions) of the lung. This subset includes the smaller coarse particles, ranging
in size up to those with a nominal aerodynamic diameter less than or equal to 10
um.
The distribution of particles that are mostly smaller than the intermodal
minimum in volume or mass distributions; this minimum generally occurs
between 1 and 3 um. This includes particles in the nucleation, Aitken, and
accumulation modes.
A subset of fine particles with diameters above about 0.1 um. Ultrafine
particles grow by coagulation or condensation and "accumulate" in this size
range.
A subset of fine particles with diameters below about 0. 1 um, encompassing the
Aitken and nucleation modes.
A subset of ultrafme particles with diameters between about 0.01 and 0. 1 um.
Freshly formed particles with diameters below about 0.01 um.
Sampling Measurements
Total Suspended Particles
(TSP)
PM10
PM25
PMia-2.5
Particles measured by a high volume sampler as described in 40 CFR Part 50,
Appendix B. This sampler has a cut point of aerodynamic diameters that varies
between 25 and 40 um depending on wind speed and direction.
Particles measured by a sampler that contains a size fractionator (classifier)
designed with an effective cut point (50% collection efficiency) of 10 um
aerodynamic diameter. This measurement includes the fine particles and a
subset of coarse particles, and is an indicator for particles that can be inhaled
and penetrate to the thoracic region of the lung; also referred to as thoracic
particles. See 40 CFR Part 50, Appendix G.
Particles measured by a sampler that contains a size fractionator (classifier)
designed with an effective cut point (50% collection efficiency of 2.5 um
aerodynamic diameter. This measurement, which generally includes all fine
particles, is an indicator for fine particles; also referred to as fine-fraction
particles. A small portion of coarse particles may be included depending on the
sharpness of the sampler efficiency curve. See 40 CFR Part 50, Appendix J.
Particles measured directly using a dichotomous sampler or by subtraction of
particles measured by a PM2 5 sampler from those measured by a PM10 sampler.
This measurement is an indicator for the coarse fraction of thoracic coarse
particles; also referred to as thoracic coarse particles or coarse-fraction particles.
See 40 CFR Part 50, Appendix O.
Source: modification of US EPA, 2005, Table 2-1
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-5
-------
1 Table 2-2. Characteristics of ambient fine (ultrafine plus accumulation-mode) and coarse
2 particles. (US EPA, 2009a, Table 3-1)
Formation
Processes
Formed by
Composed of
Solubility
Sources
Atmospheric
half-life
Removal
Processes
Travel
distance
Fine
Ultrafine
Accumulation
Combustion, high-temperature processes, and atmospheric reactions
Nucleation of atmospheric
gases including H2SO4, NH3
and some organic
compounds
Condensation of gases
Sulfate
EC
Metal compounds
Organic compounds with
very low saturation vapor
pressure at ambient
temperature
Not well characterized
High temperature
combustion
Atmospheric reactions of
primary, gaseous
compounds.
Minutes to hours
Grows into accumulation
mode
Diffuses to raindrops
<1 to 10s of km
Condensation of gases
Coagulation of smaller particles
Reactions of gases in or on particles
Evaporation of fog and cloud droplets
in which gases have dissolved and
reacted
Sulfate, nitrate, ammonium, and
hydrogen ions
EC
Large variety of organic compounds
Metals: compounds of Pb, Cd, V, Ni,
Cu, Zn, Mn, Fe, etc.
Particle-bound water
Bacteria, viruses
Largely soluble, hygroscopic, and
deliquescent
Combustion of fossil and biomass
fuels, and high temperature industrial
processes, smelters, refineries, steel
mills etc.
Atmospheric oxidation of NO2, SO2,
and organic compounds, including
biogenic organic species
(e.g., terpenes)
Days to weeks
Forms cloud droplets and rains out
Dry deposition
100s to 1000s of km
Coarse
Break-up of large solids/droplets
Mechanical disruption (crushing,
grinding, abrasion of surfaces)
Evaporation of sprays
Suspension of dusts
Reactions of gases in or on particles
Nitrates/chlorides/sulfates from
HNO3/HC1/SO2 reactions with
coarse particles
Oxides of crustal elements (Si, Al, Ti,
Fe)
CaCO3, CaSO4, NaCl, sea salt
Bacteria, pollen, mold, fungal spores,
plant and animal debris
Largely insoluble and nonhygroscopic
Resuspension of particles deposited
onto roads
Tire, brake pad, and road wear debris
Suspension from disturbed soil (e.g.,
farming, mining, unpaved roads)
Construction and demolition
Fly ash from uncontrolled combustion
of coal, oil, and wood
Ocean spray
Minutes to hours
Dry deposition by fallout
Scavenging by falling rain drops
<1 to 10s of km (100s to 1,000s of km
in dust storms for the small size tail)
Source: Wilson and Suh (1997) (adapted). US EPA, 2009a, Table 3-1 .
Note: Coarse column may not yet adequately summarize industrial processes emitting coarse PM
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-6
-------
1 Since 1987, EPA has defined indicators of PM for the PM NAAQS using cut points of
2 interest. Prior to 1987, the indicator for the PM NAAQS was total suspended particulate matter
3 (TSP), and was defined by the design of the High Volume Sampler (Hi Vol).15 TSP typically
4 includes particles with diameters less than about 40 jim, but the upper size cut varies
5 substantially with placement, wind velocity, and wind direction relative to the non-cylindrically
6 symmetrical shape of the sampler inlet. TSP samplers are still used in state regulatory
7 monitoring networks, but only to monitor for lead (Pb) and other toxic metals. When EPA
8 established new PM standards in 1987, the selection of PMio as the new indicator was intended
9 to focus regulatory attention on particles small enough to be inhaled and to penetrate into the
10 thoracic region of the human respiratory tract. In 1997, EPA established standards for fine
11 particles measured as PM2.5 (i.e., the fine fraction of PMio). Following the establishment of the
12 2.5 and 10 micron size cuts as indicators for the PM NAAQS, state monitoring networks have
13 employed samplers with those cutpoints almost exclusively. Ongoing research networks and
14 special monitoring studies for health effects research also are dominated by samplers with these
15 cutpoints, with some use of samplers with other cutpoints. {US EPA, 2005, section 2.2.1.2,
16 augmented)
17 In the PM NAAQS review completed in 2006, EPA defined but did not set a standard for
18 another PM indicator identified in Table 2-1 as PMio-2.5, which represents the subset of coarse
19 particles small enough to be inhaled and to penetrate into the thoracic region of the respiratory
20 tract (i.e., the coarse fraction of PMio, or thoracic coarse particles). The EPA also established a
21 Federal Reference Method for measuring PMi0.2.5, to facilitate consistency in future research and
22 air quality characterization relative to this indicator (71 FR 61212 to 61213, October 17, 2006).
23 2.2.2 Sources and Formation Processes
24 Parti culate matter is composed of both primary (derived directly from emissions) and
25 secondary (derived from atmospheric reactions involving gaseous precursors) components.
26 Table 2-3 summarizes anthropogenic and natural sources for the major primary and secondary
27 aerosol constituents of fine and coarse particles. Anthropogenic sources can be further divided
28 into stationary and mobile sources. Stationary sources include fuel combustion for electrical
29 utilities, residential space heating, and cooking; industrial processes; construction and
30 demolition; metal, mineral, and petrochemical processing; wood products processing; mills and
31 elevators used in agriculture; erosion from tilled lands; waste disposal and recycling; and
40 CFR Part 50, Appendix B, Reference Method for the Determination of Suspended Paniculate Matter in the
Atmosphere (High-Volume Method).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-7
-------
1 biomass combustion. Biomass combustion encompasses many emission activities including
2 burning of wood for fuel, burning of vegetation to clear land for agriculture and construction, to
3 dispose of agricultural and domestic waste, to control the growth of animal or plant pests, and to
4 manage forest resources (prescribed burning). Wildlands also burn due to lightning strikes and
5 arson. Mobile or transportation-related sources include direct emissions of primary PM and
6 secondary PM precursors from highway vehicles and non-road sources as well as fugitive dust
7 from paved and unpaved roads. Also shown in Table 2-3 are sources for several precursor gases,
8 the oxidation of which can form secondary PM. Only major sources for each constituent within
9 each broad category shown at the top of Table 2-3 are listed. Not all sources are equal in
10 magnitude (US EPA, 2009a, section 3.3).
11 In general, the sources of fine particles are very different from those of coarse particles.
12 Some of the mass in the fine size fraction forms during combustion from material that has
13 volatilized in combustion chambers and then recondensed before after emission to the
14 atmosphere. Some ambient PM2.5 forms in the atmosphere from photochemical reactions
15 involving precursor gases. Included in this category is the formation of new ultrafine particles by
16 homogeneous nucleation of precursor gases in addition to the condensation of gases on pre-
17 existing particles. Particulate matter formed by the first mechanism is referred to as primary, and
18 PM formed by the second mechanism is referred to as secondary. Biological material also exists
19 in the fine fraction including many types of microorganisms, especially viruses and bacteria and
20 fragments of pollens and fungal spores. PMio-2.5 is mainly primary in origin, as it is produced by
21 surface abrasion or by suspension of biological material and fragments of living things
22 (e.g., plant and insect debris). In addition, atmospheric reaction products condense on coarse
23 particles. Some industrial facilities employ both combustion and abrasion/suspension processes,
24 and thus can be significanat emitters of both fine and coarse particles. Because precursor gases
25 undergo mixing during transport from their sources and reactions in the atmosphere can produce
26 the same products, it is difficult to identify individual sources of secondary PM. Transport and
27 transformation of precursors can occur over distances of hundreds of kilometers. PMio-2.5 has a
28 shorter lifetime in the atmosphere, so its effects tend to be more localized. However,
29 intercontinental transport of dust from African and Asian deserts occurs, and some of this
30 material is in the PMio-2.5 size range. Major events are highly episodic but much smaller
31 contributions can be made at other times. (US EPA, 2009a, section 3.3, augmented regarding
32 industrial sources with multiple processes) (Text borrowed from the second draft ISA, but may
33 need amendment as it is not entirely clear (1) what "formed by the first mechanism " refers to
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-8
-------
1 and more generally (2) whether unreacted homogenous nucleation material and unreacted
2 material condensed on pre-existing par tides are being called primary or secondary.)
3 Precursors to secondary PM have natural and anthropogenic sources, just as primary PM
4 has natural and anthropogenic sources. A substantial fraction of the fine particle mass, especially
5 during the warmer months of the year, is secondary in nature, formed as the result of atmospheric
6 reactions involving both inorganic and organic gaseous precursors. The major atmospheric
7 chemical transformations leading to the formation of particulate nitrate (pNOs) and sulfate
8 (pSO/t) are relatively well understood, whereas those involving the formation of secondary
9 organic aerosol (SOA) are less well understood. (US EPA, 2009a, section 3.3.2}
10 The formation and growth of fine particles are influenced by several processes including:
11 (1) nucleation (i.e., gas molecules coming together to form a new particle); (2) condensation of
12 gases onto existing particles; (3) coagulation of particles, the weak bonding of two or more
13 particles into one larger particle; (4) uptake of water by hygroscopic components; and (5) gas
14 phase reactions which form secondary PM. Gas phase material condenses preferentially on
15 smaller particles since they have the greatest surface area, and the efficiency of coagulation for
16 two particles decreases as the particle size increases. Thus, ultrafme particles grow into the
17 accumulation mode, but accumulation-mode particles do not normally grow into coarse particles.
18 (USEPA, 2005, section 2.2.2)
19 Examples of secondary particle formation include:
20 o the conversion of sulfur dioxide (862) to sulfuric acid (H2SO4) droplets that
21 further react with gaseous ammonia (NHa) to form various sulfate particles (e.g.,
22 ammonium sulfate (NH^SC^ or ammonium bisulfate NtLjHSO/t);
23
24 o the conversion of nitrogen dioxide (NO2) to nitric acid (HNOs) vapor that reacts
25 further with ammonia to form ammonium nitrate (NiLiNOs) particles; and
26
27 o reactions involving gaseous volatile organic compounds (VOC) yielding organic
28 compounds with low ambient temperature (saturation) vapor pressures that
29 nucleate or condense on existing particles to form secondary organic aerosol
30 particles.
31 (US EPA, 2005 section 2.2.2)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-9
-------
1
2
Table 2-3. Constituents of atmospheric particles and their major sources
Primary (PM <2.5 um)
Aerosol species
Sulfate (SO42-)
Nitrate (NO3~)
Minerals
Ammonium
(NH4+)
Organic
carbon (OC)
EC
Metals
Bioaerosols
Natural
Sea spray
Erosion and re-
entrainment
Wildfires
Wildfires
Volcanic
activity
Viruses and
bacteria
Anthropogenic
Fossil fuel combustion
Mobile source exhaust
Fugitive dust from
paved and unpaved
roads, agriculture,
forestry, construction,
and demolition
Mobile source exhaust
Prescribed burning,
wood burning, motor
vehicle exhaust,
cooking, tire wear and
industrial processes
Mobile source exhaust
(mainly diesel), wood
biomass burning, and
cooking
Fossil fuel combustion,
smelting and other
metallurgical processes,
and brake wear
Primary (PM >2.5 um)
Natural
Sea spray
Erosion and re-
entrainment
Soil humic
matter
Erosion, re-
entrainment,
and organic
debris
Plant and insect
fragments,
pollen, fungal
spores, and
bacterial
agglomerates
Anthropogenic
Fugitive dust, paved
and unpaved road
dust, agriculture,
forestry,
construction, and
demolition
Tire and asphalt
wear, paved and
unpaved road dust
Tire and asphalt
wear, paved and
unpaved road dust
Secondary PM
Precursors
(PM <2.5 jim)
Natural
Oxidation of
reduced sulfur
gases emitted by
the oceans and
wetlands and SO2
and H2S emitted
by volcanism and
forest fires
Oxidation of NOX
produced by soils,
forest fires, and
lighting
Emissions of NH3
from wild
animals, and
undisturbed soil
Oxidation of
hydrocarbons
emitted by
vegetation
(terpenes, waxes)
and wild fires
Anthropogenic
Oxidation of SO2
emitted from fossil
fuel combustion
Oxidation of NOX
emitted from fossil
fuel combustion and
in motor vehicle
exhaust
Emissions of NH3
from motor
vehicles, animal
husbandry, sewage,
and fertilized land
Oxidation of
hydrocarbons
emitted by motor
vehicles, prescribed
burning, wood
burning, solvent use
and industrial
processes
Dash ( — ) indicates either very minor source or no known source of component.
Source: U.S. EPA,(2004).; US EPA, 2009a, Table 3-2)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-10
-------
1 In most of the ambient monitoring data displays shown later in this chapter, the first two
2 types of secondary PM are generally labeled plurally as 'sulfates' and 'nitrates' (respectively),
3 which implies that the ammonium content is encompassed. The third type of secondary PM may
4 be lumped with the directly emitted elemental or organic carbon particles and labeled "total
5 carbonaceous mass," or the two types of carbonaceous PM may be reported separately as
6 elemental carbon16 (EC) and organic carbonaceous material (OCM), sometimes also referred to
7 as organic aerosol. When only the mass of carbon atoms is being referred to, the term organic
8 carbon (OC) is used. (US EPA, 2005, section 2.2.2)
9 2.2.3 Chemical Composition
10 As summarized in Table 2-3, a number of chemical components of ambient PM are found
11 predominately in fine particles including: sulfate, ammonium, and hydrogen ions; elemental
12 carbon, secondary organic compounds, and primary organic species from cooking and
13 combustion; and certain metals, primarily from combustion processes. Chemical components
14 found predominately in coarse particles include: crustal-related materials such as calcium,
15 aluminum, silicon, magnesium, and iron; and primary organic materials such as pollen, spores,
16 and plant and animal debris. (USEPA, 2005, section 2.2.3)
17 Some components, such as nitrate and potassium, may be found in both fine and coarse
18 particles. Nitrate in fine particles comes mainly from the reaction of gas-phase nitric acid with
19 gas-phase ammonia to form ammonium nitrate particles. Nitrate in coarse particles comes
20 primarily from the reaction of gas-phase nitric acid with pre-existing coarse particles. Potassium
21 in coarse particles comes primarily from soil, with additional contributions from sea salt in
22 coastal areas. Potassium in fine particles, generally not a significant contributor to overall mass,
23 comes mainly from emissions of burning wood, with infrequent but large contributions from
24 fireworks, as well as significant proportions from the tail of the distribution of coarse soil
16 Also called light absorbing carbon and black carbon. The terms elemental carbon and black carbon are often used
interchangeably, but may be defined differently by different users. Black carbon is most often used in discussions of
optical properties and elemental carbon is most often used when referring to chemical composition. In many cases,
there is little difference between the two, but care must be taken when comparing data from studies with different
purposes. In addition, the term soot is also used in many instances to refer to either EC or BC. The differences
between soot and either EC or BC can be significant, as soot refers to elemental carbon formed from gas phase
hydrocarbons in the combustion process, and tends to be in the submicron fraction and often in the fraction of
particles that are smaller than 0.10 microns in aerodynamic diameter. Elemental carbon and BC both include
carbonaceous particles formed from incomplete burnout of solid carbonaceous fuels; these particles have distinctly
different physical characteristics compared to char. As an additive to automotive tires, commercially produced
'carbon black' and associated contaminants can also be found in resuspended urban road dust.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-11
-------
1 particles (i.e., less than 2.5 jim in diameter) in areas with high soil concentrations. (USEPA,
2 2005, section 2.2.3}
3 Many ambient particles also contain water (i.e., particle-bound water) as a result of an
4 equilibrium between water vapor and hygroscopic PM. Particle-bound water influences the size
5 of particles and in turn their aerodynamic and light scattering properties (discussed in section
6 2.2.5). Particle-bound water can also act as a carrier to convey dissolved gases or reactive
7 species into the lungs which, in turn, may cause heath consequences. The amount of particle-
8 bound water in ambient particulate matter will vary with the particle composition and the
9 ambient relative humidity. Sulfates, nitrates, and some secondary organic compounds are much
10 more hygroscopic than elemental carbon (EC), primary organic carbonaceous material (OCM),
11 and crustal material. (USEPA, 2005, section 2.2.3)
12 2.2.4 Fate and Transport
13 Fine and coarse particles typically exhibit different behaviors in the atmosphere. These
14 differences may affect several exposure-related considerations, including the representativeness
15 of central-site monitored values and the penetration of particles formed outdoors into indoor
16 spaces. The ambient residence time of atmospheric particles varies with size and composition.
17 Ultrafme particles have a very short life, on the order of minutes to hours, since they are more
18 likely to reach the accumulation mode through agglomeration or condensation mechanisms or to
19 be lost by diffusion to surfaces. However, their chemical content persists in the accumulation
20 mode. Ultrafme particles are also small enough to be removed through diffusion to falling rain
21 drops. Accumulation-mode particles remain suspended longer in the atmosphere because they
22 are too large to diffuse rapidly to surfaces or to other particles and too small to settle out or
23 impact on stationary objects. They can be transported thousands of kilometers and remain in the
24 atmosphere for days to weeks. Accumulation-mode particles serve as condensation nuclei for
25 cloud droplet formation and are eventually removed from the atmosphere in falling rain drops.
26 Accumulation-mode particles that are not involved in cloud processes are eventually removed
27 from the atmosphere by gravitational settling and impaction on surfaces. (USEPA, 2005, section
28 2.2.4)
29 By contrast, coarse particles can settle rapidly from the atmosphere with lifetimes ranging
30 from minutes to days depending on their size, atmospheric conditions, and altitude. Larger
31 coarse particles are not readily transported across urban or broader areas, because they are
32 generally too large to follow air streams, and they tend to be easily removed by gravitational
33 settling and by impaction on surfaces. Smaller coarse particles extending into the tail of the
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-12
-------
1 distribution can have longer lifetimes and travel longer distances, especially in extreme
2 circumstances. For example, dust storms in desert areas of Africa and Asia lift coarse particles
3 to high elevations, and these "dust clouds" are readily observed to undergo intercontinental
4 transport to North America. Coarse particles also are readily removed by falling rain drops. (US
5 EPA, 2005, section 2.2.4)
6 2.2.5 Optical Properties of Particles
7 As discussed in Chapter 9 of the second draft ISA, light extinction is the optical
8 characteristic of the atmosphere that best determines the impact potential of PM on perceived
9 visibility. Light extinction is the loss of light per unit of distance either by scattering out of the
10 beam of light or by absorption, which converts the light to heat. Particulate matter and gases
11 scatter and absorb light. Light scattering by gases (e.g., nitrogen, oxygen, etc.) that comprise the
12 atmosphere (also known as Rayleigh or clear-air scattering) is related to the density of the air,
13 which is sufficiently constant with elevation that it can be considered a known constant value for
14 any location. NO2 is the only atmospheric gas that absorbs visible light appreciably and its
15 effects are generally small (i.e., less than 5%) compared to PM light extinction, so its
16 contribution to ambient visibility impacts is often ignored. By this assumption, light extinction is
17 approximated as the sum of PM light extinction (includes both scattering and absorption) plus
18 Rayleigh light scattering, where the former characterizes the PM contribution to visibility
19 impacts and the latter is taken to be a time-invariant constant depending only on elevation above
20 sea level.
21 Light-absorbing carbon (e.g., diesel exhaust soot and smoke) and some crustal minerals
22 are the only commonly occurring airborne particle components that absorb light. All particles
23 scatter light, and generally particle light scattering is the largest of the four light extinction
24 components. While a larger particle scatters more light than a similar shaped smaller particle of
25 the same composition, the light scattered per unit of mass concentration is greatest for particles
26 with diameters from about 0.3 to 1.0 um. Particles composed of water soluble inorganic salts
27 (i.e., ammoniated sulfate, ammonium nitrate, sodium chloride, etc.) are hygroscopic in that they
28 absorb water as a function of relative humidity to form liquid solution droplets. The droplets
29 become larger when relative humidity increases resulting in increased light scattering, hence the
30 same particulate matter dry concentration produces greater haze levels. Because the extinction
31 efficiency of PM depends on size, composition and humidity effects, there is no simple one-to-
32 one correspondence between PM concentration and PM light extinction. The relationship
33 between ambient PM and visibility impairment is further discussed below in section 2.8.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-13
-------
1 2.2.6 Other Radiative Properties of Particles
2 As discussed in section 9.3 of the second draft ISA, PM affects climate through direct
3 and indirect effects. The direct effect is primarily realized as planet brightening when seen from
4 space because most aerosols scatter most of the visible light that reaches them. The IPCC AR4
5 reported that the radiative forcing from this direct effect was -0.5 (±0.4) W/m2 and identified the
6 level of scientific understanding of this effect as "medium-low." The global mean direct radiative
7 forcing from individual aerosol components varies from strongly negative for sulfate to positive
8 for black carbon with weaker positive or negative effects for other components, all of which can
9 vary strongly over space and time and with aerosol size. The indirect effects are primarily
10 realized as an increase in cloud brightness (termed the "first indirect" or "Twomey" effect),
11 changes in precipitation, and possible changes in cloud lifetime. The IPCC AR4 reported that the
12 radiative forcing from the Twomey effect was -0.7 (range: -1.1 to +4) and identified the level of
13 scientific understanding of this effect as "low." The other indirect effects from aerosols were not
14 considered to be radiative-forcing.
15 Taken together, direct and indirect effects from aerosols increase Earth's shortwave
16 albedo or reflectance, thereby reducing the radiative flux reaching Earth's surface from the Sun.
17 This produces net climate cooling from aerosols. The current scientific consensus reported by
18 IPCC AR4 is that the direct and indirect radiative forcing from anthropogenic aerosols computed
19 at the top of the atmosphere, on a global average, is about -1.3 (range: -2.2 to -0.5) W/ m2.
20 Although the magnitude of this negative radiative forcing appears large in comparison to the
21 analogous IPCC AR4 estimate of positive radiative forcing from anthropogenic GHG of about
22 2.9 (±0.3) W/ m2, the spatial and temporal distributions of these two very different radiative
23 forcing agents are dissimilar; therefore, they do not simply cancel, and regional differences can
24 be large. These differences result from the much shorter atmospheric lifetime of aerosols than for
25 the radiatively important trace gases, implying that the radiative effects of aerosols respond much
26 more quickly to changes in emissions than do the effects from the gas-phase forcing agents.
27 2.3 AMBIENT PM MEASUREMENT METHODS
28 The measurement of PM can be accomplished with a number of sampling and analysis
29 techniques. In the United States, much of the historical focus on PM methods has been
30 associated with measuring gravimetric mass as Total Suspended Particulate (TSP), which was
31 the PM indicator until 1987, and PMi0, which replaced TSP as the PM indicator in 1987. In
32 1997, a fine particle indicator was added, resulting in PMio becoming the coarse particle
33 indicator. PMio Federal Reference Methods (FRMs) include the high-volume method, low-
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-14
-------
1 volume method, and dichotomous method. The fine particle indicator is based on the PM2.5
2 FRM, which also provides gravimetric mass, is a low volume method (i.e., it operates at 16.67
3 1pm or one cubic meter per hour), and has been operating throughout the country since 1999.
4 Because PM is a complex mixture of substances with differing physical and chemical properties
5 that vary in time and space, measuring and characterizing particles suspended in the atmosphere
6 is a significant challenge.17 Thus, to support characterizing the composition of PM2 5, chemical
7 speciation samplers have been implemented since 2001 as part of the Chemical Speciation
8 Network (CSN). Additionally, most gravimetric samplers (i.e., the PM2.5 FRM) require post-
9 sampling laboratory analysis; therefore, PM2.5 continuous mass monitors and associated data
10 systems that provide for near real-time reporting of data have been deployed in most cities since
11 2004 to support timely reporting and forecasting air pollution to the public through the Air
12 Quality Index (AQI). In addition to gravimetric mass and composition, other PM measurements
13 include particle number, light scattering, and light absorption. Most samplers collect PM by
14 drawing a controlled volume of ambient air through a size-selective inlet, usually defined by the
15 inlet's 50 percent cut point. Measurable indicators of fine particles include PM2.5, PMi.o, British
16 or black smoke (BS), black carbon (BC), coefficient of haze (COH), and PMio (in areas
17 dominated by fine particles). Measurable indicators of coarse-mode particles include PMio-2.s,
18 PMi5_2.5, and PMio (in areas dominated by coarse-mode particles).
19 2.3.1 Particle Mass Measurement Methods
20 Ambient PM mass can be measured directly, by gravimetric methods, or indirectly, using
21 methods that rely on the physical properties of particles. Methods can also be segregated as
22 either discrete or continuous according to whether samples require laboratory analysis or the data
23 are available in real-time. Discrete methods provide time integrated data points (typically over a
24 24-hour period) that allow for post-sampling gravimetric analyses in the laboratory. These
25 methods are typically directly linked to the historical data sets that have been used in health
26 studies that provide the underlying basis for having a NAAQS. Continuous methods can provide
27 time resolution on the order of minutes and automated operation up to several weeks, facilitating
28 the cost-effective collection of greater amounts of data compared with discrete methods.
29 The most common direct measurement methods include filter-based methods where
30 ambient aerosols are collected for a specified period of time (e.g., 24 hours) on filters that are
17 Refer to second draft ISA Chapter 3 for more comprehensive assessments of particle measurement methods. A
summary of PM measurement methods is also given in Fehsenfeld et al. (2003). Significant improvements and
understanding of routine and advanced measurement methods is available through EPA's PM Supersites Program
(see www.epa.gov/ttn/amtic/supersites.htmlX
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-15
-------
1 weighed before and after collection to determine mass by difference. It should be noted that
2 while filter-based samplers can be programmed to operate over just about any time interval, they
3 are operated by state and local agencies on a schedule of midnight to midnight local standard
4 time. Examples of the most commonly used filter-based samplers include the FRM monitors for
5 PM2.5 and PMio. Dichotomous samplers contain a separator (i.e., a virtual impactor) that splits
6 the air stream from a PMio inlet into two streams so that both fine- and coarse-fraction particles
7 can be collected on separate filters. A low-volume dichotomous sampler (operated at 16.7
8 L/min), based on virtual impaction, was described in the 2004 AQCD (U.S. EPA, 2004, section
9 2.2.4.1.2). These gravimetric methods require weighing the filters after they are subjected to
10 specific equilibrium conditions (i.e., 20 - 23° C and 30 - 40 percent RH in most cases).
11 Discrete, gravimetric methodologies have been refined over the past 25 years as PM
12 monitoring networks have evolved from sampling based on the high volume TSP and PMio
13 method to the PM2 5 FRM. The inclusion of such measures as size-selective inlets and
14 separators, highly specific filter media performance criteria, active flow control to account for
15 ambient changes in temperature and pressure, and highly prescriptive filter weighing criteria
16 have reduced levels of measurement uncertainty, compared with earlier methods.
17 National quality assurance data analyzed by EPA for the latest available three-year period
18 (2005-2007) indicate that the PM2.5 FRM has been a robust indicator of ambient levels by
19 meeting the data quality objectives (DQO) established at the beginning of the PM2.5 monitoring
20 program. Three-year average estimates from reporting organizations aggregated on a national
21 basis for collocated sampler precision (7.55 %), flow rate accuracy (0.007 %), and method bias (-
22 2.97 %, from the Performance Evaluation Program)18 are well within their respective goals of
23 +10 %, +4 %, and +10 %. There are currently over 900 PM2.5 FRM discrete monitors across the
24 U.S. Of those, about 225 monitors are operating on a daily schedule.
25 The most widely deployed PM continuous technology is the Tapered Element Oscillating
26 Microbalance (TEOM®) sensor, consisting of a replaceable filter mounted on the narrow end of
27 a hollow tapered quartz tube. The air flow passes through the filter, and the aerosol mass
28 collected on the filter causes the characteristic oscillation frequency of the tapered tube to change
29 in direct relation to particle mass. This approach allows mass measurements to be recorded on a
30 near-continuous basis (i.e., every few minutes).
18 The Performance Evaluation Program (PEP) is designed to determine total bias for the PM2 5 sample collection
and laboratory analysis processes. Federally referenced audit samplers are collocated adjacent to a monitoring site's
routine sampler and run for a 24-hour period. The concentrations are then determined independently by an EPA
laboratory and compared in order to assess bias.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-16
-------
1 The next generation of the TEOM® is the Filter Dynamics Measurement System
2 (FDMS®) monitor. This method is based upon the differential TEOM that is described in the
3 ISA (ISA, Chapter 3.4.1.1). The FDMS method employs an equilibration system integrated with
4 a TEOM® having alternating measurements of ambient air and filtered air. This self-referencing
5 approach allows the method to determine the amount of volatile PM that is evaporating from the
6 TEOM sensor for 6 of every 12 minutes of operation. An hourly average is calculated from the
7 measurement of the total aerosol mass concentration, including non-volatile and volatile PM,
8 every 6 minutes.
9 The other major PM monitoring device that provides for gravimetric mass is the beta
10 attenuation monitor. A beta attenuation (or beta gauge) monitor determines the mass of particles
11 deposited on a filter by measuring the absorption of electrons generated by a radioactive isotope,
12 where the absorption is closely related to the mass of the particles.
13 In 2006, EPA finalized new performance criteria for approval of PM2.5 and PMio-2.5
14 monitors as Federal Equivalent Methods (FEMs). At the same time, EPA also finalized an
15 approach to approving PM2.5 continuous methods across an agencies network identified as an
16 Approved Regional Method (ARM). Both the FEM and ARM are based on new performance
17 criteria that utilized a data quality objective process that takes into account the existing data
18 quality objectives of the network with the higher sampling frequency (i.e., daily sampling)
19 provided for by PM continuous monitors. These new performance criteria have led to a number
20 of approved FEMs for PM2.5 including technologies that employ beta attenuation as well as use
21 of the TEOM® technology in combination with an environmental conditioning system (i.e., the
22 Thermo FDMS®). A few PMio-2.5 technologies have also been approved as FEMs including the
23 Met One BAM providing continuous measurement of coarse particles by difference and the
24 Thermo 2025D, a dichotomous sampler that utilizes the same virtual impactor approved for use
25 on its continuous dichotomous monitor, the FDMS® 1405D. No monitoring agency has applied
26 for ARM approval, although a few were pursuing it until the methods they had implemented
27 were approved as FEM's; thus making an ARM approval no longer needed. In addition to the
28 recently approved PM2.5 FEMs, many monitoring agencies continue to use conventional TEOMs
29 and other PM2 5 continuous technologies as part of their monitoring programs to report and
30 support forecasting of the AQI. However, these methods are not used for comparison with the
31 NAAQS.
32 The number of continuous PM2.s monitors across the U.S. has increased to over 700
33 stations. Although a subset of these monitors were required by regulation to be placed in
34 metropolitan areas of greater than 1 million population, a higher percentage were installed to
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-17
-------
1 provide improved temporal resolution for daily air quality index reporting and PM2.5 forecast
2 verifications through EPA's AIRNOW program. Some of the continuous PM2.5 data reported
3 through the AIRNOW program are adjusted to better match FRM results.19 The continuous data
4 used in the analyses in this chapter were obtained from EPA's Air Quality System (AQS); some
5 of these AQS data are adjusted, and some are not. There is currently an effort underway to better
6 characterize this facet of the continuous data in AQS. Still, the AQS continuous data utilized in
7 analyses here do show excellent correlation with collocated FRM measurements; over 95 percent
8 of the continuous/FRM site pairs had a correlation coefficient of over 0.72, and almost 75
9 percent had a correlation of 0.9 or higher (Schmidt et al., 2005).
10 2.3.2 Indirect Optical Methods
11 Particulate matter has also been characterized in the U.S. and elsewhere by indirect
12 optical methods that rely on the light scattering or absorbing properties of either suspended PM
13 or PM collected on a filter.20 These include BS, COH, Nephelometer, Aethalometer®, and light
14 extinction estimates derived from visibility measurements. In locations where they are calibrated
15 to standard mass units, these indirect measurements can be useful surrogates for particle mass.
16 The BS method typically involves collecting samples from a 4.5 jim inlet onto white filter paper
17 where blackness of the stain is measured by light absorption. COH is determined using a light
18 transmittance method. This involves collecting samples from a 5.0 jim inlet onto filter tape
19 where the opacity of the resulting stain is determined. This technique is somewhat more
20 responsive to non-carbon particles than the BS method. Nephelometers measure the light
21 scattered by ambient aerosols in order to calculate light extinction. This method results in
22 measurements that can correlate well with the mass of fine particles below 2 jim in diameter. The
23 Aethalometer® collects particles (typically without a size-selective inlet) onto a quartz filter tape
24 where the attenuation of transmitted light or light absorption is measured. Since the mix of
25 ambient particles varies widely by location and time of year, the correlation between optical
19 When data are sent to the AIRNOW website, they are assumed to be "FRM-like" which means that their values
are highly correlated (R2 > 0.8) with actual FRM concentrations so that values can be compared not only to the FRM
measurements but also across State boundaries. Statistical adjustments to the raw continuous data may be necessary
because some of the sampling methodologies, such as the TEOM monitors, have inlets heated from 30°C to 50°C
which causes semi-volatile fine paniculate matter including nitrates to be vaporized and never measured. The result
of this vaporization is a lower measured TEOM concentration when compared to the FRM. Adjustments have been
accomplished on a seasonal basis as well as using meteorological variables (e.g., ambient temperature) with linear
and non-linear regression techniques. The need to adjust the continuous data can depend on several factors
including the type of method, the location of the site in the country and the composition of the ambient paniculate
matter being measured.
20 See section 2.2.5 for a discussion of the optical properties of PM.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-18
-------
1 measurements and PM mass is highly site- and time-specific. The optical methods described
2 here, as well as the particle counters described below, are based on the measurement of
3 properties such as light scattering and electric mobility, which are inherently different than
4 previous methods described based on aerodynamic diameter.
5 2.3.3 Size-Differentiated Particle Number Concentration Measurement Methods
6 Recently there has been increasing interest in examining the relationship between the
7 particle number concentration by size and health effects. Several instruments are needed to
8 provide size distribution measurements (number and size) over the 5 orders of magnitude of
9 particle diameters of interest. A nano-scanning mobility particle sizer (NSMPS) counts particles
10 in the 0.003 to 0.15 jim range. A standard scanning mobility particle sizer (SMPS) counts
11 particles in the 0.01 to 1 jim range, and a laser particle counter (LPC) counts particles in the 0.1
12 to 2 |im range. An aerodynamic particle sizer measures particles in the 0.7 to 10 |im range.
13 These techniques, while widely used in aerosol research, have not yet been widely used in health
14 effects studies.
15 2.3.4 Chemical Composition Measurement Methods
16 There are a variety of methods used to identify and describe the characteristic
17 components of ambient PM.21 X-ray fluorescence (XRF) is a commonly used laboratory
18 technique for analyzing the elemental composition of primary particles deposited on filters. Wet
19 chemical analysis methods, such as ion chromatography (1C) and automated colorimetry (AC),
20 are used to measure ions such as nitrate (N(V), sulfate (SO/f), chloride (Cl"), ammonium
21 (NH4+), sodium (Na+), organic cations (such as acetate), and phosphate (PC>43"). There are several
22 thermal-optical methods for separating organic carbon (OC) and elemental carbon (EC) in
23 ambient filter samples. Thermal-optical reflectance (TOR), thermal manganese oxidation
24 (TMO), and thermal-optical transmittance (TOT) have been commonly applied in aerosol studies
25 in the United States.
26 For EPA's CSN and IMPROVE monitoring networks, energy-dispersive X-ray
27 fluorescence (EDXRF) is used for characterizing elements; 1C is used for ions; and thermal-
28 optical methods are used for OC and EC. The TOT method, which has been used primarily in the
29 CSN, uses a different thermal temperature profile and optical correction for pyrolyis than TOR,
30 which is used in the IMPROVE monitoring program. The two analysis methods yield
21 See Chapter 3, section 3.4, of the second draft ISA for a more thorough discussion of sampling and analytical
techniques for measuring PM. Methods used in EPA's National PM2 5 Chemical Speciation Network and other
special monitoring programs are summarized in Solomon et al. (2001).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-19
-------
1 comparable estimates of total carbon, but they give a different split between OC and EC when
2 analyzing filters from the same sampling device. For consistency in OC and EC measurements
3 across the U.S., CSN began transitioning the sampling and analysis methods to IMPROVE-like
4 samplers and the IMPROVE_A analysis method. The transition began in May 2007 and will be
5 completed in October 2009.
6 Commercial instruments are now available to measure elements, carbon (OC and EC),
7 and sulfate on a semi-continuous basis. These instruments provide time-resolved measurements
8 from a few minutes to a few hours. The semi-continuous methods involve a variety of
9 techniques that include particle collection on a filter tape with on-board XRF analysis; thermal
10 reduction; and thermal oxidation with non-dispersive infrared (NDIR) detection. Most of these
11 techniques have been field tested and compared through the EPA's Environmental Technology
12 Verification (ETV) program and the Supersites program (EPA, 2004a). Data are now becoming
13 available from regional planning and multi-state organizations and the EPA to understand the
14 comparison with filter-based methods and the potential limitations of these technologies. Proven
15 semi-continuous monitors will become the framework for a long-term network of up to 12 CSN
16 sites equipped with semi-continuous monitors.
17 2.3.5 Measurement Issues
18 There is no perfect PM sampler under all conditions, so there are uncertainties between
19 the mass and composition collected and measured by a sampler and the mass and composition of
20 material that exists as suspended PM in ambient air. To date, few standard reference materials
21 exist to estimate the accuracy of measured PM mass and chemical composition relative to what
22 is found in air. At best, uncertainty is estimated based on collocated precision and comparability
23 or equivalency to other similar methods, which themselves have unknown uncertainty, or to the
24 FRM, which is defined for regulatory purposes but is not a standard in the classical sense. There
25 are a number of measurement-related issues that can result in positive or negative measurement
26 artifacts which could affect the associations that epidemiologic researchers find between ambient
27 particles and health effects.
28 The semi-volatile components of PM can create both positive and negative measurement
29 artifacts. Negative artifacts arise from evaporation of the semi-volatile components of PM
30 during or after collection, which is caused by changes in temperature, relative humidity, or
31 aerosol composition, or due to the pressure drop created as collected air moves across the filter.
32 It has long been known that FRMs are subject to negative sampling artifacts or the loss of semi-
33 volatile components of PM (e.g., ammonium nitrate and some organics). In comparison with
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-20
-------
1 other sampling techniques that can measure both semi-volatile and nonvolatile PM, FRMs
2 reported PM2.5 or PMi0 mass concentrations biased low by 10-30%. The bias of the FRMs
3 depends on the components of ambient PM and the sampling conditions (e.g., ambient
4 temperature and relative humidity), which vary from day to day and from season to season (US
5 EPA, 2009a, section 3.x, p. 3-28). Positive artifacts arise when gas-phase organic compounds
6 absorb onto or react with filter media or already collected PM or when particle-bound water is
7 not removed. The chemical interaction of gases being collected with particles already on the
8 filter and conversion of PM components to gas-phase chemicals can also result in negative
9 artifacts. These interactions depend on the compounds contained in collected particles and in the
10 gas phase and also depend on both location and time. Despite these issues, the precision of the
11 FRMs are quite high, and the method bias based on the performance audit program is well within
12 the goal.
13 Particle-bound water is an important component of ambient PM mass (ISA; p. 3-30). It
14 can also represent a substantial fraction of gravimetric mass at normal equilibrium conditions
15 (i.e., 22° C, 35 % RH) when the aerosol has high sulfate content. The amount of particle-bound
16 water will vary with the composition of particles, as discussed in the Provisional Assessment of
17 Recent Studies on Health Effects of PM Exposure (US EPA, 2006). The use of heated inlets to
18 remove particle-bound water (e.g. TEOM at 50° C) can result in loss of semi-volatile compounds
19 unless corrective techniques are applied, although the newer generation TEOMs use diffusion
20 dryers rather than heating to reduce the relative humidity (US EPA, 2009a, p. 2-100, Table 2-7).
21 2.4 AMBIENT PM MONITORING NETWORKS
22 The measurement of ambient air pollution in the United States is provided through a
23 number of ambient air monitoring networks operated almost exclusively by state, local, and
24 tribal air monitoring programs. This section briefly describes the network design criteria for
25 each of the PM monitoring networks, what changes may be useful to improve the networks to
26 best meet monitoring objectives, and what the existing network coverage is across the country.
27 The ambient air monitoring networks are designed to meet three basic monitoring
28 objectives. Each objective is important and must be considered individually. The objectives are:
29 • to support compliance with ambient air quality standards and emissions strategy
30 development, including comparison to the NAAQS, assess ambient exposures,
31 development of attainment and maintenance plans, evaluation of regional air quality
32 models used in developing emission strategies, and tracking trends in air pollution
33 abatement control measures' impact on improving air quality;
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-21
-------
1 • to provide air pollution data to the general public in a timely manner such as reporting
2 the Air Quality Index (AQI) through AIRNow (www. airnow. gov), monitoring agency
3 web sites, conventional media outlets such as newspapers, radio and television news,
4 and emerging outlets such as social networking sites; and
5 • to support air pollution research studies, including atmospheric, health, and
6 epidemiological studies that are used to inform future reviews of the NAAQS.
7
8 The sections below briefly summarize the monitoring networks for PM2 5, including
9 PM2.5 speciation, PMio, PMio-2.5, and the forthcoming National Core (NCore) multi-pollutant
10 network.
11 2.4.1 PM2.5
12 The PM2.5 monitoring requirements provide for monitors in MS As based on a
13 combination of population and design value (Table D-5, 40 CFR Part 58) with higher populated
14 locations having more polluted air required to have the most monitors. Background and
15 transport monitors are also required of each state with options for utilizing IMPROVE and other
16 PM2.5 data to provide for flexibility in meeting the requirement.
17 In urban areas, required PM2 5 monitoring stations are sited to represent community-wide
18 air quality. These monitoring stations will typically be at neighborhood or urban scale; however,
19 where a population-oriented micro- or middle-scale PM2.5 monitoring station represents many
20 such locations throughout a metropolitan area, these smaller scales can be approved by the
21 applicable EPA Regional Office to also represent community-wide air quality. The EPA's
22 existing network design criteria for PM2.5 states: "(1) at least one monitoring station is to be sited
23 in a population-oriented area of expected maximum concentration and (2) for areas with more
24 than one required SLAMS, a monitoring station is to be sited in an area of poor air quality" (40
25 CFR, PART 58, Appendix D, section 4.7). Since monitors sited for either of these network
26 design criteria must represent community-wide air quality, they are also representative of
27 population exposure. The most important spatial scale to effectively characterize the emissions
28 of particulate matter from both mobile and stationary sources is the neighborhood scale for
29 PM2.s. For purposes of establishing monitoring sites to represent large homogenous areas other
30 than the above scales of representativeness and to characterize regional transport, urban or
31 regional scale sites would also be needed. Most PM2 5 monitoring in urban areas should be
32 representative of a neighborhood scale.
33 The PM2 5 network design criteria were useful for establishing the network 10 years ago
34 when the focus of the monitoring was on community-wide exposures, and the driver for attaining
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-22
-------
1 the NAAQS was the annual average; however, two important things have changed since that
2 time. First, with 10 years of data, we have a much better understanding of the spatial and
3 temporal characterization of PM2.5. Given the regional consistency of PM2.5, especially in areas
4 with relatively flat topography and uniform emission across the area, we can expect ambient
5 exposures to be relatively consistent at neighborhood and urban scales for many, but not all
6 areas. This validates devoting some portion of the required network to characterizing
7 community-wide air quality in a neighborhood or urban scale of representation where we expect
8 to find the expected maximum concentration. However, in some locations within urban areas,
9 concentrations of PM2.5 may differ from community-wide monitors due to nearby primary
10 emission sources or terrain. The second major change occurred in 2006 when the 24-hour PM2 5
11 NAAQS was finalized. This more protective daily NAAQS has led to some areas where the 24-
12 hour standard is the controlling standard for an area. To better support protection with the daily
13 PM2.5 NAAQS, a portion of the required network could be targeted to monitor short-term PM2.5
14 levels in areas that are not necessarily affirmatively designated as "community-wide" air quality.
15 Requiring monitors in areas not designated as "community-wide" would ensure protection for
16 those populations whose ambient exposures may be affected by unique local sources or terrain
17 which are often located in areas with smaller spatial scales. These areas, such as population-
18 oriented micro- and middle- scale "hot spots," are not required to be monitored in the current
19 network design.
20 In rural areas, EPA's monitoring strategy relies on IMPROVE, rural NCore stations, a
21 limited number of smaller cities, and partner monitoring agencies to provide for regional
22 characterization of PM2.5. Stations in these areas are typically sited to represent regional scale air
23 quality and are therefore located away from any local sources, should they exist.
24 2.4.1.1 PM2.5FRM Network
25 The network of PM2.5 FRMs has been operational since 1999. This network includes
26 over 900 monitoring stations throughout the country. The number of PM2.5 FRMs may decrease
27 over the coming years as PM2.s continuous FEMs are now available and can replace FRMs
28 without the loss of a data record. Figure 2-2 illustrates the locations of PM2.5 FRMs reporting to
29 the Air Quality System (AQS).
30 2.4.1.2 PM2.s Continuous Monitor Network
31 Continuous PM2.s monitors are required in metropolitan Statistical Areas (MSAs) at one
32 half (rounded up) the number of monitoring stations that are required to have an FRM/FEM
33 monitor. Since most deployed PM2.s continuous monitors are not approved as FEMs, many of
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-23
-------
1 these monitors are collocated at monitoring locations with an FRM so that the availability of data
2 from both instruments supports each of the major monitoring objectives described earlier in this
3 section. Collocation with PM2.5 FRMs and continuous monitors also ensures that reference data
4 are available to validate the performance of the continuous monitor. While PM2.5 continuous
5 monitors primarily support forecasting and reporting the AQI, they are also used in interpreting
6 the diurnal characterization of PM2 5. The network of PM2 5 continuous monitors has grown to
7 over 700 locations throughout the country. Figure 2-3 illustrates the locations of PM2.5
8 continuous monitors reporting to AQS, including those that are now approved as FEMs.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-24
-------
Figure 2-2. PM2.5FRMs Reporting to AQS
. ? I'AHO V,.,«~-—-.__, j 3
• 7-W • • ~~i j.
/i * i JWT
,' • * i • DAKI
--~J~._ * / iv*-OMIIG* /• *
Legend
• PM2.5FRM's
£3- Lakes
State Boundaries
PM2.5 FRM's Reporting to AQS
t: i : H l-.j.JId.--.
'i>l Hi F'Tf(*!••-:•
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-25
-------
Figure 2-3. PMi.s Continuous Monitors Reporting to AQS
Legend
o Thermo BAM
• FDMS
* TEOM
• Nephelometer
* M1 BAM Pre-FEM
A Ml BAM as FEM
£3> Lakes
State Boundaries
PM2.5 Continuous Monitors Reporting to AQS
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-26
-------
1 2.4.1.3 PM2.5 Chemical Speciation Network (CSN)
2 As part of the PM2.5 NAAQS review completed in 1997, EPA established a PM2.5
3 Chemical Speciation Network (CSN) consisting of 54 Speciation Trends Network (STN) sites.
4 The STN was established to conduct routine Speciation monitoring in primarily urban areas to
5 provide nationally consistent data for the assessment of trends and to provide a long-term record
6 of the characterization of PM2.5 in the United States. The initial STN monitoring began with a
7 pilot of 13 sites in February 2000. In addition to the STN, EPA also implemented a network of
8 about 200 supplemental Speciation sites for multiple monitoring objectives, including support for
9 development of modeling tools and the application of source apportionment modeling for control
10 strategy development in support of the NAAQS; support for health effects and exposure research
11 studies assessment of the effectiveness of emission reductions strategies through the
12 characterization of air quality; support for programs aimed at improving environmental welfare;
13 and state implementation plan (SIP) development. The STN and supplemental speciation
14 monitoring sites together are referred to as the CSN. The CSN sampling apparatus do not
15 include any FRM/FEMs; therefore, data produced from this network are not used for comparison
16 to the NAAQS. However, FRM's are almost always collocated with the CSN since these are
17 among the most important PM2.5 sites in a network.
18 In 2005, EPA conducted an assessment specifically focused on the PM2.5 speciation
19 monitoring network. In consultation with state and local monitoring agencies, EPA evaluated
20 CSN sites to determine which ones might be shut down so as to provide resources for future
21 monitoring needs. EPA ranked the sites according to their overall information value. The
22 ranking was based on several factors, including whether the site was in a non-attainment area and
23 whether other sites were nearby. There was general agreement that some of the sites should be
24 shut down when FY 2005 funding ran out. Other sites were identified as high value sites,
25 particularly with regard to the PM2.5 NAAQS program. EPA evaluated each of these sites when
26 FY 2006 regional funding allocations for continued operation and maintenance were developed.
27 In doing so, EPA balanced filter-based PM2 5 speciation against other uses of PM2 5 funding, such
28 as FRM site operations, filter analysis, and startup of additional precursor gas sites and
29 continuous speciation sites.
30 As of May 2008, the PM2.5 CSN consisted of approximately 52 STN sites and about
31 150 SLAMS supplemental sites. All STN sites operate on a one-in-three day sample collection
32 schedule. A majority of the SLAMS supplemental sites operate on a one-in-six day sample
33 collection schedule. These sites collect aerosol samples over 24 hours on filters that are analyzed
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-27
-------
1 for PM2.5 mass, trace elements (Al through Pb), major ions (sulfates, nitrates, and ammonium),
2 and organic and elemental carbon fractions.
3 The IMPROVE program was established in 1985 to aid the creation of federal and state
4 implementation plans for the protection of visibility in Class 1 areas (155 national parks and
5 wilderness areas) as stipulated in the 1977 amendments to the CAA and further supports goals
6 set forth in the 1999 Regional Haze Rule.22 The IMPROVE program provides PM2 5 speciation
7 data for carbon, ions, and major elements, like the CSN. IMPROVE is a cooperative
8 measurement effort managed by a steering committee composed of representatives from federal,
9 regional, and state organizations. The IMPROVE network presently comprises 110 regionally
10 representative monitoring sites, 7 sites operated collaboratively with the Clean Air Status and
11 Trends Network (CASTNET). An additional 34 CSN sites operate according to IMPROVE
12 protocols for PM2.5 mass, elements, ions and carbon. These sites are in addition to the 200 CSN
13 sites described above and referred to as IMPROVE protocol sites. See Figure 2-4 for locations of
14 IMPROVE and CSN monitors.
15 In May of 2007, the 200 CSN sites began transit!oning to a new method of sampling and
16 analysis for carbon measurements that is consistent with the IMPROVE network methodology.
17 This transition is on-going and was split into three phases with 56 CSN stations in the first phase,
18 63 stations in the second phase, which began on April 1, 2009, and 78 stations in the third phase,
19 which is scheduled to be implemented on October 1, 2009.
20 While the network of approximately 200 CSN sites provide valuable data for
21 development and tracking of control strategies, its use for supporting epidemiological studies is
22 limited. CSN sites provide data on a one-in-three or one-in-six day schedule and do not capture
23 data every day or everywhere. In April 2008, the EPA co-sponsored a workshop to discuss
24 modifications to the current ambient air quality monitoring networks that would advance our
25 understanding of the impacts of PM exposures on public health/welfare in the most meaningful
26 way. This workshop was a major step in a series of interactions to foster improved long-term
27 communication between external stakeholders, including air quality monitoring experts and
22 Additional information is available at http://www.epa.gov/visibilitv/actions.html
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-28
-------
Figure 2-4. IMPROVE and Chemical Speciation Networks
/•• / • / 6, B
OREGON S ^
':_",' /m
Legend
A Trend sCSN
• IMPROVE and IMPROVE Protocol Stations
o Supplemental CSN
£> Lakes
State Boundaries
IMPROVE and Chemical Speciation Network
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-29
-------
1 health researchers. A summary of the workshop recommendations was published in December
2 2008.23
3 2.4.2 PMio
4 PMio monitoring stations have an urban focus and are required in MSAs to characterize
5 national and regional PMio air quality trends and geographical patterns. Local considerations are
6 also a factor in determining the actual required number of monitoring sites. More stations are
7 required in larger MSAs and MSAs with more evidence of poor air quality, while monitors are
8 also required in clean MSAs of certain size. The number of monitors in areas where MSA
9 populations exceed 1,000,000 must be in the range from 2 to 10 stations, while in low population
10 urban areas, no more than two stations are required (Table D-4 of Appendix D to CFR Part 58).
11 Because sources of air pollutants and local control efforts can vary from one part of the country
12 to another, some flexibility is allowed in selecting the actual number of stations in any one
13 locale.
14 The network of PMio monitors has been operational since 1987. The network currently
15 includes over 800 monitoring stations throughout the country with most metropolitan areas
16 operating more PMio monitors than required by current monitoring requirements. The PMio
17 monitoring stations operate manual FRMs on a mix of daily, one-in-two day, one-in-three day, or
18 one-in-six day sampling, based on the relative concentration level of the site with respect to the
19 24-hour standard. There are also FEMs that are operated continuously. PMio monitors operating
20 across the country are almost exclusively FRMs or FEMs. Figure 2-5 illustrates the locations of
21 the PMio FRMs and FEMs reporting to AQS.
22 The PMio monitoring stations are currently required to collect and report monitoring data
23 under standard temperature and pressure conditions. PM2.5 and PMio-2.5 are required to be
24 collected and reported at local conditions. Correction of the sampled aerosol volume to
25 "standard" conditions may improperly report the aerosol concentration measurement. If the
26 rationale for aerosol sampling is to mimic respiratory penetration (which occurrs at local
27 conditions), a correction to standard conditions may not be appropriate. These corrections are
28 typically small (less than a few percent) except in locations at higher altitudes and those with
23 The report is available at www.epa.gov/ORD/npd/pdfs/FINAL-April-2008-AQ-Health-Research-Workshop-
Summary -Dec -2008.pdf
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-30
-------
1
2
Figure 2-5. PMio Monitoring Network
Legend
o PM10 dichot sampler
• PM1D low-volume sampler
• PM10 continuous monitor
• PM1 D hi-volume sampler
£> Lakes
State Boundaries
PM10 Monitoring Network Reporting to AQS
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-31
-------
1 large diurnal or seasonal temperature changes (1996 CD). Low-volume PMio data used for
2 2.5 determinations must be reported at local conditions.
3 2.4.3 PMio-2.5
4 The EPA is requiring PMio-2.5 mass and speciation monitoring as part of the National
5 Core (NCore) network by January 1, 2011. The NCore network is a multiple pollutant network
6 that also supports PM monitoring objectives (see section 2.4.4 below). To date, there are about
7 30 PMio-2.5 mass monitoring stations reporting data to AQS. Monitoring stations are located in
8 Arizona, Connecticut, Iowa, North Carolina, and Oklahoma. As monitoring agencies continue to
9 implement PMio-2.5 mass monitoring as required at NCore (approximately 83 stations are
10 expected), the number of stations is expected to increase.
11 For PMio-2.5 speciation, we do not expect methods to be fully developed in time to meet
12 the January 1, 2011, start date for monitoring at NCore. The EPA has been working with the
13 CAS AC Ambient Air Monitoring and Methods Subcommittee on this issue and will be
14 implementing a pilot of PMio-2.5 speciation at two locations in 2010.24 The PMio-2.5 speciation
15 pilot monitoring project will be started to further develop and field test analytical methods to be
16 used in the long-term speciation monitoring network.
17 2.4.4 National Core (NCore)
18 The NCore network is a multi-pollutant network that includes measurements of particles,
19 gases, and meteorology (71 FR 61236, October 17, 2006). The network is intended to support
20 integrated air program management needs. The NCore monitoring network will be fully
21 operational by January 1, 2011. NCore stations will include both neighborhood and urban scale
22 measurements in general, in a selection of urban areas and a limited number of more rural
23 locations. NCore stations are intended to be long-term stations useful for a variety of
24 applications. NCore stations should be placed away from direct emissions sources that could
25 substantially impact the ability to detect area-wide concentrations. NCore has both urban and
26 rural siting components. Urban NCore stations are to be generally located at a urban or
27 neighborhood scale to provide representative concentrations of exposure expected throughout the
28 metropolitan area. Rural NCore stations are to be located to the maximum extent practicable at a
29 regional or larger scale away from any large local emission source, so that they represent
24 A Febuary 2009 consultation with the CASAC AAMM subcommittee discussed issues related to coarse particle
speciation measFor more information on consultation with see:
http://yosemite.epa.gov/sab/sabproduct.nsf/264cbl227d55e02c85257402007446a4/3494de4dOccb394485257463006
4d4e4! OpenDocument
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-32
-------
1 ambient concentrations over an extensive area. States and where applicable, local monitoring
2 agencies have submitted plans for NCore during the summer of 2009. The proposed NCore
3 stations are collocated with several other well leveraged networks such as PAMS, CSN, the
4 National Air Toxics Trends Stations (NATTS), IMPROVE, and CASTNET. The recommended
5 NCore locations identified as of September 2009 are provided in Figure 2-6.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-33
-------
1
2
Figure 2-6. Proposed NCore Monitoring Network
A^UIIA Atlaiirif
Ocean
Legend
• Rural NCore Stations
• Urban NCore Stations
S3 Lakes
State Boundaries
Proposed NCore Stations
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-34
-------
1 2.5 PM CONCENTRATIONS, TRENDS, AND SPATIAL PATTERNS
2 As described in section 2.4, EPA and the states have been using a national network to
3 measure PMio concentrations since 1987 and PM2.5 concentrations since 1999. This section
4 presents summaries through the end of 2007, based on publicly available data from EPA's Air
5 Quality System (AQS) as of August 2008.
6 The Staff Paper from the previous review (US EPA, 2005, chapter 2) and the second draft
7 ISA (US EPA, 2009a, sections 3.5 and 9.3.4) present extensive characterizations of the levels,
8 composition, and temporal and spatial patterns of PM2.5 in U.S. urban areas. Both documents
9 present data summaries based on the approximately 1100 PM2 5 and 1300 PMio monitoring sites
10 in the U.S. The characterizations of recent air quality data used in the previous review were
11 based on ambient measurements collected from 2001 through 2003. In this review, EPA is
12 considering more recent air quality data collected from 2005 through 2007. Also, Chapter 6 of
13 the 2004 PM Assessment by NARSTO contains more detailed characterizations of PM in
14 different parts of the U.S. (NARSTO, 2004). While there generally have been reductions in the
15 concentrations of PM2.5 and PMio in many areas as a result of reductions in emission of PM2.5
16 and its precursors and of PMio since these documents were written, comparison of the three
17 documents indicates that the general patterns, as well as the diversity of patterns across areas,
18 noted in the two older documents still prevailed in the 2005-2007 period.
19 2.5.1 PM2.5
20 In 2005-2007, 38 urban areas violated the current annual PM2.5 NAAQS of 15 |ig/m3,
21 adopted in 1997, which level was retained in the last review completed in 2006. Seventy-six
22 areas violated the revised 24-hour NAAQS of 35 |ig/m3. There is considerable but not complete
23 overlap in the areas not meeting the annual and 24-hour NAAQS. It should be noted that in
24 many parts of the U.S., PM2 5 concentrations in 2005 were high relative to the next three years.
25 Figure 2-7 illustrates PM2.5 air quality in 2005, 2006, and 2007 by representing each
26 monitor by a symbol whose color reflects the annual mean of the concentration at that site or the
27 98th percentile of 24-hour concentrations, in both cases in a single year (US EPA, 2009a, section
28 3.5.1.1 presents maps for PM2.s air quality concentrations for the 2005-2007 period as a whole).
29 County-scale, 24-hour average concentration data for PM2 5 during 2005 through 2007 showed
30 considerable variability across the U.S. The highest reported 3-year average concentrations were
31 reported for six counties within the San Joaquin Valley and inland southern California, as well as
32 Jefferson County, AL, (containing Birmingham) and Allegheny County, PA, (containing
33 Pittsburgh). The lowest reported annual average PM2.5 concentrations occurred within 237
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-35
-------
1 counties distributed throughout many western and northern states as well as Florida and the
2 Carolinas. Of the 15 individual CSAs/CBSAs selected for detailed investigation in the second
3 draft ISA based on their geographic distribution and importance in recent health effect studies,25
4 the highest unweighted 2005-2007 mean 24-hour PM2.5 concentrations were reported for
5 Riverside (17 |ig/m3), Birmingham (16 |ig/m3), and Pittsburgh (16 |ig/m3); the lowest were
6 reported for Denver (9 |ig/m3) and Seattle (9 |ig/m3).
7 Figure 2-8 and Figure 2-9 show trend information for the annual and 24-hour PM2.5
8 NAAQS. The seasonally weighted 3-year annual average concentration fell 10 percent between
9 the 1999-2001 averaging period and the 2005-2007 averaging period. The number of sites
10 reporting concentrations above the annual average PM2.5 NAAQS fell 56 percent over these
11 same periods. Declines were the greatest in EPA Region 9 where PM2.5 concentrations fell 20
12 percent from the 1999-2001 averaging period to the 2005-2007 averaging period.
13 The second draft ISA states that in the 2004 Critiera Document, EPA concluded that that
14 PM2.5 measured in eastern cities was generally more highly correlated across monitoring sites
15 than in western cities. The higher spatial correlations in the eastern cities resulted from the more
16 regionally dispersed sources of PM2.5 in the East, and the high contribution of more regionally
17 homogeneous species (e.g. sulfates) to total PM2.5 throughout the area. Although PM2.5
18 concentrations at sites within an urban area can be highly correlated resulting from the
19 homogeneity of such components over broad regions, significant differences in concentrations
20 can occur on any given day resulting from terrain features and varying contributions of urban
21 and local sources.
22 Comparison of PM2.5 mass and PM2.5 species concentrations within and outside urban
23 areas leads to the conclusion that in the eastern areas with high sulfate concentrations, the large
24 majority of the sulfate and, thus, also a large fraction of the PM2.5 affecting any given urban area
25 originates outside that area. Inward transport and local generation of nitrate and carbonaceous
25 Fifteen metropolitan regions were chosen by EPA for closer investigation of monitor siting based on their
distribution across the nation and relevance to health studies analyzed in the second draft ISA. These regions were
Atlanta, Birmingham, Boston, Chicago, Denver, Detroit, Houston, Los Angeles, New York City, Philadelphia,
Phoenix, Pittsburgh, Riverside, Seattle, and St. Louis. Core-Based Statistical Areas (CBSAs) and Combined
Statistical Areas (CSAs), as defined by the U.S. Census Bureau were used to determine which counties, and, hence,
which monitors, to include for each metropolitan region (US EPA, 2009a, section 3.4.2.2)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-36
-------
1 Figure 2-7. Annual average and 24-hour (98 percentile 24-hour concentrations)
2 concentrations in ug/m3, 2005, 2006, and 2007.
Annual Average
24-hour
2005
(graphic to be added to external review
draft PA}
(graphic to be added to external review
draft PA}
2006
Concentration Range (fig•''
O 0-12(496 Sites)
* 121 -15<333S«*S>
O 15.1 - IB (57 Sites)
• 18 1 -21 (9 Sites)
2007
Annual
Concentration Range (pg/"m!)
* :-i it -•-:,-
O 12.1 - 15.0 (356 Steal
O 15 1 - 1B 0 (SB Sites)
• 16 1 • 22 5 (14 Sites)
2-1-hour
Concentration Range (pg ''m')
9 1 lSt38S*cs}
O 1C - IS (0(2 Sites*
O 36-S5 (168 Sites)
• :'. . ij- •--,
<0
;:,:.,-,- «„-,-,
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-37
-------
Figure 2-8. Ambient annual PM2.s concentrations in the U.S., 1999-2007,showing A)
ambient concentrations, B) number of trends sites above the annual NAAQS and C) trends
by U.S. EPA Region.
4
5
A) Ambient Concentrations
C) Trends by EPA Region3
90% of sites have
concentrations below this line
NAAQS = 35 pg/m:
10% of sites have
concentrations below this line
'99-'01 '00-'02 '01-'03 '02-'04 '03-'05 '04-'06 '05-'07
Averaging period
B) Number of Trend Sites Above NAAQS
a
'99-'01 '00-'02 '01-'03 '02-'04 '03-'05 'Q4-'06 '05-'07
Averaging period
Source: US EPA, 2009a, Figure 3-45
» E
0 CT
'•*- 2.
1^45
O P
c >>
8^
,_•§
2! 30
*r 3
S >
0 =
'•^ S> 13
| §
CU fl)
Q. 03
NAAQS = 35 ug/m3
-^=^st
>^=ii?=^==
• «^
— L- —
R1
-R2
R3
R4
~R5
R6
R7
R8
-R9
-R10
-Mat!
05 '99-'01 '00-'02 '01-'03 '02-'04 '03-'05 '04-'06 '05-W
Averaging period
Coverage: 697 monitoring sites
in the EPA Regions (out of a total
of 831 sites measuring PMj.5 in
2007) that have sufficient data to
assess PM?5 trends since 1999.
EPA Regions
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
20 o
-38
-------
Figure 2-9. Ambient 24-hour PM2.s concentrations in the U.S., 1999-2007,showing A)
ambient concentrations, B) number of trends sites above the 24-hour NAAQS and C)
trends by U.S. EPA Region.
5
6
20
A) Ambient Concentrations
15 c E
3 O ^
c u ra
C 05 3
cU as
£ t £ 10
£> a <»
«£ S 5
o S
B) °
fO C
0) O
W "
0
90% of sites have concentrations below this line
fJA_AQ_S_=1_5pg/^_
Median
A
Average
i of sites have concentrations below this line
'99-'01 '00-'02 '01-'03 '02-'04 '03-05 '04-'06 '05-'07
Averaging period
B) Number of Trend Sites Above NAAQS
'99-'01 '00-'02 '01-'03 '02-'04 '03-'05 704-'06 '05-'07
Source: US EPA, 2009a, Figure 3-44
20
C) Trends by EPA Region3
<0 (1)
1 1
« S^15
TO ~_ C
c o
C d)
2r s 10
IsS 5
CO
NAAQS = 15 pg/m3
-R1
R2
-R3
R4
Rb
RG
R7
R8
-R9
'99-'01 'G0-'02 '01-'03 '02-'fl4 '03-'05 '04-'06 '05-'07
Averaging period
"Coverage: 697 monitoring
sites in the EPA Regions (out
of a total of 802 sites
measuring PMj5 in 2007) that
have sufficient data to assess
PM2,5 trends since 1999.
EPA Regions
(D oJD
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-39
-------
1 material are more evenly balanced in eastern areas, with some differences among areas. In
2 western areas, local sources dominate for carbonaceous material and nitrate, with both local and
3 regional sources contributing to the small sulfate component (US EPA, 2009a, Figure 9-24).
4 Southeastern areas have their highest PM2.5 concentrations in the summer, when
5 conditions are most conducive to sulfate formation. More northern areas, being affected by a
6 more balanced mix of contributors, tend not to have such a strongly seasonal pattern. The
7 seasonal patterns in western areas are individual and varied, related to differences in local
8 sources and formation and dispersion conditions. In all areas, inversion conditions with low
9 wind speeds are conducive to high concentrations due to the trapping of emissions from local
10 sources. Some western areas, especially those with bowl-like topography, are especially
1 1 affected.
12 2.5.2
13 Figure 2-10 illustrates PMio air quality in 2005, 2006, and 2007 by representing each
14 monitor by a symbol whose color reflects the 98th percentile 24-hour concentration, in one year.
15 (See also US EPA, 2009a, Section 3.5.1.1 which presents maps for PMio air quality
16 concentrations for the 2005-2007 period as a whole.) Most areas of the country had
17 concentrations below the level of the 24-hour PMio standard of 150 |ig/m3, with exceptions
18 concentrated in the southwestern U.S. and isolated counties scattered across the east. The highest
19 reported 3 -year average PMio concentrations (greater than 5 1 |ig/m3) occurred in two counties in
20 southern California and five counties in southern Arizona and central New Mexico. The lowest
21 reported annual average PMio concentrations (less than or equal to 20 |ig/m3) were within 114
22 counties distributed fairly uniformly across the U.S. Of the 15 CSAs/CBSAs investigated, the
23 highest mean 24-hour PMio concentrations was reported for Phoenix (52 |ig/m3), considerably
24 higher than the means for the other CSAs/CBSAs investigated. The lowest was reported for
25 Boston (17 |ig/m3) with New York, Philadelphia and Seattle only slightly higher (19 |ig/m3).
26 Figure 2-11 shows trends in U.S. ambient 24-hour PMio concentrations from 1988 to
27 2007. In 2007, the U.S. national average second highest PMio concentration was 37% lower than
28 in 1988. Of 281 sites used in this long-term trend analysis, the number reporting concentrations
29 above the 24-hour PMioNAAQS (150 |ig/m3) fell from 23 in 1988 to 5 in 2007 with a max of 29
30 in 1989. All regions exhibited an overall decrease from 1988 to 2007. The largest decreases
3 1 occurred in EPA Region 10, which incorporates Washington, Oregon, Idaho, and Alaska. Most
32 of the decrease occurred between 1988 and 1995. (US EPA, 2009a, section 3.5.2.1)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-40
-------
1
2
3
Figure 2-10. 24-hour PMio concentrations in ug/m3, 2005- 2007
»th
(98 percentile 24-hour concentrations).
2005
(graphic to be added in external review draft PA}
2006
Concentration Range [\t gf
O 3- :•-• -~? = '^ =
* 55-154 559 Steal
O 15S-255 (21 SUES;
* >255 :125KS:
2007
Concentration Range (pg/i
• 2 • $4 (424 Sites)
O 55-154 (424 Sites)
O 155-255RQSte5)
* 256-2736 (12 Sites)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-41
-------
Figure 2-11. Ambient 24-hour PMio concentrations in the U.S., 1988-2007,showing A)
ambient concentrations, B) number of trends sites above the 24-hour NAAQS and C)
trends by U.S. EPA Region. (US EPA, 2009a, Figure 3-46)
175
i. 150
It 125
A) Ambient Concentrations
0> Q
100
-=£75
E§
S R 50
NAAQS = 1_50jg/rri3
\9Q% of sites have concentrations below this line
Average
A
Median
10% of sites have
concentrations below this line
'92 '94 '96 '98 '00 '02 '04 '05
Year
180
160
-140
a
^100
C) Trends by EPA Region3
'92 '94 '96 '98 '00 '02 '04 '06
Year
-R1
R2
R3
R4
-R5
R6
R7
R8
-R9
-R10
-Nat'!
~- ^ rtp
lumber of trend sites with second highes
4-hour concentration above the level of th
NAAQS (150 ug/m3)
3CJlQC?ioCjnOC
lll.l...l...l .11
Coverage: 2/4 monitoring sites
in the EPA Regions (out of a total
of 879 sites measuring PM^ in
2007) that have sufficient data to
assess PMio trends since 1988,
EPA Regions
0
«
©0
'88 '90 '92 '94 '96 '98 '00 '02 '04 '06
Year
Source: U.S. EPA(2008)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-42
-------
1 Being the sum of PM2.5 and PMio-2.5, PMio trends and patterns are a superimposition of
2 those for PM2 5 and PMi0-2.5, which can be quite different due to their different source/formation
3 and transport/deposition characteristics. Broadly speaking, PM2.s-like temporal and spatial
4 patterns dominate where local PMio-2.5 emissions are relatively low, such as in areas with wetter
5 climates, year-round ground cover, and an absence of large industrial sources of PMio-2.5
6 emissions; PMi0-2.5-like temporal and spatial patterns are observed in other areas, especially dry
7 areas in the west and southwest.
8 2.5.3 PMio-2.5
9 There is, at present, no systematic monitoring network in place for PMio-2.5. As outlined
10 in section 2.4.3, states have until January 1, 2011, to implement required monitoring sites for
11 PMio-2.5- Consequently, estimates of PMio-2.5 must be developed using data from PM2.5 and PMio
12 monitoring sites and equipment. In some reviews, estimates have been offered based on PM2 5
13 and PMio monitors which were within some distance of each other but not actually collocated.
14 Even when collocated, the instruments in a PM2 5 and PMio pair may not have consistent flow
15 rates, measurement principles (e.g., gravimetric filter vs. beta attenuation), filter preconditioning
16 and weighing protocols, etc.
17 In the Staff Paper for the last review, EPA used a data-inclusive approach in which we
18 used the best available data on PM2 5 and PMio concentrations in each given area - in some cases
19 not very robust data26 - to estimate 2001-2003 PMi0-2.5 concentrations for 351 counties (US
20 EPA, 205, section 2.4.3). Since the operating protocol for each monitor is not usually identical,
21 the consistency of these PMi0.2.5 measurements was relatively uncertain, and they were referred
22 to as "estimates." As reported in the last review, 98th percentile 24-hour average PMi0.2.5
23 estimated concentrations ranged from about 5 to 208 |ig/m3, with a median of about 28 |ig/m3
24 (US EPA, 2005, section 2.4.3)
25 The second draft ISA presents PMio-2.s concentration estimates in Figure 3-8 and Table 3-
26 13 of section 3.5.1.1. The second ERD ISA used a much more data-restrictive approach than the
27 Staff Paper for the last review, using only on paired (collocated) low-volume filter-based
28 samplers for both PMio and PM2 5, the method ultimately established as the basis for the Federal
29 Reference Method for measuring PMio-2.5- Only 40 counties met these requirements for paired
30 samplers. Using these available co-located PM measurements from 2005-2007, the mean 24-
26 Only in some cases was it possible to calculate PM10.2 5 concentrations based on collocated low-flow filter-based
sampler. Other data and approaches included pairing low-volume and high-volume samplers, pairing samplers at
different monitoring sites, and applying regional PM2 5 to PM10 ratios to either PM2 5 or PM10 depending on which
was available.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-43
-------
1 hour PMio-2.5 concentration in these 40 counties was 13 ug/m3. Ninety-ninth percentile values of
2 PMio-2.5 ranged from 25 to 67 ug/m3 depending on area, season, and year. Drier western areas
3 such as Phoenix and Denver tended to have the highest concentrations of PMio-2.5.
4 The IMPROVE monitoring network measures both PM2.5 and PMio and thus provides
5 PMio-2.5 data for rural areas. The PMio-2.5 levels in the relatively remote non-urban IMPROVE
6 sites were notably lower than those found in most urban areas. In all the metro areas examined
7 by comparing urban monitoring sites with nearby non-urban IMPROVE sites, the urban PMio-2.5
8 concentrations exceed those in the nearby rural locations (US EPA, 2005, section 2.4.3).
9 Both the US EPA, 2005 (section 2.7.2) and the second draft ISA (section 3.5.1.2)
10 examined the spatial variability of PMi0.2.5 across and within cities. Together, they indicated
11 that spatial distribution of PMio-2.5 is more heterogeneous than PM2.5.
12 2.5.4 Ultrafine Particles
13 The second draft ISA (US EPA, 2009a, sections 3.5.1 and 3.5.2) contains a review of the
14 current scientific information on ultrafme particles. Because there is no national network of
15 ultrafme particle samplers, only episodic and/or site-specific data sets exist. Therefore, a national
16 characterization of concentrations, temporal and spatial patterns, and trends is not possible.
17 Generally speaking, it can be said that particle number concentrations of ultrafme particles can
18 be very high close to sources of the same, but mass concentrations of ultrafme particles are
19 usually lower than background concentrations of larger particles within the PM2.s size range.
20 The number concentration falls off sharply downwind, as ultrafme particles evaporate due to
21 more mixing or accumulate into larger particles. Internal combustion engines and therefore
22 roadways are a notable source of ultrafme particles, so concentrations of ultrafme particles near
23 roadways can generally be expected to be elevated.
24 2.5.5 PM Components
25 2.5.5.1 PM2.5
26 The 2005 Staff Paper and the second draft ISA extensively summarize in considerable
27 detail 2001-2003 and 2005-2007 data, respectively, on PM2.5 composition from the rural
28 IMPROVE network and from the urban Chemical Speciation Network (CSN) operated by EPA
29 and state/local partners. The summary here mostly omits repeating specific numerical details
30 presented in those other documents (US EPA, 2005, section 2.4.5; US EPA, 2009a, sections
31 3.5.1 and 3.5.2). In general, PM2.5 in the eastern U.S. regions is dominated by sulfates and
32 carbonaceous mass. Midwestern, southeastern, and eastern urban areas have much higher PM2.s
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-44
-------
1 sulfate levels than do more western areas, attributable to the much higher emissions of 862 in
2 and upwind of the former areas. Upper midwest areas and, to a lesser extent, upper eastern areas
3 have notable nitrate concentrations in winter but not in summer, while southeastern areas
4 generally lack notable nitrate even in winter. Many western urban areas have notable nitrate year
5 round. In all areas, carbonaceous material is an important component of PM2.5 and is attributable
6 to many emission sources of organic material in PM form and of organic PM precursor gases.
7 PM2.5 at the western U.S. urban sites has a greater proportion of carbonaceous mass; in some
8 areas with high local use of wood for residential heating carbonaceous material is the dominant
9 component during the heating season. PM2.5 derived from crustal sources is generally a small
10 fraction of total mass, except during local high wind events or due to periods of intercontinental
11 transport of dust from Africa or Asia. While PM2.5 mass and all component concentrations are
12 higher in urban areas than at IMPROVE sites, in general, nitrates and carbonaceous components
13 appear to have a greater urban/rural enhancement as compared to sulfates.
14 Figure 2-12 through Figure 2-16 contain U.S. concentration maps for OC, EC, SO42 ,
15 MV, and NH4+ mass from PM2.5 measurements taken as part of the CSN network for the period
16 2005-2007. Data used in these figures are as reported to AQS: no correction was applied to OC
17 for non-carbon mass, and NOs" represents total particulate NOs". These figures show regional
18 and seasonal differences in measurements of PM2.5 components. Figure 2-12 shows regions of
19 high PM2.5 OC mass concentration with annual average concentrations greater than 5 |ig/m3 in
20 the western and the southeastern U.S. Concentrations at the western monitors peak in the fall
21 and winter while those in the Southeast peak anywhere from spring through fall. The central and
22 northeastern portions of the U.S. generally contain lower measured OC (US EPA, 2009a, section
23 3.5.7.7).
24 Figure 2-13 contains a similar map for PM2.5 EC mass concentration that exhibits smaller
25 seasonal variability than OC, particularly in the eastern half of the U.S. There are isolated
26 monitors spread throughout the country that measure high annual average EC concentrations.
27 These EC "hot spots" are primarily associated with larger metropolitan areas such as Los
28 Angeles, Pittsburgh, and New York, but El Paso, TX, also reported high annual average EC
29 concentrations (driven by a wintertime average concentration greater than 2 |ig/m3). El Paso is
30 an unusual case for its own size because of its proximity to Juarez, Mexico, a much larger city
31 with generally higher ambient PM concentrations. (US EPA, 2009a, section 3.5.1.1)
32 Figure 2-14 contains a map for PM2.5 SO42 mass concentration which shows that SO42 is
33 more prevalent in the eastern U.S. owing to the strong west-to-east gradient in SO2 emissions.
34 This gradient is magnified in the summer months when more sunlight is available for
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-45
-------
1 photochemical formation of SO42 . In contrast, PM2.5 NOs mass concentration in Figure 2-15 is
2 highest in the west, particularly in California. There are also elevated concentrations of NO3 in
3 the upper midwest. The seasonal plots show generally higher NO3 in the wintertime as a result
4 of temperature driven partitioning. Exceptions exist in Los Angeles and Riverside where high
5 NO3 readings appear year-round, (second draft ISA, section 3.5.1.7)
6 The PM2 5 NH4+ mass concentration maps in Figure 2-16 show spatial patterns related to
7 both SO42 and NO3 resulting from its presence in both (NH4)2SO4 and NH4NO3 (US EPA.
8 2009a. section 3.5.1.7).
9 No maps for PM2.5 metals are repeated here, but the second draft ISA contains such maps
10 in Annex A and it reported that there is considerably less seasonal variation in the concentration
11 profile for PM2.5 metals than for OC or the ions (US EPA, 2009a, section 3.5.1.1). This is
12 understandable in light of the less seasonal nature of emission sources for metals and the absence
13 of seasonally dependent chemical transformations.
14 In addition to providing the national maps in the figures provided, which are based
15 directly on the measurements made by the CSN network, the second draft ISA also reported the
16 results of an analysis to show the relative contributions of PM2 5 species to PM2 5 mass as
17 measured in the PM2 5 FRM network, taking into account measurement system. Figure 2-17
18 (copied from Figure 3-17 in section 3.5.1.1 of the second draft ISA) shows the resulting PM2.s
19 compositional breakdown for the fifteen CS As/CBS As identified earlier. For the fifteen
20 metropolitan areas, the contribution of the major component classes to total PM2.5 mass as
21 measured by the FRM was derived using the measured sulfate, adjusted NO3 , derived water,
22 inferred carbonaceous mass approach (SANDWICH) (Frank, 2006). This approach uses the
23 measured FRM PM2.5 mass and co-located CSN chemical constituents to perform a mass
24 balance-based estimation of the PM2.s mass fraction attributed to SO42 , NO3 , EC, OCM, and
25 crustal material. The calculations and assumptions that are used in the SANDWICH method are
26 discussed in detail in Frank (2006).27
27 On an annual average basis, SO42 is a dominant PM2 5 component in the eastern U.S.
28 cities. For the presented cities, this includes everything east of Houston. Organic carbon mass is
27 See additional information available on EPA's AirExplorer website:
(http://www.epa.gov/cgi-bin/htmSQL/mxplorer/query spe.hsql
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-46
-------
1 Figure 2-12. Three-year average 24-hour PM2.s OC concentrations measured at CSN sites
2 across the U.S., 2005-2007.
oc
Annual
• • •
Concentration (ng/m3): <1.25 >1.25-2.5C >2.50-3.7G >3.75 - 5.00 >5.00
Spring Summer
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-47
-------
Figure 2-13 Three-year average 24-hour PM2.s EC concentrations measured at CSN sites
across the U.S., 2005-2007.
o • •
Concentration (ng/m3): <0.5 >0.5-.1.C >1.0-1.5 .--1.5-2.0 >2.0
Summer
Fall
Winter
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-48
-------
1
2
2-
Figure 2-14. Three-year average 24-hour PM2.s SC>4 concentrations measured at
CSN sites across the U.S., 2005-2007.
SO/
Annual
Concentration (ng/m3): <2
Spring
- 4
>4 - 6
>6-8
>8
:
Fall
Winter
3
4
-.„'
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-49
-------
1 Figure 2-15 Three-year average 24-hour PM2.s NOs concentrations measured at CSN sites
2 across the U.S., 2005-2007.
N03
Annual
Concentration (ng/m3): <1
Spring
o
>1-2
o
>2-3
>3 - 4 >4
Summer
Fall
Winter
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-50
-------
1 Figure 2-16 Three-year average 24-hour PM2.s NH4+ concentrations measured at CSN sites
2 across the U.S., 2005-2007.
NH4+
Annual
Concentration (ng/m3): <1
Spring
3
4
Fall
>2-3 . >3-4 >A
Summer
Winter
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-51
-------
1 the next largest component in the East. In the West, OCM is the largest constituent on an annual
2 basis. SO42 , NO3 , and crustal material are also important in many of the included western cities.
3 In the West, fractional SC>42 ranges from 18% in Denver to 32% in Los Angeles while fractional
4 NOs is relatively large in Riverside (22%), Los Angeles (19%) and Denver (15%) and less
5 important on an annual basis in Phoenix (1%) and Seattle (2%). Crustal material is particularly
6 prevalent in Phoenix (28%). Elemental carbon makes up a smaller fraction of the PM2 5 mass (4
7 to 11%), but it is consistently present in all included cities regardless of region (US EPA, 2009a,
8 section 3.5.1.1).
9 The seasonal variation in PM2.5 composition across the fifteen CSAs/CBSAs is shown in
10 Figure 2-18. SC>42 dominates in most metropolitan areas in the summertime, while NOs
11 becomes important in the colder wintertime months. Notable summertime exceptions include
12 Denver, Phoenix, and Seattle, where SC>42 makes up a smaller fraction of the PM2.5 mass.
13 Likewise, NOs is less pronounced in the wintertime in Atlanta, Birmingham, Houston, Phoenix,
14 and Seattle. Los Angeles and Riverside exhibit elevated NOs from fall through spring. Crustal
15 material is a substantial summertime component in Houston (26%), much of which is attributed
16 by the state air agency to transport from Africa.28 Crustal material is generally low elsewhere in
17 the East in all seasons. In the West, crustal material represents a substantial component
18 year-round in Phoenix and Denver. (US EPA, 2009a, section 3.5.1.1)
28 See, for example, "South, Central, and Southeast Texas Saharan Dust August 27-29, 2006",
http://www.tceq.state.tx.us/assets/public/compliance/monops/air/sigevents/06/event2006-08-27txs.html)
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-52
-------
Figure 2-17. Three-year average PMi.s speciation estimates for 2005-2007 derived using the SANDWICH method for fifteen
CSAs/CBSAs.
Sulfate
Nitrate
EC
OCM
Crustal
3 Figure 2-17 notes: The fifteen areas evaluated (with the number of sites per CSA/CBSA listed in parenthesis) included: Atlanta, GA (1); Birmingham, AL (3);
4 Boston, MA (4); Chicago, IL (7); Denver, CO (2); Detroit, MI (4); Houston, TX (1); Los Angeles, CA (1); New York City, NY (7); Philadelphia, PA (6); Phoenix,
5 AZ (2); Pittsburgh, PA (4); Riverside, CA (1); Seattle, WA (4); and St. Louis, MO (3). SO42~ and NO3" estimates include NH4+ and particle bound water and the
6 circles are scaled in proportion to FRM PM2 5 mass as indicated in the legend.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-53
-------
Figure 2-18. Seasonally-stratified three-year average PM2.s speciation estimates for 2005-2007 derived using the SANDWICH
method for fifteen CSAs/CBSAs.
Spring
Summer
1 Figure 2-18 notes: The following fifteen CSAs/CBSAs were evaluated: Atlanta, GA; Birmingham, AL; Boston, MA; Chicago, IL; Denver, CO; Detroit, MI;
2 Houston, TX; Los Angeles, CA; New York City, NY; Philadelphia, PA; Phoenix, AZ; Pittsburgh, PA; Riverside, CA; Seattle, WA; and St. Louis, MO. SO42~ and
3 NO3" estimates include NH4+ and particle bound water and the circles are scaled in proportion to FRM PM2 5 mass as indicated in the legend.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-54
-------
1 2.5.5.2 PMio-2.5
2 Only speciated PM2.5 is collected routinely at CSN network sites, resulting in far less
3 information on speciated PMi0-2.5 than speciated PM2 5. Edgerton et al. (2005, 2009) published
4 speciated measurements for PM2.5 and PMio-2.5 obtained using dichotomous samplers from four
5 locations included in the Southeastern Aerosol Research and Characterization (SEARCH) study:
6 Yorkville, GA, Centreville, AL, Birmingham, AL, and Atlanta, GA. Samples were collected
7 between 1999 and 2003 on a l-in-3 day or l-in-6 day schedule, depending on site. Speciated
8 measurements for both PM2.5 and PMio-2.5 included SO42 , N(V, NIL;+, and major metal oxides
9 (MMO). In addition, OC and either black carbon (BC) or EC were reported for PM2.s over the
10 entire study period and for PMio-2.5 for a subset of samples extending from April 2003 to April
11 2004. (USEPA, 2009a, section 3.5.1.1}
12 For the Atlanta and Birmingham SEARCH sites, the annual average N(V mass fraction
13 was approximately equal for PM2.5 (5.6% and 5.0%, respectively, for Atlanta and Birmingham)
14 and PMi0-2.5 (4.9% and 3.3%). Likewise, the OC mass fraction was approximately equal for
15 PM2.5 (26% and 26%) and PMi0-2.5 (24% and 27%). MMO contributed an order of magnitude
16 smaller mass fraction to PM2.5 (2.6% and 4.7%) than PMio-2.5 (38% and 35%). In contrast, SO42
17 contributed an order of magnitude greater mass fraction to PM2.s (25.1% and 24.1%) than PMio-
18 2.5 (2.8 and 2.1%). BC also represented a slightly larger mass fraction of PM2.s (8.6% and
19 10.5%) than EC did for PMi0-2.5 (2.9% and 2.4%). Based on these findings, MMO are present
20 primarily in the thoracic coarse mode while SO42 and EC/BC are present primarily in the fine
21 mode. NOs" and OC are present in both modes in approximately equal mass fractions. These
22 results are specific to Atlanta and Birmingham and may not represent other geographic regions.
23 However, they are consistent with the current understanding of sources and formation of these
24 constituents and therefore likely resemble the general compositional split between fine and
25 thoracic coarse mode particles. (US EPA 2009a, section 3.5.1.1)
26 As noted in the previous review, primary biological aerosol particles (PBAP), which
27 include microorganisms, fragments of living things, and organic compounds of miscellaneous
28 origin in surface deposits on filters, are not distinguishable in analyses of total OC (US EPA,
29 2004, second 3.2). A clear distinction should be made between PBAP and primary OCM
30 produced by organisms (e.g., waxes coating the surfaces of organisms) and precursors to
31 secondary OCM such as isoprene and terpenes. Indeed, the fields of view of many
32 photomicrographs of PM samples obtained by scanning electron microscopy are often dominated
33 by large numbers of pollen spores, plant and insect fragments, and microorganisms. The ISA
34 summarizes several studies of the contribution of PBAP to PM in different size ranges,
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-55
-------
1 particularly PMio-2.5, generally concluding that PBAP contributes substantially to PMio-2.5 OCM
2 measurements.
3 2.5.5.3 Ultrafine Particles
4 Information about the composition of ambient ultrafine particles directly emitted by
5 sources is still sparse compared to that for the larger size modes. However, their composition is
6 expected to reflect that of the sources emitting these particles. Particle number emissions from
7 motor vehicles are predominantly in the ultrafine size range. The composition of gasoline vehicle
8 emissions consists mainly of a mix of OCM, EC, and small quantities of trace metals and
9 sulfates, with OCM constituting anywhere from 26-88% of PM. Diesel PM is generally
10 comprised of an EC and trace metal ash core, onto which organic material condenses, and
11 nucleation-mode SO42 . With the introduction of new diesel emissions standards in 2007, total
12 emissions have decreased dramatically, particularly for carbon. In areas where nucleation is the
13 dominant source of ultrafine particles, sulfate along with ammonium and secondary organic
14 compounds are the likely major components of ultrafine particles. (US EPA, 2009a, section
15 3.5.7.7)
16 In a study conducted at several urban sites in Southern California, carbon, in various
17 forms, was found to be the major contributor to the mass of ultrafine particles. However,
18 ammonium was found to contribute 33% of the mass of ultrafine particles at one site in
19 Riverside. Iron was the most abundant metal found in the ultrafine particles. Chung et al. (2001)
20 found that carbon was the major component of the mass of ultrafine particles in a study
21 conducted during January of 1999 in Bakersfield, CA. However, the contribution of
22 carbonaceous species was much lower than that found in the cities in Southern California, with
23 the crustal elements calcium, silicon, and aluminum also contributing notable mass percentages.
24 (US EPA 2009a, section 3.5.1.7)
25 Herner et al. (2005) reported a gradual increase in OC mass fraction as particle size
26 decreases from 1 |im (20% OC) to 100 nm (80% OC) in the San Joaquin Valley of California.
27 Sardar et al. (2005) found OC to be the major component of ultrafine particles at four locations
28 in California, with higher OC mass fraction in the wintertime relative to summertime. Elemental
29 carbon and SO42 were also present in the ultrafine samples but at much smaller mass fractions.
30 Furthermore, EC was present year-round whereas SO42 had a summertime preference. More
31 detailed chemical characterization of the OC fraction of ambient ultrafine particles is extremely
32 limited, but recent studies have identified specific organic molecular markers affiliated with
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-56
-------
1 motor vehicle emissions including hopanes and polycyclic aromatic hydrocarbons (Fine et al.,
2 2004; Ning et al., 2007; Phuleria et al., 2007). (USEPA 2009a, section 3.5.1.1)
3 2.5.6 Relationships among PM2.s, PMio, and PMi0-2.s
4 PMio-2.5 to PM2.5 ratios are generally lower in the east than in the west, and lower in rural
5 areas than in urban areas, except in dry rural areas. The Criteria Document from the last review
6 compared collocated PM2.5 and PMio-2.5 concentrations, averaged over the same years. The mean
7 PM2.5 to PMio-2.5 ratio was calculated for each of 17 MS As using as many sites as possible. The
8 median mean PM2.5 to PMio-2.5 ratio for the 17 MS As was 1.2. For eight MS As (Tampa, FL;
9 Columbia, SC; Louisville, KY; Chicago, IL; Gary, IN; Milwaukee, WI; Steubenville, OH;
10 Portland, OR), the mean PM2.s concentration exceeded the mean PMIO-2.5 concentration. For an
11 additional eight MSAs, the PM2.s and PMio-2.5 concentration means were the same (within one
12 SD). Salt Lake City was the only MSA for which the mean PMio-2.5 concentration exceeded the
13 mean PM2.5 concentration. (US EPA, 2004, section 3.1.2)
14 The 2005 Staff Paper presented information on the relationships among PM indicators in
15 different regions based on 2001-2003 data from the nationwide PM FRM monitoring networks.
16 The ratios of annual mean PM2.s to PMio at sites in different geographic regions were reported to
17 be highest in the eastern U.S. regions, with median ratios of about 0.6 to 0.65, and lowest in the
18 Southwest region, with a median ratio near 0.3 (US EPA, 2005, section 2.4.6). These data are
19 generally consistent with earlier findings reported in the previous review from a more limited set
20 of sites (US EPA, 1996, sections 6.3.1.4 and 6.3.2). PM2.s and PMio measured on the same days
21 at collocated monitors are fairly well correlated, on average, in the eastern regions and not as
22 well correlated in the western regions, particularly in the upper midwest. PMio was found to be
23 fairly well correlated with estimated PMio-2.5 in most regions, with the highest average
24 correlations in the southwest, upper midwest, and southern California regions. The Staff Paper
25 concluded that these data suggest that PMio might be a suitable indicator for either fine or coarse
26 particles, depending upon location-specific factors.29 However, in all locations estimated
27 PMio-2.5 and PM2.s were very poorly correlated, which should be expected due to their
28 differences in origin, composition, and atmospheric behavior.
29 In the second draft ISA, which restricted its PMio-2.5 analysis to the small network of
30 collocated low-volume sampler pairs for PMio and PM2.s, only six CS As/CBS As had sufficient
31 data for calculating PMio-2.5. In general, in the eastern metropolitan areas including Atlanta,
29 In the final action for the review completed in 2006, the existing 24-hour PM10 standard was retained, with an
explanation regarding its appropriateness as an indicator for PM10.25. See 71 FR 61185, October 17, 2006.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-57
-------
1 Boston, Chicago, and New York, most of the mass of PMio was in the PM2.5 size fraction, with
2 the highest ratio of annual average 24-hour PM2 5 to PMi0-2.5 in Chicago (14 |ig/m3 PM2 5; 5
3 |ig/m3 PMio-2.5, ratio = 2.8). In contrast, in Denver (9 |ig/m3 PM2.5, 20 |ig/m3 PMi02.5, ratio =
4 0.45) and Phoenix (10 |ig/m3 PM2.5, 22 |ig/m3 PMio-2.5, ratio = 0.45) most of PMio was in the
5 thoracic coarse mode (US EPA 2009a, section 3.9.1.1). These findings are directionally
6 consistent with those summarized above from the last review.
7 2.6 PM TEMPORAL PATTERNS
8 The 2005 Staff Paper and the second draft ISA both have extensive, data-heavy analyses
9 of PM temporal patterns, the Staff Paper for 2001 -2003 and the second draft ISA for 2006-2007.
10 Their qualitative findings are similar and can be assumed to generally apply to current
11 conditions, recognizing that ongoing reductions from large emission source categories such as
12 mobile sources and electric generating units are gradually affecting these patterns.
13 Hourly PMio and PM2.5 measurements are conducted at many sites using beta gauge or TEOM
14 monitors. Many of the hourly measurements for PMio have FRM or FEM status. The ISA used
15 all available hourly data from FRM, FEM and FRM-like monitors in fifteen CSAs/CBSAs to
16 investigate diel variation in PM. Plots of PM2 5 by hour of the day were stratified by weekdays
17 and weekends. In most cities investigated, a morning PM2 5 peak is present starting at
18 approximately 6:00 a.m., corresponding with the start of the morning rush hour just before the
19 break-up of overnight stagnation. In Pittsburgh, dispersion behavior during the night results in
20 elevated PM2 5 concentrations throughout the night that blend in with any morning peak. Six
21 other metropolitan areas (i.e., Atlanta, Chicago, Seattle, St. Louis, Houston, and New York)
22 show two distinct daily peaks on both the weekdays and weekends. The evening PM2 5
23 concentration peak is broader than the morning peak and extends to overnight hours, reflecting
24 the concentration increase caused by a drop in boundary layer height at night.
25 Figure 2-19 compares the two-peak distribution in PM2 5 for Seattle with the one-peak
26 distribution in PM2 5 for Pittsburgh. Since these figures represent the distribution of hourly
27 observations over a 3-year period, any fluctuations or changes in the timing of the daily peaks
28 would result in a broadening of the curves shown in the plot. That is, individual days would tend
29 to have sharper peaks than shown in the graphs.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-58
-------
Figure 2-19. Diel plot generated from hourly FRM-like PM2.5 data ((ig/m3)
stratified by weekday (left) and weekend (right) for Pittsburgh, PA, and Seattle, WA, from
2005 to 2007.
Pittsburgh
Weekday{N =981)
Weekend (N = 407)
58 -
38 -
19 -
n -
II ~
58 -
\
38 -
JLU-I ~~^^ - - • j!^J— .1 19 "
n
• - • - • Mean
90th & 10th
" -„ t' 95th &5th
"~ -•-•-. ' ~
x_^_ — .
^ — •--__•-"''
12
18 24
12 18
24
Seattle
Weekday (N = 5775)
Weekend (N = 2332)
4
5
6
7
8
9
10
11
12
13
14
ti -
™E 5B ~
5 38-
CM
EL 19 -
0 -
Figure 2-1
percentiles
E
if -
58 -
38 -
--- --- 19 -
Medion
• - • - • Mean
90th & 10th
95th & 5th
" --- -'"
• •" • i i i i o i •"" r .--.--,--.-., i
6 12 18 24 16 12 18 24
? notes: Included are the number of monitor days (N) and the median, mean, 5th, 10th, 90th and 95th
for each hour of the day shown on the horizontal axis.
leven cities showed a gradual morning increase in mean PMio starting at approximately
6:00 a.m. on weekdays, corresponding with the start of the morning rush hour before the
break-up of overnight stagnation. The magnitude and duration of this peak, however, varies
considerably by area. Phoenix shows the most pronounced morning PMio peak concentration,
which drops off during the day and reappears in the evening. In contrast, Chicago shows a less
pronounced peak with the PMio concentration remaining elevated throughout the day. Figure 2-
20 shows the plots of PMio for Chicago and Phoenix. (Annex A of the ISA contains similar
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-59
-------
1
2
3
4
5
9
10
11
12
13
14
15
16
17
18
plots for nine other areas.) In both instances, the weekend pattern is similar in shape to the
weekday pattern with less pronounced peaks. Once again, any fluctuations in the timing of the
daily peaks could result in a broadening of the peaks in the 3-year composite figures, and
individual days typically will have somewhat sharper peaks.
Figure 2-20. Diel plots generated from hourly FEM PMio data (ng/m3) stratified by
weekday (left) and weekend (right) for Chicago, IL, and Phoenix, AZ, from 2005 to 2007.
Chicago
Weekday (N = 1971)
Weekend (N = 793)
291 i
-T" 218 -
O)
3 145 ~
O
S. 73-
o 4
291 -i
-T" 218 -
C~>
^ 145 -
O
1 73 -
291 -i
218 -
145 -
73 -
^^^-^^--^-^^-i^-^
-.-.-.- -.-.-. T r,-. -.".-,-. - -,— o -]
6 12 18 24
Phoeni>
Weekday (N = 1532)
291 -i
218 -
145 -
'' ^v '-, ./' , - - 73 -
._ - - - _ _
6 12 18 24
'
Weekend (N = 618)
*^^^
Median
• - — Mean
90th & 10th
95th & 5th
1 6 12 18 24 1 6 12 18 24
Figure 2-20 notes: Included are the number of monitor days (N) and the median, mean, 5th, 10th, 90th and 95th
percentiles for each hour of the day shown on the horizontal axis.
Superimposed on the daily patterns described above, distinct seasonal variations in PM2.5
concentrations can be seen in air quality data but are generally not as sharp as those seen for
ozone concentrations. The months with peak urban PM2 5 concentrations vary by region and
season. The urban areas in the northeast, industrial midwest, and upper midwest regions all
exhibit peaks in both the winter and summer months. In the northeast and industrial midwest
regions, the summer peak is slightly more pronounced than the winter peak, and in the upper
midwest region the winter peak is slightly more pronounced than the summer peak. In the
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-60
-------
1 southeast, a single peak period in the summer is evident. Urban PM2.5 values tend to be higher in
2 the first (January-March) and fourth (October-December) calendar quarters in many areas of the
3 western U. S. These patterns are consistent with our understanding of dominant emission sources
4 and atmospheric processes. For example, more carbon is produced when woodstoves and
5 fireplaces are used, and particulate nitrates are more readily formed in cooler weather. In
6 addition, the effective mixing depth is restricted due to enhanced thermal stability in the
7 planetary boundary layer during the cooler seasons. The chemical components of PM2.5 also
8 exhibit seasonal patterns as described in section 2.4.5.
9 The lowest concentrations of PMio-2.5 generally occur in the winter months. Elevated
10 levels are apparent in the easternmost regions in April. In the upper midwest, northwest, and
11 southern California regions, the highest levels occur in the mid- to late- summer to mid-fall. The
12 southwest region exhibits the greatest range of variability throughout the year, with elevated
13 levels apparent in the spring, consistent with winds that contribute to windblown dust. In areas
14 subject to impact from forest fires, both PM2.5 and PMio-2.5 can have extreme peaks during the
15 local high fire hazard season, if fires are upwind.
16 The Urban-Focused Visibility Assessment, being conducted for this review, uses a
17 combination of four types of 24-hour and continuous PM monitoring to estimate hourly PM2 5,
18 PM2.5 components, and PMio-2.5 during daylight hours, for 15 study areas (US EPA, 2009c). The
19 initial results are qualitatively consistent with the characterizations described above regarding the
20 typical existence of one or two peaks in PM concentrations during the day. Visualizations of
21 these patterns are provided in the draft assessment document (US EPA, 2009c, Figures 3-9 and
22 E-l).
23 Study of the relationship for PM2.5 between the annual mean at a site and the shorter-term
24 24-hour average peaks is useful for examining the relationships between short- and long-term air
25 quality standards. The 2005 Staff Paper used box plots of 2001-2003 PM2.5 data at FRM sites
26 across the U.S to show the relationship for PM2.5 between annual mean PM concentrations and
27 peak daily concentrations as represented by the 98th percentile of the distribution of 24-hour
28 average concentrations. Sites were assigned to bins based on 1 |ig/m3 increments of annual
29 average PM2.5 concentration, and the distribution of 98th percentile values was shown for each
30 bin. Although there was a clear monotonic relationship between 98th percentiles and annual
31 means, there was considerable variability in 98th percentile values for sites with similar annual
32 means. For the bins with annual mean PM2 5 values between 10 and 15 |ig/m3, the interquartile
33 range of 98th percentile concentrations (representing a middle group of 40 to 60 sites in each bin)
34 spans about 5 to 6 |ig/m3 (US EPA, 2005, Figures 2-25 and 2-26 based on Schmidt et al., 2005).
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-61
-------
1 In a new analysis for the current PM NAAQS review, EPA staff have further studied the
2 relationship between annual average PM2 5 and 98th percentile 24-hour PM2 5, using 2005-2007
3 data. Figure 2-21 is a scatter plot of these two statistics for 2005-2007, with each point
4 representing one year at a monitoring site that met EPA completeness requirements for
5 calculating valid design values for both PM2.5 NAAQS for that period, and that is eligible for
6 comparison to both NAAQS. As was noted in the 2005 Staff Paper, there is a wide range of 98th
7 percentile 24-hour concentrations associated with the data points falling into narrow ranges of
8 annual average concentrations. (Note that in Figure 2-21 it is evident that 2005 was a worse year
9 for 98th percentile and annual average PM2.5 concentrations than the next two years, as has
10 already been noted in section 2.4.1) Figure 2-22 is similar to Figure 2-21 in concept but
11 aggregates data to a more policy relevant level, focusing on the monitoring site and the metric
12 that would determine attainment or nonattainment for the urban area as a whole. In Figure 2-21,
13 3-year design values are compared, rather than individual year statistics. In addition, Figure 2-22
14 has data points only for those monitoring sites that have the highest 3-year annual design value
15 within their CBSA, i.e., the sites that determine whether all (or part) of the CBSA is
16 nonattainment with the annual PM2.5 NAAQS. This level of aggregation and screening thins out
17 the "cloud" of data points but does not appreciably change its location or shape.
18 To provide further visualization of the diversity of location situations regarding the
19 pattern of 24-hour PM2.5 concentrations that determine the annual average concentrations, Figure
20 2-23 shows in bar chart form the history of individual 24-hour PM2.5 concentrations in 2005-
21 2007 for four example monitoring sites, in the following CBS As: Augusta-Richmond County,
22 GA-SC; El Centre, CA; Pittsfield, MA; and Beckley, WV. The sampling frequency in all four
23 CBS As was every third day. These correspond to the indicated (in red) data points in the
24 "cloud" of PM2.5 design value data points. El Centre and Pittsfield are at the upper/left edge of
25 the dense part of the cloud. In these two areas, a relatively small number of days have
26 substantially higher concentrations than other days, and the 98th percentile 24-hour PM2.5
27 concentrations are relatively high compared to their annual average concentrations. Augusta-
28 Richmond County and Beckly are at the lower/right edge of the dense part of the data cloud. In
29 these two areas, there is less range in the daily values, and the 98th percentile concentrations are
30 relatively not as high compared to their annual average concentrations.30 We note that none of
31 the PM2.5 data for these four areas for this time period have been flagged by the respective states
30 The areas represented by data points that are above the dense part of the data cloud were not selected for this
illustration because their current design values are outside the range of alternative standards examined in this review,
and hence are not of as much interest with respect the assessment of those alternatives.
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite 2-62
-------
1
2
3
4
5
6
7
to request exclusion from regulatory use under EPA' s Exceptions Events Rule, although that
does not exclude the possibility than an exceptional event did influence the PM2.5 concentration
on one or more days.
Figure 2-21. Relationship of annual average and 98th percentile 24-hour
concentrations at the site and year level, all sites with complete enough data for calculation
of valid 2005-2007 design values.
80 0
CO
-i 70 0 -1
D)
~- 60 0 -
D
2 50 n
TJ
9)
•§ 30 0 -
no
(
2005-2007 PM2.5
•
• •»
»
•" • • V - - ,-:-. "
• •• • »
- * s^^ii^v^^ir
%V^>^ ^rvf*
t ^j-flT wiT v ^ •** jfX
. «??^j^-v
•^-j
»«y
) 5 10 15 20 25
Annual Average (ug/m3)
• 2005 • 2006 2007
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-63
-------
1
2
3
4
ath
Figure 2-22 Relationship of annual average and 98 percentile 24-hour PM2.s 3-year
design values, only for sites with the highest annual design value in their CBSA.
An
ir>
£J 70
s
fin
i. bU
4=
_',. c;n
•^ ou
fS
oj an
• PN
-*^
™ ^n
u
aj on
&.
4= 10 _
00
^ 0
0
2005-2007 Design Values (|ug/m3)
.
.
.
• •
*• •..•'•
•. • ' ". '• t.~MV-l''
.-. -.' 4:.:--.r>JKis~~.-.-:"
": • •.•"{tf>:«.J"'38."" • "
•:.'r..>5i :,••>•". '
* • . •<• • / *"*.»* •**
•• ••***• **
0 5.0 10.0 15.0 20.0 2£
Annual Average PM2.5
.0
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-64
-------
1
2
Figure 2-23. Three-year history of 24-hour PMi.s concentrations at highest design values site in four example CBSAs
Own li u.
Augusta Klchmoiicl County, :H>^*" ^^
^x *^r |
0.0 ^^^.0 10.0 1^.0 ^0^.0 ^E.O
L^n —u,
PRELIMINARY DRAFT - September 2009
Do Not Quote or Cite
2-65
-------
1 2.7 PM BACKGROUND LEVELS
2 For the purposes of this document, background PM is defined as the distribution of PM
3 concentrations that would be observed in the U.S. in the absence of anthropogenic (man-made)
4 emissions of primary PM and precursor (e.g., VOC, NOX, 862, and NHs) compounds in the U.S.,
5 Canada, and Mexico. Background levels so defined are referred to as policy-relevant
6 background levels, since this definition of background facilitates separating pollution levels that
7 can be controlled by U.S. regulations (or through international agreements with neighboring
8 countries) from levels that are generally uncontrollable by the U.S. As defined here, background
9 includes PM from natural sources in the U.S. and transport of PM from both natural and
10 man-made sources outside of the U.S. and its neighboring countries.
11 There is a distinct geographic difference in background levels, with lower levels in the
12 western U.S. and higher levels in the eastern U.S. The eastern U.S. is estimated to have more
13 natural organic fine particles and more water associated with hygroscopic fine particles than the
14 western U.S. due to generally higher humidity levels.
15 Background levels of PM vary by geographic location and season and have a natural
16 component and an anthropogenic component. The natural background arises from: (1) physical
17 processes of the atmosphere that entrain coarse particles (e.g., windblown crustal material, sea
18 salt spray); (2) volcanic eruptions (e.g., sulfates); (3) natural combustion such as wildfires (e.g.,
19 elemental and organic carbon, and inorganic and organic PM precursors); and (4) biogenic
20 sources such as vegetation, microorganisms, and wildlife (e.g., organic PM, inorganic and
21 organic PM precursors). The exact magnitude of the natural portion of background PM for a
22 given geographic location cannot be precisely determined because it is difficult to distinguish
23 local sources of PM from the long-range transport of anthropogenic particles and precursors.
24 Particulate matter can be transported long distances from natural or quasi-natural events
25 occurring outside the continental U.S. The occurrence and location of these long-range transport
26 events are highly variable, and their impacts on the U.S. are equally variable. The contributions
27 to background from sources outside of the U.S., Canada, and Mexico can be significant on an
28 episodic but probably not on an annual basis.
29 Background concentrations of PM2.s, PMi0-2.5, and PMi0 may be conceptually viewed as
30 comprised of baseline and episodic components. The baseline component is the contribution
31 from natural sources within the U.S., Canada, and Mexico and from transport of natural and
32 anthropogenic sources outside of the U.S., Canada, and Mexico that is reasonably well
33 characterized by a consistent pattern of daily values each year, although they may vary by region
34 and season.
PRELIMINARYDRAFT-September 2009 2-66
Do Not Quote or Cite
-------
1 In addition to this baseline contribution to background concentrations, a second
2 component consists of more rare episodic high-concentration events over shorter periods of time
3 (e.g., days or weeks) both within the U.S., Canada, and Mexico (e.g., volcanic eruptions, large
4 forest fires) and from outside of the U.S., Canada, and Mexico (e.g., transport related to dust
5 storms from deserts in North Africa and Asia). Over shorter periods of time (e.g., days or
6 weeks), the range of background concentrations is much broader than the annual averages.
7 Specific natural events such as wildfires, volcanic eruptions, and dust storms, both of U.S. and
8 international origin, can lead to very high levels of PM comparable to or greater than those
9 driven by man-made emissions in polluted urban atmospheres. EPA has had in place for many
10 years policies that could remove consideration of data affected by such events, where
11 appropriate, from attainment decisions. In March 2007, EPA adopted a binding rule that allows
12 exclusion of data affected by so-called exceptional events, including natural events, from
13 regulatory decisions such as attainment determinations (72 FR 13560, March 22, 2007). This
14 rule has superseded the prior administrative policies.
15 There are several methods for characterizing PRB concentrations of PM within the
16 United States. Some methods rely upon analyses of measured PM concentrations at remote rural
17 locations, while other methods utilize air quality chemical transport models (CTMs) to estimate
18 PRB. The last review characterized PRB for PM2.5 on a 24-hour average basis by summarizing
19 the non-sulfate portion of PM2 5 measured at Interagency Monitoring of Protected Visual
20 Environment (IMPROVE) sites in remote areas between 1990 and 2002. Sulfate was omitted
21 because it is attributable almost entirely to anthropogenic emissions. It was noted that this
22 method likely resulted in an underestimate of PRB (US EPA, 2005, section 2.6).31
23 In this review, EPA has relied on the results of the CTM modeling described in the
24 second draft ISA to estimate PRB for PM2.5 for inclusion in estimating risks (US EPA, 2009a,
25 section 3.7). In addition, EPA staff has also used this CTM modeling to estimate PRB for PM2.5
26 components for the urban-focused visibility risk assessment (see additional description in US
27 EPA, 2009c, Appendix C). The CTM-based approach is based on a "zero-out" model simulation
28 in which anthropogenic emissions inside the U.S., Canada, and Mexico are set to zero while all
29 biogenic emissions for these areas and biogenic and anthropogenic emissions from elsewhere in
30 the world are not altered. This approach can provide more spatial and temporal resolution for
31 estimating PRB compared to the use of measurements given the sparse nature of remote
31 The statement in the 2005 Staff Paper that the described method resulted in an underestimate of PRB reflects a
judgment that the upward bias from relying on IMPROVE sites that are actually influenced by anthropogenic
sources of PM components other than sulfate is greater than the downward bias from not including the actual, small
PM sulfate component from natural sources and sources outside the US, Canada, and Mexico.
PRELIMINARYDRAFT-September 2009 2-67
Do Not Quote or Cite
-------
1 measurement sites and the concern that even remote sites are affected by non-local
2 anthropogenic sources.
3 The CTM-based approach involves coupling the global-scale circulation model GEOS-
4 Chem (Fiore, et al, 2003) with the regional scale air quality model CMAQ (Byum, et. al., 2006
5 and Byum, et. al, 1999). The GEOS-Chem model is run on a global scale and is used to provide
6 estimates of transported pollutants from emissions of natural and anthropogenic sources outside
7 the U.S., Canada, and Mexico. These transported pollutant concentrations are used to provide
8 the "boundary condition" concentrations for two CMAQ simulations covering the continental
9 U.S. and adjacent portions of Canada and Mexico (CONUS), one simulation with all emissions
10 to evaluate model performance and one to estimate PRB. In the CMAQ simulation to estimate
11 PRB, only natural emissions in the U.S., Canada, and Mexico are considered. The details of this
12 modeling approach, including the input data sets and model chemistry, are described in section
13 3.7 of the second draft ISA. The following is a brief summary.
14 The two models were applied to simulate one year of air quality data for 2004. The base
15 case CMAQ run for 2004 includes meteorology and all the anthropogenic and natural sources
16 both within and outside of the U.S., Canada, and Mexico. This run was performed to provide a
17 comparison of model predictions with measurements. The second draft ISA characterizes the
18 CMAQ performance for the annual average concentrations and most of the seasonal averages of
19 PM2 5 at remote sites as "very good" in the East and Midwest. In the West, predictions at remote
20 sites are "generally too low in all seasons." The second draft ISA notes that degraded
21 performance in the West is not unexpected because the grid resolution in the CMAQ model
22 simulation (36 km for this application) will smooth out significant variations in terrain that
23 influence measured concentrations, particularly concentrations attributable to anthropogenic
24 emissions which in the West are often concentrated in basin settings where local meteorological
25 conditions coupled with local emissions of primary particles may dominate PM2.5 concentrations.
26 However, looking across the U.S., the model does correctly reproduce broad geospatial
27 differences in that predicted PM2.5 concentrations are lower at western locations than they are in
28 the East; this finding is consistent with measured data. Also, natural emissions in the West are
29 less concentrated in basin settings, and western terrain may, therefore, have less effect on model
30 performance when estimating PRB.
31 In addition to the "base case" run which includes all anthropogenic and biogenic
32 emissions, CMAQ was also run for a second scenario to estimate PRB, with the same boundary
33 conditions but with only natural emissions from within the U.S., Canada, and Mexico. The
34 hourly outputs from this second CMAQ run were used to calculate seasonal and annual average
35 estimates of PRB within seven regions of the U.S.
PRELIMINARYDRAFT-September 2009 2-68
Do Not Quote or Cite
-------
1 For the health risk assessment, annual values presented in Table 2-4 were used in
2 modeling long-term PM2.5 exposures and estimating associated health risks (in those sensitivity
3 analysis scenarios where risk was modeled down to PRB). For estimating risks associated with
4 short-term PM2.5 exposures (which involved modeling down to PRB, exclusively), quarterly
5 values presented in Table 2-4 were used to represent the appropriate block of days within a
6 simulated year (US EPA, 2009b, section 3.2.2).
Table 2-4. Annual and quarterly mean of the CMAQ-predicted PRB PM2.s
concentrations (ug/m3) in the U.S. EPA CONUS regions in 2004.
Annual
January-March
April-June
July-September
October-December
Northeast
0.74
0.85
0.78
0.67
0.68
Southeast
1.72
2.43
1.41
1.41
1.64
Industrial
Midwest
0.86
0.89
0.89
0.94
0.73
Upper
Midwest
0.84
0.79
0.93
0.99
0.66
Southwest
0.62
0.61
0.76
0.70
0.40
Northwest
1.01
0.48
0.81
1.42
1.32
Southern
California
0.84
0.54
0.92
1.21
0.67
PRELIMINARYDRAFT-September 2009
Do Not Quote or Cite
2-69
-------
1 For the purposes of the urban focused visibility assessment, EPA extracted hour-specific
2 monthly average PM2.5 components from the PRB CMAQ run, averaging across the several
3 CMAQ grid cells that overlap with the Census-defined urbanized area of each of the visibility
4 assessment study areas. Thus, within a particular month and study area, PRB for PM2.5
5 component concentrations for a given clock hour are constant, but they vary by clock hour within
6 a day, by month within a year, and by study area. Table 2-5 summarizes them in terms of the
7 annual average PRB for each component, by study area.
8 The health risk assessment did not make use of estimates of PRB concentrations of PMio
9 or PMio-2.5, because the focus of that assessment was on evaluating risks associated with
10 exposures to fine particles, PM2.5. The urban focused visibility assessment did require estimates
11 of PRB for PMio-2.5- The second draft ISA did not present any new information on this subject.
12 The approach used in the previous two reviews was to present the historical range of annual
13 means of PMio-2.5 concentrations from IMPROVE monitoring sites selected as being least
14 influenced by anthropogenic emissions (US EPA, 2004, Table 3E-1). For sites in the lower 48
15 states, these annual means range from a low of 1.8 ug/m3 to a high of 10.8 ug/m3. For this
16 assessment, we estimated PRB for PMio-2.5 using a contour map based on average 2000-2004
17 PMio-2.5 concentrations from all IMPROVE monitoring sites, found in a recent report from the
18 IMPROVE program. (DeBell, L.J., et al., 2006). We located each study area's position on this
19 map and assigned it the mid-point of the range of concentrations indicated by the contour band
20 for that location.
21 The results are shown in Table 2-6. Lacking any other information, these PRB values
22 have been assumed to apply to every hour of the year. While the contour map and, thus, these
23 PRB values are influenced by data from IMPROVE sites that were not considered in the last
24 review to be the sites most isolated from the influence of anthropogenic emissions, including
25 three IMPROVE sites in urban areas, these values are generally within the range of values
26 presented for such isolated sites in the previous science assessment (US EPA 2004, Appendix E,
27 Table 3E-1).
28
PRELIMINARYDRAFT-September 2009 2-70
Do Not Quote or Cite
-------
Table 2-5. Summary of PRB estimates for the five PMi.5 components: average 1-hour values
across 2005-2007
Study Area
Tacoma
Fresno
Los Angeles-South
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia-
New York-N.New
Jersey -Long Island
Average 1 -Hour PRB Concentration Across 2005-2007 (ug/m3)
Sulfate
(dry, no
ammonium)
0.45
0.4
0.36
0.31
0.25
0.27
0.3
0.31
0.29
0.3
0.34
0.3
0.34
0.34
0.36
Nitrate
(dry, no
ammonium)
0.026
0.00062
0.0037
0.000052
0.00028
0.0022
0.0055
0.0027
0.007
0.016
0.00062
0.00052
0.0016
0.00097
0.0038
Elemental
Carbon
0.15
0.08
0.028
0.02
0.025
0.055
0.091
0.047
0.099
0.1
0.024
0.029
0.039
0.03
0.026
Organic
Carbonaceous
Material
1.3
0.74
0.3
0.26
0.26
0.59
0.86
0.53
1.1
1.1
0.32
0.36
0.44
0.36
0.31
Fine
Soil/Crustal
0.31
0.19
0.036
0.015
0.034
0.092
0.17
0.07
0.19
0.19
0.018
0.034
0.054
0.032
0.022
PRELIMINARYDRAFT-September 2009
Do Not Quote or Cite
2-1 \
-------
Table 2-6. PRB Concentrations of PMi0-2.s Used in the Urban Focused Visibility
Assessment, Based on Measurements at IMPROVE Sites
Study Area
Tacoma
Fresno
Los Angeles-South Coast Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia- Wilmington
New York-N.New Jersey -Long Island
PRB PMio-2.5 Mass (ug/m3)
4.5
5.5
4.5
5.5
4.5
8.5
5.5
7.5
5.5
5.5
9.5
3.5
3.5
6.5
3.5
2 2.8 RELATIONSHIP BETWEEN AMBIENT PM MEASUREMENTS AND HUMAN
3 EXPOSURE
4 The statutory focus of the primary NAAQS for PM is protection of public health from
5 the adverse effects associated with the exposure to ambient PM - that is, the focus is on particles
6 in the outdoor atmosphere that are either emitted directly by sources or formed in the atmosphere
7 from precursor emissions. We refer to PM in the ambient air as ambient PM. An understanding
8 of human exposure to ambient PM helps inform the evaluation of underlying assumptions and
9 interpretation of results of epidemiologic studies that characterize relationships between
10 monitored ambient PM concentrations and observed health effects discussed in Chapter 3.
11 An important exposure-related issue for this review is the characterization of the
12 relationships between ambient PM concentrations measured at one or more centrally located
13 monitors and personal exposure to ambient PM, as characterized by particle size, composition,
14 source origin, and other factors. Information on the type and strength of these relationships,
15 discussed below, is relevant to the evaluation and interpretation of associations found in
16 epidemiologic studies that use measurements of PM concentrations at centrally located monitors
PRELIMINARYDRAFT-September 2009
Do Not Quote or Cite
2-72
-------
1 as a surrogate for exposure to ambient PM.32 The focus here is on particle size distinctions; the
2 second draft ISA also discusses exposure relationships related to compositional differences (US
3 EPA, 2009a, section 3.8.6.6).
4 2.8.1 Definitions
5 Exposure to a contaminant is defined as contact at a boundary between a human and the
6 environment (e.g., the breathing zone) at a specific contaminant concentration for a specific
7 interval of time; it is measured in units of concentration(s) multiplied by time (or time interval)
8 (National Research Council, 1991). An individual's total personal exposure to PM results from
9 breathing air containing PM in different types of environments (e.g., outdoors near home,
10 outdoors away from home, indoors at home, indoors at office or school, commuting, restaurants,
11 malls, other public places). These environments may have different concentrations of PM with
12 particles originating from a wide variety of sources.
13 Ambient PM is comprised of particles emitted by anthropogenic and natural sources and
14 particles formed in the atmosphere from emissions of gaseous precursors. This includes
15 emissions not only from outdoor sources such as smokestacks, industrial sources, and
16 automobiles, but also from sources located indoors with emissions vented outdoors, such as
17 fireplaces, wood stoves, and some cooking appliances. Exposure to ambient PM can occur both
18 outdoors and indoors to the extent that ambient PM penetrates into indoor environments - we use
19 the term PM of ambient origin to refer to both outdoor and indoor concentrations of ambient PM.
20 We use the term nonambient PM to refer to concentrations of PM that are only due to indoor
21 sources of particles that are not vented outdoors such as smoking, cooking, other non-vented
22 sources of combustion, cleaning, mechanical processes, and chemical interactions producing
23 particles. In characterizing human exposure to PM concentrations relevant to setting standards
24 for ambient air quality, the ISA conceptually separates (US EPA, 2009a, Section 3.8.1,) an
25 individual's total personal exposure to PM into exposure to PM of ambient origin and exposure
26 to all other sources of PM (i.e., nonambient PM exposure).
27 Outdoor concentrations of PM are affected by emissions, meteorology, topography,
28 atmospheric chemistry, and removal processes. Indoor concentrations of PM are affected by
29 several factors, including outdoor concentrations, processes that result in infiltration of ambient
30 PM into buildings, indoor sources of PM, aerosol dynamics and indoor chemistry, resuspension
31 of particles, and removal mechanisms such as particle deposition, ventilation, and air-
32 conditioning and air cleaning devices (US EPA, 2009a, section 3.x). Concentrations of PM
33 inside vehicles are subject to essentially the same factors as concentrations of PM inside
32 Consideration of exposure measurement error and the effects of exposure misclassification on the interpretation
of the epidemiologic studies are addressed in section 3.4.
PRELIMINARYDRAFT-September 2009 2-73
Do Not Quote or Cite
-------
1 buildings. Personal exposure to PM also includes a component which results specifically from
2 the activities of an individual that typically generate particles affecting only the individual or a
3 small localized area surrounding the person, such as walking on a carpet, referred to as the
4 personal cloud.
5 Epidemiologic studies generally use measurements from central monitors to represent the
6 ambient concentrations in an urban or rural area. We use the term central site to mean the site of
7 a PM monitor centrally located with respect to the area being studied. In many cases,
8 epidemiologic studies combine the measurements from more than one monitor to obtain a
9 broader representation of area-wide PM concentrations than a single monitor provides.
10 2.8.2 Centrally Monitored PM Concentration as a Surrogate for Particle Exposure
11 Prior reviews focused on two interconnected PM exposure issues: (1) the ability of
12 central fixed-site PM monitors to represent population exposure to ambient PM and (2) how
13 differences between fine and coarse particles affect population exposures (US EPA, 2005,
14 section 2.7.2). Distinctions between PM size classes and components were found to be
15 important considerations in addressing the representativeness of central monitors. In particular,
16 fine particles were observed to have a longer residence time and generally to exhibit less spatial
17 variability in the atmosphere than coarse fraction particles. In the PM NAAQS review
18 completed in 1997, EPA concluded that measurements of daily variations of PM at nearby
19 ambient monitoring stations have a plausible linkage to daily variations of human exposures to
20 PM of ambient origin, and that this linkage was stronger for fine particles than for PMio or the
21 coarse fraction of PMio (US EPA, 1996a, section x.x). Furthermore, evidence in the current PM
22 NAAQS review is consistent with findings in previous reviews concluding that central
23 monitoring can be a useful index for representing the average exposure of people in a community
24 to PM of ambient origin in epidemiological studies assessing PM associated health effects (US
25 EPA, 1996b, section x.x,; US EPA, 2009a , section 3.9.2.4).
26 As outlined in the last review, an individual's total personal exposure to PM may differ
27 from the ambient concentration measured at the central site monitor because: (1) spatial
28 differences in ambient PM concentrations exist across a city or region; (2) generally only a
29 fraction of the ambient PM is present in indoor or in-vehicle environments, whereas individuals
30 generally spend a large percentage of time indoors; and (3) a variety of indoor sources of PM
31 contribute to total personal exposure. Thus, the amount of time spent outdoors, indoors, and in
32 vehicles and the types of activities engaged in (e.g., smoking, cooking, vacuuming) also will
33 heavily influence personal exposure to PM. The first two factors are important for determining
34 the strength of the relationship between ambient PM and ambient personal exposure (US EPA,
35 2005, section 2.7.2).
PRELIMINARYDRAFT-September 2009 2-74
Do Not Quote or Cite
-------
1 With regard to the first factor that influences the relationship between total personal
2 exposure and concentrations measured at central sites, the spatial variability of PM plays a large
3 role. As discussed in Section 2.5, for many areas PM2 5 concentrations are fairly uniform
4 spatially, with higher concentrations near roadways and other direct sources of PM2.5. Analyses
5 of PM2.5 data for X urban areas indicate that differences in annual mean concentrations between
6 monitoring sites in an urban area range from less than x |ig/m3 to as much as x |ig/m3. However,
7 the correlations of daily PM2.5 between sites are typically greater than O.XX. Daily mean PM2.5
8 concentrations exhibit much higher spatial variability than annual means, even when the daily
9 concentrations at sites are highly correlated. Although the spatial variability of PM2.5 varies for
10 different urban areas, overall, some degree of uniformity results from the widespread formation
11 and long lifetime of the high regional background of secondary PM2 5.
12 In summarizing the key findings related to spatial variability in PM2.5 concentrations, the
13 second draft ISA states, "In general, PM2 5 has a longer atmospheric lifetime than PMio-2.s
14 because larger particles have a higher gravitational settling velocity. For PM2.5, most
15 metropolitan areas exhibited high correlations (generally greater than 0.75) out to a distance of
16 100 km (US EPA, 2009a, Figures 3-25 through 3-27). Notable exceptions were Denver, Los
17 Angeles and Riverside where correlations dropped below 0.75 somewhere between 20 and 50
18 km.. .Furthermore, correlations between PMio concentrations exhibited substantially more scatter
19 relative to PM2 5... .Although the general understanding of PM differential settling leads to an
20 expectation of greater spatial heterogeneity in the PMi0-2.5 fraction relative to the PM2.5 fraction
21 in urban areas, deposition of particles as a function of size depends strongly on local
22 meteorological conditions, in particular on the degree of turbulence in the mixing layer."
23 Therefore, the findings reported for 15 CS As/CBS As33 in the second draft ISA "may not apply
24 to all locations or at all times. Population density and associated building density are also
25 important determinants of the spatial distribution of PM concentrations. Inter-sampler
26 correlations as a function of distance between monitors obtained for sampler pairs located less
27 than 4 km apart (i.e., on a neighborhood scale) showed a shallower slope for PM2 5 than for
28 PMio. The average correlation was 0.93 for PM2.5, but it dropped to 0.70 for PMio" (US EPA,
29 2009a, section 3.9.1.1).
30 Relative to fine particles, coarse and ultrafine particles are likely to be more variable
31 across urban scales. Daily mean PMio-2.s concentrations tend to be more variable and have lower
32 inter-site correlations than PM2 5, possibly due to their shorter atmospheric lifetime (travel
33 distances < 1 to 10s of km) and the more sporadic nature of PMio-2.s sources (US EPA, 2009a,
34 Section 3.2). Ultrafine particles also have shorter atmospheric lifetimes (travel distances < 1 to
PRELIMINARYDRAFT-September 2009 2-75
Do Not Quote or Cite
-------
1 10s of km, compared with 100s to 1000s of km for PM2 5) and spatially variable sources. High
2 concentrations of ultrafine particles have been measured near roadways, but with concentrations
3 falling off rapidly with increasing distance from the roadway. Both coarse and ultrafine particles
4 also have reduced concentrations indoors compared to PM2.5, due to lower infiltration rates,
5 greater deposition rates, and coagulation of ultrafine particles into larger particles. These
6 differences make it more difficult to find a relationship between ambient concentrations and
7 personal exposures to these size fractions than for PM2.5.
8 The second factor influencing the relationship between ambient PM concentrations
9 measured at central sites and total personal exposure to PM is the extent to which ambient PM
10 penetrates indoors and remains suspended in the air. If the flow of ambient PM into the home
11 from the outdoors is very restricted, the relationship between ambient PM concentrations
12 measured at a central site and total exposure to PM will tend to be weaker than in a situation
13 where ambient PM flows more readily into the home and is a greater part of the overall indoor
14 PM concentrations. This is heavily dependent on the building air exchange rate, and also on
15 penetration efficiency and deposition or removal rate, both of which vary with particle size. Air
16 exchange rates (the rates at which the indoor air in a building is replaced by outdoor air) are
17 influenced by building structure, the use of air conditioning and heating, opening and closing of
18 doors and windows, and meteorological factors (e.g., difference in temperature between indoors
19 and outdoors). Based on physical mass-balance considerations, usually the higher the air
20 exchange rate the greater the fraction of PM of ambient origin found in the indoor and in-vehicle
21 environments. Higher air exchange rates also dilute the concentration of indoor- generated PM.
22 Rates of infiltration of outdoor PM into homes through cracks and crevices are higher for PM2 5
23 than for PMio, PMio-2.5, or ultrafine particles (US EPA, 2009a, section 3.8.4.3). Since PMi0-2.5
24 and ultrafine particles penetrate indoors less readily than PM2 5 and deposit to surfaces more
25 rapidly than PM2.5, a greater proportion of PM2.5 of ambient origin is found indoors than PMio-2.s
26 and ultrafine particles, relative to their outdoor concentrations. Thus, the particle size
27 distribution influences the amounts of PM of ambient origin found indoors.
28 Since people typically spend a large part of their time indoors at home, the air exchange
29 rate of the home has a large impact on exposures to ambient pollution. Homes with low air
30 exchange rates are more protected from outdoor sources, and vice-versa. Homes in regions with
31 moderate climate tend to be better ventilated and have higher air exchange rates than areas which
32 have very cold or very hot climates. Thus, climate plays an important role in regional population
33 exposure to ambient pollution.
34 The third factor influencing the relationship between ambient concentrations measured at
35 central sites and total personal exposure is the contribution of indoor sources to total personal
36 exposure. On average, individuals spend nearly 90 percent of their time indoors. The
PRELIMINARYDRAFT-September 2009 2-76
Do Not Quote or Cite
-------
1 contribution of indoor sources to indoor concentrations of PM is significant, and can be quite
2 variable on different days and between individuals. Indoor sources such as combustion devices
3 (e.g., stoves and kerosene heaters) generate predominantly fine particles; cooking produces both
4 fine and coarse particles; and resuspension (e.g., dusting, vacuuming, and walking on rugs)
5 generates predominantly coarse particles (US EPA, 2004, section 5.3.3.2.2). This factor,
6 however, does not influence exposure to PM of ambient origin.
7 These three factors related to total personal exposure can give rise to measurement error
8 in estimating exposures to fine and coarse PM (US EPA, 2009a, Section 3.8.6), thus making the
9 quantification of relationships between concentrations measured at central site monitors and
10 health effects more difficult due to reduction in statistical power. Moreover, exposure
11 measurement errors can also affect the magnitude and the precision of the health effects
12 estimates. However, as discussed in the second draft ISA and below in chapter 3, exposure
13 measurement errors under most ordinary circumstances are not expected to influence the overall
14 interpretation of findings from either the long-term exposure or time-series epidemiologic studies
15 that have used ambient concentration data (US EPA, 2009a, sections 3.8.6.3 and 3.8.6.4). They
16 will more likely affect the magnitude of the effects found from these studies and result in higher
17 effect estimates, since exposure measurement errors tend to bias towards the null hypothesis.
18 Exposure measurement errors also tend to decrease the precision of health effects estimates,
19 resulting in wider confidence intervals.
20 The prior review discussed the finding by some researchers that some epidemiologic
21 studies yield statistically significant associations between ambient concentrations measured at a
22 central site and health effects even though there is a very small correlation between ambient
23 concentrations measured at a central site and total personal exposures (US EPA, 2004, sections
24 5.3.3.1.5 and 5.6). The explanation of this finding is that total personal exposure includes both
25 ambient and nonambient generated components, and while the nonambient portion of personal
26 exposure is not generally correlated with ambient concentrations, the exposure to concentrations
27 of ambient origin is correlated with ambient concentrations (US EPA, 2004, section 5.5.2.1; US
28 EPA, 2009a, section 3.x). Thus, it is not surprising that health effects might correlate with
29 central site PM concentrations, because exposure to PM of ambient origin correlates with these
30 concentrations, and the lack of correlation of total exposure with central site PM concentrations
31 does not statistically alter that relationship. By their statistical design, time-series epidemiologic
32 studies of this type only address the ambient component of exposure, since the impact of day-to-
33 day fluctuations in ambient PM on health effects associated with short-term exposures is
34 examined.
35 When indoor sources only have minor contributions to personal exposures, total exposure
36 is mostly from PM of ambient origin. In these cases high correlations are generally found
PRELIMINARYDRAFT-September 2009 2-77
Do Not Quote or Cite
-------
1 between total personal exposure and ambient PM measured at a central site (US EPA, 2004,
2 section 5.3.3.1.3). For example, measurements of ambient sulfate, which is mostly in the fine
3 fraction, have been found to be highly correlated with total personal exposure to sulfate (p. 3-
4 195, US EPA, 2009a, section 3.8.4.1). Since in these studies there were minimal indoor sources
5 of sulfate, the relationship between ambient concentrations and total personal exposure to sulfate
6 was not weakened by possible presence of small indoor-generated sulfates in some
7 environments.
8 It is recognized that existing PM exposure measurement errors or uncertainties most
9 likely will reduce the statistical power of PM health effects analyses, thus making it more
10 difficult to detect a true underlying association between the exposure metric and the health
11 outcome of interest. However, the use of ambient PM concentrations as a surrogate for personal
12 ambient exposures is not expected to change the principal conclusions from PM epidemiological
13 studies that use community average health and pollution data (US EPA, 2009a, sections 2.x and
14 3.x). Based on these considerations and on the review of the available exposure-related studies,
15 the ISA concludes that for epidemiologic studies, ambient PM2.s concentration as measured at
16 central site monitors is a useful surrogate for exposure to PM2.5 of ambient origin. However, for
17 coarse and ultrafine PM, such ambient concentrations are not likely to be as good a surrogate for
18 personal ambient exposure. While nonambient PM may also be responsible for health effects,
19 since the ambient and nonambient components of personal exposure are independent, the health
20 effects due to nonambient PM exposures generally will not bias the risk estimated for ambient
21 PM exposures (US EPA, 2009a, section 3.x).
22 2.9 RELATIONSHIP BETWEEN AMBIENT PM AND VISIBILITY
23 Visual air quality (VAQ) is defined as the visibility effects caused solely by air quality
24 conditions and excluding those associated with meteorological conditions like fog and
25 precipitation. Light extinction measures the ability of particles and gases in the atmosphere to
26 scatter and absorb light traveling between an object and a person (or camera). Because it is the
27 factional loss of light per unit of distance, it is measured in inverse length units (i.e. I/length,
28 commonly 1/106 meter written Mm"1 and pronounced inverse megameters).
29 The extent to which any amount of light extinction affects a person's ability to view a
30 scene depends on both scene and light characteristics. For example the appearance of a nearby
31 object (i.e., a building) is generally less sensitive to a change in light extinction than the
32 appearance of a similar object at a greater distance. For a scene with known characteristics, the
33 amount of degradation in the scene associated with a change in light extinction can be
34 determined, and the change in appearance can be realistically displayed on a digital photograph
35 of the scene using the WinHaze system.
PRELIMINARYDRAFT-September 2009 2-78
Do Not Quote or Cite
-------
1 Survey studies have used sets of actual or computer modified (WinHaze) photographs
2 depicting a range of visibility conditions on urban scenes to assess the public's opinion on what
3 constitutes acceptable conditions. However, differences in scene sensitivities caused by changed
4 scene and/or lighting characteristics (e.g. differing shadow patters or clouds cover) can affect
5 people's perceptions of the haze level shown for a given photo. For the WinHaze modified
6 scenes which maintain constant lighting and scene characteristics, there is a
7 known/predetermined one-to-one correspondence between the perceived haze in the WinHaze-
8 generated photographs and the amount of light extinction. Visibility effects are related to the
9 sightpath-averaged light extinction between the observer (or camera) and each object viewed in
10 the scene. Ambient measurements that are typically available at only one or a few locations near
11 the surface are not always representative of the sightpath-averaged light extinction, especially for
12 elevated sightpaths. This can be an important source of uncertainty for the results of preference
13 survey studies that use actual photographs of a scene that are paired with ambient measurements
14 of light extinction. The WinHaze program assumes a specific distribution of light extinction that
15 is invariant from one haze image to the next so their use in preference survey studies eliminates
16 that source of uncertainty in the results. For visibility preferrence studies, visibility levels are
17 generally characterized using the haze index in units of deciviews (similar to the decibel scale for
18 sound), which is a simple logarithmic transformation of light extinction.
19 Light extinction (&ext) is the sum of the light scattering and absorption by particles and
20 gases in the atmosphere as shown below (US EPA, 2009a, Equation 9-x):
21
22 bext = 6ap + bag + bsg + bsp (Equation 2-1)
23 where:
24 &ap = light absorption by particles
25 bag = light absorption by gases
26 6sg = light scattering by gases (also known as Rayleigh scattering)
27 bsp = light scattering by particles
28
29 Light scattering by gases (e.g., nitrogen, oxygen, etc.) that comprise the atmosphere (also
30 known as Rayleigh or clear-air scattering) is related to the density of the air, which is sufficiently
31 constant with elevation that it can be considered a known constant value for any location. NC>2 is
32 the only atmospheric pollutant gas that absorbs visible light appreciably and its effects are
33 generally small (i.e., less than 5%) compared to PM light extinction (defined as the sum of
34 particulate light scattering and paniculate light absorption), so its contributions to ambient
35 visibility impacts are often ignored (as is done here). Thus light extinction is approximated as
36 the sum of PM light extinction plus Rayleigh light scattering, where the former characterizes the
PRELIMINARY DRAFT-September 2009 2-79
Do Not Quote or Cite
-------
1 PM contribution to visibility impacts and the latter is taken to be a time-invariant constant
2 depending only on elevation above sea level. In the same way PM light extinction is a good
3 measure of VAQ.
4 2.9.1 Particle Mass and Light Extinction
5 If the relative composition of the PM as a function of particle size were held constant and
6 the ambient relative humidity is low (RH<60%) or held constant, there would be a direct linear
7 relationship between PM light extinction and PM mass concentration. Since the PM composition
8 and ambient relative humidity vary in both time and space, there is in general no simple one-to-
9 one correspondence between PM concentration and PM light extinction. These variations result
10 in significant uncertainty (i.e., greater than a factor of two) in predicting PM light extinction
11 from PM mass concentration. However the PM light extinction can be more accurately
12 estimated from PM composition and relative humidity data, using a simple algorithm with
13 assumed light extinction efficiencies for each of the major PM species and water growth factors
14 for the hygroscopic species. In addition PM light extinction can be accurately determined by
15 direct measurements using an integrating nephelometer to measure particle light scattering and
16 an aethalometer or similar instrument to measure particle light absorption.
17 The EPA guidance for tracking progress under the regional haze rule specifies an
18 algorithm for calculating total light extinction as the sum of aerosol light extinction for each of
19 the five major fine particle components and for the coarse fraction mass, plus 10 Mm"1 for light
20 extinction due to Rayleigh scattering as discussed below. This algorithm is represented by the
21 following equation (US EPA, 2009a, equation 9-1):
22
23 bext= (3)f(RH) [SULFATE]
24 + (3)f(RH) [NITRATE]
25 + (4) [ORGANIC CARBON]
26 + (10) [LIGHT ABSORBING CARBON] (Equation 2-2)
27 + (1) [SOIL]
28 +(0.6) [COARSE PM]
29 +10 (for Rayleigh scattering by gases)
30
31 The estimated mass for each component is multiplied by its dry extinction efficiency and,
32 in the case of sulfate and nitrate, by a relative humidity adjustment factor, f(RH), to account for
33 their hygroscopic behavior. The relative humidity adjustment factor increases significantly with
34 higher humidity, ranging from about 2 at 70 percent, to 4 at 90 percent, and over 7 at 95 percent
35 relative humidity.
PRELIMINARYDRAFT-September 2009 2-80
Do Not Quote or Cite
-------
1 Rayleigh scattering represents the degree of natural light scattering found in a particle-
2 free atmosphere, caused by the gas molecules that make up "blue sky" (e.g., N2, 02). The
3 magnitude of Rayleigh scattering depends on the wavelength or color of the light being scattered,
4 as well as on the density of gas in the atmosphere, and varies by site elevation, generally from 8
5 to 12 Mm"1 corresponding to elevations from about 4km above sea level to sea level for green
6 light at about 550 nm. A standard value of 10 Mm-1 is often used to simplify comparisons of
7 light extinction values across a number of sites with varying elevations (Malm, 2000).
8 A refined version of the algorithm described by the relationship above (equation 2-2) was
9 developed for remote area application to better fit the high and low light extinction extremes as
10 needed for the regional haze rule (ISA, equation 9-2). Neither the original or revised version of
11 the algorithm has been explicitly evaluated for urban conditions. However for remote areas, the
12 results are comparable except at the extremes where the revised version produces larger
13 estimates of PM light extinction by having larger dry extinction efficiency terms for some PM
14 components when their concentrations are large. The original version (2-xx) was selected for
15 use in the UF VA because of its simplicity and the judgment that it may be more appropriate for
16 freshly produced PM as would be expected in urban settings. The light extinction coefficient
17 integrates the effects of aerosols on visibility, yet it is not dependent on scene-specific
18 characteristics. It measures the changes in visibility linked to emissions of gases and particles.
19 By apportioning the light extinction coefficient to different aerosol constituents, one can estimate
20 changes in visibility due to changes in constituent concentrations (Pitchford and Malm, 1994).
21 2.9.2 Other Measures of Visibility
22 Light extinction can be converted to two other parameters that have often been used as
23 measures of visibility, visual range and the haziness index expressed in deciview units. Visual
24 range is a measure of visibility that is inversely related to the extinction coefficient. Visual range
25 can be defined as the maximum distance at which one can identify a large black object against
26 the horizon sky. The colors and fine detail of many objects will be lost at a distance much less
27 than the visual range, however. Visual range has been widely used in air transportation and
28 military operations where its characterization as the maximum distance that an object can be seen
29 is helpful. It is less useful for characterizing air quality impacts, because it can be misinterpreted
30 as the distance over which good visibility conditions exist. Conversion from the extinction
31 coefficient to visual range can be made with the following equation (NAPAP, 1991):
32
33 Visual Range (km) = 3912/6exf(Mm"1) (Equation 2-3)
34
PRELIMINARY DRAFT-September 2009 2-81
Do Not Quote or Cite
-------
1 The haziness index with deciview units (abbreviated as dv) was developed specifically to
2 produce a visibility metric that has a near-linear response with respect to perceived visual
3 changes over its entire range in a way that is analogous to the decibel scale for sound (Pitchford
4 and Malm, 1994). Neither visual range nor the extinction coefficient has this property. For
5 example, a 5 km change in visual range or 10 Mm"1 change in light extinction can result in a
6 change that is either imperceptible or very apparent depending on baseline visibility conditions.
7 The haziness index allows one to more effectively express perceptible changes in visibility,
8 regardless of baseline conditions. This is particularly useful in the design and assessment of
9 results of perception, preference, and valuations of visibility effects studies. A one deciview
10 change is a small but perceptible scenic change under many conditions, approximately equal to a
11 10 percent change in the extinction coefficient (Pitchford and Malm, 1994). The deciview value
12 can be calculated from the light extinction (bext) by the equation:
13
14 Haziness (dv) = 10 \n(bext/\0 Mm"1) (Equation 2-4)
15
16 Figure 2-24 graphically illustrates the relationship among the three VAQ metrics
17 described above.
18
19 Figure 2-24. Graphical representation of the relationships between the three commonly
20 used metrics of VAQ.
21
391 km 2OO km 1OO km 5O km 25 km 1O km 5 km
lOMmr1 2O Mm-1 5O Mm-1 1OO Mm-1 2OO Mm-1 4OQ Mirr1 SOO Mm-1
O dv 1O dv 2O dv 3O dv 4O dv SO dv
I •
l%ht Unction
22 2.9.3 Visibility at PM Background Conditions
23 Light extinction caused by PM from natural sources can vary significantly from day to
24 day and location to location due to natural events such as wildfire, dust storms, and volcanic
25 eruptions. It is useful to consider estimates of natural background concentrations of PM on an
26 annual average basis, however, when evaluating the relative contributions of anthropogenic
27 (man-made) and non-anthropogenic sources to total light extinction. Background PM is defined
28 and discussed in detail in section 2.6, and Table 2-x provides the annual average regional
29 background PM2.5 mass ranges for the eastern and western U.S. The National Acid Precipitation
30 Assessment Program report (NAPAP, 1991) provides estimates of extinction contributions from
31 background levels of fine and coarse particles, plus Rayleigh scattering. In the absence of
PRELIMINARYDRAFT-September 2009 2-82
Do Not Quote or Cite
-------
1 anthropogenic emissions of visibility-impairing particles, these estimates are 26 + 7 Mm"1 in the
2 East and 17 + 2.5 Mm"1 in the West. Excluding light extinction due to Rayleigh scattering,
3 annual average background levels of fine and coarse particles are estimated to account for
4 approximately 14 Mm"1 in the East and about 6 Mm"1 in the West. The primary non-
5 anthropogenic substances responsible for natural levels of visibility impairment are naturally
6 occurring organics, suspended dust (including coarse particles), and water associated with
7 naturally occurring hygroscopic particles.
8 At the ranges of fine particle concentrations associated with background conditions,
9 discussed above in section 2.6, small changes in fine particle mass have a large effect on total
10 light extinction. Thus, higher levels of background fine particles and associated average
11 humidity levels in the East result in a fairly significant difference between naturally occurring
12 visual range in the rural East as compared to the rural West. Fine particles originate from both
13 natural and anthropogenic, or man-made, sources. Natural background concentrations of fine
14 particles are those originating from natural sources. On an annual average basis, concentrations
15 of natural background fine particles are generally small when compared with concentrations of
16 fine particles from anthropogenic sources (NRC, 1993). The same relationship holds true when
17 one compares annual average light extinction due to background fine particles with light
18 extinction due to background plus anthropogenic sources. Table VIII-4 in the 1996 Staff Paper
19 (US EPA, 1996) makes this comparison for several locations across the country by using
20 background estimates from Table VIII-2 and light extinction values derived from monitored data
21 from the IMPROVE network. These data indicate that anthropogenic emissions make a
22 significant contribution to average light extinction in most parts of the country, as compared to
23 the contribution from natural background fine particle levels. Anthropogenic contributions
24 account for about one-third of the average extinction coefficient in the rural West and more than
25 80 percent in the rural East (NAPAP, 1991).
26 The draft Urban Focused Visibility Assessment (UFVA) for this review (US EPA,
27 2009c) has included the development of PRB estimates for light extinction during each daylight
28 hour in 15 study areas for 2005-2007 (for days with available PM monitoring data), based on the
29 PRB estimates of PM2.5 components and PMio described in section 2.7, along with data on hour-
30 specific relative humidity.
31 Table 2-4Table 2-4 presents the unweighted 2005-2007 average of the hourly daylight
32 PM light extinction values (in units of Mm"1) for reach of the 15 visibility assessment areas. In
33 the East, these values are somewhat lower than the value cited above for eastern rural areas (26
34 Mm"1), but the latter estimates were for 24-hour average light extinction which includes hours
35 with higher relative humidity. In the West, the value cited above (17 Mm"1) falls within the range
36 of these estimates for the western study areas.
PRELIMINARYDRAFT-September 2009 2-83
Do Not Quote or Cite
-------
1
2
Table 2-4. 2005-2007 Average PRB Daylight PM Light Extinction
Study Area
Tacoma
Fresno
Los Angeles-South Coast Air Basin
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit-Ann Arbor
Pittsburgh
Baltimore
Philadelphia-Wilmington
New York-N.New Jersey-Long Island
2005-2007 Average Policy Relevant Background
Daylight Total Light Extinction, Mm *
22
21
18
18
15
18
20
19
19
19
17
17
19
18
18
4
5 It is important to note that, even in areas with relatively low concentrations of
6 anthropogenic fine particles, such as the Colorado plateau, small increases in anthropogenic fine
7 particle concentrations can lead to significant decreases in visual range. As discussed in the
8 second draft ISA (US EPA, 2009a, section 9.x), visibility in an area with lower concentrations of
9 air pollutants (such as many western Class I areas) will be more sensitive to a given increase in
10 fine particle concentration than visibility in a more polluted atmosphere. Conversely, to achieve
11 a given amount of visibility improvement, a larger reduction in fine particle concentration is
12 required in areas with higher existing concentrations, such as the East, than would be required in
13 areas with lower concentrations. This relationship between changes in fine particle
14 concentrations and changes in visibility (in deciviews) also illustrates the relative importance of
15 the overall extinction efficiency of the pollutant mix at particular locations. At a given ambient
16 concentration, areas having higher average extinction efficiencies, due to the mix of pollutants,
17 would have higher levels of impairment. In the East, the combination of higher humidity levels
18 and a greater percentage of sulfate as compared to the West causes the average extinction
19 efficiency for fine particles to be almost twice that for sites on the Colorado Plateau.
PRELIMINARYDRAFT-September 2009
Do Not Quote or Cite
2-84
-------
1 2.10 REFERENCES
2 Byun D; Schere KL. (2006). Review of the Governing Equations, Computational Algorithms, and Other
3 Components of the Models-3 Community Multiscale Air Quality (CMAQ) Modeling System. Applied
4 Mechanics Reviews 59: 51.
5 Byun DW; Ching JKS. (1999). Science Algorithms of the EPA Models-3 Community Multiscale Air Quality
6 (CMAQ) Modeling System: US Environmental Protection Agency, Office of Research and Development.
7 Chung A; Herner JD; Kleeman MJ. (2001). Detection of alkaline ultrafine atmospheric particles at Bakersfield,
8 California. Environ Sci Technol, 35: 2184-2190.
9 DeBell L. (2006). Spatial and seasonal patterns and temporal variaibility of haze and its constituents in the United States: Report
10 IV. 1 Cooperative Institute for Research in the Atmosphere. Colorado State University Fort Collins, CO.
11 Edgerton ES; Hartsell BE; Saylor RD; Jansen JJ; Hansen DA; Hidy GM. (2005). The Southeastern Aerosol
12 Research and Characterization Study: Part II Filter-based measurements of fine and coarse paniculate
13 matter mass and composition. J Air Waste Manag Assoc, 55: 1527-1542.
14 Edgerton ES; Casucclo GS; Saylor RD; Lersch TL; Hartsell BE; Jansen JJ; Hansen DA. (2009). Measurements of
15 OC and EC in coarse paniculate matter in the southeastern United States. J Air Waste Manag Assoc, 59:
16 78-90.
17 Fine PM; Chakrabarti B; Krudysz M; Schauer JJ; Sioutas C. (2004). Diurnal Variations of Individual Organic
18 Compound Constituents of Ultrafine and Accumulation Mode Paniculate Matter in the Los Angeles Basin.
19 Environ Sci Technol, 38: 1296-1304.
20 Fiore, A.M., D.J. Jacob, H. Liu, R.M. Yantosca, T.D. Fairlie, and Q.Li (2003). Variability in surface ozone
21 background over the United States: Implications for air quality policy. Journal of Geophysical Research.
22 Vol. 108 (D24), 4787.
23 Frank NH. (2006). Retained nitrate, hydrated sulfates, and carbonaceous mass in federal reference method fine
24 paniculate matter for six eastern U.S. cities. J Air Waste Manag Assoc, 56: 500-11.
25 Herner JD; Aw J; Gao O; Chang DP; Kleeman MJ. (2005). Size and composition distribution of airborne paniculate
26 matter in Northern California: I - Paniculate mass, carbon, and water-soluble ions. J Air Waste Manag
27 Assoc, 55:30-51.
28 Malm, W.C. (2000) Introduction to Visibility, ISSN 0737-5352-40, available from the Cooperative Institute of
29 Research in the Atmosphere (CIRA), Colorado State University, Fort Collins, CO, 80523, April 2000.
30 NAPAP (1991) The Experience and Legacy of NAPAP, Report of the Oversight Review Board. National Acid
31 Precipitation Program, Washington, DC.
32 NARSTO (2004) Paniculate Matter Science for Policy Makers: A NARSTO Assessment. P. McMurry, M.
33 Shepherd, and J. Vickery, eds. Cambridge University Press, Cambridge, England. ISBN 0 52 184287 5.
34 National Research Council. 1993. Protecting Visibility in National Parks and Wilderness Areas. National
35 Academy Press, Washington, D.C.
36 Ning Z; Geller MD; Moore KF; Sheesley R; Schauer JJ; Sioutas C. (2007). Daily Variation in Chemical
3 7 Characteristics of Urban Ultrafine Aerosols and Inference of Their Sources. Environ Sci Technol, 41:
38 6000-6006.
PRELIMINARYDRAFT-September 2009 2-85
Do Not Quote or Cite
-------
1 Phuleria HC; Sheesley RJ; Schauer JJ; Fine PM; Sioutas C. (2007). Roadside measurements of size-segregated
2 paniculate organic compounds near gasoline and diesel-dominated freeways in Los Angeles, CA. Atmos
3 Environ, 41:4653-4671.
4 Sardar SB; Fine PM; Mayo PR; Sioutas C. (2005). Size-fractionated measurements of ambient ultrafine particle
5 chemicalcomposition in Los Angeles using the NanoMOUDI. Environ Sci Technol, 39: 932-944.
6 Sardar SB; Fine PM; Sioutas C. (2005). Seasonal and spatial variability of the size-resolved chemical composition
7 of paniculate matter (PM10) in the Los Angeles Basin. J Geophys Res, 110D07S08.
8 Schmidt. M; Frank, N.; Mintz, D.; Rao, T.; McCluney, L. (2005). Analyses of paniculate matter (PM) data for the
9 PM NAAQS review. Memorandum to PMNAAQS review docket EPA-HQ-OAR-2001-0017. June 30,
10 2005. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_td.html.
11 US EPA (1996a). Air Quality Criteria for Paniculate Matter. Research Triangle Park, NC: National Center for
12 Environmental Assessment-RTF Office; report no. EPA/600/P-95/001aF-cF. 3v. Available:
13 http://www.epa.gov/ttn/naaqs/standards/pm/sjmjr cd.html.
14 US EPA (1996b). Review of the National Ambient Air Quality Standards for Paniculate Matter: Policy Assessment
15 of Scientific and Technical Information, O AQPS Staff Paper. Research Triangle Park, NC 27711: Office
16 of Air Quality Planning and Standards; report no. EPA-452/R-96-013. Available:
17 http://www.epa.gov/ttn/naaqs/standards/pm/sjmjr sp.html.
18 US EPA (2004). Air Quality Criteria for Paniculate Matter. National Center for Environmental Assessment, Office
19 of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711;
20 report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October 2004. Available:
21 http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_cr_cd. html
22 US EPA (2005). Review of the National Ambient Air Quality Standards for Paniculate Matter: Policy Assessment
23 of Scientific and Technical Information, OAQPS Staff Paper. Research Triangle Park, NC 27711: Office
24 of Air Quality Planning and Standards; report no. EPA EPA-452/R-05-005a. December 2005.
25 Available: http://www.epa.gov/ttn/naaqs/standards/pm/sjm cr sp.html
26 US EPA (2008a). Integrated Review Plan for the National Ambient Air Quality Standards for Paniculate Matter.
27 National Center for Environmental Assessment and Office of Air Quality Planning and Standards, U.S.
28 Environmental Protection Agency, Research Triangle Park, NC. Report No. EPA 452/R-08-004. March
29 2008. Available at:
30 http://www.epa.gov/ttn/naaqs/standards/pm/data/2008 03 final integrated reviewjlaapdf
31 U.S. EPA (2008b). Integrated Science Assessment for Paniculate Matter: First External Review Draft. National
32 Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
33 Research Triangle Park, NC. EPA/600/R-08/139 and 139A. December 2008. Available:
34 http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_2007_isa.html.
35 US EPA (2009a). Integrated Science Assessment for Paniculate Matter: Second External Review Draft. National
3 6 Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
37 Research Triangle Park, NC. EPA/600//R-08/139B. July 2009. Available:
3 8 http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_2007_isa.html.
39 US EPA (2009b). Risk Assessment to Support the Review of the PM Primary National Ambient Air Quality
40 Standards - External Review Draft. Office of Air Quality Planning and Standards, U.S. Environmental
41 Protection Agency, Research Triangle Park, NC. EPA-452/P-09/006. September 2009. Available:
42 http://www.epa.gOv/ttn/naaqs/standards/pm/s pm index.html.
PRELIMINARYDRAFT-September 2009 2-86
Do Not Quote or Cite
-------
1 US EPA (2009c). Particulate Matter Urban-Focused Visibility Assessment - External Review Draft. Office of Air
2 Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC.
3 EPA-452/P-09/005. September 2009. Available:
4 http://www.epa.gOv/ttn/naaqs/standards/pm/s pm index.html.
PRELIMINARYDRAFT-September 2009 2-87
Do Not Quote or Cite
-------
1 3 POLICY-RELEVANT ASSESSMENT OF HEALTH EFFECTS
2 EVIDENCE
3 3.1 INTRODUCTION
4 This chapter assesses key policy-relevant information on the known and potential health
5 effects associated with short- and long-term exposures to ambient particulate matter (PM). In the
6 last two PM NAAQS reviews, EPA focused on particle mass and primarily distinguished
7 between two categories of particle pollution based on size (i.e., fine- and thoracic coarse-fraction
8 particles) and conducted parallel evaluations of the available scientific evidence relating to each
9 category. The importance of specific PM components and sources was evaluated within the
10 context of this basic size differentiation. In those reviews, it was determined that size-
11 fractionated particle mass, rather than particle composition, remained the most appropriate
12 approach for addressing ambient PM. Building upon lessons learned in the last two reviews,
13 EPA's review of the available scientific evidence, as presented in the second draft Integrated
14 Science Assessment (ISA, US EPA, 2009a), is based on particle size, considering fine and
15 thoracic coarse-fraction particles separately. Within this basic structure, EPA is also evaluating
16 relevant, but more limited, scientific evidence on additional size fractions (e.g., ultrafme particles
17 (UFPs) and specific components, sources, and environments (e.g., urban and non-urban
18 environments).
19 The presentation of the policy-relevant health effects evidence in this chapter first
20 summarizes the qualitative assessment of health evidence contained in the second draft ISA, as a
21 basis for the evidence-based assessment of the primary standards for PM to be presented in
22 chapter 5 of the external review draft Policy Assessment (PA). Secondly, this assessment
23 addresses key issues relevant to the quantitative assessment of the epidemiologic health evidence
24 available in this review so as to provide a foundation for the quantitative health risk assessment
25 to be discussed in chapter 4 and used as the basis for the risk-based assessment of the primary
26 standards for PM to be presented in chapter 5.
27 In the last review of the PM NAAQS, a broad range of health effects associated with
28 ambient PM were evaluated with emphasis placed on those health endpoints for which the Air
29 Quality Criteria Document (AQCD, US EPA, 2004) concluded that the associations were likely
30 to be causal. The body of scientific evidence reviewed included consideration of hundreds of
31 new studies conducted in many countries around the world. Of special importance in the last
32 review were (1) a large number of new single- and multi-city time-series studies that evaluated
33 the relationship between short-term PM exposures and mortality and morbidity, including new
34 evidence of health endpoints related to cardiovascular morbidity; (2) extensive reanalyses of a
35 number of important time-series studies to address issues associated with statistical model
PRELIMINARY DRAFT-September 2009 3-1
Do Not Quote or Cite
-------
1 specification (HEI, 2003); (3) reanalysis and validation of critical long-term exposure cohort
2 studies, specifically, the Harvard Six Cities Study and the American Cancer Society (ACS)
3 study, which provided evidence of generally robust associations between PM2 5 and mortality and
4 provided a basis for greater confidence in the reported associations than in the review completed
5 in 1997 (Krewski et. al, 2000); and (4) additional toxicological and controlled human exposure
6 evidence to support the plausibility of the epidemiologic evidence and provide insights into
7 potential mechanisms for PM-related effects.
8 The EPA's conclusion in the last review that fine and thoracic coarse particles should
9 continue to be considered as separate pollutants was based on long-established information on
10 differences in physical and chemical properties, sources, and atmospheric formation and
11 transport (US EPA, 2005, section 2.2). In this review, EPA has evaluated the newly available
12 evidence related to the physics and chemistry of PM, exposure relationships, and particle
13 dosimetry and concludes that the chemical and physical distinctions between fine and coarse
14 particles recognized in the previous reviews remain generally unchanged; recent studies continue
15 to show that fine and coarse particles generally have different sources and composition and
16 different formation processes and deposition rates (See Table 2-1).
17 The assessment of health evidence in this chapter, therefore, focuses on health effects
18 associated with fine and thoracic coarse particles, with more limited evidence presented for PM
19 components and UFPs. Partly as a consequence of EPA's decision in the previous reviews to
20 consider fine particles and thoracic coarse particles separately, much new information is now
21 available on PM air quality and human health effects directly in terms of PM2.5 and, to a much
22 more limited degree, PMio -2.5. Since the purpose of this review is to evaluate the adequacy of
23 the current primary PM standards, which separately address fine and thoracic coarse particles,
24 staff has focused this policy assessment and associated quantitative analyses primarily on the
25 evidence related directly to PM2.5 and PMio -2.5. In so doing, staff has considered PMio-related
26 evidence primarily to help inform our understanding of key issues and to help interpret and
27 provide context for understanding the public health impacts of ambient fine and thoracic coarse
28 particles.
29 This assessment draws from the second draft ISA's evaluation and conclusions on the
30 full body of evidence from health studies, summarized in chapters 6 through 8 of the ISA, with
31 consideration of the integrative synthesis presented in chapter 2. Chapter 2 of the second draft
32 ISA focuses on synthesizing newly available scientific information with evidence available from
33 the previous PM NAAQS reviews, integrated from various disciplines, so as to inform
34 consideration of the policy-relevant questions that are central to EPA's assessment of scientific
35 information upon which this review of the PM NAAQS is to be based (US EPA, 2008a, section
36 3.1 and Chapter 4). It is intended to provide an integrative overview of human health effects
37 elicited by ambient PM in the U.S., and to facilitate consideration of the key policy-related issues
PRELIMINARY DRAFT-September 2009 3 -2
Do Not Quote or Cite
-------
1 that will inform staff conclusions related to indicators, averaging times, levels, and forms in
2 chapters.
3 As presented and discussed in chapters 6 through 8 of the second draft ISA, a large
4 number of new studies containing further evidence of health effects associated with PM
5 exposures have been published since the last review, with important new information coming
6 from epidemiologic, lexicological, controlled human exposure, and dosimetric studies, including
7 information on effects in at-risk populations. As was true in the previous two reviews, evidence
8 from epidemiologic studies plays a key role in the second draft ISA's evaluation of the scientific
9 evidence. As discussed further in section 3.3, some highlights of the new evidence include:
10
11 • New multi-city time-series studies that use uniform methodologies to investigate the
12 effects of PM2.5 on health with data from multiple locations representing varying
13 regions and seasons representative of different climate and air pollution mixes. These
14 studies provide more precise estimates of the magnitude of an effect of exposure to
15 PM2 5 than most smaller-scale individual city studies that were more commonly
16 available in the last review. The new studies also contribute to improving our
17 understanding of the role of various potential confounders, including gaseous co-
18 pollutants, on observed associations.
19
20 . Recent studies investigating cardiovascular morbidity and mortality provide some of
21 the strongest evidence for cardiovascular effects related to PM2.5 exposures. A
22 number of large multi-city studies have been conducted throughout the U.S. and have
23 reported consistent increases in cardiovascular morbidity and mortality related to
24 ambient PM2.5 concentrations. These effects are biologically plausible and coherent
25 with available toxicological studies.
26
27 . Extended analyses of key long-term exposure cohort studies that were critical in
28 evaluating PM2.5-related mortality in the last review (e.g., ACS and Harvard Six
29 Cities Studies) continue to provide evidence that the data are robust and that reducing
30 PM2 5 improves public health. Results of new studies employing within-city exposure
31 gradients suggest that reduced exposure misclassification may result in greater
32 magnitudes of association.
33
34 . New evidence has become available documenting the population health benefits of
35 reducing ambient air pollution by correlating past reductions in ambient PM2 5
36 concentrations with increased life expectancy and by providing evidence for
37 improvement in community health following reduction in fine particle exposures.
38 These studies add further support to the results of the hundreds of other
39 epidemiologic studies linking ambient PM exposure to an array of health effects.
40 Such studies showing improvements in health with reductions in emissions of
41 ambient PM and/or gaseous co-pollutants provide strong evidence that reducing
42 emissions of PM and gaseous pollutants has beneficial public health impacts.
43
44
PRELIMINARY DRAFT-September 2009 3 -3
Do Not Quote or Cite
-------
1 . Recent studies of long-term exposure to PM2.5 have expanded the evidence for effects
2 on lung development in children. The extended follow-up for the Southern California
3 Children's Health Study includes evidence that decreased lung function effects
4 associated with long-term exposure to PM2.5 persist into early adulthood, and are
5 more robust and larger in magnitude than reported in the prior review. Supporting
6 these results are new longitudinal cohort studies conducted by other researchers in
7 other locations using different methods provide enhanced evidence for respiratory
8 effects related to long-term fine particle exposures.
9
10 . New evidence for respiratory symptoms and aggravation of preexisting respiratory
11 disease strengthens evidence from the last review and provides increased consistency
12 as results have reported by several different researchers in different countries using
13 different designs. The most recent studies have focused on effects in children, but a
14 few studies have also reported associations in adults.
15
16 . Additional evidence supporting no discernible thresholds below which PM-related
17 health effects are absent.
18
19 . New analyses informing our understanding of issues related to potential confounding
20 by gaseous co-pollutants, exposure misclassification (e.g.., measurement error, lag
21 period), model specification, seasonal and regional influences, and the shape of the
22 concentration-response function.
23
24 . Emerging, though still limited evidence on effects associated with exposure to PMi0.
25 2.5, PM components, and UFPs.
26
27 In addition, the body of evidence on PM-related effects has greatly expanded with
28 findings from studies that help inform modes of action, including important new dosimetric,
29 lexicological, and controlled human exposure studies.
30
31 .In general, potential pathways or modes of action do not appear to be specific to a
32 particular PM size fractions, however emerging evidence related to the characteristics
33 of UFPs may inform unique modes of action or effects disproportionate to their mass.
34 In addition, for UFPs, particle number rather than mass may be the appropriate metric
35 to evaluate, however, UFP number concentrations are highly dependent on monitor
36 location (i.e., concentrations drop off quickly from the road compared to
37 accumulation mode particles, and, therefore, more subject to exposure error than
38 accumulation mode particles).
39
40 . Animal and controlled human exposure studies using concentrated ambient particles
41 (CAPs), various indicators of response (e.g., heart rate variability), and animal models
42 simulating at risk populations, continue to demonstrate plausibility of the
43 epidemiologic evidence and provide insights into potential mechanisms for PM-
44 related effects.
45
PRELIMINARY DRAFT-September 2009 3 -4
Do Not Quote or Cite
-------
1 . The findings of new toxicological and controlled human exposure studies continue to
2 provide support that a number of potential biologic mechanisms or pathways for PM-
3 related cardiovascular and respiratory effects exist.
4
5 . New evidence has developed to inform our understanding of the role of certain
6 factors that may affect our understanding of at-risk populations including: gender,
7 age, and pre-existing lung disease on deposition and clearance.
8
9 In presenting the evidence and conclusions based on it, this chapter first summarizes
10 information from the second draft ISA's evaluation of health evidence integrated across different
11 disciplines (e.g., epidemiology, toxicology, controlled human exposure). Sections 3.2 and 3.3
12 provide overviews of the EPA's integrative findings and conclusions regarding dosimetry and
13 modes of action (MOA), causality findings, and on the nature of effects associated with PM
14 exposures. Section 3.4 addresses several key issues relevant to the staffs interpretation and
15 quantitative assessment of the health evidence, including: (1) questions related to exposure time
16 periods used in associations between air quality and health effects, including lag periods used in
17 short-term exposure studies and the selection of time periods used to represent exposures in long-
18 term exposure studies; (2) questions related to the shape of concentration-response relationships
19 and potential threshold levels; (3) considerations related to air quality measurements and data
20 used in the health studies; (4) measurement error and exposure error in fine and thoracic coarse
21 particle studies; (5) specification of models used in epidemiologic studies; and (6) approaches to
22 evaluating the role of co-pollutants and potential confounding in PM-effects associations. In this
23 section, staff builds upon the second draft ISA's and 2004 AQCD's detailed evaluation and
24 integration of the scientific evidence on these issues to reach conclusions regarding the use of the
25 health study results in quantitative evaluation and the PM risk assessment to be discussed in
26 chapter 4. Section 3.5 draws from chapters 2 and 8 of the second draft ISA to characterize
27 potential at-risk populations and potential public health impacts of exposure to ambient PM.
28 3.2 MECHANISMS
29 This section provides an overview of evidence related to potential mechanisms by which
30 exposure to PM may result in adverse health effects, drawing upon evidence presented in
31 Chapters 4 through 7 of the second draft ISA (US EPA, 2009a). Evidence from dosimetric
32 studies has played a key role in previous PM NAAQS reviews, especially in the decision to
33 revise the indicator from total suspended particulates (TSP) to PMio to focus on thoracic particles
34 (52 FR 24634, July 1, 1987). In the last review, EPA recognized that much new evidence was
35 available on potential mechanisms or pathways for PM-related effects, ranging from effects on
36 the respiratory system to indicators of cardiovascular response. The new findings advanced our
37 understanding of the complex and different patterns of particle deposition and clearance in the
38 respiratory tract and provided insights into potential modes of action (MOA) for PM-related
PRELIMINARY DRAFT-September 2009 3 -5
Do Not Quote or Cite
-------
1 effects as well as supported the plausibility of the epidemiologic evidence (US EPA, 2004,
2 Chapter 7; US EPA, 2005, section 3.2).
3 In this review, although the basic understanding of the mechanisms governing deposition
4 and clearance of inhaled particles has not generally changed, there is additional information
5 available to provide insights on the role of certain biological determinants such as gender, age,
6 and lung disease on deposition and clearance. In addition, new evidence provides further
7 characterization of the retention and translocation of UFPs following deposition in the
8 respiratory tract (US EPA, 2009a, chapter 4) as well as providing additional insights into
9 possible MOA for pulmonary and extra-pulmonary effects related to PM exposures, in particular,
10 cardiovascular effects (US EPA, 2009a, chapter 5). Policy-relevant information is presented
11 below considering first dosimetric studies and second studies evaluating modes of action
12 (MOA).
13 3.2.1 Dosimetry
14 An evaluation of the ways by which inhaled particles might ultimately affect human
15 health must take into account particle dosimetry which refers to patterns of deposition,
16 translocation, clearance, and retention of particles and their constituents within the respiratory
17 tract and extrapulmonary tissues (US EPA, 2009a, Chapter 4). Briefly, the human respiratory
18 tract can be divided into three main regions: (1) extrathoracic (upper airways), (2)
19 tracheobronchial, and (3) alveolar34 (US EPA, 2009a, Figure 4-1). The regions differ markedly
20 in structure, function, size, mechanisms of deposition and removal, and sensitivity or reactivity
21 to deposited particles. Overall, the site of particle deposition within the respiratory tract has
22 implications related to lung retention and surface dose of particles as well as potential systemic
23 distribution of particles and/or PM components (US EPA, 2009a, section 4.1.2). .
24 Inhaled particles may be either exhaled or deposited in the extrathoracic,
25 tracheobronchial, or alveolar region. The dose from inhaled particles deposited and retained in
26 the respiratory tract is governed by a number of factors including: (1) exposure concentration
27 and duration; (2) activity and ventilatory parameters; and (3) particle properties (e.g., particle
28 size, hygroscopic!ty35, and solubility in airway fluids and cellular components) (US EPA, 2009a,
29 section 4.1). Particle number is most highly concentrated in the ultrafme size range and volume
30 (or mass) is most concentrated in the larger size ranges (US EPA, 2009a, section 3.2). The basic
31 characteristics of particles as they relate to deposition and retention, as well as anatomical and
32 physiological factors influencing particle deposition and retention have been discussed in detail
33 in previous science assessments (US EPA, 1996, Chapter 10; US EPA, 2004, Chapter 6). In this
34 review, EPA has focused on (1) dosimetry evidence that informs our understanding of issues that
34 The term "lower airways" is used to refer to the intrathoracic airways, that is, the combination of the
tracheobronchial region which is the conducting airways and the aleveolar region which is the functional part of the
lung (US EPA, 2009a, section 4.1.2)
35Hygroscopicity relates to the ability of particles to absorb moisture from the atmosphere.
PRELIMINARY DRAFT-September 2009 3 -6
Do Not Quote or Cite
-------
1 may affect the susceptibility of an individual to adverse effects and (2) evidence providing new
2 insights into possible modes of action including evidence informing our ability to extrapolate
3 findings between studies and between species.
4 Fine particles, including and ultrafine particles, and thoracic coarse particles can all
5 penetrate into and be deposited in the tracheobronchial and alveolar regions of the respiratory
6 tract (i.e., the "thoracic" regions). Differences in the distribution of these size fractions has been
7 observed. Since the nasal passages (extrathoracic region) are more efficient at removing inhaled
8 particles than the oral passage (mouth), an individual's mode of breathing (i.e., oral vs. nasal)
9 influences the quantity of particles penetrating to the lung. Once past the extrathoracic region,
10 particle deposition in the tracheobronchial and alveolar regions varies with different exertion
11 levels or breathing patterns. Furthermore, breathing patterns may change with increasing age.
12 As shown in Figures 4-3 and 4-4 of the second draft ISA (reproductions of Figures 6-16 and 6-
13 17, US EPA, 2004), mathematical models estimate that the deposition fractions in these regions
14 are largest for particles in the thoracic coarse fraction and ultrafine modes. Note that the
15 fractional deposition in the alveolar region of the respiratory system for healthy individuals is
16 greatest for particles in the size ranges of approximately 2.5 to 5 |im and 0.02 to 0.03 jim, and
17 fractional deposition to the tracheobronchial region is greatest for particles in the size range of
18 approximately 3 to 6 |im (US EPA, 2009a, section 4.2). Both experimental and modeling
19 techniques are based on many assumptions that may be relatively good for the healthy lung but
20 not for the diseased lung (US EPA, sections 4.2.4.4 and 4.2.4.5).
21 Interspecies similarities and differences in deposition were described in detail in the
22 previous two PM reviews (US EPA, 1996, section X.X; US EPA, 2004, section x.x). More
23 recently, Brown et al. (2005) conducted a thorough evaluation of extrapolations between rats and
24 humans in relation to PM exposures. One of the many factors they considered was the choice of
25 a dose metric appropriate for comparison between species. For example, deposited mass may be
26 an appropriate PM indicator for health effects associated with soluble PM constituents. For
27 health effects associated with insoluble PM, the particle number, surface area, or mass may be
28 appropriate indicators. Given interspecies differences in deposition patterns and clearance rates,
29 the question of retained versus deposited dose, Brown et al. (2005) concluded that for acute
30 effects, the maximum deposited incremental dose may be the appropriate type of dose metric.
31 For chronic effects, long-term burden may be more appropriate (US EPA, 2009a, section 4.2.3).
32 Evaluation of factors affecting particle deposition is important to help understand at-risk
33 populations. Differences in biological response following PM exposures may be caused by
34 dosimetry differences as well as by differences in innate sensitivity. The effects of different
35 biological factors on particle deposition are summarized in section 4.2.4 of the second draft ISA
36 building upon a more complete discussion of these factors prepared for the previous PM NAAQS
37 review (US EPA, 2004, section x.x). These factors include:
PRELIMINARY DRAFT-September 2009 3 -7
Do Not Quote or Cite
-------
1
2 • Physical activity - the activity level of an individual is well recognized to affect their
3 minute ventilation and route of breathing. Humans are oronasal breathers tending to
4 breathe through the nose when at rest and increasingly through the mouth with
5 increasing activity level. There is considerable inter-individual variability in both the
6 route by which people breathe and the way the breathing pattern changes occur with
7 changes in activity levels. Changes in breathing patterns and flow rates may alter the
8 regional deposition of particles (US EPA, 2009a, section 4.2.4.1).
9
10 • Age - airway structure and respiratory conditions vary with age, and these variations
11 may alter the amount and site of particle deposition in the respiratory tract. Limited
12 evidence suggests that children may tend to have more oral breathing both at rest and
13 during exercise and that children may also display more variability than adults with
14 respect to their oronasal pattern of breathing with exercise (Becquemin et al 1999;
15 Bennett et al., 2008; James et al., 1997). In addition, models suggest that nasal
16 deposition of particles is less efficient in children. These findings suggest that
17 children's lower respiratory tract (tracheobronchial and alveolar regions) may receive
18 a higher dose of ambient PM compared to adults and that for a given height and age,
19 children with higher body mass index (BMI) may have increased deposition of fine
20 particles (Bennett and Zeman, 2004). Bennett et al. (2008) also recently reported
21 measures of fine particle (1 and 2 um) deposition associated with light exercise in
22 children and adults and showed that, like with resting breathing, deposition fraction
23 was predicted by breathing pattern and did not differ by age or tended to be less in
24 children compared to adults. On the other hand, children generally have a faster
25 breathing rate than adults relative to their lung volumes, therefore, the rate of
26 deposition of fine particles normalized to lung surface area may be greater in children
27 than in adults (Bennett and Zeman, 1998). While older adults have been recognized
28 as an at-risk population for PM-related effects, limited information is available to
29 assess particle deposition in this population (US EPA, 2009a, section 4.2.2.2 and
30 section 4.2.4.2).
31
32 • Gender - Males and females differ in body size, size of their upper airways, and
33 ventilator parameters. Studies indicate variability in deposition efficiencies due to
34 inter-individual difference in lung size and anatomical variability in airway
35 dimensions and branching patterns. Deposition of thoracic coarse particles and UFPs
36 was greater in females than in males (Kim and Hu, 1998; Kim and Jaques, 2000; as
37 summarized in US EPA, 2009a, Figure 4-5 and section 4.2.4.3).
38
39 • Anatomical Variability - Nasal anatomy may influence the efficiency of particle
40 uptake and deposition. Individual differences may occur in both the nasal airways as
41 well as within the branching structures of the lung (US EPA, 2009a, section 4.2.4.4).
42
43 • Respiratory Disease - Respiratory disease status can also affect regional particle
44 deposition patterns. The presence of respiratory tract disease can affect airway
45 structure and ventilator parameters, thus altering particle deposition in individuals
PRELIMINARY DRAFT-September 2009 3 -8
Do Not Quote or Cite
-------
1 with pre-existing respiratory disease compared to healthy individuals. Both
2 experimental and modeling techniques are based on many assumptions that may be
3 relatively good for the healthy lung but not representative of the diseased lung (US
4 EPA, 2004, sections 4.2.4.4 and 4.2.4.5). Studies have indicated that, in general,
5 enhanced deposition of particles occurs at airway bifurcations (US EPA, 2004,
6 section 6.2.2.4). Evidence available in the last review indicated that people with
7 chronic obstructive lung diseases can have increased total lung deposition and can
8 also show increases in local deposition ("hot spots") due to uneven airflow in
9 diseased lungs (US EPA, 2004, section 6.2.3.4). Individuals with chronic obstructive
10 pulmonary disease (COPD) have very heterogenous deposition patterns and
11 differences in regional deposition compared to healthy individuals and generally
12 show greater particle deposition in the tracheobronchial region than do healthy
13 individuals. Furthermore, there is, in general, an inverse relationship between
14 bronchoconstriction and the extent of deposition in the alveolar region, whereas total
15 respiratory tract deposition generally increases with increasing degrees of airway
16 obstruction (US EPA, 2009a, section 4.2.4.5; US EPA, 1996, section x.x; US EPA,
17 2004, section x.x). In such cases, the respiratory condition can enhance sensitivity to
18 inhaled particles by increasing the delivered dose overall as well as increasing doses
19 to localized regions of the respiratory tract. Such dosimetry studies are of obvious
20 relevance to identifying at-risk populations (see section 3.6 below).
21
22 • Hygroscopicity of Aerosols - Hygroscopicity relates to the ability of particles to
23 absorb moisture from the atmosphere. The high relative humidity in the lungs
24 contributes to rapid growth of hygroscopic particles and may dramatically alter the
25 deposition characteristics of hygroscopic aerosols relative to nonhygroscopic
26 aerosols. Examples of hygroscopic ambient aerosols include sulfates, nitrates, some
27 organics, and aerosols containing significant amounts of sodium or potassium (US
28 EPA, 2009a, section 4.2.4.6).
29
30 The potential effects of deposited particles are influenced by the rate and nature of
31 removal. The predominant clearance and translocation mechanisms vary across the three regions
32 of the respiratory tract and also vary with solubility of the particles.36 For example, particles
33 deposited in the extrathoracic region are cleared by several mechanisms. Particles deposited in
34 the mouth are generally assumed to be swallowed or removed by expectoration. Particles
35 deposited in the posterior portions of the nasal passages are moved via mucociliary transport
36 towards the nasopharynx and swallowed. Mucus flow in the most anterior portion of the nasal
37 passages is forward, toward the opening of the nose where removal occurs by sneezing, wiping,
36 The term "clearance" is used here to refer to the processes by which deposited particles are removed by
mucociliary action or phagocytosis from the respiratory tract. "Translocation" is used mainly to refer to the
movement of free particles across cell membranes and to extrapulmonary sites. In the literature, translocation may
also refer to the extra- and intracellular dissolution of particles and the subsequent transfer of dissociated material to
the blood through extra- and intracellular fluids and across the various cell membranes and lung tissues.
Translocation can occur across cell membranes within the lung as well as to extrapulmonary sites. For example,
particles deposited on the epithelial surface of the lung may translocate to the interstitium.
PRELIMINARY DRAFT-September 2009 3 -9
Do Not Quote or Cite
-------
1 or nose blowing (US EPA, 2009a, section 4.3.1.1). The primary mechanism for particle
2 clearance or translocation from the tracheobronchial region is mucociliary transport and
3 coughing (US EPA, 2009a, section 4.3.1.2; US EPA, 2004, Table 6-2). Dissolution or
4 absorption of particles or particle constituents and endocytosis37 by cells such as macrophages38
5 where the cells are cleared by the mucociliary escalator is the primary clearance mechanism
6 operating in the alveolar region. Under normal conditions, a small fraction of ingested particles
7 may also be cleared through the lymphatic system (US EPA, 2009a, section 4.3.1.3).
8 Many factors may influence particle clearance. The movement of soluble materials
9 depends on the site of deposition in the lung, the rate of material dissolution from particles, and
10 the molecular weight of the solute. The rate of soluble material clearance from the lungs
11 depends on epithelial permeability which may be affected by age, physical activity, respiratory
12 disease, and concurrent exposures (US EPA, 2009a, section 4.4.2). For example, limited
13 evidence shows that particle clearance may be decreased throughout the respiratory tract with
14 increasing age (US EPA, 2009a, section 4.3.4.1). Respiratory disease may influence both the site
15 of initial particle deposition and rate of mucociliary clearance from the airways (U.S. EPA,
16 2009a, section 4.3.4.3). There is no evidence to support that gender affects clearance rates (US
17 EPA, 2009a, section 4.3.4.2).
18 Particles depositing in the mouth are generally assumed to be swallowed or removed by
19 expectoration. Particles deposited in the posterior portions of the nasal passages or the
20 tracheobronchial regions are moved via mucociliary transport towards the nasopharynx and
21 swallowed. Clearance from the extrathoracic and tracheobronchial regions generally occurs over
22 hours to days, however, clearance from the alveolar region is much slower, occurring over
23 months to years (US EPA, 2009a, section 5.1.9.1). Although clearance from the
24 tracheobronchial region is generally rapid, there appears to be a fraction of material deposited in
25 this region in humans and dogs that is retained much longer. The underlying sites and
26 mechanisms of long-term tracheobronchial retention are not known though they may be
27 accounted for by differences in deposition patterns (US EPA, 2009a, section 4.3.1.2). The
28 primary alveolar clearance mechanism is macrophage phagocytosis and migration to terminal
29 bronchioles where the cells are cleared by the mucociliary escalator (US EPA, 2009a, section
30 4.3.1.3).
31 There are differences between species in both the rates of particle clearance from the lung
32 and the manner in which particles are retained in the lung. In contrast to humans, mice and rats
33 appear to have negligible long-term retention of particles, including UFPs, in the
37 Endocytosis is the process of cellular ingestion by which the plasma membrane folds inward to bring a
substance(s) into the cell.
38 A macrophage is a type of white blood cell. Macrophages develop from circulating monocytes that migrate from
the blood into tissues throughout the body, including the lungs.
PRELIMINARYDRAFT-September 2009 3-10
Do Not Quote or Cite
-------
1 tracheobronchial airways (US EPA, 2009a, sections 4.3.1.2, 4.3.2). In addition, clearance from
2 both the tracheobronchial and alveolar region is more rapid in rodents than humans.
3 Soluble particles and soluble components of particles may be retained in the lung or
4 absorbed through the epithelium and distributed throughout the body. The rate of dissolution
5 depends on a number of factors including particle surface area and chemical structure. In the
6 ciliated airways, soluble particles and soluble components of particles are cleared by mucociliary
7 transport and diffuse into underlying tissues and the blood. In the alveolar regions, the thin
8 barrier between the air and blood allows for rapid transport of these soluble particles/components
9 into the blood. While enhanced clearance of insoluble particles acts to reduce dose to airway
10 tissue, increased transport of soluble matter into the bloodstream and, thus, throughout the body
11 potentially enhance effects on extra-pulmonary organs (US EPA, 2009a, section 4.4).
12 The importance of particle translocation to the brain is not yet understood. Studies,
13 primarily in rodents, have demonstrated the translocation of soluble and poorly soluble particles
14 from the olfactory mucosa via the axons to the olfactory bulb in the brain. The rate of
15 translocation appears to be rapid, perhaps less than an hour. It is unclear to what extent
16 translocation to the olfactory bulb and other brain regions may vary between species.
17 Interspecies differences may predispose rats, more so than humans, to the deposition of particles
18 in the olfactory region with subsequent translocation to the olfactory bulb (US EPA, 2009a,
19 section 4.3.3.2 and 4.3.5)
20 In summary, new evidence from dosimetry studies has advanced our understanding of the
21 complex and different patterns of particle deposition and clearance in the respiratory tract
22 exhibited by fine particles, ultrafine particles, and thoracic coarse particles. The evidence
23 continues to show that all three size fractions (PM2.s, PMi0-2.5, and UFPs) can enter the
24 tracheobronchial or alveolar regions of the respiratory system and potentially cause effects.
25 Additional information has become available in this review to provide insights on the role of
26 certain biological determinants such as gender, age, and lung disease on particle deposition and
27 clearance. As summarized in section 4.4.3 of the second draft ISA:
28
29 The healthy airway and alveolar epithelium is generally impermeable to very large
30 insoluble macromolecules and particles. Water and acid soluble particles may more
31 rapidly move through the epithelium as they dissolve on the airway surface or within the
32 phagolysomes of macrophages. The presence of airway inflammation in a variety of
33 airway diseases (e.g., asthma, fibrosis, ARDS, pulmonary edema, inflammation from
34 smoking) alters epithelial integrity to allow more rapid movement of these solutes into
35 the bloodstream. While diabetics are another group recently shown to have increased
36 susceptibility to particulate air pollution (Zanobetti and Schwartz, 2002), it is unclear
37 whether transport of soluble particles across the epithelium is affected in these patients.
38 In general, it appears that co-exposure to irritant pollutants results in a disruption of
39 epithelial integrity and macrophage function which, on the one hand, retards mucociliary
40 and alveolar clearance, but also allows for a more rapid movement of soluble constituents
PRELIMINARYDRAFT-September 2009 3-11
Do Not Quote or Cite
-------
1 across the epithelial surface into the interstitium and blood stream. Alterations in
2 epithelial permeability by disease, pollutant exposure, or infection may partially explain
3 increased susceptibility to PM associated with these co-conditions.
4 3.2.2 Possible Pathways/Modes of Action
5 A major research need identified in the PM review completed in 1997 was improvement
6 of our understanding of the potential biologic MO A by which deposited particles could result in
7 the varying effects observed in epidemiologic studies with PM exposure (ADD REF). In the last
8 review, new evidence from toxicological and controlled human exposure studies was available
9 that provided insights to advance our understanding of potential MOA for PM-related effects and
10 to support the plausibility of findings observed in epidemiologic studies. New evidence in the
11 last review from toxicological and controlled human exposure studies helped to identify and
12 provide support for a number of potential pathways by which particles could elicit biological
13 effects. Fully defining the MOA for PM would involve description of the pathogenesis or origin
14 and development tof any related diseases or processes resulting in premature mortality. While the
15 evidence summarized in the last review provided important insights that contributed to the
16 plausibility of effects observed in epidemiological studies, the more ambitious goal of fully
17 understanding fundamental MOA for how particles produce specific health effects had yet to be
18 attained. However, some of the more important findings presented in last review including those
19 related to the cardiovascular system, were generally described as intermediate responses
20 potentially caused by PM exposure rather than complete mechanisms. Based upon the evidence
21 available in the last review, EPA concluded that it appeared unlikely that the complex mixes of
22 particles that are present in ambient air would act alone through any single pathway of response.
23 Accordingly, EPA concluded that it was plausible that several physiological responses might
24 occur in concert to produce the reported health endpoints (US EPA, 2004, Chapter 7; US EPA,
25 2005, section 3.2).
26 In assessing the more recent animal, controlled human exposure, and epidemiologic
27 information, the second draft ISA developed a summary of current thinking on
28 pathophysiological MOA for the effects related to PM exposure. Chapter 5 of the second draft
29 ISA discusses a series of potential MOA or potential general pathways for pulmonary and
30 extrapulmonary effects. Prior to this review, much of the evidence for the proposed modes of
31 action was obtained from animal studies involving intratracheal (IT) instillation or inhalation of
32 high concentrations of PM and from cell culture experiments. In many cases, the types of PM
33 used were of questionable relevance to ambient exposures (i.e., high concentrations of ROFA,
34 metals and ambient PM collected on filters). Since then, many inhalation studies have been
35 conducted using a variety of particles including concentrated ambient particles (CAPs) from
36 various urban environments, combustion-derived PM (including gasoline and diesel exhaust
37 studies), carbon black, and metals, generally using concentrations of PM lower than 1 mg/m3.
38 Much of this research has been conducted in animal models of disease, however, the levels of
PRELIMINARYDRAFT-September 2009 3-12
Do Not Quote or Cite
-------
1 PM2.5 evaluated in these studies are not generally representative of current ambient levels of
2 PM2.5. These key new studies, described in detail in Chapters 6 and 7 of the second draft ISA,
3 add to the understanding of modes of action which are relevant to ambient PM exposure. A
4 compilation of the pertinent results of recent inhalation studies is presented in section 5.6 of the
5 second draft ISA.
6 The second draft ISA's conclusions on the evidence supporting different types of effects
7 is briefly summarized below. The characterization of evidence is presented for PM in general,
8 since most of the potential pathways or modes of action do not appear to be specific to a
9 particular size class of PM. However, characteristics of UFPs may allow for unique modes of
10 action or effects disproportionate to their mass and, where appropriate, evidence specific to UFPs
11 is briefly highlighted below.
12 Respiratory Effects. Modes of action underlying pulmonary effects associated with PM
13 exposures have been well-studied in the laboratory and, in general, there is agreement regarding
14 the key roles played by cellular injury and inflammation. These pathways are initiated following
15 deposition of inhaled particles on respiratory tract surfaces (US EPA, 2009a, section 5.1).
16 Evidence that supports hypotheses on direct pulmonary effects includes toxicological and
17 controlled human exposure studies using both sources of ambient particles and combustion-
18 related particles, including from gasoline and diesel exhaust. A great deal of research has
19 focused on the role of reactive oxygen species (ROS) in the initiation of pulmonary injury and
20 inflammation following PM exposures. Particles may also interact with cells leading to the
21 induction of nitric oxide synthase and the production of nitric oxide and other reactive nitrogen
22 species (RNS). Particles may act as a direct or indirect source of ROS/RNS in the respiratory
23 system. Although all size fractions of PM may contribute to oxidative and nitrosative stress,39
24 UFPs may contribute disproportionately to their mass due to their large surface/volume ratio and
25 greater particle number compared to larger size fractions (US EPA, 2009a, section 5.1.1).
26 Particle exposure may alter pulmonary function by a variety of different MOA including
27 activation of irritant receptors, production of ROS/RNS and pulmonary inflammation and injury
28 (US EPA, 2009a, Figure 5-4). These potential modes of actions can lead to the development of a
29 range of respiratory effects including:
30
31 . Airway hyperresponsiveness (AHR) - generally considered a short-term response
32 (US EPA, 2009a, section 5.1.6).
33
39 .In general, high levels of intracellular ROS/RNS can lead to irreversible protein modifications, loss of cellular
membrane integrity, dNA damage and cellular toxicity. Lower levels of ROS/RNS may cause reversible protein
modifications that trigger intracellular signaling pathways and/or adaptive responses. Thus, PM-dependent
generation of ROS may be responsible for a continuum of responses from cell signaling to cellular injury (US EPA,
2009a, section 5.1.1).
PRELIMINARYDRAFT-September 2009 3-13
Do Not Quote or Cite
-------
1 . Allergic immune responses and impaired host defense and infections - affecting
2 both innate immunity through effects on macrophage pathogen handling as well as
3 adaptive immunity by altering cellular responses potentially resulting in impaired
4 lung defense mechanisms leading to frequent or persistent infections (US EPA,
5 2009a, sections 5.1.7, 5.1.8). Studies conducted since the last review confirm and
6 extend previous findings that PM can modulate immune reactivity in both humans
7 and animals to promote allergic sensitization and exacerbate allergic responses (US
8 EPA, 2009a, sections 6.3.6). A number of long-term exposure studies have reported
9 associations between PM and allergic responses (US EPA, 2009a, section 7.3.6).
10 Allergy is a major driver of asthma, which has been associated with PM in a number
11 of studies (US EPA, 2009a, section 7.3). In the case of allergic asthma, acute PM
12 exposure may provoke asthmatic responses through oxidative stress and
13 inflammatory pathways (US EPA, 2009a, section 5.5). Toxicological studies
14 demonstrate that short-term inhalation exposures to CAPs and disesel exhaust, but not
15 gasoline exhaust or wood smoke, can increase susceptibility to bacterial and viral
16 infections. Furthermore, impaired host defense against the etiological agents of
17 influenza, pneumonia (S. pneumoniae\ and bronchiolitis (RSV), which are
18 commonly reported respiratory morbidities associated with PM exposures, have been
19 documented in toxicological studies (US EPA, 2009a, section 6.3.7). While most of
20 the evidence of respiratory infections associated with PM exposures is available from
21 short-term exposure studies, limited evidence from long-term exposure studies
22 suggest an associated with general respiratory symptoms often caused by infection,
23 such as bronchitis (US EPA, 2009a, section 7.3.7).
24
25 . Progression of pre-existing respiratory disease - Many factors likely contribute to
26 the resolution of pulmonary inflammation and injury and the progression or
27 exacerbation of respiratory disease. These factors "are likely to include the updtake
28 and clearance of PM by macrophages, the retention of PM in parenchymal cells and
29 tissues; the balance of pro/anti-inflammatory soluble mediators, oxidants/anti-
30 oxidants and proteases/antiproteases, and the presence of pre-existing disease" (US
31 EPA, 2009a, section 5.1.9). These factors may impact the resolution of pulmonary
32 effects associated with PM exposure including effects associated with inflammatory
33 lung diseases including asthma and COPD (US EPA, 2009a, section 5.1.9). Particle
34 exposures are likely to contribute to an unfavorable balance through its oxidative
35 potential and capacity to promote cellular production of ROS. Exacerbations of
36 asthma and COPD resulting from bacterial and viral infections are also associated
37 with oxidative stress (Barnes et al, 2007) (US EPA, 2009a, section 5.1.9.2).
38
39 . DNA damage and lung cancer - Pulmonary DNA damage can occur primarily or
40 secondarily to PM exposure (US EPA, 2009a, section 5.1.10). New, but limited
41 evidence, has become available indicating a role for PM in promoting epigenetic40
40 Epigenetic mechanisms regulate the transcription of genes without altering the nucleotide sequence of DNA.
These mechanisms generally involve DNA methylation and histone modifications, leading to alternations which
may have long-term consequences or are heritable. Early life exposures to environmental pollutants such as PM
may be especially important since periods of rapid cell division and epigenetic remodeling are likely to occur at this
PRELIMINARYDRAFT-September 2009 3-14
Do Not Quote or Cite
-------
1 changes (US EPA, 2009a, section 5.1.11; Baccarelli et al., 2009; Liu et al., 2008;
2 Read et al., 2008; Yauk et al., 2008).
O
4 Changes in the function of the respiratory tract barrier may modify response related to
5 PM exposures. For example, epithelial injury can lead to an increase in permeability of the
6 airway epithelial and alveolar-capillary barriers. Enhanced transport of the soluble and possibly
7 insoluble PM components into the circulation may occur under these conditions (US EPA,
8 2004a, section 5.1.4).
9 Resolution of pulmonary inflammation and injury has been demonstrated in experimental
10 models using higher than ambient concentrations of PM. Factors contributing to this complex
11 process are likely to include the uptake and clearance of PM by macrophages; the retention of
12 PM in cells and tissues; the balance of pro/anti-inflammatory soluble mediators,
13 oxidants/antioxidants and proteases/anti-proteases; and the presence of pre-existing disease.
14 These factors may also influence the resolution of pulmonary responses to ambient PM
15 exposures. The long-term consequences of prolonged inflammation may be harmful and may
16 lead to remodeling of the respiratory tract and to the progression or exacerbation of disease (US
17 EPA, 2009a, section 5.1.9). In addition, lung development is a multi-step process which begins
18 in embryogenesis and continues to adult life (Pinkerton and load, 2006). This allows a long
19 period of potential vulnerability to environmental and other stressors. Disruption of cell signaling
20 during development could affect cellular differentiation, branching morphogenesis, and overall
21 lung growth, possibly leading to life-long consequences. Although very little is know about the
22 effects of maternal exposure to PM on the developing fetus or the effects of exposure during
23 childhood, recent animal studies demonstrate respiratory and immune system effects of perinatal
24 exposure to sidestream cigarette smoke (Pinkerton and load, 2006; Wang and Pinkerton, 2007).
25 Cardiovascular and Other Systemic Effects. In addition to pulmonary effects,
26 extrapulmonary effects resulting from lung damage may occur. For example, pulmonary
27 inflammation may trigger systemic inflammation through the action of cytokines and other
28 soluble mediators which leave the lung and enter the circulation (see US EPA, 2009a, section
29 5.2). In the last review, limited evidence from toxicological studies suggested that injury or
30 inflammation in the respiratory system could lead to changes in heart rhythm, reduced
31 oxygenation of the blood, changes in blood cell counts, or changes in the blood that could
32 increase the risk of blood clot formation, a risk factor for heart attacks or strokes (US EPA, 2004,
33 section 7.10.1.2). More specifically, in the last review, emerging evidence was available that
34 provided some initial evidence that particles could have direct cardiovascular effects. Two types
35 of hypothetical pathways by which particle deposition in the respiratory system could lead to
36 cardiovascular effects were identified: (1) PM-induced pulmonary reflexes resulting in changes
time (Foley et al., 2009; Keverne and Curley, 2008; Wright and Baccarelli, 2007; see also US EPA, 2009a, section
5.1.11).
PRELIMINARYDRAFT-September 2009 3-15
Do Not Quote or Cite
-------
1 in the autonomic nervous system that could affect heart rhythm and (2) effects on the heart or
2 other organs if particles or particle constituents are released into the circulatory system from the
3 lungs (US EPA, 2004, section 7.2, Figure 7-1). Some new evidence indicated that UFPs or their
4 soluble constituents could move directly from the lungs into the systemic circulation (US EPA,
5 2004, section 6.3.2.3). The EPA concluded that the data evaluated in the last review remained
6 limited but provided some new insights into MOA by which particles, primarily fine particles,
7 could affect the cardiovascular system (US EPA, 2004, section 7.2.4, section 7.10.1.3).
8 In this review, MOA underlying extrapulmonary effects are still incompletely
9 understood. Pulmonary inflammation can lead to systemic inflammation and pulmonary reflexes
10 can active the autonomic nervous system (ANS). These latter responses may mediate
11 cardiovascular and other systemic effects. In addition, it has been proposed the PM or soluble
12 components of PM may enter the circulation by translocating across the epithelial and
13 endothelial barriers of the respiratory tract, thus allowing PM or its components to interact
14 directly with cells in the vasculature and blood and be transported to the heart and other organs.
15 New evidence available in the current review has not advanced the hypothesis discussed in the
16 previous review, that UFPs or other PM size fractions may access the circluaton by transversing
17 the epithelial barrier of the respiratory tract (US EPA, 2009a, sections 5.2 and 5.3).
18 Pulmonary inflammation resulting from PM exposure may trigger systemic
19 inflammation. Systemic inflammation is generally seen under conditions of mild pulmonary
20 inflammation - and sometimes under conditions of no measurable pulmonary inflammation. The
21 time-dependent nature of pulmonary and systemic inflammatory responses may, in part, explain
22 these findings since biomarkers of inflammation are frequently measured at only one point in
23 time. In addition, chronic exposures may lead to adaptive responses. In general, systemic
24 inflammation is associated with changes in circulating white blood cells, the acute phase
25 response, pro-coagulation effects, endothelial dysfunction, and the development of
26 atherosclerosis41 (US EPA, 2009a, Figure 5-5). The development of atherosclerosis involves
27 inflammation and remodeling of the blood vessel wall. Factors contributing to this process
28 include systemic inflammation, endothelial dysfunction, oxidative stress, and high levels of
29 circulating lipids. PM exposure is associated with the first three of these four processes. The
30 role of PM in initiating, promoting, or complicating atherosclerosis or its outcomes has yet to be
31 determined (US EPA, 2009a, section 5.5). Furthermore, inflammation has the potential to
32 promote thrombosis which can complicate this disorder leading to adverse effects on the
33 cardiovascular and cerebrovascular systems such as thrombosis, plaque rupture, myocardial
41 Atherosclerosis is a chronic progressive disease which contributes greatly to cardiovascular disease. Primarily a
disease of the large arteries, it is characterized by the accumulation of lipid and fibrous tissue in atheromas, or
swellings of the vessel wall. Although a strong link is known to exist between hypercholesterolemia and
atherogenesis, recent studies demonstrate a key role for inflammation in the initiation and progression of
atherosclerosis (Libby et al., 2002).
PRELIMINARYDRAFT-September 2009 3-16
Do Not Quote or Cite
-------
1 infarction (MI), and stroke may result. Systemic inflammation may also affect other organ
2 systems such as the liver or the central nervous system (US EPA, 2009a, section 5.2).
3 Epidemiological studies conducted worldwide consistently demonstrate that both short-
4 and long-term exposures to PM, are associated with a large variety of CV events: myocardial
5 ischemia and infarction, heart failure, stroke, sudden death and arrhythmia, hospitalization for
6 cardiovascular disease, and increased overall cardiovascular-related mortality (US EPA, 2009a,
7 sections 6.2 and 7.2). The MOA(s) by which PM exposures may increase the risk of
8 cardiovascular morbidity and mortality remains uncertain and much research is focusing on
9 improving our understanding of this issue. In the current review, new evidence expands our
10 understanding of PM-related effects that can lead to the development of a range of
11 cardiovascular effects (see US EPA, 2009a, Figure 5-5 and section 5.4) including:
12 • Alterations in vasomotor function - the most noteworthy new health-related
13 revelation in the past six years with regards to PM exposure is that the systemic
14 vasculature may be a target organ. The second draft ISA presents alterations in
15 vasomotor function which has been demonstrated following exposure to concentrated
16 ambient particles (CAPs) and diesel exhaust (DE) (US EPA, 2009a, section 6.2.4)
17 • Myocardial ischemia and myocardial infarction (MI) - altered vasoreactivity of
18 coronary vessels may lead to myocardial ischemia and Mis; Mis may alter conduction
19 and depolarization properties of the heart potentially leading to arrhythmias
20 • Thrombosis -procoagulation effects or plaque destabilization or rupture may lead to
21 stroke and/or thromboembolic disease
22 • Arrhythmia - alterations in the conduction and depolarization properties of the heart
23 may lead to arrhythmic events
24 As noted in the second draft ISA, many of these process may be interlinked and
25 responses to ambient PM exposures may involve multiple MOA simultaneously with some
26 variability depending on PM composition. Furthermore, it is not clear at this time whether PM
27 initiates cardiovascular disease or whether it perturbs existing disease (US EPA, 2009a, section
28 5.4)
29 Central Nervous System Effects. Recent studies have demonstrated PM-dependent
30 effects on the central nervous system (CNS) though, at this time, it is unknown if these effects
31 are related to direct or indirect effect of PM exposure. Translocation of soluble and poorly
32 soluble particles from the olfactory mucosa via the axons to the olfactory bulb of the brain has
33 been proposed as a possible MOA by which PM or its components may directly access the CNS.
34 Given their small size, UFPs deposited onto nasal epithelium may be more efficiently
35 translocated by this mechanism than other PM size fractions (US EPA, 2009a, sections 5.4 and
36 4.3.2.2).
37 Overall, the findings reported in the second draft ISA continue to support the overall
38 conclusions from the previous PM NAAQS review indicating that different health responses are
PRELIMINARYDRAFT-September 2009 3-17
Do Not Quote or Cite
-------
1 linked with different particle characteristics and that many biologic responses may be responsible
2 for the morbidity/mortality effects associated with these exposures.
3 3.3 NATURE OF EFFECTS
4 An extensive body of new scientific evidence has been published since completion of the
5 2004 PM AQCD including policy-relevant information from epidemiologic, controlled human
6 exposure, lexicological, and exposure studies. In the last review, EPA concluded that recent
7 epidemiologic studies continued to report associations between various indicators of ambient PM
8 and effects such as premature mortality, hospital admissions or emergency department visits for
9 respiratory and cardiovascular disease, and effects on lung function and symptoms. In addition,
10 emerging evidence identified several new types of health outcomes reported to be associated
11 with exposure to various indicators of PM, including physicians' office or clinic visits,
12 cardiovascular health indicators such as heart rate variability (HRV) or increased C-reactive
13 protein (CRP) levels, developmental effects such as low birth weight, and infant mortality, and
14 lung cancer mortality (US EPA, 2004, p. 9-23 - 9.24).
15 The second draft ISA, along with its annexes, integrates newly available, policy-relevant
16 epidemiologic, controlled human exposure, and animal toxicological evidence with consideration
17 of key findings and conclusions from prior reviews to draw inferences about the relationship
18 between short- and long-term exposures to PM and a range of human health effects. The second
19 draft ISA evaluates evidence from the full body of health effect studies conducted world-wide
20 and summarizes results of all such mortality and morbidity studies.
21 The discussions that follow draw primarily from evidence evaluated in chapters 6 and 7
22 of the second draft ISA as well as the integration of evidence across disciplines presented in
23 chapter 2. Annexes B through F provide supplemental information. For the purposes of this
24 preliminary draft Policy Assessment (PA), staff has drawn from the qualitative evaluation of all
25 studies considered in the second draft ISA, with focus on those studies conducted in the U.S. and
26 Canada for supporting quantitative assessments.42 Effect estimates for mortality and morbidity
27 effects associated with increments of PM2.5 and PMio-2.s from multi-city and single-city U.S. and
28 Canadian studies are summarized in Chapters 6 and 7 of the second draft ISA for short-term and
29 long-term exposure studies, respectively, as a consolidated reference for the following
30 discussions.43 For each outcome considered, EPA staff has placed emphasis on multi-city
31 studies concluding that these studies have a number of advantages compared to single-city
42 Findings of U.S. and Canadian studies are more directly applicable for quantitative considerations in this review,
since studies conducted in other countries may well reflect quite different population and air pollution
characteristics.
43 Effect estimates are presented using standardized PM increments to allow for comparison across studies.
Increments of 10 ug/m3 were generally used.
PRELIMINARYDRAFT-September 2009 3-18
Do Not Quote or Cite
-------
1 studies including: (1) multi-city studies reflect ambient PM levels and potential health impacts
2 across a range of diverse locations; (2) multi-city studies "clearly do not suffer from potential
3 omission of negative analyses due to 'publication bias'" (EPA, 2004, p. 8-30); and (3) multi-city
4 studies generally have higher statistical power. Although the single-city studies may lack
5 statistical power needed to evaluate interactions and detect some of the subtle effects of air
6 pollution, they inform the interpretation of the heterogeneous effect estimates that have been
7 observed across North America. Therefore, key single-city studies are also discussed in the
8 second draft ISA and briefly discussed below.
9 In general, conclusions from the 2004 PM AQCD are summarized briefly at the
10 beginning of each section, and the evaluation of evidence from recent studies builds upon the
11 evidence that was available during the previous review. In the discussions that follow, an
12 individual study result is considered to be statistically significant if the 95 percent confidence
13 interval does not include zero. Positive effect estimates indicate increases in the health outcome
14 with PM exposure. In considering these results as a whole, it is important to consider not only
15 whether statistical significance at the 95 percent confidence level is reported in individual
16 studies, but also the general pattern of results, focusing in particular on studies with greater
17 statistical power that report relatively more precise results.
18 3.3.1 Causal Determinations
19 The EPA has developed a consistent and transparent framework for evaluating the causal
20 nature of air pollution-induced health effects drawing upon the evaluation and synthesis of
21 evidence from across scientific disciplines. This framework was developed considering guidance
22 used by other regulatory and scientific agencies (US EPA, 2009a, section 1.5). In the second
23 draft ISA, EPA assesses the new health evidence, building upon evidence available during
24 previous PM NAAQS reviews, and integrates findings from epidemiologic studies with
25 experimental (e.g., dosimetric and lexicological) studies, to make judgments about the extent to
26 which causal determinations can be made between health endpoints and various indicators or
27 constituents of ambient PM. More specifically, EPA has outlined a framework for causal
28 determination that:
29 • describes the kinds of scientific evidence used in establishing a general causal
30 relationship between exposure and health effects;
31 • defines cause, in contrast to statistical association;
32 • discusses the sources of evidence necessary to reach a conclusion about the existence
33 of a causal relationship;
34 • highlights the issue of multi-factorial causation;
35 • identifies issues and approaches related to uncertainty; and
PRELIMINARYDRAFT-September 2009 3-19
Do Not Quote or Cite
-------
1 • provides a framework for classifying and characterizing the weight of evidence in
2 support of a general causal relationship (US EPA, 2009a, section 1.5).
3 The EPA has used a two-step approach to evaluate the scientific evidence on health or
4 environmental effects of PM. The first step determines the weight of evidence in support of
5 causation and characterizes the strength of any resulting causal classification (US EPA, 2009a,
6 section 1.5.5). The second step includes further evaluation of the quantitative evidence regarding
7 the concentration-response relationships and the loads or levels, duration, and pattern of
8 exposures at which effects are observed (US EPA, section 1.5.6).
9 As outlined in Table 1-2 of the second draft ISA, the aspects, or criteria, of causality
10 considered include:
11 • consistency of the observed association
12 • coherence
13 • biological plausibility
14 • biological gradient (exposure-response relationship)
15 • strength of the observed association
16 • experimental evidence
17 • temporal relationship of the observed association
18 • specificity of the observed association
19 • analogy44
20
21 Aspects that EPA considers to generally play a larger role in determination of causality
22 are "consistency of results across studies, coherence of effects observed in different study types
23 or disciplines, biological plausibility, exposure-response relationship, and evidence from
24 "natural" experiments." In addition, EPA recognizes "that no one aspect is either necessary or
25 sufficient for drawing inferences of causality" (US EPA, 2009a, section 1.5.4).
26 In addition to controlled human exposure and epidemiologic studies, the tools of
27 experimental biology have been valuable for developing insights into human physiology and
28 pathology. Animal toxicological studies explore the effects of pollutants on human health,
29 especially through the study of model systems in other species. These studies evaluate the
30 effects of exposures to a variety of pollutants in a highly-controlled laboratory setting and allow
31 exploration of MO As or mechanisms by which a pollutant may cause effects. In making causal
32 determinations, EPA has considered differences between animal species with regard to pollutant
33 absorption and distribution profiles based on breathing patterns, exposure dose, and differences
34 in lung structure and anatomy. Background knowledge of the biological mechanisms by which
44 Structure activity relationships and information on the agent's structural analogs can provide insights into whether
an association is causal. Similarly, information on the mode of action for a chemical, as one of many structural
analogs, can inform decisions regarding likely causality.
PRELIMINARY DRAFT-September 2009 3 -20
Do Not Quote or Cite
-------
1 an exposure might or might not cause disease can prove crucial in establishing, or negating a
2 causal claim (US EPA, 2009a, section 1.5.3).
3 The second draft ISA uses a five-level hierarchy that classifies the weight of evidence for
4 causation, not just association into a qualitative statement about the overall weight of evidence
5 and causality (US EPA, 2009a, section 1.5.5):
6
7 • Causal relationship
8 • Likely to be a causal relationship
9 • Suggestive of a causal relationship
10 • Inadequate to infer a causal relationship
11 • Not likely to be a causal relationship (see US EPA, 2009a, Table 1-3)
12
13 Beyond judgments regarding causality are questions relevant to quantifying health risks
14 based on our understanding of the quantitative relationships between pollutant exposure and
15 health effects. The strength of associations most directly refers to the magnitude of the reported
16 relative risk estimates. This requires evaluation of levels of PM and exposure durations at which
17 effects were observed, including effects observed in at-risk populations. "This integration of
18 evidence results in identification of a study or set of studies that best approximates the
19 concentration-response relationships between health outcomes and PM indicators for the U.S.
20 population or subpopulations, given the current state of knowledge and the uncertainties that
21 surrounded these estimates." (US EPA, 2009a, section 1.5.6.1) Concentration-response
22 relationships vary depending on specific health outcomes evaluated. Of particular relevance for
23 the review of the PM NAAQS, is information on whether the concentration-response relationship
24 is linear across the full concentration range evaluated, or if nonlinear relationships exist along
25 any part of the range, in particular, at levels that are at or below the levels of the current PM
26 standards (US EPA, 2009a, section 1.5.6). Information from toxicological and controlled human
27 exposure studies may also be used to inform the evaluation of concentration-response
28 relationships, specifically related to MOA and characteristics of at-risk populations. In general,
29 when associations are strong in terms of yielding large relative risk estimates, it is less likely that
30 the association could be completely accounted for by a potential confounder or some other
31 source of bias (ISAC, 2004). With associations that yield small relative risk estimates it is
32 especially important to consider potential confounding and other factors in assessing causality.
33 In considering both the magnitude and statistical strength of the associations, the second
34 draft ISA observes a pattern of positive and often statistically significant associations for a
35 number of different endpoints with the strongest evidence associated with exposures to fine
36 particles, PM2.5, and more limited evidence for thoracic coarse particles, PMio-2.5, additional size
37 fractions (e.g., UFPs), and specific PM components, sources, and environments (e.g., urban and
38 non-urban environments). In considering the strength of the associations between short- and
PRELIMINARY DRAFT-September 2009 3 -21
Do Not Quote or Cite
-------
1 long-term exposures to PM and mortality or morbidity, EPA has developed a number of causal
2 determinations as summarized in Table 3-1 (US EPA, 2009a, sections 2.3.1.1, 2.3.1.2, 2.3.3.1,
3 2.3.4.1).
4 As was true in the last review, there is coherence in the epidemiologic evidence with
5 supportive evidence from toxicological and controlled human exposure studies linking short- and
6 long-term exposures to fine particles with mortality and morbidity effects on the cardiovascular
7 and respiratory systems. The strongest evidence has been observed for cardiovascular effects
8 associated with short-term fine particle exposures. More limited information is available linking
9 PM2.5 exposures to reproductive/developmental and cancer/mutagenic/genotoxic effects. Less
10 information is available to allow conclusions to be drawn about coherence or plausibility for
11 associations with PMi0-2.5 or UFPs. Table 3-1 summarizes the causal determinations by PM size
12 fraction and exposure duration (i.e., short- and long-term exposure) as presented in the second
13 draft ISA (US EPA, 2009a, section 2.3).
14
15 Table 3-1. Summary of Causality Determinations by PM Size Fraction
Size Fraction
PM25
PM1025
UFPs
Outcome
Cardiovascular effects
Respiratory effects
Mortality
Central nervous system
(CNS) effects
Reproductive and
developmental effects
Cancer, Mutagenicity, and
Genotoxicity
Cardiovascular effects
Respiratory effects
CNS effects
Mortality
Cardiovascular effects
Respiratory effects
CNS effects
Mortality
Causality Determination
Short-term
Exposure
Causal
Likely to be causal
Likely to be causal
Inadequate
Suggestive
Suggestive
Inadequate
Suggestive
Suggestive
Suggestive
Inadequate
Inadequate
Long-Term
Exposure
Causal
Likely to be causal
Likely to be causal
Suggestive
Suggestive
Inadequate
Inadequate
Inadequate
Inadequate
Inadequate
Inadequate
PRELIMINARYDRAFT-September 2009
Do Not Quote or Cite
3-22
-------
1 3.3.2 Fine Particles
2 The nature of the effects that have been reported to be associated with fine particle
3 exposures include premature mortality, aggravation of respiratory and cardiovascular disease (as
4 indicated by increased hospital admissions and emergency department visits), changes in lung
5 function and increased respiratory symptoms, evidence for more subtle indicators of
6 cardiovascular health, and more limited evidence of developmental effects (e.g., low birth
7 weight, infant mortality) and death from lung cancer. Many studies using PMio as an indicator
8 have been conducted in areas where fine particles are the dominant fraction of PMio; results of
9 these studies, where appropriate, have been considered in evaluating the effects associated with
10 fine particle exposures. Evidence is first presented for PM2.5 mass followed by discussion of
11 PM2.5 components and UFPs.
12 3.3.2.1 Effects Associated with Short-term PM2.5 Exposures
13 Numerous epidemiologic studies as well as supportive evidence from controlled human
14 exposure and toxicological studies have demonstrated statistical associations between short-term
15 exposure to fine particles and health outcomes related to cardiovascular effects, respiratory
16 effects, and premature mortality. This information is presented and discussed in chapter 6 of the
17 second draft ISA and briefly described below. Figure 2-1 in the second draft ISA (US EPA,
18 2009a, section 2.3.2) provides a summary of U.S. and Canadian studies examining the
19 association between short-term exposure to PM2.5 and cardiovascular and respiratory morbidity
20 and mortality.
21 The majority of epidemiological studies provide limited information on the air quality
22 data used in their analyses, generally reporting average concentrations rather that a range of
23 upper percentile values making it difficult to understand the air quality distribution considered in
24 the analyses. Being mindful that the form of the current 24-hour PM2.5 standard is based on a
25 98th percentile value averaged over three years to provide protection against short-term PM2.5
26 exposures, EPA has taken steps to obtain additional air quality data from study authors to better
27 understand the air quality distributions associated with the observed effects. In the external
28 review draft Policy Assessment Document staff will evaluate 98th percentile values of specific
29 studies that reported statistically significant associations.
30 3.3.2.1.1 Cardiovascular and Systemic Effects
31 Several studies considered in the previous PM NAAQS review reported positive
32 associations between short-term PM2.5 exposures and hospital admissions or ED visits for
33 cardiovascular disease, although few studies reported statistically significant effects. U.S. and
34 Canadian studies available in the last review provided evidence of cardiovascular-related
35 mortality that was consistent with evidence of all-cause (non-accidental) mortality. This
PRELIMINARY DRAFT-September 2009 3-23
Do Not Quote or Cite
-------
1 evidence was relatively stronger than that observed for respiratory-related mortality. In addition,
2 significant associations were also observed between MI and short-term PM2.5 exposures averaged
3 over 2 or 24-hours, as well as evidence of decreases in heart rate variability (HRV) related to
4 short-term PM2.5 exposures in controlled human studies and animal toxicological studies. In the
5 last review, evidence was emerging regarding more subtle indicators of cardiovascular health
6 (US EPA, 2004; US EPA, 2005; see also summary at 71 FR 2626 to 2637, January 17, 2006).
7 In this review, the body of scientific evidence has been expanded greatly by the
8 publication of a large number of new multi-city epidemiological studies that provide consistent
9 evidence of a positive association between short-term PM2.5 exposures and hospital admissions
10 or ED visits as well as premature mortality related to cardiovascular outcomes. Hospital
11 admissions and ED visits were reported, predominantly for ischemic heart disease (MD) and
12 congestive heart failure (CHF), with the majority of studies reporting increases ranging from 0.5
13 to 3.4% per 10 ug/m3 increase in PM2.5. The excess risk of cardiovascular disease (CVD)
14 hospitalization may be somewhat greater in Europe and Australia/New Zealand than in the U. S.
15 Evidence for a range of cardiovascular and systemic effects include: changes in heart rate and
16 heart rate variability, arrthymia, ischemia, vasomotor function, changes in blood pressure,
17 cardiac contractility, systemic inflammation, hemostasis, thrombosis and coagulation factors,
18 systemic and cardiovascular oxidative stress. Controlled human exposure studies as well as
19 toxicological studies provide evidence to provide biological plausibility for the findings observed
20 in the epidemiologic studies (US EPA, 2009a, section 6.2.10.1).
21 In the second draft ISA, EPA concludes that, taken together, the collective evidence from
22 epidemiological, controlled human exposure studies, and toxicological studies is sufficient to
23 conclude that a causal relationship exists between short-term PM2.5 exposures and cardiovascular
24 effects (US EPA, 2009a, sections 2.3.1.1 and 6.2.12.1). The policy-relevant evidence supporting
25 this causal determination is presented and discussed in section 6.2 of the second draft ISA and
26 briefly summarized below. Excess risk estimates for studies of hospitalizations and ED visits
27 associated with various cardiovascular endpoints are presented in figures 6-1 through 6-5.
28 Hospital admissions and emergency department visits for cardiovascular disease.
29 Epidemiologic studies reviewed in the second draft ISA substantiate prior findings of a
30 positive association between exposure to PM2 5 and hospital admissions and ED visits from
31 cardiovascular disease. The 2004 PM AQCD reviewed more than 25 publications relating PM
32 and risk of CVD hospitalizations. Results from a handful of larger multi-city studies were
33 emphasized, with the greatest emphasis placed on findings from the U.S. National Morbidity,
34 Mortality, and Air Pollution Study (NMMAPS) (Samet et al., 2000) and a subsequent reanalysis
35 o f NMMAPS II data (Zanobetti and Schwartz, 2003). The NMMAPS II study evaluated the
36 effect of daily changes in ambient PM levels on total CVD hospitalizations among elderly
PRELIMINARY DRAFT-September 2009 3 -24
Do Not Quote or Cite
-------
1 Medicare beneficiaries in 14 U.S. cities and found a -1% excess risk per 10 |ig/m3 increase in
2 PMio. The 2004 PM AQCD concluded that these results, along with those of the other single-
3 and multi-city studies reviewed "generally appear to confirm likely excess risk of CVD-related
4 hospital admissions for U.S. cities in the range of 0.6 to 1.7% per 10 |ig/m3 PMio, especially
5 among the elderly" (US EPA, 2004, section x.x). The 2004 PM AQCD also concluded that there
6 was some evidence from single-city studies suggesting an excess risk specifically for
7 hospitalizations related to ischemic heart disease and heart failure. Furthermore, the 2004 PM
8 AQCD found that "insufficient data existed from the time-series CVD admissions studies [...] to
9 provide clear guidance as to which ambient PM components, defined on the basis of size or
10 composition, determine ambient PM CVD effect potency" (U.S. EPA, 2004, section x.x). The
11 key studies reviewed in the 2004 PM AQCD on this topic included those by Burnett et al.(1997),
12 Lippman et al. (2000), Ito (2003), and Peters et al. (2001).
13 Recent large studies conducted in the U.S., Europe, and Australia and New Zealand have
14 confirmed these findings for PMio, and have also observed consistent associations between PM2.5
15 and cardiovascular hospitalizations. However, findings from single-city studies have
16 demonstrated regional heterogeneity in effect estimates. It is apparent from these recent studies
17 that the observed increases in cardiovascular hospitalizations are largely due to admissions for
18 ischemic heart disease (IHD) and congestive heart failure (CHF) rather than cerebrovascular
19 disease (CBVD, such as stroke). Building upon the evidence available in the last review, the
20 new literature on hospitalizations and ED visits for cardiovascular causes published since 2002 is
21 reviewed in section 6.2.10 of the second draft ISA.
22 Almost all of the published time-series studies of cardiovascular hospitalizations and ED
23 visits identified cases based on administrative discharge diagnosis codes as defined by the
24 International Classification of Disease 9th revision (ICD-9) or 10th revision (ICD-10) (NCHS,
25 2007). A complicating factor in interpreting the results of these studies is the lack of consistency
26 in both defining specific health outcomes and in the nomenclature used. 45
27 Recently, multiple research groups in the U.S., Europe, and Australia have created large
28 datasets to evaluate specific CVD and respiratory endpoints using more detailed and relevant
29 measures of PM concentration. In the U.S., the MCAPS analyses of Dominici et al. (2006), Bell
30 et al. (2008) and Peng et al. (2008) are large, comprehensive and informative studies based on
45 The second draft ISA outlines major groups of diagnostic codes used in air pollution studies for diseases of the
circulatory system. The codes ICD-9: 390-459 are frequently used to identify all CVD morbidity. Note that this
definition of CVD includes diseases of the heart and coronary circulation, CBVD, and PVD. In contrast, the term
cardiac disease specifically excludes diseases not involving the heart or coronary circulation. While this distinction
is conceptually straightforward, the implementation of the definition of cardiac disease in terms of ICD-9 or ICD-10
codes varies among authors. Even greater heterogeneity can be found among studies in the implementation of
definitions related to CBVD (US EPA, 2009a, Table 6-5, section 6.2.10)
PRELIMINARY DRAFT-September 2009 3-25
Do Not Quote or Cite
-------
1 Medicare hospitalization data. Likewise, the Atlanta-based SOPHIA study (Metzger et al., 2004;
2 Peel et al., 2005; Tolbert et al., 2007) is the largest and most comprehensive study of U.S.
3 cardiovascular and respiratory ED visits. In Europe, the APHEA initiative (Le Tertre et al., 2002;
4 Le Tertre et al., 2003), the more recent HEAPSS study (Von Klot et al., 2005), and the French
5 PSAS program (Host et al., 2008; Larrieu et al., 2007) are similarly noteworthy for their large
6 sample size, geographic diversity, and consideration of specific CVD and/or respiratory
7 endpoints. These studies contain adequate data to examine interactions by season and region; the
8 effects of different size fractions, components and sources of PM; or the effect of PM on at-risk
9 populations.
10 The 2004 PM AQCD (U.S. EPA, 2004) incorporated the results of a large number of
11 time-series studies in the U.S. and elsewhere relating ambient PM levels and risk of
12 hospitalization for CVD. Since then, the U.S. MCAPS study has reported new evidence, in older
13 adults, evaluating the association between PM2.5 and the risk of CVD hospitalization in 202 U.S.
14 counties between 1999 and 2005. Specifically, investigators reported a 0.7% increase in risk per
15 10 |ig/m3 increase in PM2.5 on day 0 (same day)46 (Peng et al., 2008). Bell et al. (2008) found
16 evidence of substantial and statistically significant variability in the effects of PM2.5 on
17 cardiovascular hospitalizations by season and region, with the highest national average estimates
18 occurring in the winter (i.e., estimates for the nation (1.49%) and northeast (2.01%) were highest
19 in the winter) and the highest regional estimates observed in the northeastern U.S. (1.08% per
20 10 |ig/m3 increase in PM2.5).
21 Several large multi-city studies conducted outside and within the U.S., provide support
22 for a PM effect on CVD hospitalizations including the European APHEA2 study (Le Tertre et
23 al., 2002),the Spanish EMECAS study (Ballester et al., 2006), a study conducted in multiple
24 cities across New Zealand and Australia (Barnett et al., 2006) as well as the Atlanta-based
25 SOPHIA study (Metzger et al., 2004).
26 A number of single-city studies have been published showing positive associations
27 between hospital admissions and ambient PM in Europe and the western U.S. In some of these
28 studies the associations did not reach statistical significance potentially because the studies likely
29 lacked the statistical power to find effects of the expected magnitude. In fact, it is possible that
30 studies conducted outside of large metropolitan areas have sufficient statistical power to detect
31 associations of the expected magnitude. For example, Delfino et al. (2009) evaluated the effects
32 of the 2003 California wildfires and observed a slightly larger excess risk of total cardiovascular
33 disease admissions during the wildfire period compared to the period prior to the wildfire
34 although excess risk estimates were generally weak and non-significant. In addition, some of the
46 See discussion of lag structure in section 3.4.
PRELIMINARY DRAFT-September 2009 3 -26
Do Not Quote or Cite
-------
1 PM2.5-related associations may not be statistically significant potentially because of regional and
2 seasonal heterogeneity of the PM2.5 ambient mix.
3 In summary, large studies from the U.S., Europe, and Australia/New Zealand published
4 since the 2004 PM AQCD (U.S. EPA, 2004) provide strong support for an association between
5 short-term increases in ambient levels of PM2.5 and PMio and increased risk of hospitalization
6 for total CVD. Studies of specific CVD outcomes indicate that IHD and CHF may be driving the
7 observed associations (US EPA, 2009a, sections 6.2.10.3 and 6.2.10.5). Although estimates
8 from studies of cerebrovascular diseases are less precise and consistent, ischemic diseases appear
9 to be more strongly associated with PM2.5 compared to hemorrhagic stroke47 (US EPA, 2009a,
10 section 6.2.10.7). The available evidence suggest that these effects occur at very short lags (0-1
11 days), although effects at longer lags have rarely been evaluated. The results from hospital
12 admissions and ED visits studies are supported by the associations observed between short-term
13 PM2.5 exposures and cardiovascular-related mortality, which also provide additional evidence for
14 regional and seasonal variability in PM2.5 risk estimates (see discussion in section 3.3.2.1.4).
15 Furthermore, limited evidence from the studies that examined the C-R relationship between PM
16 and CVD hospital admission and ED visits supports a no-threshold, log-linear model, which is
17 also consistent with the observations made in studies that examined the PM-mortality
18 relationship (US EPA, 2009a, section 6.2.10.10)
19 Out-of-hospital cardiac arrest. Only a few studies have examined out-of-hospital
20 cardiac arrest or deaths (see US EPA, section 3.2.10.11). Two studies from Seattle, WA
21 consistently found no association (also consistent with other cardiac effects and mortality studies
22 conducted in that locale (add ref)). One study in Indianapolis, IN found an association with
23 hourly PM2.5 but not daily PM2.5 (add ref). A study in Rome found an association with PMio but
24 also with particle numbers and CO, implicating combustion sources (add ref). Because multi-
25 city mortality studies examining this association found heterogeneity in PM risk estimates across
26 regions, future studies of out-of-hospital cardiac arrest may need to consider location and the air
27 pollution mixture during their design. There have also been very few PM studies that have used
28 hourly PM measurements and further studies are needed to confirm associations at shorter time
29 scales.
30 Cardiovascular health outcomes. Studies evaluating changes in cardiovascular function
31 provide support for the epidemiological evidence showing associations between short-term PM2.5
47 The epidemiology literature provides inconsistent support for an association between short-term increases in
ambient levels of PM10 and PM2 5 and risk of hospitalization and ED visits for CBVD. The heterogeneity in results
is likely partly attributed to: 1) differences in the sensitivity and specificity of the various outcome definitions used
in the relevant studies; 2) lag structures between PM exposure and stroke onset which may vary by stroke type and
patient characteristics; and 3) exposure misclassification due to the use of hospital admission date rather than stroke
onset time, which may vary by region, population characteristics, and stroke type (US EPA, 2009a, section 6.2.10.7)
PRELIMINARY DRAFT-September 2009 3 -27
Do Not Quote or Cite
-------
1 exposure and cardiovascular-related hospital admissions, ED visits, and mortality. These
2 changes have been reported in epidemiological, controlled human exposure, and/or toxicological
3 studies conducted by multiple independent laboratories. In the second draft ISA, the following
4 effects are discussed: heart rate and heart rate variability (section 6.2.1); arrhythmia (section
5 6.2.2); ischemia (section 6.2.3); vasomotor function (section 6.3.4); changes in blood pressure
6 (section 6.2.5); cardiac contractility (section 6.2.6); systemic inflammation (section 6.2.7);
7 hemostasis, thrombosis, and coagulation factors (section 6.2.8); and systemic and cardiovascular
8 oxidative stress (section 6.2.9). Highlights of this evidence is summarized below.
9 Vasomotor function. The most noteworthy new health-related revelation in the past six
10 years with regards to PM exposure is that the systemic vasculature48 may be a target organ. The
11 second draft ISA presents alterations in vasomotor function which has been demonstrated
12 following exposure to concentration ambient particles (CAPS) and diesel exhaust (DE) (US
13 EPA, 2009a, section 6.2.4).
14 The majority of the new evidence of particle-induced changes in vasomotor function in
15 controlled human exposure studies comes from studies of exposures to DE (US EPA, 2009a,
16 section x.x). None of these studies have evaluated the effects of DE with and without a particle
17 trap. However, it is important to note that a study by Peretz et al. (2008) used a newer diesel
18 engine with lower gaseous emissions and reported significant DE-induced decreases in BAD. In
19 addition, increasing the particle exposure concentration from 100 to 200 ug/m3, without
20 proportional increases in NO, NC>2, or CO, resulted in an approximate 100% increase in
21 response. Further evidence of a particle effect on vasomotor function is provided by significant
22 changes in BAD demonstrated in healthy adults following controlled exposure to CAPs with Os
23 (Brook et al., 2002). These findings are consistent with epidemiologic studies of various
24 measures of vasomotor function (e.g., FMD and BAD were the most common), which have
25 demonstrated an association with short-term PM2.5 exposures in healthy and diabetic populations
26 (US EPA, 2009a, section x.x). A limited number of epidemiologic studies examined multiple
27 lags and the strongest associations were with either the 6 day mean concentration (O'Neill et al.,
48 The vasculature of all tissues is lined with endothelial cells that will naturally encounter any systemically
absorbed toxin. The endothelium (1) maintains barrier integrity to ensure fluid compartmentalization, (2)
communicates dilatory and constrictive stimuli to vascular smooth muscle cells, and (3) recruits inflammatory cells
to injured regions. Smooth muscle cells lie within the layer of endothelium and are crucial to the regulation of blood
flow and pressure. In states of injury and disease, both cell types can exhibit dysfunction and even pathological
responses.
Endothelial dysfunction is a factor in many diseases and may contribute to the origin and/or exacerbation of
perfusion-limited diseases, such as MI or IHD, as well as hypertension. Endothelial dysfunction is also a
characteristic feature of early and advanced atherosclerosis. A primary outcome of endothelial dysfunction is
impaired vasodilatation, frequently due to uncoupling of NOS. It is this uncoupling that appears central to impaired
vasodilation and thus endothelial dysfunction.
PRELIMINARY DRAFT-September 2009 3-28
Do Not Quote or Cite
-------
1 2005) or the concurrent day (Schneider et al., 2008) either the 6 day mean concentration (O'Neill
2 et al., 2005) or the concurrent day (Schneider et al., 2008).
3 The toxicological findings with respect to vascular reactivity are generally in agreement
4 and demonstrate impaired dilation following PM2.5 exposure that is likely endothelium
5 dependent (US EPA, 2009a, section x.x). These effects have been demonstrated in varying
6 vessels and in response to different PM2.5 types, albeit using IT exposure in most studies. Further
7 support is provided by IT studies of ambient PMi0 that also demonstrate impaired vasodilation
8 and a PM2.5 CAPs study that reported decreased LAV ratio of the pulmonary artery. An
9 inhalation study of Boston PM2.5 CAPs reported increases in coronary vascular resistance during
10 ischemia, which indicated a possible role for PM-induced coronary vasoconstriction. The
11 mechanism behind impaired dilation following PM exposure may include increased ROS and
12 RNS production in the microvascular wall that leads to altered NO bioavailability and
13 endothelial dysfunction. Despite the limited number of inhalation studies conducted with
14 concentrations near ambient levels, the toxicological studies collectively provide coherence and
15 biological plausibility for the myocardial ischemia observed in controlled human exposure and
16 epidemiologic studies.
17 Ischemia. Epidemiologic studies demonstrate associations between PM2.5 pollution and
18 ST-segment depression at lags of 1 hour to 2 days. Morever, these findings demonstrate a
19 potential role for traffic (Chuang et al., 2008; Gold et al., 2005) and long-range transported
20 PM2.5 (Lanki et al., 2006). Toxicological studies have reported reduced myocardial blood flow
21 during ischemia and altered vascular reactivity (add ref). These studies provide coherence and
22 biological plausibility for the myocardial ischemia that has been observed in both controlled
23 human exposure and epidemiological studies (US EPA, 2009a, section 6.2.3).
24 Heart rate and heart rate variability. Heart rate (HR), HRV49, and BP are all regulated,
25 in part, by the sympathetic and parasympathetic nervous systems. Changes in one or more may
26 increase the risk of cardiovascular events (e.g., arrhythmias, MI, etc.). Decreases in HRV have
27 been associated with cardiovascular mortality/morbidity in older adults and those with
28 significant heart disease (TFESC, 1996). While HRV is commonly described as being a
49 HRV is measured using electrocardiograms (EGG) and can be analyzed in the time domain (e.g.,
standard deviation of all NN intervals [SDNN], square root of the mean squared successive NN
interval differences [r-MSSD]), and/or the frequency domain measured by power spectral analysis
(e.g., high frequency [HF], low frequency [LF], ratio of LF to HF [LF/HF]). SDNN generally reflects
the overall modulation of HR by the autonomic nervous system, whereas r-MSSD generally reflects
parasympathetic activity and high frequency variations in HR. Thus, r-MSSD is generally well
correlated with HF, which also reflects the parasympathetic modulation of HR. LF is predominately
dictated by sympathetic tone and increased LF/HF indicates sympathoexcitation, which correlates
overall with decreased overall HRV (SDNN, rMSSD). Thus LF/HF is thought to estimate the ratio of
sympathetic influences on HR to parasympathetic influences (see US EPA, 2009a, section 6.2. l).
PRELIMINARY DRAFT-September 2009 3-29
Do Not Quote or Cite
-------
1 reflection of vagal and adrenergic input to the heart, there is clearly a more complex
2 phenomenon reflected in HRV parameters. Rowan et al. (2007) provide a review of HRV and its
3 use and interpretation with respect to air pollution studies. To summarize, HRV indices are
4 excellent measures of extrapulmonary effects from inhaled pollutants, but the characterization of
5 the acute, reversible responses to air pollution as being either parasympathetic or sympathetic in
6 origin, much less predictive of some adverse outcomes such as ventricular arrhythmia, is
7 relatively unsupported by the clinical literature. This is consistent with the conclusions presented
8 in the 2004 PM AQCD which stated that there is inherent variability in the minute-to-minute
9 spectral measurements, but long-term HRV measures demonstrate excellent day-to-day
10 reproducibility (U.S. EPA, 2004, section x.x).
11 The 2004 PM AQCD (U. S. EPA, 2004) presented limited evidence of PM-induced
12 changes in HRV. However, findings from epidemiologic, controlled human exposure, and
13 toxicological studies were seemingly contradictory, with reports of both decreases and increases
14 in HRV following PM exposure. Recent epidemiologic studies have demonstrated a more
15 consistent decrease in HRV (SDNN and r-MSSD), which is supported by several controlled
16 human exposure studies published since 2003, however, the observed changes are often variable.
17 In these studies, decreases in HRV were observed among healthy adults following short-term
18 exposures to PM2.5 CAPs, but not in adults with asthma or COPD. Several studies have
19 investigated the association between acute changes in multiple HRV parameters and ambient air
20 pollutant concentrations in the U.S., Canada, Europe, Mexico, and Asia. Features and results of
21 these studies are presented in Table 6-1, and the discussed in section 6.2.1 of the second draft
22 ISA.
23 Arrhythmia. Epidemiologic and toxicological studies presented in the 2004 PM AQCD
24 (U.S. EPA, 2004) provided some evidence of arrhythmia following exposure to PM. However, a
25 positive association between PM and ventricular arrhythmias among patients with implantable
26 cardioverter defibrillators was only observed in one study conducted in Boston, MA (Peters et al,
27 2000), while toxicological studies reported arrhythmogenesis in rodents following exposure to
28 ROFA, diesel exhaust, or metals. Recent epidemiologic studies have confirmed the findings of
29 PM-induced ventricular arrhythmias in Boston, MA, and have also reported increases in ectopic
30 beats in studies conducted in the Midwest and Pacific Northwest regions of the U.S. In addition,
31 two studies from Germany have demonstrated positive associations between traffic and
32 combustion particles and changes in repolarization parameters among patients with ischemic
33 heart disease (US EPA 2009, section 6.2.3).
34 Since 2004, only two studies (in Boston and Sweden), reported adverse associations of
35 PM2.5, other size fractions and components with ICD detected ventricular arrhythmias (Dockery
36 et al., 2005a; Dockery et al., 2005b; Ljungman et al., 2008; Rich et al., 2005), while other studies
PRELIMINARYDRAFT-September 2009 3-30
Do Not Quote or Cite
-------
1 done elsewhere did not (Metzger et al., 2007; Rich et al., 2004; Vedal et al., 2004). A range in
2 exposure lags was evaluated in the Boston study (3 hours to 3 days) (Dockery et al., 2005a;
3 Dockery et al., 2005b; Rich et al., 2005) and Sweden study (2 hours and 24 hours) (Ljungman et
4 al., 2008). Given the unique and homogenous nature of the study populations, it is not clear why
5 there is not more consistency across these studies. Rich et al. (2005) reported that use of the
6 mean pollutant concentration from the specific 24 hours before the arrhythmia rather than just
7 the day of the arrhythmia, resulted in less exposure misclassification and less bias towards the
8 null, possibly explaining the lack of association when using just the day of ICD discharge and
9 daily PM concentrations (Rich et al., 2005). Other reasons for the inconsistent findings may
10 include differing degrees of exposure misclassification within each study or city due to
11 differences in PM composition and pollutant mixes (e.g., less transition metals and sulfates in the
12 Pacific Northwest than the Northeast U. S.), and differences in the size of study areas. Studies of
13 ventricular arrhythmia and PM concentration in patients with ICDs are summarized in Table 6-1
14 of the second draft ISA.
15 A few panel studies have used ECG recordings to evaluate associations between ectopic
16 beats (ventricular or supraventricular)50 and mean PM concentrations in the previous hours
17 and/or days (Berger et al., 2006; Ebelt et al., 2005; Sarnat et al., 2006). Four studies of ectopic
18 beats and runs of supraventricular and ventricular tachycardia, captured using ECG
19 measurements, all report at least one positive association. Further, they report findings in
20 regions other than Boston and Sweden (i.e., Midwest U.S., Pacific Northwest, 24 U.S. cities, and
21 Erfurt, Germany). A range of lags and/or moving averages were investigated (0-30 days) with
22 the strongest effects observed for either the 5-day mean, same day, or 1-day lagged PM
23 concentrations. Taken together, these ICD studies and ectopy studies provide evidence of an
24 arrhythmic response to PM, although further study is needed to understand the variable ICD
25 study findings.
26 No reported investigations of the relationship of PM concentration and ECG
27 abnormalities indicating arrhythmia were conducted prior to 2002 and thus were not included in
28 the 2004 PM AQCD (U.S. EPA, 2004). Abnormalities in the myocardial substrate, myocardial
50 Ectopic beats are defined as heart beats that originate at a location in the heart outside of the sinus node. They are
the most common disturbance in heart rhythm. Ectopic beats are usually benign, and may be present with or without
symptoms, such as palpitations or dizziness. Such beats can arise in the atria or ventricles. When the origin is in the
atria the beat is called an atrial or supraventricular ectopic beat. When such a beat occurs earlier than expected it is
referred to as a premature supraventricular or atrial premature beat. Likewise, when the origin is in the ventricle the
beat is defined as a ventricular ectopic beat, or when early a premature ventricular beat. When three or more occur
ectopic beats occur in succession, this is called a non-sustained run of either supraventricular or ventricular origin.
When the rate of the run is greater than 100 beats per minute it is defined as a tachycardia. Sustained ventricular
tachycardias are the arrhythmias investigated in the ICD studies described in the section above.
PRELIMINARY DRAFT-September 2009 3 -31
Do Not Quote or Cite
-------
1 vulnerability, and resulting repolarization abnormalities are believed to be key factors
2 contributing to the development of arrhythmogenic conditions. These abnormalities include
3 ECG measures of repolarization such as QT duration (time for depolarization and repolarization
4 of the ventricles), T-wave complexity (a measure of repolarization morphology), and T-wave
5 amplitude (height of the T-wave). Abnormalities in repolarization may also identify subjects
6 potentially at risk of more serious events such as sudden cardiac death (Atiga et al., 1998; Berger
7 et al., 1997; Chevalier et al., 2003; Okin et al., 2000; Zabel et al., 1998). Two recent studies of
8 changes in these measures following acute increases in air pollution have been conducted in
9 Erfurt, Germany (Henneberger et al., 2005; Yue et al., 2007). These two analyses demonstrated
10 associations between PM pollution and repolarization changes (QT duration, T-wave complexity,
11 T-wave amplitude, T-wave amplitude variability), at lags of 5 hours to 2 days. Moreover, the
12 findings from the Yue et al. (2007) study demonstrate a potential role of traffic
13 particles/pollution (US EPA, 2009a, section 6.2.x)
14 Changes in Blood Pressure. One of the potential outcomes of air pollution-mediated
15 alterations in vascular tone is its impact on variable BP or hypertension. Blood pressure is
16 tightly regulated by autonomic (central and local), cardiac, renal, and regional vascular
17 homeostatic mechanisms with changes in arterial tone being countered by changes in cardiac
18 contractility, HR, or fluid volume. The evidence of PM-induced changes in BP presented in the
19 2004 PM AQCD (U.S. EPA, 2004) was limited and inconsistent. Recent epidemiologic, human
20 clinical, and toxicological studies have similarly reported conflicting results regarding the effect
21 of PM on BP. A significant increase in DBF was observed in the only human clinical study that
22 evaluated BP during exposure (concomitant exposure to CAPs and Os). In addition, evidence
23 from toxicological studies suggests that the effect of PM on BP may be modified by health
24 status, as PM-induced increases in BP have been more consistently observed in SH rats.
25 These studies are not entirely consistent with regard to the assessed associations between
26 BP and PM exposure. Most have reported increases in SBP and DBF associated with increases
27 in either PM2.5, PMio, or UFPs (Choi et al., 2007; Chuang et al., 2005; Mar et al., 2005;
28 Zanobetti et al., 2004). However, two studies reported small decreases in BP associated with
29 multiple particulate pollutants (Ibald-Mulli et al., 2004; Mar et al., 2005). Dales et al. (2007)
30 reported no change in BP associated with a 2-hour exposure to bus stop PM2 5 and Jansen at al.
31 (2005) reported null findings among older adults in Seattle, WA. Exposure lags ranging from
32 1-3 hours (Chuang et al., 2005), to the same day (Ebelt et al., 2005; Mar et al., 2005; Rich et al.,
33 2008), to the mean across the previous 5 days (Zanobetti et al., 2004) were reported as having the
34 strongest associations with BP.
35 Several recent studies have reported increased risk of hospital admissions for congestive
36 heart failure (CHF) associated with increased PM concentration on the same day (Wellenius et
PRELIMINARYDRAFT-September 2009 3-32
Do Not Quote or Cite
-------
1 al., 2005; Wellenius et al., 2006). As a possible mechanism for these reported associations, Rich
2 et al. (2008) hypothesized that these hospital admissions for decompensation of heart failure
3 would be preceded by more subtle increases in pulmonary arterial (PA) and right ventricular
4 (RV) diastolic pressures. Wellenius et al. (2007) conducted a panel study of 28 subjects living in
5 the greater Boston metropolitan area, each with chronic stable heart failure and impaired systolic
6 function. They hypothesized that circulating levels of B-type natriuretic peptide (BNP),
7 measured in whole blood at 0, 6, and 12 weeks, were associated with acute changes in ambient
8 air pollution, as a possible mechanistic explanation for the observed association between hospital
9 admissions for heart failure and ambient PM concentration (Wellenius et al., 2005; Wellenius et
10 al., 2006). During the study, the mean PM2.5 concentration was 10.9 |ig/m3. Using linear mixed
11 models, they reported no association between any pollutant (PM2 5, CO, SC>2, NC>2, Os, and BC)
12 and BNP at any lag (Wellenius et al., 2007). However, BNP the active peptide has a very short
13 half-life and might not be the best biomarker for such a study. Thus the absence of a correlation
14 between PM and BNP may not suggest that PM does not have an impact on RV or LV function
15 in individuals with impaired cardiac mechanics.
16 Systemic inflammation. The evidence presented in the 2004 PM AQCD (U.S. EPA,
17 2004) of increases in markers of systemic inflammation associated with PM was limited and not
18 sufficient to reach a definitive conclusion. These studies were found to offer limited support for
19 mechanistic explanations of the associations between PM concentration and heart disease
20 outcomes. Recent controlled human exposure and toxicological studies continue to provide
21 mixed results for an effect of PM on markers of systemic inflammation including cytokine
22 levels, C-reactive protein (CRP), and white blood cell count. While results from recent
23 epidemiologic studies have also been inconsistent across studies, there is some evidence to
24 suggest that PM levels may have a greater effect on inflammatory markers among populations
25 with preexisting diseases.
26 In recent studies, the most commonly measured marker of inflammation was CRP.
27 Several other markers were examined in only a few studies, in relation to a wide range PM size
28 fractions and components. These markers included IL-6, TNF-a, vascular cell adhesion
29 molecule-1 (VCAM-1), intercellular adhesion molecule-1 (ICAM-1), soluble CD40 ligand
30 (sCD40L), white blood cells (WBC) and soluble adhesion molecules (sP-selectin and e-selectin).
31 CRP was not consistently associated with short-term PM concentrations (PM2.5, PMio, SC>42 ,
32 EC, OC, PNC). Several other markers of inflammation have been examined in relation to
33 several PM size fractions and components but the number of studies examining the same
34 marker/PM metric combination is too few to allow results to be compared across epidemiologic
35 studies (US EPA, 2009a, section 6.2.7).
36 Hemostasis, thrombosis and coagulation factors. The 2004 PM AQCD presented
PRELIMINARY DRAFT-September 2009 3-33
Do Not Quote or Cite
-------
1 limited and inconsistent evidence from epidemiologic, controlled human exposure, and
2 toxicological studies of PM-induced changes in blood coagulation markers (US EPA, 2004,
3 section x.x). The body of scientific literature investigating hemostatic effects of PM has grown
4 significantly since the publication of the 2004 PM AQCD, with a limited number of
5 epidemiologic studies demonstrating consistent increases in von Willebrand factor (vWf)
6 associated with PM and less consistent associations with fibrinogen. Recent human clinical and
7 toxicological studies have also observed changes in blood coagulation markers (e.g., fibrinogen,
8 vWf, factor VII, t-PA) following exposure to PM. However, the findings of these studies are
9 somewhat inconsistent, which may be due in part to differences in the post-exposure timing of
10 the assessment (US EPA, 2009a, section 6.2.8).
11 Systemic and cardiovascular oxidative stress. Very little information on systemic
12 oxidative stress associated with PM was available for inclusion in the 2004 PM AQCD. Since
13 2002, numerous studies have examined whether short-term increases in mean PM concentrations
14 are associated with adverse changes in systemic markers of oxidative stress. Recent
15 epidemiologic studies have provided consistent evidence of PM-induced increases in markers of
16 systemic oxidative stress including plasma TEARS, CuZn-SOD, 8-oxodG, and total
17 homocysteine. This is supported by a limited number of controlled human exposure studies that
18 observed PM-induced increases in free-radical mediated lipid peroxidation as well as
19 upregulation of the DNA repair gene hOGGl (US EPA, 2009a, section x.x). In addition, recent
20 toxicological studies have demonstrated an increase in cardiovascular oxidative stress following
21 PM exposure in rats, see section 6.2.9 of the second draft ISA (US EPA, 2009a, section x.x).
22 3.3.2.1.2 Respiratory Effects
23 Epidemiological studies reviewed in the second draft ISA substantiate prior findings of a
24 positive relation between exposure to fine particles and respiratory morbidity, particularly for
25 such endpoints as hospital admissions and emergency department and physician visits, and for
26 respiratory symptoms, pulmonary function and inflammation among asthmatic children. In the
27 second draft ISA, EPA concludes that, taken together, the collective evidence from
28 epidemiological studies, controlled human exposure studies, and toxicological studies is
29 sufficient to conclude that a likely causal relationship exists between short-term PM2 5 exposures
30 and respiratory effects (sections 2.3.1.1 and 6.3.10.1, US EPA, 2009a). The information
31 supporting this causal determination is presented and discussed in section 6.3 of the second draft
32 ISA and briefly summarized below.
33 Hospital admissions or emergency department (ED) and physician visits for respiratory
34 diseases. Several large multi-city studies conducted in the US and other countries since the last
35 review report a positive association between short-term increases in PM2.5 levels and ED and
PRELIMINARYDRAFT-September 2009 3-34
Do Not Quote or Cite
-------
1 physician visits and hospital admissions for respiratory diseases. Among older adults evaluated
2 in the multi-city Medicare Air Pollution Study (MCAPS),, investigators reported null findings
3 for national estimates of PM2 5 associated respiratory hospitalizations, but observed seasonal and
4 regional heterogeneity in effect estimates, with a statistically significant increased risk of
5 respiratory hospitalization during the winter in the Northeast (Bell et al., 2008). In addition,
6 significant increases in respiratory admission were observed at lag 0 and lag 2.
7 Among children, effect estimates were consistently positive between short-term exposure
8 to PM2.5 and ED visits and hospitalizations for respiratory diseases (US EPA, 2009a, section
9 6.3.10.1). Single-city studies generally report positive, non-significant associations for PM2.5
10 exposures in relation to respiratory hospitalizations (section 6.3.8.1, US EPA, 2009a). Recent
11 studies considering adults and children combined report short-term PM2 5 exposures that are, in
12 general, positively associated with specific disease endpoints, including asthma, COPD, and
13 respiratory infection (US EPA, 2009a, sections 6.3.8.2, 6.3.8.3, 6.3.8.4). However, for asthma-
14 related hospitalizations, studies conducted among only children are inconsistent across varying
15 ages and lag periods. In contrast to asthma-related hospitalizations, several studies of children
16 report positive, statistically non-significant associations between short-term PM2.5 exposure and
17 hospitalizations related to respiratory infections (section 6.3.8.4, US EPA, 2009a). These results
18 are similarly found in studies conducted among adults hospitalized for respiratory infections (US
19 EPA, 2009a, Figure 6-14). These recent epidemiological studies are consistent with the prior
20 review in finding an association between short-term exposure to PM2.5 and ED visits and hospital
21 admissions related to respiratory diseases.
22 Respiratory symptoms and lung function changes. Several single city studies included
23 in the second draft ISA (section 6.3.1.1 .,USA EPA, 2009a) report positive associations between
24 short-term exposures to PM2.5 and respiratory symptoms (e.g., cough, shortness of breath, chest
25 tightness) among asthmatic children. However, most of these associations failed to achieve
26 statistical significance and a multi-city U.S. based study (O' Connor et al., 2008) reported a
27 negative association between PM2.5 and wheeze among asthmatic children. Among asthmatic
28 adults, the findings from recent studies are inconsistent, with several single-city studies including
29 one US based study reporting negative associations between PM2.5 and respiratory symptoms
30 (Figure 6-8, US EPA, 2009a). The few studies examining the relationship between short-term
31 PM2.5 exposures and respiratory symptoms among healthy children and adults have not observed
32 a consistent association (sections 6.3.1 and 6.3.10.1, US EPA, 2009a). In addition, healthy
33 individuals have not exhibited an association between respiratory symptoms and medication use
34 with exposure to PM2 5.
35 Several panel studies discussed in the second draft ISA report statistically significant
36 associations at varying exposure periods (e.g., 5-day average, lag 1, lag 3-5) to PM2.5 and
PRELIMINARY DRAFT-September 2009 3-35
Do Not Quote or Cite
-------
1 reduced lung function (e.g.,. FEVi) among asthmatic children that were generally robust to
2 adjustment by gaseous co-pollutants (section 6.3.2.1, US EPA, 2009a). The few US based
3 studies conducted among healthy adults and adults with COPD report inconsistent findings for
4 the association between short term PM2.5 exposure and lung function, although several studies
5 conducted outside the U.S. and Canada all reported a decrease in lung function (US EPA, 2009a,
6 section 6.8.2.1). Overall, the recent epidemiological evidence is consistent with the prior review
7 which reported statistically significant associations between short-term exposures to PM2.5 and
8 respiratory symptoms and small decrements in lung function, respectively, among asthmatics.
9 Oxygen Saturation. Oxygen saturation measures the percentage of hemoglobin binding
10 sites in the bloodstream occupied by oxygen. Two panel studies have evaluated the effect of
11 PM2 5 on oxygen saturation in older adults with different results. DeMeo et al. (2004) estimated
12 the change in oxygen saturation and mean PM2.5 concentration in the previous 24 hours. They
13 used the same panel of elderly Boston residents (n = 28) and study protocol and analytic methods
14 (12 weeks of repeated oxygen saturation measurements) as Gold et al. (2005) and Schwartz et al.
15 (2005). At each clinic visit, subjects had 5 minutes each of rest, standing, post-exercise rest, and
16 20 cycles of paced breathing. PM2 5 levels were associated with decreases in oxygen saturation
17 in each of the 5-minute periods except exercise. The authors suggest that these oxygen
18 saturation reductions may result from pulmonary vascular and inflammatory changes. In a
19 similar study, Goldberg et al. (2008) examined the association between oxygen saturation and
20 ambient PM2.5, NO2, and SO2 concentrations in a panel of 31 subjects in Montreal, with NYHA
21 Class II or III heart failure who were aged 50-85 years. Only SO2 was significantly associated
22 with reduced oxygen saturation in multivariate adjusted models.
23 3.3.2.1.3 Central Nervous System Effects
24 There are relatively few epidemiological studies assessing PM2.5 associated central
25 nervous system (CNS) effects in comparison to other health outcomes, and no evidence was
26 presented of an effect of PM on the CNS in the prior review (, USA EPA, 2009a section 6.4) . In
27 the single epidemiological study discussed in the second draft ISA, utilizing the Third National
28 Health and Nutrition Examination Survey (NHANES III), investigators found PMio was
29 associated with reduced performance neurobehavioral tests of reaction time, visuomotor speed,
30 attention, and short-term memory. However, these effects were greatly attenuated with
31 adjustment for race/ethnicity and socio-economic factors, suggesting confounding of the
32 association due to residential segregation in areas of high PMio levels (Chen and Schwartz,
33 2009). The contribution of the fine fraction to the PMio levels observed in this study are
34 unknown, and therefore represent a challenge in determining the independent effect of PM2 5 on
35 neurobehavioral function. The single human clinical study that evaluated CNS effects included
PRELIMINARYDRAFT-September 2009 3-36
Do Not Quote or Cite
-------
1 in the current review is suggestive of acute cortical stress response to diesel exhaust exposure,
2 but the independent effect of PM2.5 as compared to the ultrafine fraction and diesel gases is
3 unclear (Cruts et al., 2008). Recent inhalation toxicological studies provide potential
4 mechanistic evidence, indicating inflammation in the brain and neurotransmitter modulation in
5 response to concentrated air particles (CAPs). However, the levels of PM2.5 observed in these
6 studies are not representative of exposures to ambient levels of PM2.5. Additional evidence
7 beyond the currently available toxicological data is needed to identify potential biological
8 mechanisms and assess the effects of specific PM size fractions.
9 Overall, in the second draft ISA, EPA concludes that the evidence is inadequate to
10 determine if a causal relationship exists between short-term exposures to PM2.5 and CNS
11 outcomes (US EPA, 2009a, section 6.4.4)
12 3.3.2.1.4 Mortality
13 The prior review reported consistent evidence of a relation between short-term exposure
14 to PM2.5 and mortality in locations with varying climates, PM composition, and levels of gaseous
15 co-pollutants (US EPA, 2004, Criteria Document, section 8.2). Two multi-city studies assessing
16 the PM2.s-daily mortality relation were discussed in the prior review, one of which was
17 conducted among participants of the Six Cities Study cohort (Schwartz et al., 2003, Klemm and
18 Mason, 2003) and the other was conducted in eight Canadian cities (Burnett and Goldberg,
19 2003). A reanalysis of Six Cities Study data reported results consistent with the findings of the
20 original study, with statistically significant increases ranging from 2% to over 3% reported for all
21 cause mortality associated with a 25 |ig/m3 increment increase in PM2 5 (Schwartz, 2003a;
22 Klemm and Mason, 2003). In a study using data from the eight largest Canadian cities, a
23 positive and statistically significant association was reported between PM2.5 exposure and the
24 daily mortality that was similar to the aforementioned findings from the Six Cities Study and
25 generally remained statistically significant in a number of analyses when gaseous co-pollutants
26 and 0- and 1-day lags were included in the models (Burnett and Goldberg, 2003).
27 A number of new multi-city and single-city studies and additional analysis of existing
28 data sets since the last review assessed the association between short-term exposure to PM2.5, and
29 mortality using time-series and case-crossover study designs. Of particular note is the National
30 Morbidity, Mortality and Air Pollution Study (NMMAPS) which included analyses of PM2.5
31 effects on mortality in 96 U.S. cities (Dominici et al., 2007). PM2.5 at lag 1-day was positively
32 associated with all-cause [0.29% (PI: 0.01, 0.57)] and cardio-respiratory mortality [0.38% (PI: -
33 0.07, 0.82)], respectively, per 10 |ig/m3, although the association achieved statistical significance
34 only for all-cause mortality. In addition, these associations were sensitive to the degree of
35 adjustment for temporal confounding, although a smaller sample size for the PM2.s analysis (as
PRELIMINARYDRAFT-September 2009 3-37
Do Not Quote or Cite
-------
1 compared to the PMio analysis) may have contributed to this observation (section 6.5.2.2 - 2nd
2 draft ISA - USA EPA, 2009a). In another multi-city study, Franklin et al. analyzed daily
3 mortality data from 27 U.S. cities using a case-crossover study design and reported PM2 5 was
4 positively and statistically significantly associated with all-cause [1.2% (CI: 0.29, 2.1)],
5 respiratory, [1.8% (CI: 0.20, 3.4)] and stroke deaths [1.0% (CI: 0.02, 2.0)], respectively, per 10
6 |ig/m3 (Franklin et al., 2007). PM2.5 was positively but non-significantly associated with
7 cardiovascular deaths [0.94% (CI; -0.14, 2.0)]. In a subsequent time-series analysis Franklin et
8 al. investigated the association between PM2.5 and daily mortality between 2000 and 2005 in 25
9 U.S. cites, of which 15 cities overlap with the 27 cities examined in Franklin et al. 2007
10 (Franklin et al., 2008). A seasonal pattern was observed between PM2.5 and daily mortality, with
11 the highest PM2 5 risk estimates reported for the spring and summer seasons. The PM2 5 risk
12 estimates for all-cause, cardiovascular, and respiratory deaths were generally similar to their 27
13 city study, and the regional pattern (i.e., eastern city PM2 5 risk estimates higher than western city
14 PM2.5 risk estimates) was consistent across both studies.
15 In a multi-city time-series analysis of 112 U.S. cities, Zanobetti and Schwartz (2009)
16 reported that mean PM2.s levels from lag 0-1 days was associated with 0.98% (CI: 0.75, 1.22)
17 for all-cause, 0.85% (CI: 0.46, 1.24) for cardiovascular disease, 1.18% (CI: 0.48, 1.89) for
18 myocardial infarction, 1.78% (CI: 0.96, 2.62) for stroke, and 1.68% (CI: 1.04, 2.33) for
19 respiratory mortality per 10|ig/m3 PM2 5 In co-pollutant models including PMi0-2.5, PM2 5 risk
20 estimates declined in magnitude but remained positive and statistically significant for all-cause,
21 cardiovascular disease, and respiratory-related mortality, respectively. Similarly to the seasonal
22 analysis in the aforementioned 25 city study (Franklin et al., 2008), the risk estimate for all-cause
23 mortality was highest during the spring season [2.57% (CI: 1.96, 3.19)] in comparison to the risk
24 estimates for summer, fall, and winter seasons, which ranged from 0.25% to 0.95%.
25 In an analysis of risk estimates aggregated by climate regions, the authors reported PM2.5
26 risk estimates generally similar across the regions with the exception of the Mediterranean
27 region51, which exhibited a lower risk estimates in comparison to the other regions. However,
28 intra-regional risk estimates were significantly heterogeneous for the Mediterranean region,
29 potentially contributing to the lower effect estimate observed in comparison to the regions. In a
30 study estimating the daily mortality risk associated with short-term exposure to PM2 5 for a single
31 U.S. region (i.e., 9 counties in California) Ostro et al. reported mean PM2.5 levels from lag 0-1
32 days were significantly associated with all-cause 0.6% (CI: 0.2, 1.0), cardiovascular 0.6% (CI:
33 0.0,1.1) respiratory [2.2% (CI: 0.6, 3.9)], and diabetes 2.4% (CI: 0.6, 4.2) mortality, respectively,
34 per 10 |ig/m3 (Ostro et al., 2006). Although utilizing different lag periods, the results presented
51 The Mediterranean climate considered in this analysis included California, Oregon, and Washington (Zanobetti
and Schwartz, 2009).
PRELIMINARY DRAFT-September 2009 3-38
Do Not Quote or Cite
-------
1 by Ostro et al. and Franklin et al. are generally consistent for the 5 California counties included
2 in both studies. In addition to U.S. based studies, the second draft ISA highlighted results from
3 an analysis of PM2 5 associated daily mortality for 12 Canadian cities, reporting that PM2 5 risk
4 estimates were negative in models adjusting for NO2 for years in which PM2.5 data was only
5 available on every 6th day (Burnett et al., 2004). In s subsequent analysis, using data for years in
6 which daily PM2 5 levels were available for 11 of the 12 cities, in models including NO2 the risk
7 estimates for PM2.5 remained positive, but statistically non-significant.
8 Multi-city studies support a largely positive and frequently statistically significant
9 relationship between short-term exposure to PM2.5 and increased risk of mortality. In general, it
10 can be seen in Figure 6-23 of the second draft ISA that the effect estimates for associations
11 between mortality and short-term exposure to PM2 5 are positive and a number are statistically
12 significant. The second draft ISA concludes that risk estimates from multi-city U.S. based
13 studies fall in the range of 0.29% to 1.21% per 10 |ig/m3 PM2 5 for all-cause mortality, which is
14 similar to the risk estimates for cardiovascular mortality (0.30% to 1.03% vs. 1.2% to 2.7% in
15 the last review) per 10 |ig/m3 PM2.5, but less than the risk estimates for respiratory mortality
16 (1.01% to 2.2% vs. 0.8% to 2.7% in 2004 review) using the same exposure period (i.e., lag
17 period) and averaging time (e.g., 24-hour averaging time) (USA EPA, 2009a, section 6.5.x.x).
18 The somewhat larger effect estimates reported for associations with respiratory mortality in
19 comparison to cardiovascular and all-cause mortality included larger confidence intervals (i.e.
20 less precision) since respiratory deaths comprise only a small proportion of total deaths. In
21 addition, the current range of risk estimates for PM2 5 associated cardiovascular mortality is
22 narrower than the 2004 estimates of cardiovascular mortality risk, a reflection of improved
23 consistency in results from recent studies.
24 Seasonal and regional analysis of PM2 5 associated daily mortality indicated risk was
25 highest in the spring and generally in the Eastern U.S., although PM2.5 risk estimates are
26 sensitive to varying designation methods (i.e., climate-based vs. solely spatial (add ref)). A key
27 limitation of the multi-city U.S. based studies is the absence of evidence on gaseous co-
28 pollutants potentially influencing the association between short-term exposure to PM2.5 and
29 mortality. However, the additional information on regional and season influences on the
30 association between short-term exposure to PM2 5 and mortality is a key piece of new evidence in
31 the current review. Taken together, the epidemiological evidence provides support for a likely
32 causal relation between short term exposure to PM2 5 and mortality (US EPA, 2009a, sections
33 2.3.1.1 and 6.5.3.2).
PRELIMINARY DRAFT-September 2009 3-39
Do Not Quote or Cite
-------
1 3.3.2.2 Effects Associated with Long-term PM2.s Exposures
2 This section includes an overview of the second draft ISA's findings on effects associated
3 with long-term PM2.5 exposures with emphasis on extended follow-ups of long-term
4 epidemiological cohort studies that were important in the previous PM NAAQS review as well
5 as several new multi-city studies conducted in the U.S. and Europe. In the last review, the new
6 evidence available included an extensive reanalysis of data from the Six Cities and ACS studies,
7 new analyses using updated data from the ACS and California Seventh Day Adventist
8 (AHSMOG) studies, and a new analysis using data from a cohort of veterans. In addition, new
9 studies had been published on the association between long-term exposure to fine particles and
10 respiratory morbidity using data from a cohort of schoolchildren in Southern California.
11 Epidemiologic and toxicological studies in the current review provide evidence on the adverse
12 effects of long-term exposure to PM2.5 similar to effects observed in short-term exposure studies,
13 especially for respiratory and cardiovascular morbidity and mortality. This information is
14 presented and discussed in Chapter 7 of the second draft ISA. Figure 2-2 of the second draft ISA
15 provides a summary of U.S. and Canadian studies examining the association between long-term
16 exposure to PM2.5 and cardiovascular and respiratory morbidity and mortality as well as all-cause
17 mortality. In addition, new evidence, although limited, is available on developmental and
18 reproductive effects including low birth weight and infant.
19 3.3.2.2.1 Cardiovascular and Systemic Effects
20 Studies examining associations between long-term exposure to ambient PM (over months
21 to years) and cardiovascular disease (CVD) morbidity had not been conducted and thus were not
22 included in the previous science assessments (U.S. EPA, 2004; US EPA, 1996). A number of
23 studies considered in the last review evaluated the effect of long-term PM2.5 exposure on
24 cardiovascular mortality and found consistent associations. One subchronic study evaluated
25 atherosclerosis progression in hyperlipidemic rabbits (Suwa et al., 2002) providing the
26 foundation for subsequent work considered in this review. .
27 Several new epidemiologic studies have examined the long-term PM-CVD association in
28 multi-city studies conducted in the U.S. and Europe. New epidemiologic and toxicological
29 studies have provided evidence of the adverse effects of long-term exposure to PM2.5 on
30 cardiovascular outcomes, including atherosclerosis, clinical and subclinical markers of
31 cardiovascular morbidity, and cardiovascular mortality. Some of the strongest evidence of CVD
32 health effects related to long-term exposure to PM2 5 has been in found in recent studies
33 investigating cardiovascular mortality. As discussed in section 3.3.2.2.x below, this evidence
34 includes extended follow-up to the ACS and Harvard Six Cities studies as well as new
35 information from the Women's Health Initiative (WHI) cohort study and Seventh-Day Adventist
PRELIMINARY DRAFT-September 2009 3 -40
Do Not Quote or Cite
-------
1 cohort (AHSMOG) which reported a positive association with coronary heart disease mortality
2 in females but not males (Chen et al., 2005).
3 With respect to cardiovascular morbidity, a number of endpoints have been evaluated
4 ranging from subtle indicators of cardiovascular health to serious clinical events associated with
5 coronary heart disease (CHD) and cerebrovascular discease (CVD) including myocardial
6 infarction (MI), coronary artery revascularization (e.g., bypass graft, angioplasty, stent,
7 atherectomy), congestive heart failure, and stroke, This information is presented and discussed in
8 section 7,2 of the second draft ISA and summarized below. Taken together, the second draft ISA
9 concludes that the evidence from epidemiologic and toxicological studies is sufficient to
10 conclude that a causal relationship exists between long-term exposures to PM2.5 and
11 cardiovascular effects (US EPA, 2009a, section 2.3.1.2)
12 Atherosclerosis. Atherosclerosis is a progressive disease that contributes to several
13 adverse outcomes, including myocardial infarction and aortic aneurysm. Although no study has
14 examined the association between long-term PM exposure and longitudinal change in subclinical
15 markers of atherosclerosis, several cross sectional studies have been conducted. Markers of
16 atherosclerosis used in these studies include coronary artery calcium (CAC), carotid intima-
17 media thickness (CIMT), ankle-brachial index (ABI), and abdominal aortic calcium (AAC).
18 These measures are described briefly below and in more detail in Section 7.2.1.1 of the ISA
19 (EPAa, 2009).
20 CAC is a measure of atherosclerosis of the coronary arteries in the heart. The prevalence
21 of CAC is strongly related to age. Few people have detectable CAC in their second decade of
22 life but the prevalence of CAC rises to approximately 100% by age 80 (Ardehali et al., 2007).
23 Previous studies suggest that while the absence of CAC does not rule out atherosclerosis, it does
24 imply a very low likelihood of significant arterial obstruction. Conversely, the presence of CAC
25 confirms the existence of atherosclerotic plaque and the amount of calcification varies directly
26 with the likelihood of obstructive disease. CAC is a quantified using the Agatston method
27 (Agatston et al., 1990), and the Agatston scores are frequently used to classify individuals into
28 one of five groups (zero; mild; moderate; severe; extensive) or according to age- and sex-specific
29 percentiles of the CAC distribution (Erbel et al., 2007).
30 CIMT is a measure of atherosclerosis assessed by ultrasonography of the carotid arteries
31 in the neck, the walls of which have inner (intimal), middle (medial) and outer (adventitial).
32 CIMT estimates the distance in mm or jim between the innermost (blood-intima) and outermost
33 (media-adventitia) interfaces. CIMT has been associated with atherosclerosis risk factors,
34 prevalent coronary heart disease, and incident coronary and cerebral events. Several studies
35 have indicated that CIMT measurements are accurate and reproducible, especially for the
36 common carotid artery (EPA 2009a, Section 7.2.1.1).
PRELIMINARY DRAFT-September 2009 3 -41
Do Not Quote or Cite
-------
1 ABI, which is also known as the ankle-arm or resting (blood) pressure index, is a
2 measure of lower extremity arterial occlusive disease commonly caused by advanced
3 atherosclerosis. ABI is defined as the unitless ratio of ankle to brachial systolic blood pressures
4 measured in mmHg. As ankle pressure is normally equal to or slightly higher than arm pressure
5 (resulting in an ABI > 1.0), epidemiologic studies typically define the normal ABI range as 0.90
6 to 1.50 (Resnick et al., 2004). Low ABI has been associated with all-cause and CVD mortality,
7 as well as myocardial infarction and stroke (EPA 2009a, Section 7.2.1.1)
8 AAC is a measure of atherosclerosis of the abdominal aorta. It is scored much like CAC
9 (Agatston et al., 1990), but the age-specific prevalence and extent of AAC is greater, particularly
10 among women and at ages >50 years. Although AAC has not been studied as extensively as
11 CAC, it is associated with carotid and coronary atherosclerosis as well as cardiovascular
12 morbidity and mortality and measurements are sufficiently reproducible to allow serial
13 investigations over time (EPA 2009a, Section 7.2.1.1).
14 The effect of long-term PM2.5 exposure on pre-clinical measures of atherosclerosis
15 (CIMT, CAC, AAC or ABI) has been studied in several populations using a cross-sectional study
16 design. The magnitude of the PM2.5 effects and their consistency across different measures of
17 atherosclerosis in these studies varies widely, and they may be limited in their ability to discern
18 small changes in these measures. Kunzli et al. (2005) observed a non-significant 4.2% increase
19 in CIMT associated with long-term PM2 5 exposure among participants of a clinical trial in
20 greater Los Angeles, which was several fold higher than the 0.5% increase observed by Diez-
21 Roux et al. (2008) in their analyses of MESA baseline data. The associations in MESA of CAC
22 and ABI with long-term PM2 5 exposure were largely null (Diez Roux et al., 2008), while an
23 increase in AAC with long-term PM2.5 exposure was reported (Chang et al., 2008). By contrast,
24 a 43% increase in CAC was associated with long-term PM2 5 exposure in a German study but no
25 similar association with ABI was observed (Hoffmann et al., 2009). Although the number of
26 studies examining these relationships is limited, effect modification by use of hyperlipidemics
27 and smoking status was reported in more than one study of long-term PM exposure.
28 Evidence of enhanced atherosclerosis development was demonstrated in new
29 toxicological studies that demonstrate increased plaque and lesion areas, lipid deposition, and TF
30 in aortas of ApoE" mice exposed to CAPs (Section 7.2.1.2). However, PM25 CAPs derived from
31 traffic in Los Angeles did not affect plaque size (Araujo et al., 2008).
32 Venous thromboembolism. One epidemiologic study examined the relationship between
33 long term PMio concentration, venous thromboembolism, and laboratory measures of hemostasis
34 (prothrombin and activated partial thomboplastin times [PT; PTT]). PT and PTT measure the
35 extrinsic and intrinsic blood coagulation pathways, the former activated in response to blood
36 vessel injury, the latter, key to subsequent amplification of the coagulation cascade and
PRELIMINARY DRAFT-September 2009 3 -42
Do Not Quote or Cite
-------
1 propagation of thrombus. Decreases in PT and PIT are consistent with a hypercoagulable,
2 prothrombotic state (EPA 2009a, Section 7.2.1.1). Baccarelli et al. (2007) found decreases in
3 standardized correlation coefficients for PT as well as for PTT among cases and controls per
4 10 ug/m3 increase in PMio.
5 Systemic inflammation, immune function, and blood coagulation. Only two
6 cross-sectional analyses of PMio concentration and markers of immune function or inflammation
7 have been conducted with significant changes observed in the NHANES population (stronger
8 effects among those with metabolic disorders) but not in a relative large survey of adults, which
9 was conducted in England. Ecological analyses comparing children in high versus low pollution
10 regions in Mexico show differences in unadjusted blood markers that may be related to PM
11 concentration or other unmeasured risk factors that differs across the communities studied. The
12 few toxicological studies included do not indicate robust systemic inflammation or coagulation
13 responses in F344 rats following 6-month exposures to diesel, HWS, or gasoline exhaust. The
14 limited effects that were observed could possibly be due to the varying gas concentrations in the
15 exposure mixtures.
16 Renal and vascular function. There is limited evidence for the effects of PM2.5 on renal
17 or vascular function. Cross-sectional and longitudinal epidemiologic analyses of PM2.5 and
18 UACR revealed no evidence of an effect (O'Neill et al., 2007) while small non-statistically
19 significant increases in BP with 30- and 60-day avg PM2 5 concentrations were reported
20 (Auchincloss et al., 2008). A toxicological study did not show changes in MAP with CAPs, but
21 indicated a CAPs-related potentiation of experimentally-induced hypertension (Sun et al., 2008).
22 In addition, CAPs has induced changes in insulin resistance, visceral adiposity, and inflammation
23 in a diet-induced obesity mouse model (Sun et al., 2009), indicating that diabetics may be a
24 potentially susceptible population to PM exposure.
25 Clinical outcomes in epidemiologic studies. Several epidemiologic studies of U.S. and
26 European populations have examined associations between long-term PM exposures and clinical
27 CVD events.
28 Coronary Heart Disease (CHD). Epidemiologic studies examining the association of
29 coronary heart disease (CHD) with long-term PM exposure are discussed below. Cases of CHD
30 were variably defined in these studies to include history of angina pectoris, MI, coronary artery
31 revascularization (bypass graft; angioplasty; stent; atherectomy), and congestive heart failure
32 (CHF).
33 Miller et al. (2007) studied incident, validated MI, revascularization, and CHD death,
34 both separately and collectively, among 58,610 post-menopausal female residents of 36 U.S.
35 metropolitan areas enrolled in the Women's Health Initiative Observational Study (WHI OS,
36 1994-1998). In this prospective cohort study of participants free of CVD at baseline (median
PRELIMINARY DRAFT-September 2009 3-43
Do Not Quote or Cite
-------
1 duration of follow-up = 6 yr), the authors used arithmetic averaging of year 2000 EPA AQS
2 PM2.5 data available at the monitor nearest to each participant's geocoded U.S. Postal Service
3 five-digit ZIP code centroid to estimate one-yr average exposures. They found 6%, 20% and
4 21% increases in the overall risk of MI, revascularization, and their combination with CHD death
5 per 10 ug/m3 increase in PM2.5, respectively. Hazards were higher within than between cities
6 and in the obese. For the combined CVD outcome (MI, revascularization, stroke, CHD death,
7 cerebrovascular disease), authors reported a 24% increase in risk that was higher among
8 participants at higher than lower quintiles of body mass index, waist-to-hip ratio, and waist
9 circumference. The PM2.5.CVD association was stronger among non-diabetic than diabetic
10 participants.
11 Puett et al. (2008) studied incident, validated CHD, CHD death, and non-fatal MI among
12 66,250 female residents of metropolitan statistical areas in thirteen northeastern U.S. states who
13 were enrolled in the Nurses' Health Study (NHS, 1992-2002). In this prospective cohort study
14 of women without a history of non-fatal MI at baseline (maximum duration of follow-up = 4
15 years), the authors used two-stage, spatially smoothed, land use regression to estimate residence-
16 specific, 1-year moving average PMio exposures from U.S. EPA AQS and emissions,
17 IMPROVE, and Harvard University monitor data. They found a 10% increase in risk of first
18 CHD event per 10 u/m3 increase in 1-year average PMio exposure, while the association with MI
19 was close to the null value.
20 Rosenlund et al. (2006) studied 2,938 residents of Stockholm County, Sweden. In this
21 case-control study of 1,085 patients with their first, validated non-fatal MI and an age-, gender-
22 and catchment-stratified random sample of 1,853 controls without MI (1992-1994), the authors
23 used street canyon-adjusted dispersion modeling of emissions data to estimate 30-year average
24 exposure to PMio (median = 2.4 ug/m3). They found that the OR for prevalent MI per 10 ug/m3
25 increase in PMio was 0.85. In a more recent study, Rosenlund et al. (2009) evaluated 554,340
26 residents of Stockholm County, Sweden (1984-1996). In this population-based, case-control
27 study of 43,275 cases of incident, validated MI, the authors used dispersion modeling of traffic
28 emissions and land use data to estimate 5-year average exposure to PMio. They found that after
29 adjustment for demographic, temporal, and socioeconomic characteristics, the OR for MI per 5
30 ug/m3 increase in PMio was 1.04. ORs were higher after restriction to fatal cases, in- or out-of-
31 hospital deaths, and participants who did not move between population censuses. Authors state
32 that control for confounding was superior in their previous study (Rosenlund et al., 2006)
33 although the size of the population was larger in this recent study (Rosenlund et al., 2009).
34 Zanobetti and Schwartz (2007) studied ICD-coded recurrent MI and post-infarction CHF
35 among 196,131 Medicare recipients discharged alive following MI hospitalization in 21 cities
36 from 12 U.S. states (1985-1999). In this ecologic, open cohort study of re-hospitalization among
PRELIMINARY DRAFT-September 2009 3 -44
Do Not Quote or Cite
-------
1 MI survivors (mean duration of follow-up = 3.6 and 3.7 years for MI and CHF, respectively), the
2 authors used arithmetic averaging of EPA AQS PMio data available in the county of
3 hospitalization to estimate one-year average exposures. They found 17% and 11% increases in
4 the risk of recurrent MI and post-infarction CHF, respectively, per 10 ug/m3 increase in PMio
5 exposure. Hazards were somewhat higher among persons aged >75 years.
6 Hoffmann et al. (2006) studied self-reported CHD (MI or revascularization) among 3,399
7 residents of two cities in Germany at the baseline exam of the Heinz Nixdorf Recall Study
8 (2000-2003) discussed above. In this cross-sectional ancillary study, the authors used dispersion
9 modeling of emissions, climate and topography data to estimate 1-year average exposure to
10 PM2.5 (mean = 23.3 ug/m3). They found little evidence of an association between PM2.5 and
11 CHD in these data.
12 In the study of 1030 census enumeration districts in Sheffield, U.K. discussed above,
13 Maheswaran et al. (2005) studied 11,407 ICD-10-coded emergency hospitalizations for CHD
14 among 199,682 residents. In this ecologic study, the authors used dispersion modeling of
15 emissions and climate data to estimate 5-year average exposure to PMi0. They found that after
16 adjusting for smoking prevalence, controlling for socioeconomic factors, and smoothing, the age-
17 and gender-standardized rate ratios for CHD admission were 1.01, 1.04, 0.97, and 1.07 across
18 PMio quintiles.
19 The study of post-menopausal women enrolled in the WHI OS by Miller et al. (2007) was
20 the only U.S. study to examine the effect of PM2.5 rather than PMio. This study, which provides
21 strong evidence of an association, was distinguished by its prospective cohort design, validation
22 of incident cases and large population. Puett et al. (2008), the other U.S. study with comparable
23 design features, provides evidence of an association of incident CHD with long-term PMio
24 exposure. Findings from Swedish case control studies of incident validated cases of MI were not
25 consistent. A cross-sectional study of self-reported CHD did not provide evidence of an
26 association with PM2 5 while findings from two ecologic studies of PMio indicated positive
27 associations of CHD hospitalizations with PMio (Maheswaran et al., 2005; Zanobetti and
28 Schwartz, 2007).
29 Stroke. Miller et al. (2007) found 28% and 35% increases in the overall risk of validated
30 stroke and cerebrovascular disease, respectively, per 10 ug/m3 increase in one-year average
31 PM2.5 exposure. Risks were higher within than between cities. In the study of 1030 Census of
32 enumeration districts in Sheffield, U.K. described previously, Maheswaran et al. (2005) studied
33 5,122 ICD-10-coded emergency hospital admissions for stroke among approximately 200,000
34 residents of 1,030 census enumeration districts in Sheffield, U.K. (1994-1999). In this ecologic
35 study, the authors used dispersion modeling of emissions and climate data to estimate five-
PRELIMNARY DRAFT-September 2009 3-45
Do Not Quote or Cite
-------
1 year average exposure to PMi0. They found that the age- and gender-standardized rate ratios for
2 stroke admission were 1.05, 1.07, 1.06, and 1.15 across PMio quintiles.
3 These studies examining the long-term PM-stroke relationship provide evidence of
4 association. Maheswaran et al. (2005) examined emergency room HAs in Sheffield, U.K. using
5 an ecologic design while results reported by Miller et al. (2007) are based on the prospective
6 cohort study of the WHI OS population.
7 Peripheral Arterial Disease. The German Heinz Nixdorf Recall cross-sectional study
8 described above (Hoffmann et al., 2009) also evaluated the association between one-year average
9 exposure to PM2.5 and peripheral arterial disease (self-reported history of a surgical or procedural
10 intervention or an ABI <0.9 in one or both legs). The authors found no evidence of an increase
11 in risk. However, evidence of an association with traffic exposure was present in these data.
12 ORs were higher among participants with CAC scores < 75th percentile, women, and smokers.
13 Deep Vein Thrombosis. The Italian case-control study discussed above also examined
14 the chronic PMi0-DVT association (Baccarelli et al., 2008). The authors found a 70% increase in
15 the odds of deep vein thrombosis (DVT) per 10 ug/m3 increase in one-year average PMi0
16 exposure. This finding was consistent with the decreases in PT and PTT also observed among
17 controls in this context as well as the 47% increase in the odds of DVT per inter-decile range
18 (242 meters) increase in the residence-to-major-roadway distance observed among a subset of
19 cases and controls (Baccarelli et al., 2009). The PMi0-DVT and distance-DVT associations were
20 both weaker among women and among users of oral contraceptives or hormone therapy.
21 3.3.2.2.2 Respiratory Effects
22 The prior review observed long-term exposure to PM2.5 was associated with increased
23 prevalence of respiratory disease (e.g. asthma, bronchitis, chronic cough) and lung function. The
24 evidence from the prior review suggested children were an at-risk population, with the US-based
25 Southern California Communities' Children's Health Study (CHS) reporting decreases in lung
26 function growth associated with exposure to PM2.5 over four-year follow-up periods (Gauderman
27 et al., 2002). The associations were statistically significant for some measures of lung function
28 among one cohort of 4th graders, whereas in a second cohort of 4th graders the associations
29 generally did not achieve statistical significance. Similarly, among a group of older children a
30 positive but statistically non-significant association was detected between long-term PM2.5
31 exposure and a decrease in lung function growth (Gauderman et al., 2000).
32 A prospective study conducted among CHS participants that focused on participants
33 moving to other locations during the study period, investigators reported that those participants
34 moving into areas of lower PMio exhibited an increase in lung function growth in comparison to
35 children moving to areas with higher PMio concentrations (Avol et et al., 2001). Another study
PRELIMINARY DRAFT-September 2009 3 -46
Do Not Quote or Cite
-------
1 conducted among participants of the CHS, but with PM2.5 levels, reported an association between
2 long-term exposure to PM2.5 and bronchitis symptoms (McConnell et al., 2003). However, this
3 study also observed that the single-pollutant effect of PM2 5 markedly decreased with adjustment
4 for NC>2 and OC, suggesting other traffic-related pollutants may also be etiologically relevant for
5 the development of bronchitic symptoms. These findings were similarly observed in studies of
6 long-term exposure to PMio in areas where the fine fraction predominates (US EPA, 2009a,
7 section 7.3.1.1). These findings are consistent with the cross-sectional results reported from the
8 24-cities study (Dockery et al., 1996; Raizenne et al., 1996) which reported associations between
9 increased bronchitis rates and decreased peak flow with fine particle sulfate and fine particle
10 acidity. However, as noted in the study by McConnell et al. (1999), the high degree of
11 correlation between PMio and co-pollutants obscures the assessment of an independent effect of
12 PM. In the last review, the 2004 AQCD "concluded that the evidence for an association between
13 long-term exposure to PM and respiratory effects may be confounded by other pollutants" (US
14 EPA, 2009a, section 7.3).
15 Several new studies analyzing the CHS report positive associations between long-term
16 exposure to PM2.5 and respiratory morbidity, particularly for such endpoints as lung function
17 growth, respiratory symptoms, and disease incidence (US EPA, 2009a, section 7.3.9.1). Recent
18 epidemiological evidence building upon earlier findings from the CHS is generally supportive of
19 a longitudinal association between PM2 5 and pulmonary function that is observed during the lung
20 development period among children with effects and extending into adulthood. A prospective
21 study utilizing the CHS cohort reported that among children between the ages of 10 and 18
22 years, PM2 5 was significantly associated with deficits in FEVi attained at the age of 18 years
23 (Gauderman et al., 2004). Furthermore, the 8 year time period evaluated provides evidence for
24 "the persistence of effect, but in addition the strength and robustness of the outcomes were more
25 positive, larger, and more certain than previous CHS studies of shorter follow-up" (section
26 7.3.2.1, US EPA, 2009a). These results are consistent with positive findings from studies of
27 PMio exposure that were conducted in Mexico (Rojas-Martinez et al., 2007) and Sweden
28 (Nordling et al., 2008), and a national cystic fibrosis cohort in the US (Goss et al., 2004). In
29 addition, investigators of the CHS suggest PM2.5 may also act as a modifier of the association
30 between lung function with asthma, inducing declines in lung function and a concomitant
31 increase in new onset asthma (Islam et al 2007). Furthermore, the second draft ISA concludes
32 that:
33 The CHS (McConnell et al., 2003) provides evidence in a prospective longitudinal cohort
34 study that relates PM2.5 and bronchitic symptoms and reports larger associations for
35 within-community effects that are less subject to confounding than between-community
36 effects. Several new studies report similar findings with long-term exposure to PMio in
37 areas where fine particles predominate PMio. In England, in a cohort of 4,400 children
PRELIMINARY DRAFT-September 2009 3 -47
Do Not Quote or Cite
-------
1 (aged 1-5), an association is seen with an increased prevalence of cough without a cold.
2 Further evidence includes a reduction or respiratory symptoms corresponding to
3 decreasing PM levels in "natural experiments" in both a cohort of Swiss school children
4 (Bayer-Oglesby et al., 2005) and adults (Schindler et al., 2009) (US EPA, 2009a, section
5 7.3.1.1).
6
7 Overall, the second draft ISA notes that study results for respiratory symptoms and lung
8 function changes associated with PM2.5 exposure were consistent across different study designs,
9 multiple measures of lung function, and varying locations and researchers. In addition, recent
10 evidence is supportive of an association between long-term exposure to PM2.5 and allergy and
11 pulmonary inflammation (US EPA, 2009a, sections 7.3.3, 7.3.4, 7.3.5, 7.3.6, 7.3.7). However,
12 the evidence for respiratory symptoms and lung function remain stronger as studies of allergic
13 indicators are inconsistent, and comparatively few studies assess long-term PM2.5 exposure in
14 association with pulmonary inflammation.
15 The prior review concluded the evidence for the association between long-term exposure
16 to PM2.5 and respiratory morbidity may be confounded by other pollutants. However, the second
17 draft ISA notes consistent epidemiological evidence in concurrence with toxicological evidence
18 providing a biological mechanism supports an independent effect of long-term exposure to PM2.5
19 in association with respiratory morbidity, with children at particular risk (add ref).
20 3.3.2.2.3 Reproductive, Developmental, Prenatal, and Neonatal Outcomes
21 In the prior review, the limited evidence available for PM associated developmental
22 effects suggested exposure to PMio during early (i.e. first month) and late (i.e. 6 weeks prior to
23 birth) pregnancy was associated with a higher risk of pre-term birth, and exposure to PM2.5
24 during the first month of pregnancy was associated with intrauterine growth restriction.
25 However, an association was not detected between PMio exposure and low birth weight. In
26 contrast, the current review includes new studies reporting a significant association between
27 exposure to PM2.5 during pregnancy and lower birth weight, pre-term birth, and intrauterine
28 growth restriction, respectively, and post-natal exposure to PM2.5 associated with an increased
29 risk of infant mortality. Although a multi-city US based study did not observe an association
30 between quarterly estimates of exposure to PM2.5 and lower birth weight (Parker and Woodruff,
31 2008) other US based studies report an association between PM2 5 and lower birth weight (Bell et
32 al., 2007; Parker et al., 2005) with higher effect estimates reported for PM2 5 in comparison to
33 PMio (Bell et al., 2007). The influence of region variability (i.e., fine particles were a stronger
34 predictor of lower birth weight in the Northeast) may have obscured an association in the multi-
35 city US based study.
36 Among the few studies conducted in the US since the last review, investigators reported
37 statistically significant associations between exposure to PM2 5 at 1st trimester, 1st month, 2 and 6
PRELIMINARY DRAFT-September 2009 3-48
Do Not Quote or Cite
-------
1 weeks prior to birth, and throughout pregnancy with pre-term birth (Wilhem and Ritz, 2005; Ritz
2 et al., 2007; Huynh et al., 2005). In addition, studies conducted outside the US support an
3 association between exposure to PM25 and greater risk of pre-term birth (Brauer et al., 2008;
4 Jalaludin et al., 2007). Since the last review, two U.S. based studies have examined intrauterine
5 growth (e.g.., small for gestational age, intrauterine growth retardation), an earlier indicator of
6 fetal development compared to other birth outcomes (e.g., pre-term birth, low birth weight) in
7 relation to PM2.5 exposure. These studies reported an statistically significant increase in risk of
8 small for gestational age associated with exposure to PM2.5 during the first, second, and third
9 trimester (Rich et al., 2009; Parker et al., 2005). Furthermore, two Canadian studies reports
10 similar findings of a significant association between exposure to PM2.5 throughout pregnancy
11 and greater risk of small for gestational age (Brauer et al., 2008) and intrauterine growth
12 retardation (Liu et al., 2007), respectively, although the effect diminished with adjustment for co-
13 pollutants in the latter study of intrauterine growth retardation.
14 At the acute end of the birth outcomes spectrum, the current review indicates evidence of
15 an association between post-natal exposure to PM2.5 and infant mortality is limited, particularly
16 for studies assessing PM2.5 exposure, and with mixed findings observed across differential ages
17 (i.e. less than 1 month, 1 month to 1 year, less than 1 year) and for cause-specific deaths (i.e.,
18 respiratory mortality, sudden infant death, all-cause). Two U.S. based studies conducted since
19 the last review report exposure to PM2 5 over the first two months or entire lifetime of the infant
20 is significantly associated with increased risk of respiratory-cause mortality (Woodruff et al.,
21 2006; Woodruff et al., 2008). Moreover, although statistically non-significant, the authors report
22 a positive association between exposure to PM2 5 and increased risk of all-cause mortality. The
23 results for PM2.5 associated risk of sudden infant death are inconsistent, potentially due to the
24 changes in diagnosis between 1999 and 2001.
25 There remain significant limitations in the evidence assessing the relation between PM2 5
26 exposure and birth outcomes (US EPA, 2009a, section 7.4). Specifically, characterizing PM2.5
27 exposure at the etiologically relevant time period for developmental effects, evaluating the
28 influence of co-pollutants (refer to section 3.4.6.1), exposure misclassification (refer to section
29 3.4.4), accounting for differences in fetal developmental stage in association with PM exposure
30 prior to delivery between cases and controls in studies of pre-term birth, limitations in the
31 toxicological data available to understand the biological mechanisms underpinning these
32 relations, and fewer studies in comparison to the evidence available for other endpoints (e.g.,
33 cardiovascular/respiratory morbidity) remain important challenges. Despite these limitations
34 recent studies included in the current review are suggestive of an association between PM2.5
35 exposure and birth outcomes.
PRELIMINARY DRAFT-September 2009 3-49
Do Not Quote or Cite
-------
1 3.3.2.2.4 Cancer Incidence, Mutagenicity, and Genotoxicity
2 The previous review concluded that long-term exposure to PM was associated with an
3 increase in the risk of lung cancer incidence and mortality based on evidence from three
4 prospective studies (i.e. ACS, Six Cities, AHSMOG) as well as results from case-control and
5 ecological studies (US EPA, 2004, p. 8-318; Beeson et al., 1998; Dockery et al., 1993; Pope et
6 al., 2002). Furthermore, the EPA concluded in 2002 that diesel exhaust, a source of PM in the
7 ambient air is a likely human carcinogen (U.S. EPA, 2002). Toxicological evidence of
8 mutagenic or genotoxic potential from exposures to PM presented in the prior review provide
9 support for a biological mechanism underpinning this relation (US EPA, 2004, section 7.10.1).
10 The epidemiological evidence presented in the current ISA is largely an extended
11 analyses of the prospective cohort studies included in the prior review, primarily focusing on the
12 association between long term exposure to PM and lung cancer incidence and mortality. A
13 recent re-analysis of the ACS cohort reported a statistically significant association between PM2.5
14 and lung cancer mortality (Krewski, 2009). Similarly, the extended follow-up of the Six Cities
15 Study yielded a significantly higher risk of lung cancer mortality in association with increasing
16 levels of PM2.5 (Laden et al., 2006). A ten-year follow-up study conducted among all
17 participants of the Netherlands cohort study (NLCS) observed a positive non-significant
18 association between PM2.s and lung cancer mortality among (Brunekreef et al., 2009). However,
19 in the same study investigators also observed a negative association between PM2.5 and lung
20 cancer mortality in a case-cohort analysis, and observed a negative association between PM2.5
21 and lung cancer incidence among the full cohort and case-cohort.
22 In addition to the inconsistencies in evidence between lung cancer mortality and
23 incidence from prospective cohort studies, the current review also reports differing findings for
24 the relation between PM and biomarkers of susceptibility (e.g., chromosomal aberrations,
25 oxidative stress-induced DNA damage, frequency of micronuclei). However, recent in vitro
26 studies confirm findings from the prior review that PM originating from urban and combustion-
27 derived sources is mutagenic and genotoxic, and that fine particles may be more mutagenic and
28 genotoxic than larger particles. The results of these in vitro studies provide supportive evidence
29 for biological plausibility of an association between PM and lung cancer, however, two in vivo
30 inhalation toxicological studies evaluating carcinogenic endpoints reported negative results (US
31 EPA, 2009a, section 7.5.2.3). Initial epigenetic investigations also provide evidence that long-
32 term exposure to PM2.5 may affect DNA methylation content, but further epigenetic research is
33 required to elucidate whether long term exposure to PM2 5 results in heritable DNA mutations.
34 Overall, current evidence is suggestive of an association between long term exposure to PM2.5
35 and cancer, although limitations in the evidence include the carcinogenicity of the fine PM
36 fraction in comparison to ultrafine particles (< 2.5|im) in diesel exhaust and studies utilizing
PRELIMINARY DRAFT-September 2009 3-50
Do Not Quote or Cite
-------
o, and inconsistencies in studies assessing the relation between PM2.5 and lung cancer
2 incidence and mortality, respectively.
3 3.3.2.2.5 Mortality
4 The 2004 CD reported evidence from four prospective studies (i.e., Six Cities Study,
5 ACS Study, the Veterans cohort, and AHSMOG cohort) (section 8.2.3, US EPA, 2004). In
6 particular, the Health Effects Institute reanalysis of ACS and Six Cities Studies provided the
7 strongest evidence of a statistically significant association between chronic exposure to PM2.5
8 and mortality (Krewski et al., 2000). Furthermore, an extended analysis for the ACS cohort
9 included in the prior review reported significant associations between long-term exposure to fine
10 particles (using various averaging periods for air quality concentrations) and mortality from all
11 causes, cardiopulmonary diseases, and lung cancer, respectively (Pope et al., 2002; CD p. 8-102,
12 US EPA, 2004). Overall, the prior review reported a 10 |ig/m3 increase in PM2.5 levels was
13 significantly associated with increased risk of death ranging from 6 to 13% and 6 to 19%, for all
14 causes and cardiopulmonary causes, respectively, and 13% for lung cancer mortality.
15 New studies discussed in the current ISA include an extended analysis with updated data
16 of the Six Cities and ACS cohorts, further analysis of the AHSMOG and Veterans study cohorts,
17 and a new analysis among participants of the Cystic Fibrosis and WHI cohort. In an extended
18 analysis of the ACS cohort utilizing new details on cardiovascular and respiratory cause
19 mortality, investigators reported a positive association between PM2.5 and specific cardiovascular
20 diseases (i.e., IHD, dysrhythmia, heart failure, cardiac arrest). However, a null association was
21 reported between PM2 5 and respiratory mortality (Pope et al., 2004).
22 Another new analysis of the ACS cohort extended the follow-up period to 18 years,
23 reporting parameter estimates generally higher but not statistically different than earlier ACS
24 effect estimates (Krewski 2009; section 7.6.1 2nd draft ISA - USA EPA, 2009a). Krewski et al.
25 employed a Cox proportional-hazards regression model incorporating multiple levels of
26 information to adjust for the similarity of observations among people living in the same
27 neighborhoods and cities, and for social and economic confounders at two spatial levels (i.e. zip
28 code and city). This reanalysis also assessed the robustness of the model results to alternative
29 forms (i.e. log-log) compared to the linear function for the PM-mortality relation. The
30 investigators reported that as observed in the reanalysis of the original ACS study (Krewski et
31 al., 2000), education level was found to be an effect modifier, with PM2.5 associated mortality
32 from all causes excluding IHD higher for those with lower education levels. These results are
33 consistent with the prior review noting educational attainment may serve as proxy for lower
34 socioeconomic status and thus greater vulnerability to PM2.s-associated effects (AQCD - US
35 EPA 2004, p. 8-94). The authors found that adjustment for individual-level variables did not
PRELIMINARY DRAFT-September 2009 3 -51
Do Not Quote or Cite
-------
1 alter the results for the association between long-term PM2.5 exposure and mortality, and
2 addressing spatial patterns in the data generally increased the size of the association between
3 PM2 5 and mortality (Krewski et al., 2009). In consideration of the results of the sensitivity
4 analyses, the authors report that the reanalysis are consistent with earlier findings of an
5 association between long term exposure to PM2.5 and mortality.
6 The extended analyses for the ACS cohort study included additional follow-up health
7 data for 1990-1998 and PM2.5 levels for 1985-1998 were estimated from visibility data (Laden et
8 al., 2006). The investigators reported statistically significant associations between long-term
9 exposure to PM2.5 and increased risk of mortality over two distinct time periods. Of particular
10 note is the statistically significant 27% reduction in mortality risk from cardiovascular and
11 respiratory causes associated with a decline in long-term PM2 5 levels over the follow-up period.
12 Although the use of the estimated PM2.5 levels in place of observed measurements introduces
13 uncertainty in the derivation of the risk estimates, the investigators note the high correlation
14 (0.93) between estimated and measured PM2.5 levels during years when both were available
15 (Laden et al., 2006), and these results are consistent with the findings from the original analysis
16 of a significant association between long term exposure to PM2 5 and increased risk of mortality
17 (Dockery et al., 1993).
18 There are new analyses among participants of the AHSMOG and WHI cohort that
19 provide evidence of a statistically significant association between long term exposure to PM2 5
20 and increased risk of mortality among women. Positive, statistically significant associations
21 were reported for coronary heart disease mortality among females participants of the AHSMOG
22 cohort, but not among men, suggesting women may be at greater risk for acute PM health effects
23 in comparison to men (Chen et al., 2005). Furthermore, among participants of the WHI,
24 investigators observed long term exposure to PM2 5 was significantly associated with an
25 increased risk of a cardiovascular related mortality (Miller et al., 2007). Effect estimates from
26 the WHI study are higher than those from the ACS cohort, which has eight times the
27 cardiovascular mortality rate as the WHI, suggesting the PM2 5 impact may be higher among
28 populations with a lower prevailing risk of cardiovascular death (section 7.6.1 2nd draft ISA -
29 USA EPA, 2009a). These results are consistent with the findings of long- term exposure to
30 PM2 5 associated risk of cardiovascular morbidity among WHI participants (add ref), providing
31 support for PM related chronic and acute effects on the cardiovascular system.
32 Since the last review several new studies were conducted among the elderly. Among
33 participants of the California Cancer Prevention Study, a cohort of elderly individuals in 11
34 California counties, a statistically significant association was reported between long term
35 exposure to PM2 5 and all cause deaths during the time period at which observed PM2 5 levels
36 were highest (1973-1982), but a null association was reported for the subsequent time period
PRELIMINARY DRAFT-September 2009 3-52
Do Not Quote or Cite
-------
1 when PM2.5 levels had declined in counties which previously exhibited the highest PM2.5 levels
2 (Enstrom et al., 2005). Among participants of Medicare data-based cohort studies, investigators
3 report statistically significant associations between long term exposure to PM2 5 and increases in
4 all-cause mortality for the ACS and Six Cities Study counties, respectively (Eftim et al., 2008).
5 Although effect estimates among the Medicare cohort are higher than the ACS and Six Cities
6 studies, these results are potentially sensitive to the lack of adjustment for individual-level
7 covariate data on potential confounders (e.g. smoking). Another study of Medicare data from
8 113 counties reported variability in the PM2.5 mortality effect estimates at the national and local
9 level from 2000 to 2002, and noted that trends in PM2.5 and mortality at the national scale are
10 more likely to be confounded than trends at the local scale (Janes et al., 2007). Overall, the
11 utilization of long-term time trends as a source of exposure variability is problematic due to
12 potential confounding of the effect estimates (Pope and Burnett, 2007).
13 Zeger et al. (2007) reported that long-term exposure to PM2.5 was significantly associated
14 with an increase in mortality among Medicare participants, and stronger associations were
15 observed in eastern counties compared to a national estimate, and no association was observed
16 among Western counties (Zeger et al., 2007). However, effect estimates decreased by 50% with
17 adjustment for spatial confounding. In a subsequent retrospective cohort study (MCAPS), Zeger
18 et al., reported than average 6-year exposure to PM2.5 was significantly associated with increased
19 risk of mortality in the eastern and central regions, respectively, and (Zeger et al., 2008)52. In
20 addition, the results indicated that risk declined with increasing age, and similar to their earlier
21 study, no associated was observed between PM2.5 and mortality in the western region.
22 Moreover, risk estimates were similar to effect estimates reported in the ACS and Six Cities
23 study, provide coherence across prospective studies for an association between long-term
24 exposure to PM2.5 and mortality.
25 In addition, an analysis of PM2.s associated mortality and the influence of traffic density
26 among the Veterans Administration (VA) cohort reported a statistically significant association
27 between PM2.5 and mortality in a single-pollutant model. However, with adjustment for co-
28 pollutants the PM2.s effect estimates decreased and was no longer statistically significant (Lipfert
29 et al., 2006). In another investigation of the effects of traffic-related air pollution on mortality
30 among participants of the Netherlands Cohort Study, Brunekreef et al., reported a statistically
31 significant 6% increase for all-cause mortality associated with a 10 g/m3 increase in PM2.s
32 levels, similar to results reported in the ACS (Brunekreef et al., 2009; US EPA, 2009a, section
33 7.6.1).
52 The Zeger et al. (2008) analysis included COPD as a proxy for smoking status. The investigators reported the risk
estimate for the eastern region declined and the central region increased using this adjustment. This result may
possibly be related to bias introduced in using COPD as a proxy for smoking.
PRELIMINARYDRAFT-September 2009 3-53
Do Not Quote or Cite
-------
1 Recent studies assessed the potential for an unknown confounder driving the PM-
2 mortality association at the local scale by considering the within-city effects of PM2.5 exposure.
3 Two new analyses using ACS considered intra-urban variability in PM2 5 in Los Angeles and
4 New York City, respectively (Krewski et al., 2009). In Los Angeles, a refined exposure
5 assessment using interpolation and land use regression methods yielded generally similar results,
6 with a significant association reported between PM2.5 and mortality from all causes and
7 cardiopulmonary diseases. In New York City, Krewski et al. (2009) used land use regression
8 and annual average PM2.5 levels to assign estimated PM2.5 exposures to ACS participants. A
9 positive association was observed between exposure to PM2.5 and IHD related mortality, but
10 negative associations were observed between PM2.5 exposure and all cause, CPD, and lung
11 cancer mortality. The limited range of average PM2 5 concentrations potentially contributed to
12 the limited power to detect an association for these outcomes.
13 Overall, in the current review, the updated epidemiological evidence from the Six Cities
14 and ACS cohorts, provides support for a likely causal relation between long-term exposure to
15 PM2.s and mortality (US EPA, 2009a, section 7.6.5.1). For epidemiological studies reporting
16 associations between long-term exposure to PM2.5 and increased risk of mortality, the mean
17 PM2.5 concentrations in study areas ranged between 13.2 to 29 |ig/m3, with more precise and
18 consistently positive findings in locations at or above 13.5 |ig/m3 (US EPA, 2009a, Figure 2-2).
19 Recent ACS analyses indicate mortality effects related to cardiovascular disease, as compared to
20 respiratory- related mortality, are the predominant cause of mortality in association with PM2.5.
21 New study findings from the Six Cities and ACS cohorts review suggest the magnitude of the
22 association between long term exposure to PM2 5 and mortality is potentially larger than
23 previously reported (US EPA, 2009a, section 7.6.1). Moreover, updated analyses using the Six
24 Cities study, and a new study examining the relation between life expectancy and PM2 5 report a
25 decreased risk of mortality and increase in life expectancy with declining levels of PM2.5 (Laden
26 et al., 2006; Pope et al., 2009).
27 3.3.2.3 PM2.5 Components and Sources
28 One of the major research priorities defined by the National Academy of Sciences
29 National Research Councils (NRC) related to assessing the health effects of PM components and
30 sources (NRC, 2004). In the last review, EPA recognized the availability of a limited number of
31 epidemiologic studies implicating various components within the mix of fine particles (e.g.,
32 sulfates, nitrates, carbon, organic compounds, and metals) as being associated with adverse
33 effects (EPA, 2004, section 9.2.2.1.1, Table 9-3) as well as several studies that used PM2.5
34 speciation data to evaluate the association between mortality and particles from different sources
35 (Schwartz, 2003; Mar et al., 2003; Tsai et al., 2000; EPA, 2004, section 8.2.2.5). Overall, EPA
PRELIMINARY DRAFT-September 2009 3-54
Do Not Quote or Cite
-------
1 concluded that, "[t]hese studies suggest that many different chemical components of fine
2 particles and a variety of different types of source categories are all associated with, and
3 probably contribute to, mortality, either independently or in combinations" (EPA, 2004, section
4 9.2.2.1.1). Conversely, scientific evidence available in the last review provided no basis to
5 conclude that any individual fine particle component cannot be associated with adverse health
6 effects (EPA, 2005, p. 5-17). In short, there was not sufficient evidence that would lead EPA to
7 select one or more PM2.5 component as being primarily responsible for effects associated with
8 fine particles, nor was there sufficient evidence to suggest that any component should be
9 eliminated from the indicator for fine particles (71 FR 61163, October 17, 2006).
10 In this review, once again EPA is considering to what extent evidence has become
11 available to inform the understanding of the role and relative public health importance of specific
12 components within the ambient mix of particles. More specifically, we are interested in how the
13 new scientific evidence can inform our understanding of potential differences in toxicity for
14 individual PM components as well as the interactions of these components in the ambient mix.
15 New epidemiological evidence available in this review builds upon ambient measurement data
16 available through the Chemical Speciation Network (CSN) to support initial analyses of fine
17 particles components in addition to available toxicological evidence.
18 As presented in section 6.6 of the second draft ISA, "[f]rom a mechanistic perspective, it
19 is highly plausible that the chemical composition of PM would be a better predictor of health
20 effects than particle size. This would be consistent with observed regional heterogeneity in PM-
21 related health effects in some epidemiologic studies. Also, data from the CSN demonstrate
22 gradients in a number of PM2 5 components, including EC, OC, nitrate, and SO42"." Recent
23 epidemiologic, toxicological, and controlled human exposure studies have evaluated the health
24 effects associated with ambient PM constituents and sources, using a variety of quantitative
25 methods applied to a broad set of PM2.5 constituents, rather than selecting a few constituents a
26 priori. In this review, EPA has considered approximately 40 new studies evaluating the health
27 effects associated with chemical components and sources of PM, including factors for PM from
28 crustal and soil, traffic, secondary sulfates, power plants, and oil combustion sources (US EPA,
29 2009a, section 6.6; Table 6-17). There is some evidence for trends and patterns that link
30 particular ambient PM constituents or sources with specific health outcomes, but there is
31 insufficient evidence to determine whether these patterns are consistent or robust.
32 Overall, EPA continues to find support for general conclusions presented in the last
33 review that the available evidence for particle components "indicate[s] that many constituents of
34 PM can be linked with differing health effects and that the evidence is not yet sufficient to allow
35 differentiation of those constituents or sources that are more closely related to specific health
36 outcomes" (US EPA, 2009a, sections 2.3.2 and 6.6.3.
PRELIMINARYDRAFT-September 2009 3-55
Do Not Quote or Cite
-------
1 3.3.2.3.1 Effects Associated with Short-term Exposures to PM2.5 Components and
2 Sources
3 As discussed in the last review, associations were reported between mortality and short-
4 term exposure to a number of PM components, especially fine particle components (e.g.,
5 sulfates, nitrates, metals, organic compounds, elemental carbon). Different patterns of
6 associations between various components or source categories of fine particles and total or
7 cardiovascular mortality were seen in these various studies (US EPA, 2004, section 8.2.2.5.5,
8 Tables 8-3, 8-4). Three studies evaluated in the last review used PM2.5 speciation data to evaluate
9 the effects of air pollutant combinations or mixtures using factor analysis or source
10 apportionment methods to link effects with different PM2.5 source types (Schwartz, 2003; Mar et
11 al., 2003; and Tsai et al., 2000). These studies reported that fine particles from combustion
12 sources, including motor vehicle emissions, coal combustion, oil burning and vegetative burning,
13 were associated with increased mortality. No significant increase in mortality was reported with
14 a source factor representing crustal material in fine particles (US EPA, 2004, section 8.2.2.5.3).
15 The EPA concluded that these studies indicated that exposure to fine particles from combustion
16 sources, but not crustal material, was associated with mortality (US EPA, 2004, p. 8-85; US
17 EPA, 2005, p. 3-16).
18 Recent epidemiologic, lexicological, and controlled human exposure studies have
19 evaluated the health effects associated with ambient PM constituents and sources, using a variety
20 of quantitative methods evaluating a range of PM constituents. Evidence presented in the second
21 draft ISA indicate numerous ambient PM2 5 source categories have been associated with health
22 effects, including factors for PM from crustal and soil, traffic, secondary SC>42 , power plants,
23 and oil combustion sources (US EPA, 2009a, Table 6-17). There is some evidence for trends and
24 patterns that link particular ambient PM components or sources with specific health outcomes,
25 but "there is insufficient evidence to determine whether these patterns are consistent or robust"
26 (US EPA, 2009a, section 6.6.3).
27 As outlined in the second draft ISA, evaluating a potentially large number of ambient PM
28 constituents with a large number of health outcomes "presents difficulties that are related to both
29 the nature of PM, and the methods of quantitative analysis. First, the number of constituents that
30 comprise PM is not only large, but the correlations between them are inherently high. Reducing
31 the correlation between constituents has been accomplished in most of the recent studies through
32 various forms of factor analysis, which limits the correlations between constituents by grouping
33 the most highly correlated ambient PM constituents into less correlated groups or factors. Some
34 studies identify the resulting groups or factors with named sources of ambient PM, but many do
35 not draw explicit links between factors and actual sources" (US EPA, 2009a, section 6.6). The
36 methods for estimating source contributions to ambient PM are reviewed in Section 3.5.4.
PRELIMINARY DRAFT-September 2009 3-56
Do Not Quote or Cite
-------
1 Most epidemiologic, lexicological, and controlled human exposure studies considered
2 between 7 and 20 ambient PM2.5 constituents, with EC, OC, SC>42", and NOs most commonly
3 measured. Many of the studies reduced the number of ambient PM constituents by grouping
4 them with various factorization or source apportionment techniques to examine the relationship
5 between the grouped PM constituents and various health effects. However, not all studies labeled
6 the constituent groupings according to their presumed source and a small number of controlled
7 human exposure and toxicological studies did not apply any kind of grouping to the ambient PM
8 speciation data. In addition, there were differences in the type of PM constituent data used in the
9 various studies (US EPA, 2009a, section 6.6.2.1).
10 In epidemiologic studies, ambient PM2.5 speciation data were obtained from atmospheric
11 monitors; while all of the controlled human exposure and the majority of the toxicological
12 studies used CAPs. According to the second draft ISA, there are important limitations in
13 interpreting the body of scientific evidence for PM components largely because "few, if any of
14 the results are easily comparable, due to: differences in the sets of ambient PM constituents that
15 make up each of the factors53; the subjectivity involved in labeling factors as sources; the
16 numerous potential health effects examined in these studies, including definitive outcomes (e.g.,
17 HAs) as well as physiological alterations (e.g., increased inflammatory response); and the
18 various statistical methods and analytical approaches used in the studies. There are no well-
19 established, objective methods for conducting the various forms of factor analysis and source
20 apportionment, leaving much of the model operation and factor assignment open to judgment by
21 the individual investigator" (US EPA, 2009a, section 6.6.1).
22 A few recent epidemiological studies have examined the association between mortality
23 and morbidity and components of PM2.5. This endeavor has been undertaken by some
24 investigators through the use of data collected for PM2.5 components through the Chemical
25 Speciation Network. As outlined in section 2.x, the CSN includes more than 250 monitors
26 across the country collecting over 40 chemical species. A limited number of CSN monitors
27 began collecting ambient data in 2000 with the majority of sites collecting data starting in 2001.
28 One limitation of the current CSN is that the sampling frequencies of the monitors are either
29 every third day or every sixth day, therefore, in general, reducing the statistical power available
30 to examine associations with mortality in time-series studies.54 As outlined in the second draft
31 ISA, some investigators have circumvented this issue by using the PM2.5 chemical species data in
53 For example, the Al/Si factor identified in one study may differ from the Al/Ca/Fe/Si factor from another study,
and the "Resuspended Soil" factor from a third study.
54 The EPA recognizes that not having daily speciated PM2 5 is a limitation for understanding the health effects
associated with specific fine particle components. The Agency has been taking steps to address this issue (US EPA,
2008).
PRELIMINARY DRAFT-September 2009 3-57
Do Not Quote or Cite
-------
1 a second stage regression55 to explain the heterogeneity in PMio or PM2.5 mortality risk estimates
2 across cities and assuming that the relative contributions of PM2.5 have remained the same over
3 time (Bell et al., 2009; Dominici et al., 2007; Franklin et al., 2008; Lippmann et al., 2006). There
4 have also been some studies that directly analyzed speciated PM2.5 data (e.g., Klemm et al.,
5 2004; Ostro et al., 2007).
6 Lippmann et al. (2006) and Dominici et al. (2007) evaluated the heterogeneity of
7 response associated with PMi0 as evaluated in NMMAPS by analyzing the PM2.5 speciation data.
8 Nickel (Ni) and Vanadium (V) were identified as significant predictors of variation in PMio-
9 related mortality across cities, with Ni levels in New York City being reported as particularly
10 high (US EPA, 2009a, section 6.5.2.5). Bell et al. (2009) and Peng et al. (2009) conducted
11 similar analyses focusing on the variation in PM25-related cardiovascular and respiratory
12 hospital admissions in older adults. Bell et al. (2009) and Peng et al. (2009) used data from the
13 MCAPS study and the CSN to identify the components of PM2 5 that are most strongly associated
14 with hospitalizations for cardiovascular disease. Peng et al. (2009) focused on the components
15 that make up the majority of PM2.5 mass (SO42 , NO3", Si, EC, OC, NA+ and NH4+) and found
16 that in multi-pollutant models only EC and OC were significantly associated with risk of
17 hospitalization for cardiovascular disease. Bell et al. (2009) used data from 20 PM2.5 components
18 and found that EC, Ni, and V were most positively and significantly associated with the risk of
19 PM2 s-related hospitalizations suggesting that the observed associations between PM2 5 and
20 hospitalizations may be primarily due to particles from oil combustion and traffic (US EPA,
21 2009a, section 6.2.10.1). However, as noted in the second draft ISA, in a sensitivity analysis
22 when selectively removing cities from the overall estimate, the significant association between
23 the PMio mortality risk estimate and the PM2 5 Ni fraction was diminished upon removing New
24 York City from the analysis, which is consistent with the results presented by Dominici et al.
25 (2007) (US EPA, 2009a, section 6.5.2.5).
26 In a study of 25 U.S. cities, Franklin et al. (2008) focused on a time-series regression of
27 mortality related to PM2 5 mass by season. In addition, the authors considered the PM2 5 risk
28 estimates to examine effect modification due to various PM2.5 species. They concluded that Al,
29 As, Ni, Si and SO42" were significant effect modifiers of PM2 5 mortality risk estimates, and
30 "simultaneously including Al, Ni, and SO42"_together or Al, Ni, and As together further
31 increased explanatory power. Of all the species examined, Al and Ni explained the most residual
As noted in section 6.5.2.5 of the second draft ISA, "[i]n light of the results presented in speciation studies it must
be noted that second stage analyses that use PM chemical species as effect modifiers have some limitations. Unlike
analyses that directly examine the associations between chemical species and mortality, if an effect modification is
observed it may be confounded if the variations of the mean levels of the chemical species examined are correlated
with other demographic factors that vary across cities. Thus, more concrete conclusions could be formulated if direct
associations are found between mortality and PM chemical components in time-series analyses (US EPA, 2009a).
PRELIMINARYDRAFT-September 2009 3-58
Do Not Quote or Cite
-------
1 heterogeneity." Of note, New York City was not included in the 25 cities examined by Franklin
2 et al. (2008) (US EPA, 2009a, section, 6.5.2.5 and Table 6-16).
3 For cardiovascular effects, the second draft ISA concludes that:
4 ... multiple outcomes have been linked to a PM crustal/soil/road dust source, including
5 cardiovascular mortality in Washington D.C. (Ito et al., 2006) and Santiago, Chile
6 (Cakmak et al., 2009) and ST-segment changes in Helsinki (Lanki et al., 2006), Los
7 Angeles (Gong et al., 2003), and Boston (Wellenius et al., 2003). Interestingly, the ST-
8 segment changes have been observed in an epidemiologic panel study, a controlled
9 human exposure study, and a toxicological study, although the majority of the CAPs in
10 the controlled human exposure study was PMio-2.5. Further support for a crustal/soil/road
11 dust source associated with cardiovascular health effects comes from a PMio source
12 apportionment study in Copenhagen that reported increased cardiovascular hospital
13 admissions (Andersen et al., 2007).
14 PM2.5 traffic and woodsmoke/vegetative burning sources have also been linked to
15 cardiovascular effects. Cardiovascular mortality in Phoenix (Mar et al., 2000; Mar et al.,
16 2006) and Santiago, Chile (Cakmak et al., 2009) was associated with traffic at lag 1.
17 Gasoline and diesel sources were associated with ED visits in Atlanta for cardiovascular
18 disease at lag 0 (Sarnat et al., 2008). Studies that only examined the effects of individual
19 PM2.5 constituents linked EC to cardiovascular hospital admissions in a multi-city
20 analysis (Peng et al., 2009) and cardiovascular mortality in California (Ostro et al., 2007;
21 2008). (US EPA, 2009a, section 6.6.3).
22
23 In a more recent analysis from a single-city study in Atlanta, with an additional 4 years of
24 data and analysis of PM2 5 components, ED visits for cardiovascular disease were not
25 significantly associated with PMio or PM2.5, but were significantly associated with total carbon,
26 EC and organic carbon components of PM2.5 (Tolbert et al., 2007).
27 Thus, the second draft ISA presents evidence that suggests cardiovascular effects
28 (hospitalizations or ED visits) may be associated with PM2.5 from multiple source categories
29 including: wood burning, oil burning, traffic and crustal material, but the best evidence suggests
30 that, in the U.S., oil combustion, wood burning, and traffic are likely the sources of PM2.5 most
31 strongly associated with cardiovascular hospitalizations or ED visits. In addition, the second
32 draft ISA concludes "there are many studies that observed associations between other sources
33 (i.e., salt, secondary SC>42 long-range transport, other metals) and cardiovascular effects, but at
34 this time, there does not appear to be a consistent trend or pattern of effects for those factors"
35 (US EPA, 2009a, section 6.6.3).
36 There is less consistency in observed associations between PM sources and respiratory
37 health effects, which may be partially due to the fact that fewer studies have been conducted that
38 evaluated respiratory-related outcomes and measures. However, there is some evidence for
39 associations with secondary SO42~.and increased respiratory-related ED visits in Atlanta (Sarnat
40 et al., 2008). Decrements in lung function in Helsinki (Lanki et al., 2006) and Los Angeles
PRELIMINARYDRAFT-September 2009 3-59
Do Not Quote or Cite
-------
1 (Gong et al., 2005) in asthmatic and healthy adults, respectively, were also linked to secondary
2 SO42 (US EPA, 2009a, section 6.6.3). Respiratory effects relating to the crustal/soil/road dust
3 and traffic sources of PM included observations of increased respiratory symptoms in asthmatic
4 children (Gent et al., 2009) and decreased peak expiratory flow (PEF) in asthmatic adults
5 (Penttinen et al., 2006). Inconsistent results were observed in studies that evaluated associations
6 with respiratory morbidity and mortality and individual PM2.5 constituents, although Cu, Pb, OC,
7 and Zn were related to respiratory health effects in two or more studies (US EPA, 2009a, section
8 6.6.3).
9 A few studies have identified PM2.5 sources associated with total mortality. These studies
10 found an association between mortality and a PM2.5 coal combustion factor (Laden et al., 2000),
11 while others linked mortality to a secondary SC>42long-range transport PM2 5 source (Ito et al.,
12 2006; Mar et al., 2006).
13 Recent studies have shown that source apportionment methods have the potential to add
14 useful insights into which sources and/or PM constituents may contribute to different health
15 effects. Of particular interest are several epidemiologic studies that compared source
16 apportionment methods and the associated results. These studies are discussed in section 6.6.2.1
17 of the second draft ISA. One set of studies compared epidemiologic associations with PM2.5
18 source factors using several methods - PCA, PMF, and UNMIX - independently analyzed by
19 separate research groups (Hopke et al., 2006; Ito et al., 2006; Mar et al., 2006; Thurston et al.,
20 2005). Schreuder et al. (2006) compared UPM and two versions of UNMIX to derive tracers and
21 Sarnat et al. (2008) compared PMF, modified CMB, and a single-species tracer approach. In all
22 analyses, epidemiologic results based on the different methods were generally in close
23 agreement. The variation in risk estimates for daily mortality between source categories was
24 significantly larger than the variation between research groups (Ito et al., 2006; Mar et al., 2006;
25 Thurston et al., 2005). Additionally, the variation in risk estimates based on the source
26 apportionment model used had a much smaller effect than the variation caused by the different
27 source constituents. Further, the most strongly associated source types were consistent across all
28 groups. This supports the general validity of such approaches, though integration of results
29 would be simpler if the methods employed for grouping PM constituents were more consistent
30 across studies and disciplines. Further research would aid understanding of the contribution of
31 different factors, sources, or source tracers of PM to health effects by increasing the number of
32 locations where similar health endpoints or outcomes are examined.
33 In summary, these findings presented in the second draft ISA are consistent with the
34 conclusions of the 2004 PM AQCD, that is, a number of source types, including motor vehicle
35 emissions, coal combustion, oil burning, and vegetative burning, are associated with health
36 effects (U.S. EPA, 2004, section x.x, US EPA, 2009a, section 6.6.3). Differences observed
PRELIMINARY DRAFT-September 2009 3 -60
Do Not Quote or Cite
-------
1 across studies may be due to the difference in geographic coverage, PM size (PM2.5 may
2 represent more secondary aerosols than PMio), or the difference in the analytical methods used in
3 each study. One difference noted is that although the crustal factor of fine particles was not
4 associated with mortality in the last review "recent studies have suggested that PM (both PM2.5
5 and PMio-2.s) from crustal, soil or road dust sources or PM tracers linked to these sources are
6 associated with cardiovascular effects." In addition, the second draft ISA reports that secondary
7 SO42~has been associated with both cardiovascular and respiratory effects. To summarize, the
8 second draft ISA concludes, "Overall, the results displayed in Table 6-17 indicate that many
9 constituents of PM can be linked with differing health effects and the evidence is not yet
10 sufficient to allow differentiation of those constituents or sources that are more closely related to
11 specific health outcomes" (US EPA, 2009a, section 6.6.3).
12 3.3.2.3.2 Effects Associated with Long-term Exposures to PM2.5 Components and
13 Sources
14 Limited evidence is available to evaluate the health effects associated with long-term
15 exposures to PM2.5 components. This evidence is presented and discussed in section 7.6.2 of the
16 second draft ISA (US EPA, 2009a). The most significant new evidence is provided by a study
17 that evaluated multiple PM2 5 components and an indicator of traffic density in an assessment of
18 effects related to long-term exposure to PM2.5 (Lipfert et al., 2006). Using health data from a
19 cohort of U.S. military veterans and PM2.5 data from EPA's CSN, Lipfert et al. (2006) reported a
20 positive association for mortality with sulfates. Positive associations were found between
21 mortality and long-term exposures to nitrates, EC, Ni and V, as well as traffic density and peak
22 Os concentrations. In multi-pollutant models, associations with traffic density remained
23 significant, as did nitrates, Ni and V in some models. Additional evidence from a long-term
24 exposure study conducted in a Dutch cohort provides supportive evidence that long-term
25 exposure to traffic-related particles is associated with increased mortality (Breelen et al., 2008).
26 3.3.2.4 Ultrafine Particles
27 In evaluating ultrafme particles (UFPs), the second draft ISA includes evaluation of a
28 limited number of controlled human exposure studies that examined the association between
29 UFPs and cardiovascular morbidity as well as a larger body of evidence from controlled human
30 exposures studies that exposed subjects to fresh diesel exhaust (DE), which is typically
31 dominated by UFPs. As summarized in section 2.3.6 of the second draft ISA, "[t]he controlled
32 human exposure studies evaluated have consistently demonstrated effects on vasomotor function
33 and systemic oxidative stress with additional evidence for alterations in HRV parameters in
34 response to exposure to ultrafme CAPs. The toxicological studies provide coherence for the
35 changes in vasomotor function observed in the controlled human exposure studies...More limited
PRELIMINARY DRAFT-September 2009 3 -61
Do Not Quote or Cite
-------
1 evidence is available regarding the effect of UFP on respiratory effects. Controlled human
2 exposure studies have not extensively examined the effect of UFPs on respiratory measurements,
3 but a few studies have observed small decrements in pulmonary function. Additional effects
4 including oxidative, inflammatory, and pro-allergic outcomes have been demonstrated in
5 toxicological studies, but the lack of coherence with the controlled human exposure studies limits
6 the interpretation of these findings."
7 Section 2.3.6 of the second draft ISA makes note that "[e]pidemiologic studies are
8 limited because a national network is not in place to measure UFPs in the U.S. UFP
9 concentrations are spatially variable, which would increase uncertainty and make it difficult to
10 detect associations between health effects and UFPs in epidemiologic studies. In addition, data
11 on the composition of UFPs and potential effects of UFP constituents are sparse." Furthermore,
12 the second draft ISA notes"[o]verall, a limited number of studies have examined the association
13 between exposure to UFP and morbidity and mortality. Of the studies evaluated, controlled
14 human exposure studies provide the most evidence for UFP-induced cardiovascular and
15 respiratory effects; however, these studies focus on exposure to DE. As a result, it is unclear if
16 the effects observed are due to UFP[s], larger particles (i.e., PM^.s), or the gaseous components
17 of DE. Additionally, ultrafine CAPs systems are limited as the atmospheric ultrafme PM
18 composition is modified when concentrated, which adds uncertainty to the health effects
19 observed in controlled human exposure studies (US EPA, 2009a, chapter 1) (US EPA 2009a,
20 section 2.3.6).
21 Collectively, EPA has determined that this evidence is suggestive of a causal relationship
22 between short-term exposures to UFPs and cardiovascular and respiratory effects, with stronger
23 evidence for cardiovascular than respiratory effects (US EPA, 2009a, section 2.3.5). In addition,
24 EPA has determined that data are inadequate to infer whether a causal relationship exists for
25 short-term exposures to UFPs and premature mortality and central nervous system effects as well
26 as inadequate to infer whether a causal relationship exists for long-term exposures to UFPs and
27 all health effects and mortality (US EPA, 2009a, section 2.3). Evidence of cardiovascular and
28 respiratory effects associated with UFPs is briefly summarized below.
29 3.3.2.4.1 Cardiovascular Effects
30 As summarized in section 2.3.5 of the second draft ISA, "[Controlled human exposure
31 studies provide the majority of the evidence for cardiovascular health effects in response to
32 short-term exposure to UFPs. While there are a limited number of studies that have examined the
33 association between UFPs and cardiovascular morbidity, there is a larger body of evidence from
34 studies that exposed subjects to fresh DE, which is typically dominated by UFPs. These studies
35 have consistently demonstrated effects on vasomotor function (Section 6.2.4.2). Markers of
PRELIMINARY DRAFT-September 2009 3 -62
Do Not Quote or Cite
-------
1 systemic oxidative stress have also been observed to increase after exposure to various particle
2 types that are predominantly in the UFP size range. In addition controlled human exposure
3 studies have observed alterations in HRV parameters in response to exposure to ultrafine CAPs,
4 with inconsistent evidence for changes in markers of blood coagulation following exposure to
5 ultrafine CAPs and DE (Sections 6.2.1.2 and 6.2.8.2). A few toxicological studies have also
6 observed consistent changes in vasomotor function, which provides coherence with the effects
7 demonstrated in the controlled human exposure studies (Section 6.2.4.3). Additional UFP-
8 induced effects observed in toxicological studies include alterations in HRV, with less consistent
9 effects observed for systemic inflammation and blood coagulation. Only a few epidemiologic
10 studies have examined the effect of UFP on cardiovascular morbidity and collectively they found
11 inconsistent evidence for an association between UFPs and CVD hospital admissions, but some
12 positive associations for subclinical measures of CVD (i.e., arrhythmias and supraventricular
13 beats) (Section 6.2.2.1). These studies were conducted in the U.S. and Europe at mean particle
14 number concentration ranges of-8,500-36,000 particles/cm3. However, UFP number
15 concentrations are highly dependent on monitor location (i.e., concentrations drop off quickly
16 from the road compared to accumulation mode particles), and therefore, more subject to
17 exposure error than accumulation mode particles. In conclusion, the evidence from the studies
18 evaluated is suggestive of a causal relationship between short-term exposures to UFPs and
19 cardiovascular effects."
20 3.3.2.4.2 Respiratory Effects
21 As summarized in section 2.3.5 of the second draft ISA "[a] limited number of
22 epidemiologic studies have examined the potential association between short-term exposure to
23 UFPs and respiratory morbidity. Of the studies evaluated, there is limited, and predominately
24 inconsistent evidence for an association between short-term exposure to UFPs and respiratory
25 symptoms, as well as asthma hospital admissions at a median particle number concentration of
26 -6,200 to a mean of 38,000 particles/cm3 (Section 6.3.8). The spatial and temporal variability of
27 UFPs also affects these associations. Although controlled human exposure studies have not
28 extensively examined the effect of UFPs on respiratory outcomes, a few studies have observed
29 small UFP-induced decreases in pulmonary function. However, these studies have not reported
30 an increase in respiratory symptoms and the observed effects on pulmonary inflammation are not
31 consistent. Toxicological studies have also reported mixed results when examining the effect of
32 UFPs on respiratory effects, but several studies demonstrate oxidative, inflammatory, and
33 allergic responses (Section 6.3). Some effects, such as inflammation or pulmonary
34 histopathology, are only observed when using particular animal models (e.g., immature or
35 compromised). Additionally, although a number of controlled human exposure and toxicological
PRELIMINARY DRAFT-September 2009 3-63
Do Not Quote or Cite
-------
1 studies that used controlled exposures to fresh DE report respiratory effects, the relative
2 contributions of gaseous copollutants to the health effects observed remain unresolved. Thus, the
3 current collective evidence is suggestive of a causal relationship between short-term exposures to
4 UFPs and respiratory effects."
5 3.3.3 Thoracic Coarse Particles
6 In the PM NAAQS review completed in 1997, little new toxicologic evidence was
7 available on potential effects of thoracic coarse particles and there were few epidemiologic
8 studies that had included direct measurements of thoracic coarse particles. Evidence of
9 associations between health outcomes and PMio that were conducted in areas where PMio was
10 predominantly composed of thoracic coarse particles was an important part of EPA's basis for
11 reaching conclusions about the requisite level of protection from coarse particles provided by the
12 final standards (62 FR 38652, July 18, 1997).
13 The evidence from a growing, but still limited, body of evidence available in the PM
14 NAAQS review completed in 2006 included epidemiologic studies that had reported associations
15 with health effects using direct measurements of PMio-2.5, as well as number of new toxicological
16 studies on known or potential effects associated with exposure to thoracic coarse particles and
17 their major constituents. In the last review, EPA highlighted that the nature of the effects
18 associated with short-term exposures to ambient thoracic coarse particles, particularly in urban
19 and industrial settings, included aggravation of respiratory and cardiovascular disease (as
20 indicated by increased hospital admissions), increased respiratory symptoms in children, and
21 premature mortality. Populations that appeared to be at greater risk for experiencing effects
22 related to thoracic coarse particles exposures included individuals with preexisting lung diseases
23 such as asthma, as well as children and older adults.56
24 In the current review, new epidemiologic, controlled human exposure, and toxicological
25 studies provide evidence that is suggestive of associations between short-term PMio-2.5 exposures
26 and cardiovascular effects, respiratory effects, and mortality (US EPA 2009a, section 2.3.3) and
27 inadequate for inferring whether a causal relationship exists between PMi0-2.5 and central
28 nervous system effects (US EPA, 2009a, section 2.3). Causal determinations presented in the
29 second draft ISA "were made for PMio-2.5 as a whole regardless of origin, since PMi0-2.5-related
30 effects have been demonstrated for a number of different environments" (US EPA, 2009a,
31 section 2.3.4). These causal determinations were made based upon short-term PMio-2.5 exposure
32 studies. With respect to evaluating long-term exposures, the second draft ISA concludes that
56 See summary of the evidence of health effects related to thoracic coarse particle exposures in the 2006 proposed
rule - 71 FR 2654 to 2662, January 17, 2006; and a more comprehensive and detailed discussions of the scientific
evidence in chapters 6-9 of the Criteria Document (US EPA, 2004) and chapter 3 of the Staff Paper (US EPA,
2005).
PRELIMINARY DRAFT-September 2009 3 -64
Do Not Quote or Cite
-------
1 available evidence is inadequate (US EPA, 2009a, section 2.3). Specifically, the second draft
2 ISA states, "[t]o date, a sufficient amount of evidence does not exist in order to draw conclusions
3 regarding the health effects and outcomes associated with long-term exposure to PMio-2.5" (US
4 EPA, 2009a, section 2.3.4).
5 Measurement error and exposure error are issues that may be distinctly more important
6 for interpretation of results for assessments of PMio-2.5 than for PM2.5. This exposure
7 measurement error may bias effect estimates for PMio-2.5 toward the null. These and other issues
8 relevant to the interpretation of epidemiological evidence of PM-related effects are discussed in
9 section 3.4 below.
10 The majority of epidemiological studies evaluating thoracic coarse particles provide
11 limited information on the air quality data used in the analyses, generally reporting average
12 concentrations rather that upper percentile values making it difficult to understand the air quality
13 distribution considered in the analyses. The statistical form of the current PMio standard is set as
14 one expected exceedence and evidence has supported looking carefully at short-term peak
15 exposures to evaluate public impacts of thoracic coarse particles. Without additional air quality
16 data, we are limited to interpret the policy relevancy of the scientific body of evidence for
17 thoracic coarse particles. As noted in section 3.3.2.1 above, we have requested additional air
18 quality data from study authors, including information on coarse particles, in order to better
19 interpret the policy relevancy of the scientific body of evidence, including air quality data for
20 evaluating short-term exposures to thoracic coarse particles.
21 3.3.3.1 Effects Associated with Short-term PMio-2.5 Exposures
22 In the PM NAAQS review completed in 2006, emphasis was placed on evaluating a
23 range of respiratory morbidity effects associated with thoracic coarse particle exposures (US
24 EPA, 2004, section 8.4.6.4) with consideration of new findings on potential cardiovascular
25 effects of thoracic coarse particles (US EPA, 2004, section 8.3.1.4). New evidence from studies
26 of mortality available in the last review indicated effect estimates for associations between
27 mortality and short-term exposures to PMio-2.5 that were generally positive and similar in
28 magnitude to those for PM2.5 and PMio though most were not statistically significant and, in
29 general, had greater confidence intervals (indicating greater uncertainty). Furthermore, effect
30 estimates were somewhat larger for cause-specific mortality (e.g., respiratory and cardiovascular
31 mortality) than for total mortality (US EPA, 2005, Figures 3-1 and 3-2).
32 In the current review, additional epidemiological evidence as well as limited new
33 toxicological and controlled human exposure studies are available. Collectively, as noted above,
34 EPA has determined that this evidence is suggestive of a causal relationship between short-term
35 exposures to PMio-2.5 and cardiovascular effects, respiratory effects, and mortality (US EPA,
PRELIMINARY DRAFT-September 2009 3-65
Do Not Quote or Cite
-------
1 2009a, section 2.3.3, Table 2-3). However, limited PMio-2.5 air quality data, variability in the
2 chemical and biological composition of PMio-2.5, limited evidence regarding effects of the
3 various components of PMio-2.5, and lack of clearly defined biological mechanisms for PMio-2.s-
4 related effects are important sources of uncertainty (US EPA, 2009a, section 2.3.3.2). The key
5 evidence supporting these causal determinations is briefly described below.
6 3.3.3.1.1 Cardiovascular and Systemic Effects
7 Limited evidence was available in the last review regarding PM10-2.5-related
8 cardiovascular effects. Two single-city epidemiologic studies found generally positive
9 associations of PMio-2.5 and cardiovascular hospital admissions in Toronto (Burnett et al., 1997)
10 and Detroit, MI (Ito, 2003). A study in Tucson, AZ, an urban area where thoracic coarse
11 particles are a much greater fraction of PMio than are fine particles, reported a statistically
12 significant association between PMio and increased hospitalizations for cardiovascular disease
13 (Schwartz, 1997). In addition, one study in the last review reported no significant association
14 between onset of myocardial infarction (MI) and short-term PMio-2.5 exposures (Peters et al.,
15 2001).
16 In this review, new epidemiological evidence includes a recent multi-city study
17 evaluating hospital admissions and emergency department (ED) visits for cardiovascular disease
18 in older adults. In a study of Medicare patients (MCAPS), Peng et al. (2008) reported a
19 significant association between PMio-2.5 and cardiovascular disease hospitalizations in a single
20 pollutant model using air quality data for 108 U.S. counties with one or more co-located PM2.5
21 and PMio monitors.57 However, this association diminished slightly and was no longer
22 statistically significant after adjustment for PM2.5 (US EPA, 2009a, section 6.2.12.2).
23 In contrast, associations of cardiovascular outcomes with PMio-2.5 were weak for CHF
24 and null for fflD in the Atlanta-based SOPHIA study (Metzger et al., 2004). Results from
25 single-city studies were generally positive but effect sizes were heterogeneous and estimates
26 were imprecise (US EPA, 2009a, section 6.2.10). Crustal material from a dust storm in the Gobi
27 desert that was largely coarse PM (generally indicated using PMio) was associated with
28 hospitalizations for cardiovascular diseases including IHD and CHF in most studies (US EPA,
29 2009a, section 6.2.10.1). A few epidemiologic studies that examined the association between
30 short-term exposure to PMio-2.5 and cardiovascular mortality (US EPA, 2009a, section 6.2.11)
57 Analyses with PM10.2 5 were carried out using similar methods to those of Dominici et al. (2006). Peng et al.
(2008) evaluated the robustness of PM25 associations to adjustment for thoracic coarse PM (Peng et al., 2008).
Gaseous pollutants were not considered in these analyses. In 108 U.S. counties with co-located PM10 and PM10.2.5
monitors, the authors found a 0.4% (95% PI, 0.1 to 0.7, lag 0) increase in risk per 10 ug/m3 PM10_2.5 and no
associations at lags of 1 and 2 days. In a 2-pollutant model adjusted for PM2 5, the association between PM10.2 5 and
CVD hospitalization lost precision (0.3% [95% PI: -0.1 to 0.6, lag 0]) (Peng et al., 2008).
PRELIMINARY DRAFT-September 2009 3 -66
Do Not Quote or Cite
-------
1 provide supporting evidence for the cardiovascular-related hospital admission and ED visit
2 studies. A multi-city study reported risk estimates for cardiovascular mortality of similar
3 magnitude to those for all-cause (nonaccidental) mortality (Zanobetti and Schwartz, 2009).
4 However, the single-city studies evaluated (Wilson et al., 2007; Villeneuve et al., 2003) reported
5 substantially larger effect estimates, possibly related to differences between cities and
6 compositional differences in PMio-2.5 across regions. Of note is the lack of analyses within the
7 studies to evaluate potential confounders of the PMi0-2.5-cardiovascular mortality relationship
8 (US EPA, 2009a, section 6.2.12.2).
9 As presented in the second draft ISA, EPA concludes that collectively, the evidence from
10 epidemiologic studies, along with the more limited evidence from controlled human exposure
11 and toxicological studies is suggestive of a causal relationship between short-term exposure to
12 PMio-2.5 and cardiovascular effects (US EPA, 2009a, section 2.3.3, section 6.2.12.2). The second
13 draft ISA concludes, "[i]n multi-city epidemiologic studies, associations between short-term
14 exposure to PMio-2.5 and cardiovascular outcomes (i.e., IHD, HAs, supraventricular ectopy, and
15 changes in HRV) have been found that are similar in magnitude to those observed in PM2.5
16 studies (US EPA, 2009a, section 2.3.4). Controlled human exposure studies have also observed
17 alterations in HRV, providing consistency and coherence for the effects observed in the
18 epidemiologic studies. To date, only a limited number of toxicological studies have been
19 conducted to examine the effects of PMio-2.5 on cardiovascular outcomes. All of these studies
20 involved IT instillation due to the technical challenges of using PMio-2.5 for rodent inhalation
21 studies.... As a result, the toxicological studies evaluated provide limited biological plausibility
22 for the PMio-2.5 effects observed in the epidemiologic and controlled human exposure studies"
23 (US EPA, 2009a, section 2.3.4).
24 3.3.3.1.2 Respiratory Effects
25 In the last PM NAAQS review, epidemiologic studies analyzing the effect of PMio-2.5
26 exposures on hospitalizations or ED visits for respiratory diseases (i.e., pneumonia, COPD and
27 respiratory diseases combined) reported generally positive associations (US EPA, 2004, section
28 x.x). The few mortality studies that examined cause-specific mortality associated with PMio-2.5
29 exposures suggested somewhat larger risk estimates for respiratory mortality compared to all-
30 cause (non-accidental) mortality. Several epidemiologic studies of respiratory symptoms and
31 thoracic coarse particles provided limited evidence for cough and effects on morning PEF (ADD
32 REF). Toxicology data for PMio-2.5 were extremely limited, and there were no controlled human
33 exposure studies considered in the last review that evaluated the effect of PMio-2.5 on respiratory
34 symptoms, pulmonary function, or inflammation.
PRELIMINARY DRAFT-September 2009 3 -67
Do Not Quote or Cite
-------
1 In this review, several new epidemiologic studies report associations between short-term
2 PMio-2.5 exposures and hospital admissions, ED visits, or outpatient visits related to respiratory
3 effects with the most consistent evidence reported in children, including asthmatic children (US
4 EPA, 2009a, Figures 6-10 through 6-14). Overall, the second draft ISA reports the most
5 compelling new evidence showing significant associations between an increase in respiratory-
6 related hospital admissions and ED visits and short-term exposure to PMio-2.5 has been observed
7 in studies conducted in Canada and Europe (e.g., Host et al, 2008; Fung et al., 2006; Chen et al.,
8 2005; Yang et al., 2004). In these studies, the strongest associations were reported for children,
9 with less consistent evidence of effects in adults, including older adults (i.e., 65 years of age and
10 older) (US EPA, 2009a, section 2.3.3). Mar et al., (2004) report an association between short-
11 term PMio-2.5 exposures and increased respiratory symptoms in asthmatic children but not
12 asthmatic adults. Several Canadian studies of respiratory hospital admissions reported larger
13 effects for PMio-2.5 compared to PM2.5 that were robust to adjustment for gaseous pollutants
14 (Chen et al., 2005; Lin et al., 2002; Yang et al., 2004). The COPD associations with short-term
15 PMio-2.5 exposures reported by Chen et al. (2004) remained positive but were diminished slightly
16 after adjustment for NC>2. Sinclair and Tolsma (2004) reported a significant association between
17 PMio-2.5 exposures and childhood but not adult asthma-related outpatient visits in Atlanta, GA.
18 In contrast, Slaughter et al. (2005) reported no associations with ED visits or hospitalizations for
19 asthma among all ages in Spokane, WA. Although a number of studies provide evidence of
20 respiratory effects in older adults, a recent analysis of MCAPS data reports that weak
21 associations of PMio-2.5 with respiratory hospitalizations are further diminished after adjustment
22 forPM2.5.58
23 Limited evidence for PMio-2.5- related effects is available from controlled human studies
24 and toxicological studies. Increases in markers of pulmonary inflammation in the absence of
25 lung function effects or respiratory symptoms have been reported in healthy adults (US EPA,
26 2009a, section x.x). In toxicological studies, PMio-2.5 from both rural and urban environments
27 has induced inflammation and injury responses in rats or mice following IT instillation, "making
28 it difficult to distinguish effects of PMio-2.5 from different environments" (US EPA, 2009a,
29 section 2.3.3). The majority of toxicological evidence, described in section 6.3.7.1 of the second
30 draft ISA and in the 2004 PM AQCD (U.S. EPA, 2004, section x.x), suggests that PM impairs
31 innate immunity, the first line of defense in preventing infection.
58 In an analysis of PM10.25, MCAPS investigators observed small imprecise increases in respiratory admissions with
24-h PMj0-2s concentration (0.33% [95% PI: -0.21 to 0.86, per 10 ug/m3, lag 0]) (Peng et al., 2008), which decreased
after adjustment for PM25 (0.26% [95% PI: -0.32 to 0.84 per 10 ug/m3 lag 0]) (US EPA, 2009a, section 6.3.8.1)
PRELIMINARY DRAFT-September 2009 3-68
Do Not Quote or Cite
-------
1 Overall, based on epidemiologic studies, along with the limited number of controlled
2 human exposure and toxicological studies that examined PMio-2.5 respiratory effects the second
3 draft ISA determines that the overall scientific data base for thoracic coarse particles provides
4 evidence that is suggestive of a causal relationship between short-term exposures to PMi0-2.5 and
5 respiratory effects (US EPA, 2009a, section 2.3.3.1).
6 3.3.3.1.3 Mortality
7 Evidence available in the last review evaluating short-term exposure to PMio-2.5 and
8 premature mortality was limited and indicated mixed results primarily from single-city studies,
9 where two studies reported a positive and statistically significant association while other studies
10 reported associations that were generally positive but not statistically significant. Effect
11 estimates were similar in magnitude to those for PM2.5 and PMio (US EPA, 2005, section
12 3.3.1.1). Staff noted that on a unit mass basis, the effect estimates for both PM2.5 and PMio-2.5
13 [were] generally larger than those for PMio, which is consistent with PM2.5 and PMio-2.5 having
14 independent effects (US EPA, 2004, p. 9-25). In general, cause-specific effect estimates reported
15 in the last review were somewhat larger for respiratory and cardiovascular mortality than for
16 total mortality. Two out of five studies reported effect estimates for cardiovascular mortality
17 with short-term PMio-2.5 exposures as positive and statistically significant (Mar et al., 2003; Ostro
18 et al., 2003) while none of the effect estimates for total mortality or respiratory mortality reached
19 statistical significance (US EPA, 2005, Figure 3-1). The magnitude of the effect estimates for
20 PMio-2.5 were similar to those for PM2.5, generally falling in the range of 3 to 8 % for
21 cardiovascular mortality and 3 to 16% per 25 |ig/m3 PMio-2.5 (US EPA, 2004, p. 8-306).59 These
22 studies were considered to have multiple limitations including measurement and exposure issues
23 for PMio-2.5 correlations between PM2.5 and PMio-2.5. "These limitations increased the uncertainty
24 surrounding the concentrations at which PMio-2.5 mortality associations were observed" (US
25 EPA, 2009a, section 6.5.2.3).
26 In this review, the majority of new studies continue to provide evidence of an association
27 between premature mortality and short-term PMio-2.5 exposures. This evidence continues to
28 show positive, but, not generally statistically significant results with generally stronger evidence
29 reported for cardiovascular-related and respiratory-related mortality rather than all-cause,
30 nonaccidental mortality (US EPA, 2009, Figures 6-28 and 6-30). Two new multi-city studies
31 have been published since the last review. A Canadian 12-city study provides evidence for an
32 association between short-term exposure to PMio-2.5 and mortality using direct measurements of
59 As noted in the last review, effect estimates for respiratory mortality are often larger than those for either total or
cardiovascular mortality, but they are often less precise, which would be expected since respiratory deaths comprise
a small proportion of total deaths (US EPA, 2005, section 3.3.1.1).
PRELIMINARY DRAFT-September 2009 3-69
Do Not Quote or Cite
-------
1 PMio-2.5 with an every 6th day sampling schedule (Burnett et al., 2004, US EPA, 2009a, section
2 6.5.2.3). In addition, an examination of PMio-2.5 mortality associations in a new multi-city U.S.
3 study found associations between PMio-2.5 and cardiovascular, respiratory, and total mortality,
4 but this association varied when examining city-specific risk estimates (Zanobetti and Schwartz,
5 2009; US EPA, 2009a, Figure 6-29; section 6.5.2.3). The study authors reported seasonal (i.e.,
6 larger in spring) and regional differences in PMio-2.5 respiratory mortality risk estimates. In this
7 study, PMio-2.5 ambient concentrations were estimated by calculating the difference in
8 countywide averages of PMio and PM2.5. As reported in section 6.5.2.3 of the second draft ISA
9 (US EPA, 2009a), "[i]t is not clear how the computed PMio-2.5 concentrations used by Zanobetti
10 and Schwartz (2009) compare with the PMio-2.5 concentrations obtained by directly measuring
11 PMio-2.5 using a dichotomous sampler, or the PMio-2.5 concentrations computed using the
12 difference of PMio and PM2.5 measured at co-located samplers." Furthermore, copollutant
13 analyses were not conducted in the Zanobetti and Schwartz (2009) study, and the associations
14 observed were inconsistent with those reported in single-city studies, for example, single-city
15 studies previously conducted in Phoenix (Mar et al., 2003; Wilson et al., 2007) reported a PMi0.
16 2.5 mortality association in this "dry" region city in contrast to a lack of association reported by
17 Zanobetti and Schwartz in this region (see US EPA, 2009a, Figure 6-28). In addition, single-city
18 studies conducted in Atlanta, GA (Klemm et al., 2004), and Vancouver, Canada (Villeneuve et
19 al., 2003) reported no associations between short-term exposure to PMio-2.5 and respiratory
20 mortality where the overall multi-city analysis conducted by Zanobetti and Schwartz (2009)
21 indicated an association. Villeneuve et al. (2008) observed a statistically significant association
22 between PMio-2.5 exposures and cardiovascular mortality in Vancouver, Canada. The
23 investigators reported a stronger association for PMio-2.5 than for PM2.5 (US EPA, 2009a, section
24 6.5.2.3)
25 The second draft ISA notes limitations "exist in the PMio-2.5 associations reported due to
26 the small number of PMio-2.5 studies that have investigated confounding by gaseous copollutants
27 or the influence of model specification on PMio-2.5 risk estimates. Additionally, more data [are]
28 needed to characterize the chemical and biological components that may modify the potential
29 toxicity of PMi0-2.5 (US EPA, 2009a, section 6.5.2.3) Specifically, the second draft ISA notes
30 that the "difference in the results observed between the multi- and single- city studies could be
31 due to a variety of factors including differences between cities and compositional differences in
32 PMio-2.5 across regions" (see also, US EPA, 2009a, Figure 6-28). As was reported in the last
33 review, the strongest evidence for cardiovascular related mortality associated with thoracic
34 coarse particles has been reported for Phoenix (Mar et al., 2003) and Coachella Valley (Ostro et
35 al., 2003).
PRELIMINARY DRAFT-September 2009 3 -70
Do Not Quote or Cite
-------
1 3.3.3.2 Effects Associated with Long-term PMi0-2.s Exposures
2 Evidence to evaluate health impacts of long-term PMi0-2.5 exposures is extremely limited.
3 In the last PM NAAQS review, no association was observed between mortality and long-term
4 exposure to PMi0-2.5 in the ACS study (Pope et al., 2002), and a positive but nonsignificant
5 association was reported in males in the AHSMOG cohort (McDonnell et al., 2000). In that
6 review, EPA concluded that the available studies provided no evidence for associations between
7 long-term exposure to coarse fraction particles and mortality. Furthermore, there was little
8 evidence available to draw conclusions about long-term PMio-2.5 exposures and morbidity (US
9 EPA, 2009a, section 9.2.2.1.2, p. 9-34). In 2006, EPA revoked the annual PMio standard stating:
10
11 The long-term exposure studies of mortality and morbidity that permit comparisons of
12 fine and coarse particles continue to suggest that, at current ambient levels in the US, fine
13 particles are associated with health effects and coarse particles are not.60 (71 FR 61198,
14 October 17, 2006).
15
16 In the current review, evidence is still limited for evaluating long-term exposures to
17 thoracic coarse particles. The second draft ISA concludes that "[t]o date, a sufficient amount of
18 evidence does not exist in order to draw conclusions regarding the health effects and outcomes
19 associated with long-term PMio-2.5 exposures" (US EPA, 2009a, section 2.3.3.2). New findings
20 from the California Seventh Day Adventist Study (AHSMOG) and Veterans cohort studies
21 provide limited evidence of associations between long-term exposure to PMio-2.5 and mortality
22 (US EPA, 2009a, sections 7.6.1 and 7.6.5.2). The AHSMOG study reported a positive
23 association for PMio-2.5 and coronary heart disease mortality among females, but not males as
24 was found with fine particles; associations were strongest in the subset of postmenopausal
25 women (Chen et al., 2005). A recent reanalysis of the Veterans cohort study (Lipfert et al, 2006)
26 focused on traffic-related air pollution reported a significant association between long-term
27 PMio-2.5 exposures and total mortality in a single-pollutant model, however, that association
28 became negative and not statistically significant in a model that included traffic density. The
29 second draft ISA concluded "[a]s it would be expected that traffic would contribute to the
30 thoracic coarse particle concentrations, it is difficult to interpret the results of these multi-
31 pollutant analyses" (US EPA, 2009a, section 7.6.1). No supporting toxicological or clinical
60 See US EPA 2004, section 8.4.6.2 - "no statistically significant associations have been reported between long-
term exposure to coarse fraction particles and cause-specific mortality" and section 8.4.6.4 "[tjhe recent studies
suggest that long-term exposure to fine particles is associated with development of chronic respiratory disease and
reduced lung function growth; little evidence is available on potential effects of exposure to coarse fraction
particles."
PRELIMINARY DRAFT-September 2009 3 -71
Do Not Quote or Cite
-------
1 studies of long-term exposure to ambient PMio-2.5 and cardiovascular effects have been
2 conducted to date.
3 Studies evaluating the association between long-term exposure to PMi0 and respiratory
4 morbidity were primarily in areas where PM is predominantly fine particles. Consequently,
5 studies utilizing the PMio fraction provide limited information on the association between long-
6 term exposure to thoracic coarse particles and respiratory morbidity.
7 Overall, the second draft ISA determined that evidence is inadequate to infer the
8 presence or absence of a causal relationship between long-term PMio-2.5 exposures and a range of
9 health effects, specifically, cardiovascular and systemic effects (US EPA, 2009a, section
10 7.2.12.2), respiratory effects (US EPA, 2009a, section 7.3.9.2), cancer (US EPA, 2009a, section
11 7.5.4) and mortality (US EPA, 2009a, section 7.5.5.2).
12 3.3.3.3 PMio-i.s Components
13 One of the major research priorities defined by the National Academy of Sciences
14 National Research Councils (NRC) related to assessing the health effects of PM components and
15 sources (NRC, 2004). In the last review, the issue of differential toxicity related to particles
16 originating in urban versus non-urban areas was a central focus. The strongest evidence of
17 effects associated with short-term exposures to ambient thoracic coarse particles were reported in
18 studies evaluating urban and/or industrial settings. In general, there was evidence of toxicity of
19 certain components of thoracic coarse particles, such as metals and endotoxins, as well as
20 evidence that natural crustal materials of geologic origin, such as Mt. St. Helens volcanic ash,
21 may have very little toxicity. There was largely an absence of evidence regarding the presence
22 or absence of toxicological effects associated with other types of coarse particles in non-urban
23 areas.
24 With respect to thoracic coarse particles, PMio-2.5 may contain iron, silica, aluminum, and
25 base cations from soil, plant and insect fragments, glucans, pollen, fungal spores, bacteria,
26 animal waste, and viruses, as well as fly ash, brake lining particles, debris, and automobile tire
27 fragments, toxic trace elements and other components from previously deposited fine PM e.g.
28 metals from smelters and steel mills, polycyclic aromatic hydrocarbons from automobile
29 exhaust, and pesticides from agricultural activities (US EPA, 2009a, section 2.3.4; EPA 2004 8-
30 344). It is important to note that the chemical composition of PMio-2.5 can vary considerably by
31 location, but city-specific speciated PMio-2.5 data are limited (ADD REF). Staff observes that no
32 epidemiologic evidence is available to evaluate effects of different components or sources of
33 thoracic coarse particles and that, overall, EPA has not considered the scientific data to be
34 extensive enough to support making causal determinations for thoracic coarse particle
35 components in this review. Causal determinations presented in the second draft ISA "were made
PRELIMINARY DRAFT-September 2009 3 -72
Do Not Quote or Cite
-------
1 for PMio-2.5 as a whole regardless of origin, since PMi0-2.5-related effects have been demonstrated
2 for a number of different environments" (US EPA, 2009a, section 2.3.4).
3 3.4 ISSUES RELATED TO INTERPRETING EPIDEMIOLOGIC EVIDENCE
4 The 2004 AQCD included extensive discussions of methodological issues for
5 epidemiologic studies, including questions about model specification or selection, co-pollutant
6 confounding, measurement error in pollutant measurements, and exposure misclassification.
7 Based on information available in the last review, the 2004 PM AQCD concluded that PM-health
8 effects associations reported in epidemiologic studies were not likely to be an artifact of model
9 specification, since analyses or reanalyses of data using different modeling strategies reported
10 similar results (ADD REF). The 2004 review provided the opportunity to evaluate these and
11 other related methodological issues.
12 A large number of studies now available in this review have provided new insights on
13 these and other issues as evaluated in chapters 2, 3, 6, 7, and 8 of the ISA. The following
14 discussion builds upon the ISA's evaluation of key methodological issues related to
15 epidemiologic studies as a basis for staff conclusions specifically regarding the use of
16 epidemiologic evidence in quantitative risk assessments, as discussed in chapters 4 and 5 of this
17 document. This section addresses a number of key methodological issues. Two key policy-
18 relevant issues related to interpreting epidemiologic evidence have been highlighted in the
19 second draft ISA and are presented first. Section 3.4.1 includes discussion of several topics
20 related to the exposure periods used in epidemiologic studies, with an emphasis on the issue of
21 lag periods. In section 3.4.2, the form of concentration-response (C-R) relationships in both
22 short-term and long-term exposure studies is discussed, as is evidence related to the potential
23 existence of population threshold levels for effects. Additional issues discussed below draw
24 primarily from discussions in the 2004 AQCD. Section 3.4.3 discusses issues related to air
25 quality data used in epidemiologic studies and section 3.4.4 discusses the potential impact of
26 measurement error and exposure error, related to the use of ambient air concentrations as
27 indicators of population exposures. Section 3.4.5 addresses statistical modeling and alternative
28 model specifications used in epidemiologic studies. Section 3.4.6 addresses the issue of potential
29 confounding by co-pollutants and effect-modification, as it relates to staff conclusions about the
30 use of specific study results in quantitative assessments.
31 3.4.1 Issues Related to Alternative Exposure Periods in Epidemiological Studies
32 3.4.1.1 Lag Structure in Short-term Exposure Studies
33 In epidemiological studies of short-term exposure to PM, many investigators have
34 examined a range of lag periods between health outcomes and PM concentrations to identify the
PRELIMINARY DRAFT-September 2009 3-73
Do Not Quote or Cite
-------
1 etiologically relevant time period for PM exposure (US EPA, 2009a section 2.4.2 ). As
2 discussed in the second draft ISA, it is important to consider the pattern of results that is seen
3 across the series of lag periods. Specifically, if there is an apparent pattern of results across the
4 different lags indicating a cumulative effect from PM exposure, then selecting the single-day lag
5 with the largest effect from a series of positive associations is likely to underestimate the overall
6 effect size, since single-day lag effect estimates do not fully capture the risk that may be
7 distributed over adjacent or other days (US EPA, 2004, p. 8-270). The prior review noted that
8 the evidence supported the use of lags of 0-1 days for cardiovascular effects and longer moving
9 averages or distributed lags for respiratory disease (USA EPA, 2004). However, the current
10 review reports there is limited consensus on the etiologically relevant lag period for PM
11 associated health effects, and recommends examining a cumulative lag structure of associations
12 between PM concentration and health outcome instead of focusing on a priori lag times.
13 Moreover, where effects are found for a series of lag periods, a distributed lag model will more
14 accurately characterize the effect estimate size (add ref). However, if there is no apparent pattern
15 or reported effects vary across lag days, any result for a single day may well be biased (US EPA,
16 2004 p. 9-42). Staff also observes that the high degree of autocorrelation in PM measurements
17 complicates the assessment of various lag periods.
18 For selecting effect estimates from studies for use in quantitative risk assessment or for
19 evaluation of potential revisions to the standards, staff considered patterns of results for PM25
20 across lag periods from U.S. and Canadian studies. Many short-term studies reviewed in the
21 second draft ISA assess effects associated with a range of lag periods, with some authors
22 reporting effect estimates for one lag period based on this evaluation. However, a number of
23 studies selected lag periods a priori. For example, for their multi-city US based study,
24 Zanobetti et al. (2009) used an average of 0-day and 1-day lagged PM2.5 in analyses of
25 associations with mortality.
26 Most authors report testing associations across a range of lag periods, and in many cases
27 the authors reported a pattern of positive associations across several lag periods. In the case of
28 cardiovascular morbidity, Figure 6-2 in the second draft ISA presents associations for PM2.5
29 levels over a series of days preceding hospitalization for ischemic heart disease and myocardial
30 infarction, and positive associations can be seen over several adjacent lag periods (US EPA,
31 2009a, section 6.2.10). However, the second draft ISA notes that most studies rarely report
32 results for alternative lag structures (US EPA, 2009a, section 2.4.2.1). In general, studies
33 evaluating the association between short-term exposure to PM and cardiovascular
34 hospitalizations and emergency department visits reported strong results for lags of 0 to 2 days,
35 with limited evidence at shorter time periods (i.e., hours). Although human clinical and
36 toxicological data provide evidence of biological plausibility for cardiovascular effects
PRELIMINARY DRAFT-September 2009 3 -74
Do Not Quote or Cite
-------
1 associated with PM levels at immediate or concurrent day lag, a recent study provides support
2 for a longer lag time (i.e., 14-day distributed lag model) associated with non-ischemic
3 cardiovascular disease (US EPA, 2009a Section 6.2.10). Moreover, cumulative PM exposures
4 over multi-day lags that are associated with the development of risk factors (e.g., atherosclerosis)
5 can potentially lead to individuals being at risk for experiencing an acute event (e.g., myocardial
6 infarction) from a comparatively short-term PM exposure over a lag of 0 to 2 days.
7 Consequently, the second draft ISA concludes "effects have been observed at a range of lag
8 periods from a few hours to several days with no clear evidence for any lag period having
9 stronger associations then another" ( US EPA, 2009a, section 2.4.2.1).
10 In section 2.4.2.2 the second draft ISA discusses the influence of lag on studies
11 evaluating the association between short-term exposure to PM and respiratory morbidity. The
12 prior review noted that researchers reported testing associations for 3 day lags and beyond and
13 reporting consistent patterns across lags for associations between asthma hospitalization and
14 PMio, PM2.5, or PMio-2.5 (US EPA, 2005 p. 3.-47). Furthermore, the prior review indicated that
15 effect estimates should be considered in the context of adjacent lags (i.e. distributed lags) instead
16 of a single day lag due to the correlated nature of daily pollutant exposures. In the current
17 review, studies of respiratory hospitalizations associated with PM exposure over multiple lags
18 report a higher magnitude of effect with distributed lag models. Longer lags (e.g. 2 day lag)
19 were strongly associated with asthma hospitalizations and emergency department visits among
20 children, but the second draft ISA notes inconsistent evidence on PM associated asthma
21 hospitalizations for shorter lags (i.e. lag 0 or 1 day) among older adults (US EPA, 2009a, section
22 6.3.8) is potentially due to the etiological complexity of asthma. However, toxicological and
23 human clinical data on PM effects on allergic sensitization and immune responses provide
24 biological plausibility for a longer lag period.
25 In assessments of the influence of lag on the association between short-term exposure to
26 PM and mortality the prior review reported that among U.S. and Canadian studies, the AQCD
27 observed that many authors report finding a pattern of PM-related effects across adjacent lags
28 (US EPA, 2004 p. 8-279). However, other studies reported inconsistent results across lags, in
29 particular for COPD mortality (US EPA, 2005 p. 3-50). The second draft ISA notes that most
30 recent studies focus on short-term exposure to PM in association with mortality (i.e., all-cause,
31 cardiovascular, and respiratory) with an a priori lag structures of either 1 or 0-1 days, with the
32 strongest PM-mortality associations observed for these lag periods (US EPA, 2009a, section
33 2.4.2.3). However, recent evidence, although limited, suggests larger effects of PM associated
34 mortality occur over a distributed lag period (ADD REF).
35 In summary, the second draft ISA concludes that it is likely that the most appropriate lag
36 period for a study will vary, depending on the health outcome and the specific pollutant under
PRELIMINARY DRAFT-September 2009 3-75
Do Not Quote or Cite
-------
1 study. Some general observations can be made about lag periods for different health outcomes.
2 For total and cardiovascular mortality, it appears that the greatest effect size is generally reported
3 for the 0-day lag and 1-day lags. For cardiovascular disease, PM levels at 0 to 2 day lag are
4 strongly associated with hospitalization, although effects were observed for a range of lag
5 periods. For respiratory symptoms, many studies report effects over a series of lags, with larger
6 effect estimates for moving average or distributed lag models. Similarly, for asthma
7 hospitalization among children, there appear to be larger effects over longer average time
8 periods. As noted in the second draft ISA, these results should be interpreted with caution as the
9 etiologically relevant temporal period for PM exposure may also be influenced by individual
10 susceptibility, which could increase or decrease the lag times associated with a specific health
11 effect (US EPA, 2009a, section 2.4.2). For quantitative assessment, staff concludes that it is
12 appropriate to use results from lag period analyses consistent with those reported in the second
13 draft ISA, focusing on shorter lag periods for cardiovascular effects and multi-day lag periods for
14 respiratory effects, depending on availability of results.
15 3.4.1.2 Seasonal and Regional Differences in Time-Series Epidemiological
16 Results
17 Epidemiological studies of health effects associated with short-term exposure to PM (e.g.,
18 time-series studies) generally use temporal or seasonal terms in regression models to control for
19 potential confounding by seasonal changes in health outcomes. In addition to adjusting PM risk
20 estimates for seasonal influence, epidemiologic studies also evaluate PM-health associations
21 stratified by different seasons, and assess seasonal effect modification of PM risk in models
22 using seasonal interaction terms for the exposure surrogate and meteorology (e.g., Dominici et
23 al., 2000). The second draft ISA reports that there can be seasonal differences in correlations
24 between PM and other pollutants (US EPA, 2009a, section 3.9.1.3), or in PM levels across
25 seasons (US EPA, 2009a section 3.8.6.4).
26 The limited evidence presented in the prior review indicated that PM2.s associated effects
27 were stronger in the winter season in southern California than in the summer season (Ostro et al.,
28 2000), but an earlier study also conducted in southern California found stronger effects only in
29 the summer months (Ostro et al., 1995). Furthermore, a study conducted in San Jose found no
30 significant differences in PM2 5 associated mortality were observed between the four seasons
31 (Fairley, 2003). In Phoenix, the association between PMio-2.5 and mortality was reported to be
32 highest in spring and summer, when PMio-2.5 concentrations were lowest (Mar et al., 2003). In
33 Seattle, associations between PM2.5 and PMio-2.5 and asthma hospitalization were positive in all
34 seasons, but effect estimates were larger in spring and fall (Sheppard et al., 2003).
PRELIMINARY DRAFT-September 2009 3 -76
Do Not Quote or Cite
-------
1 Section 3.8.6.4 of the second draft ISA discusses the influence of season and region on
2 PM2.5 risk estimates. Bell et al. (2008) recently reported higher PM2.5 risk estimates for
3 cardiovascular and respiratory morbidity in the winter and in the northeast compared to other
4 seasons in the rest of the country in a large, national multi-city study, and posits that these results
5 are potentially attributable to seasonally varying PM composition and concomitant toxicity. In
6 addition, the second draft ISA observes that the limited influence of seasonality on PM risk
7 estimates in regions other than the northeast may be due to exposure misclassification arising
8 from variability in time spent indoors vs. outdoors in different regions, and the higher prevalence
9 of infectious disease in the winter (US EPA, 2009a, section 3.8.6.4). Furthermore, increased air
10 conditioning in the warmer months is associated with a decrease in risk of cardiovascular
11 morbidity outcomes (Bell et al., 2009). In the case of mortality, Zanobetti and Schwartz (2009)
12 reported in their multi-city US based study a 4-fold higher effect estimate for PM2.5 associated
13 mortality for the spring as compared to the winter. These results suggest individuals are at
14 greater risk of dying from higher exposures to PM in the warmer months, and at greater risk of
15 PM associated hospitalization for cardiovascular and respiratory diseases during colder months
16 of the year.
17 Overall, staff observes that there are few studies presently available to deduce a general
18 pattern in PM risk across seasons and regions. However, the two aforementioned multi-city
19 studies showed a seasonal and regional influence for cardiovascular and respiratory morbidity
20 and mortality, respectively (Schwartz et al., 2000). These studies report generally positive
21 results across all seasons tested, with some reporting larger effect estimates in one or more
22 season(s). Therefore, staff concludes the available evidence provides an opportunity to conduct
23 a quantitative assessment of seasonal differences in relationships between PM and health
24 outcomes.
25 3.4.1.3 Health Effects Related to Different Short-term Exposure Time
26 Periods
27 While most time-series epidemiologic studies use 24-hour average PM measurements, a
28 few studies have used ambient PM concentrations averaged over shorter time intervals, such as
29 1- or 2-hour averages. The prior review noted that several epidemiological studies evaluating
30 associations with cardiovascular health biomarkers or physiological changes reported statistically
31 significant associations between 2- to 4-hour PMio or PM2.5 concentrations and cardiovascular
32 health endpoints, including myocardial infarction incidence and heart rate variability (US EPA,
33 2004 pp. 8-162 to 8-165). In particular, Peters et al (2001) reported effect estimates for
34 myocardial infarction incidence with PM2 5 averaged over 2- and 24 hours that were similar in
35 magnitude and statistically significant (US EPA, 2004, p. 8-165). However, two studies
PRELIMINARY DRAFT-September 2009 3 -77
Do Not Quote or Cite
-------
1 discussed in the current review that assessed sub-daily exposures to PM2.5 did not observe an
2 association between PM2.5 and risk of myocardial infarction (Sullivan et al., 2005; Peters et al.,
3 2005).
4 For respiratory health outcomes, the prior review reported on results from two panel
5 studies of symptoms in asthmatic subjects (US EPA, 2004, section 8.3.3.1.1). One study in a
6 small Southern California community, reported larger effect estimates for 1- or 8-hour
7 concentrations than for 24-hour PMi0 concentrations (Delfino et al., 1998), while the other, in
8 Los Angeles, reported larger effect estimates for 24-hour PMio concentrations (Ostro et al., 2001;
9 US EPA, 2004 p. 8-206). In the current review, a study in two New York City communities
10 assessing the association between 24-hour and 1-hour maximum PM2.5 levels and ED visits for
11 asthma reported an increase in ED visits that was similar for 24-hour and 1-hour PM2 5 (New
12 York State Department of Health, 2006). These results were robust to adjustment by co-
13 pollutants. However, the evidence on sub-daily PM levels associated with respiratory diseases is
14 limited, and the current review concludes that for f several studies of hospital admissions or
15 medical visits for respiratory diseases, the strongest associations were observed with several-day
16 average PM concentrations (US EPA, 2009a, section 6.3.7.1). Staff observes that the very
17 limited information available in the second draft ISA continues to suggest that cardiovascular
18 effects may be associated with acute exposure time periods on the order of an hour or so.
19 3.4.1.4 Exposure Periods Used in Prospective Cohort Studies
20 Prospective cohort studies use air quality measurements averaged over long periods of
21 time (e.g. one to several years) and varying spatial scales to characterize the long-term ambient
22 levels in the community. The prior review noted that exposure comparisons were basically
23 cross-sectional in nature, which does not provide evidence of a temporal relationship between
24 exposures to PM and resulting effects (US EPA, 2004 p. 9-42). Difficulty in disentangling
25 effects associated with historic exposures as compared to more recent exposures in these studies
26 can lead to potential exposure measurement error (US EPA, 2004, p. 5-118). This potential
27 misclassification of exposure is exacerbated under conditions where average PM concentrations
28 change differentially over time between areas. Therefore, long-term exposure effect estimates
29 are less likely to be influenced by exposure error for pollutants that do not exhibit high spatial
30 variability within a specified geographical area (US EPA, 2009a section 2.2.3).
31 The challenge in distinguishing between PM associated effects due to past and recent
32 exposures is also relevant to the identification of the latency period for long-term exposure to
33 PM and resultant health effects, and as the second draft ISA notes, studies assessing the
34 relationship between long-term exposure to PM and mortality have also tried to identify the
35 latency period between exposure to PM and death (US EPA, 2009, section 7.6.4). In the prior
PRELIMINARY DRAFT-September 2009 3-78
Do Not Quote or Cite
-------
1 review, EPA assessed several studies that used different air quality periods for estimating long-
2 term exposure and tested associations with mortality for the different exposure periods (US EPA,
3 2005, p. 3-55. In the first extended analysis of the ACS study, Pope et al. (2002) reported
4 associations between mortality and PM2.5 using the original air quality data (1979-1983), data
5 from the new fine particle monitoring network (1999-2000), and the average PM2.5
6 concentrations from both time periods. The authors reported that the PM2.5 concentrations for
7 the different time periods were well correlated, indicating that the ordering of the cities from low
8 to high pollution levels had changed little. When using average PM2.5 levels from all years, the
9 associations for total, cardiopulmonary and lung cancer were slightly larger in size, though not
10 significantly so, than for either individual air quality data set. In another study of the ACS that
11 was discussed in the prior review, Lipfert et al. (2000) analyzed mortality and PM data from
12 different time segments, and reported varied results, with some statistically significant negative
13 associations. Furthermore, the authors report that the strongest positive associations were found
14 with air quality data from the earliest time periods, as well as the average across all data (US
15 EPA, 2004, p. 8-115).
16 In the recent extended follow-up analysis of ACS data, Krewski et al. (2009) utilized
17 similar air quality data and assigned ACS participants to PM2.5 exposure periods as average
18 concentrations from three successive five-year periods prior to the date of death (i.e., years 1-5;
19 years, 6-10; years 11-15). The investigators reported that the differences in mortality risk
20 estimates between the 5-year exposure periods were minor, and no pattern was evident from the
21 results. The second draft ISA notes the highly correlated nature of the air quality data across the
22 3 exposure periods (as similarly noted in the aforementioned discussion of two exposure periods
23 evaluated by Pope et al., 2002) may have obscured potential differences in effect estimates (US
24 EPA, 2009a, section 7.6.4).
25 In addition to the ACS cohort, investigators have also utilized data from the Six Cities
26 study to evaluate exposure periods within the context of a prospective study. The prior review
27 discussed an analysis of the Six Cities data evaluating mortality risk with different estimates of
28 long-term PM2.5 exposure (US EPA, 2005, section 3.6.5.4). Villeneuve et al. (2002) conducted a
29 follow-up analysis to the original Six Cities study (Dockery et al., 1993) that included PM2.5 data
30 from more recent years and evaluated associations with PM2 5 averaged over a range of time
31 periods, such as 2 or 3-5 years preceding the individual's death. The authors reported effect
32 estimates for mortality that were lower with time-dependent PM2.5 exposure indicators (e.g., 2
33 years before individual's death), than with the longer-term average concentrations. The authors
34 observed that the fixed average concentration window may be more representative of cumulative
35 exposures, and, thus, a more important predictor of mortality, than a shorter time period just
36 preceding death.
PRELIMINARY DRAFT-September 2009 3-79
Do Not Quote or Cite
-------
1 In a recent analyses of the Six Cities Study, Schwartz et al. (2008) used model averaging
2 (i.e., multiple models were averaged and weighted by probability of accuracy) to assess exposure
3 periods prospectively. The exposure periods were estimated across a range of unconstrained
4 distributed lag models (i.e., same year, year prior, two years prior, etc.). In comparing lags (US
5 EPA, 2009a, Figure 7-10) the authors reported the effects of changes in exposure to PM2.5 on
6 mortality are strongest within a two year period. Similarly, a large multi-city study of the elderly
7 found that the mortality risk associated with long-term exposure to PMi0 reported cumulative
8 effects that extended over the follow-up year and 3 preceding years (Zanobetti et al., 2008). In a
9 study of two locations that experienced an abrupt decline in PM levels (i.e. Utah Steel Strike,
10 coal ban in Ireland), Roosli et al. (2005) reported that approximately 75% of health benefits were
11 observed in the first 5 years (US EPA, 2009a Table 7-9). Furthermore, Schwartz et al. (2008)
12 and Puett et al. (2008) found, in a comparison of exposure periods ranging from 1 month to 48
13 months prior to death, that exposure to PMio 24 months prior to death exhibited the strongest
14 association, and the weakest association was reported for exposure in the time period of 1 month
15 prior to death.
16 Overall, the evidence for determining the window over which the mortality effects of
17 long-term pollution exposures occur suggests a latency period of up to five years, with the
18 strongest results observed in the first two years. In addition, the evidence from one study
19 indicated that the exposure to PM exhibited a larger cumulative effect on mortality that was
20 spread over the follow-up year and 3 preceding years (US EPA, 2009a, section 2.4.2.3). For use
21 in quantitative assessments, staff concludes that it is appropriate to use results from analyses that
22 are based on averaging PM levels over longer time periods, since the recent studies indicate that
23 this provides a better indicator of long-term PM exposure. Thus, as described in chapter 4, the
24 results from the extended ACS analyses using average PM2.5 concentrations from both the
25 original and more recent time periods are used in the PM2.5 quantitative risk assessment. Staff
26 notes that this is consistent with the HEI Subcommittee's recommendation to use the results of
27 ACS cohort analyses (i.e. air quality data averaged over the full study time period), indicating
28 that this represented the best period to use in order to reduce measurement error (HEI, 2009).
PRELIMINARY DRAFT-September 2009 3 -80
Do Not Quote or Cite
-------
1 3.4.2 Concentration-Response Relationships and Potential Thresholds
2 In assessing or interpreting public health risk associated with exposure to PM, the form of
3 the concentration-response (C-R) function is a critical component. The prior review recognized
4 that there are likely to be biologic thresholds for different health effects in individuals or groups
5 of individuals with similar innate characteristics and health status. Furthermore, individual
6 thresholds would presumably vary substantially from person to person due to individual
7 differences in genetic-level susceptibility and pre-existing disease conditions (and could even
8 vary from one time to another for a given person). Thus, it would be difficult to detect a distinct
9 threshold at the population level, below which no individual would experience a given effect,
10 especially if some members of a population are unusually sensitive even down to very low
11 concentrations. Inter-individual variability in the relation between personal exposure to PM of
12 ambient origin and the PM concentration measured at a monitor may also contribute to
13 differences in observed C-R R relationships, further obscuring potential population thresholds
14 within the range of observed concentrations (US EPA, 2004, p. 9-43, 9-44).
15 The prior review indicated there was no strong evidence to support a threshold for PM
16 mortality effects (US EPA, 2009a, section 6.5.2.7). However, the prior review noted several
17 challenges needed to be addressed in determining the shape of the C-R R function, namely the
18 (1) limited range of PM levels; (2) heterogeneity of susceptibility in at-risk populations; and (3)
19 the influence of error in exposure assessment. The current review continues to present evidence
20 on the PM C-R relationship largely in the context of the associations between PM and mortality
21 and cardiovascular morbidity endpoints, respectively (US EPA, 2009a, section 2.4.3 and
22 6.2.10.10). As noted in second draft ISA, short and long-term exposure studies examining the
23 relation between PM and mortality have "consistently found no evidence for deviations from
24 linearity or a safe threshold" (US EPA, 2009a, section 6.2.10.10).
25 Among the new epidemiologic studies of short-term PM exposure are several that use
26 different modeling methods to investigate alternative forms of C-R functions and potential
27 threshold levels. In a multi-city analysis of three C-R models (log-linear, spline, threshold),
28 Daniels et al. (2004) a reported that the spline model, indicated a linear relation with no evidence
29 of a threshold for risk of death for all-causes and for cardiovascular-respiratory causes in relation
30 to PMio. However, the other cause deaths (i.e., all cause minus cardiovascular-respiratory)
31 exhibited a threshold at around 50 ug/m3 PMio (US EPA, 2009a, Figure 6-35).
32 For short-term studies of cardiovascular hospital admissions and ED visits, the limited
33 studies investigating the PM C-R relation, largely as part of the model selection process, suggest
34 a log-linear, non-threshold C-R relationship. In the last review, Schwartz and Morris (1995)
35 found no evidence for a threshold in the dose-response relationship between short-term PMio
36 exposure and hospital admissions for IHD. Recent single- and multi-city studies of hospital
PRELIMINARY DRAFT-September 2009 3 -81
Do Not Quote or Cite
-------
1 admissions and ED visits for CVD provide additional support for this finding. In an extensive
2 multi-city analysis of the nature of the C-R relationship and potential for a threshold between
3 PMio and cardiovascular hospital admissions and ED visits, Zanobetti and Schwartz (2005)
4 reported a near linear concentration-response between PMio and MI (with a steeper slope
5 occurring below 50 ug/m3) and no evidence for a threshold. Overall, this limited evidence
6 supports the use of a no-threshold, log-linear model, consistent with the observations made in
7 studies that examined the PM-mortality relationship (US EPA, 2009a, sections 2.4.3 and 6.x)
8 For long-term exposure to PM and mortality, a study by Schwartz et al (2008) using Six
9 Cities Study data (previously noted in section 3.4.x) reported that a Cox proportional hazards
10 model fit with a penalized spline yielded a concentration-response curve between PM2.5 and
11 mortality that was linear, and was "clearly continuing below the level of the current U.S. air
12 quality standard of 15 ug/m3" (US EPA, 2009a, section 7.6.4) Although Figure 6-37 of the ISA
13 suggests a no-threshold relation between PMio and mortality based on data from the APHEA
14 project (Samoli et al, 2008), the potential influence of exposure error and heterogeneity of
15 concentration-response across cities (i.e., regions) and seasons (i.e., within-year variability)
16 deserves further consideration (US EPA, 2009a, section 6.5.2.7).
17 In summary, while staff recognizes that there likely are individual biologic thresholds for
18 specific health responses; the overall evidence from existing studies do not support the existence
19 of thresholds in PM-mortality relationships at the population level, for either long-term or short-
20 term PM exposures within the range of air quality observed in the studies. While epidemiologic
21 analyses have not identified thresholds in observed associations in the range of air quality
22 concentrations in the studies, it is possible that such thresholds exist toward lower end of these
23 ranges (or below these ranges) but cannot be detected due to variability in susceptibility across a
24 population. Based on the above considerations, staff concludes, that it is appropriate to use the
25 linear or log-linear concentration-response models reported in epidemiologic studies in the
26 quantitative risk assessment. Therefore, staff has included in the quantitative risk assessment,
27 discussed in chapter 4, analyses incorporating a linear slope without an imposed cut point.
28 3.4.3 Air Quality Data in Epidemiologic Studies
29 In general, epidemiologic studies use ambient measurements obtained at central site
30 monitors to represent population exposures to PM of ambient origin. Many epidemiological
31 studies use different monitoring methods for measuring PM levels, including dichotomous
32 samplers or Harvard impactors, as well as from co-located Tapered Element Oscillating
33 Microbalances (TEOMs) or beta attenuation monitors (BAMs), and other methods (refer to
34 section 2.3 for more detailed descriptions of ambient PM measurement methods). In reviewing
PRELIMINARY DRAFT-September 2009 3 -82
Do Not Quote or Cite
-------
1 results from studies using various PM monitoring methods, staff finds that there appear to be no
2 systematic differences in the effect estimates related to the use of differing monitoring methods.
3 In consideration of the sampling frequency of PM data collection, the second draft ISA
4 notes sampling often involves collecting filter samples once every three or six days, with
5 approximately 17% of FRMs operating in 2007 scheduled to sample each day (USEPA, 2009a,
6 section 3.7). Staff observes that this missing data can have a systematic effect on the results
7 reported from epidemiologic analyses. The absence of daily monitoring data as a source of
8 uncertainty is of particular concern for short-term studies that evaluate lag structures and
9 distributed lags between ambient levels and associated health effects. Data collection frequency
10 is one component of statistical power for time-series studies, and missing data would result in
11 increased uncertainty and reduced precision in study results. The second draft ISA notes the
12 challenge in using less-than-everyday monitoring data to determine concentration-response (C-
13 R) relationships in time-series analyses (US EPA, 2009a, section 3.4.1.1). Many such studies
14 were conducted in areas where PM was monitored on a daily basis; and the availability of every -
15 day monitoring is often cited as a basis for study location in a number of reports. Panel studies
16 in particular typically assess the relation between a health outcome and PM levels on the day of
17 exposure (i.e., lag 0). However, staff observes that a small number of the recent studies have
18 been based on less frequently collected data. Multi-city studies analyzing a large number of
19 counties with less-than-every-day-monitoring data may still have sufficient power to detect a PM
20 associated health effect, but an absence of daily PM monitoring data for single-city studies can
21 limit statistical significance (US EPA, 2009a, section 3.8.6.4). Consistent with the increased
22 uncertainty in study results using less than everyday collected PM data, staff judges that greater
23 weight should be placed on those studies with daily or near-daily PM data collected with
24 preference for multi-city studies conducted in the U.S. and Canada for use in drawing
25 quantitative conclusions.
26 3.4.4 Measurement Error and Exposure Error
27 Error in the measurement of ambient PM levels is an important source of uncertainty in
28 epidemiological studies evaluating health effects associated with ambient exposures to PMio-2.s
29 or PM2.5. As noted in the prior review (US EPA, 2004 p. 8-282, 8-283), an important source of
30 exposure-related uncertainty occurs when the effect estimate for the etiologically relevant
31 pollutant decreases if measured with error and its significance is transferred to a surrogate.
32 Specifically, a transfer of association from the causal pollutant to the confounder can arise if
33 there are high levels of measurement error in the causal variable and collinearity between the two
34 variables (US EPA, 2009a, section 3.8.x). Consequently, the presence of measurement error and
35 collinearity between pollutants can result in the underestimation of the effects of the poorly
PRELIMINARY DRAFT-September 2009 3-83
Do Not Quote or Cite
-------
1 measured pollutant. However, an analysis discussed in the prior review noted that due to the
2 high underlying high ratio of fine and coarse particle toxicities is it is unlikely that error in one
3 PM measurement will result in "false negative" or "false positive" results for fine and coarse
4 particles, respectively (US EPA, 2005, p. 3-39). Consequently, PM measurement error is
5 unlikely to systematically attribute effects from one pollutant to another pollutant. Thus, while
6 the potential remains for differential error in pollutant measurements to influence the results of
7 epidemiologic studies, it is unlikely that the levels of measurement error and correlation between
8 pollutants reported in existing studies would result in transfer of apparent causality from one
9 pollutant to another.
10 However, it must be recognized that measurement error is a larger issue for PMio-2.5 than
11 for fine particles, especially when PMi0-2.5 concentrations are calculated as the difference
12 between PMio and PM2.5 measurements instead of directly measured using a dichotomous
13 sampler. It is likely that measurement error would increase the uncertainty of an epidemiologic
14 association, and with greater error in PMio-2.5 monitoring methods, any reported epidemiologic
15 associations would be less likely to reach statistical significance (US EPA, 2005, p 3-40).
16 Decreases in study precision would also occur even if gravimetric PMio-2.5 mass was measured
17 accurately, but the sources and relative composition of coarse particles are highly variable.
18 Exposure error is an issue that is closely linked with the preceding discussion of error in
19 measuring PM air quality. Concentrations measured at ambient monitoring stations are generally
20 used to represent a community's exposure to ambient PM. For time-series studies, the emphasis
21 is on the short-term temporal (usually daily) changes in ambient PM. In contrast, cohort or
22 cross-sectional studies use air quality data averaged over a period of months to years as
23 indicators of a community' s long-term exposure to ambient PM and/or other pollutants.
24 Exposure misclassification can result from utilizing ambient PM levels as a proxy for community
25 average personal exposures to ambient PM. However, the second draft ISA notes that "the use
26 of the community average ambient PM2.s concentration as a surrogate for the community average
27 personal exposure to ambient PM2.5 is not expected to change the principal conclusions from
28 time-series and most panel epidemiologic studies that use community average health and
29 pollution data. Several recent studies ... show that the non-ambient component of personal
30 exposure to PM2 5 is basically uncorrelated with ambient PM2 5 concentrations." Furthermore,
31 with respect to long-term exposure studies, the second draft ISA concludes that "[f]or long-term
32 studies that use differences in long-term community average ambient PM concentrations as an
33 exposure metric, the effect of possible community-to-community differences in the average
34 ambient exposure factor or in the average non-ambient exposure are less understood." In
35 addition, for panel studies, the second draft concludes "the most appropriate exposure metric
36 may depend on the health outcome measured. However, sufficient information should be
PRELIMINARY DRAFT-September 2009 3 -84
Do Not Quote or Cite
-------
1 obtained to enable determining the association of the health outcome with ambient concentration,
2 ambient exposure, non-ambient exposure, and total personal exposure (US EPA, 2009a, section
3 3.9.2.4).
4 As discussed in section 3.8.6 of the second draft ISA, one component of exposure error
5 (i.e., misclassification) is how evenly distributed PM is across a community, as indicated by
6 levels at different monitoring sites; another component is how well particles penetrate from
7 ambient air into indoor environments. Several factors affect how readily particles can move into
8 buildings and remain suspended in indoor air. In general, fine particles move indoors and remain
9 suspended more easily than do thoracic coarse particles. In time-series analyses, measurements
10 of PM2.5 made at a central site are found to be better correlated with indoor measurements than
11 are measurements of PMi0-2.5 (see section 2.7.2). In general, the result of studies evaluating
12 exposure misclassification related to the use of PM data from central monitoring sites suggest it
13 is likely to result in underestimation of the effect of PM exposure on health (US EPA, 2004, p. 8-
14 288).
15 However, misclassification of PMi0-2.5 exposure is of particular concern since PMi0-2.5
16 disperses over a shorter distance than PM2.5, and therefore is likely to incur greater exposure
17 error over larger spatial scales as compared to ambient PM2.s levels that remain spatially
18 homogenous over larger areas (US EPA, 2009a, section 3.8.4.1). For example, as noted in
19 section x.x above, Zanobetti and Schwartz (2009) estimated PMi0-2.5 by taking the difference in
20 county-wide averages of PMio and PM2.s, as compared to taking the difference in PMio and
21 PM2.5 levels from co-located monitors. The absence of a significant association between PMio-2.s
22 and mortality for the "dry" region in Zanobetti and Schwartz (2009), a finding observed in the
23 same area of the U.S. by (ADD REF) may be due to measurement error arising from the use of
24 county-wide average differences in PMio and PM2.5 levels. Therefore, a set of positive but
25 generally not statistically significant associations between PMio-2.s and a health outcome could
26 be reflecting a true association that is measured with error.
27 In summary, there are several key exposure-related distinctions between PM2.5 and PMio-
28 2.5. In section 3.8.6.3, the second draft ISA notes that PM2.s spatial variability is generally low at
29 urban regional levels, with frequently high site-to-site correlations among monitors; as
30 summarized above, although there can be differences in some specific locations. In contrast, the
31 second draft ISA notes that PMio-2.s spatial variability is higher than PM2.s owing to the more
32 rapid gravitational settling of larger particles, and the aforementioned low site-to-site correlations
33 between monitors. In such situations, while the epidemiologic associations may be illustrating
34 true time-series relationships between PM and a health outcome, it is more difficult to draw
35 inferences about the population exposure levels at which those effects are seen. From studies in
36 which significant associations are reported with PMio-2.5, the distribution of ambient monitoring
PRELIMINARY DRAFT-September 2009 3-85
Do Not Quote or Cite
-------
1 data available for the study may reflect levels that are higher or lower than those experienced by
2 neighborhoods in other parts of the community.
3 An additional source of exposure-related uncertainty is noted in studies of PM
4 constituents reporting inconsistent or null associations with health effects that may be, in part,
5 due to exposure misclassification, wherein the measured PM species are acting as a surrogate, or
6 another component is the causal agent. The second draft ISA discusses the results of two studies
7 that indicate the health effects associated with improved exposure data on constituents is of
8 greater significance in comparison to poorly characterized PM compositional data. Differences
9 in PM size distribution and constituents between indoor ambient PM and outdoor ambient PM
10 can also introduce differential or non-differential exposure misclassification, depending on the
11 differences in the toxicity of the unmeasured PM constituents (US EPA, 2009a, section 3.9.2.4).
12 As was recognized in the last review, there is greater uncertainty in the relationship
13 between ambient PMio-2.5 measured at central monitors and individuals' exposure to ambient
14 PMio-2.5, based on both variability in PMio-2.5 concentrations across an area and a decreased
15 ability for thoracic coarse particles to penetrate into buildings. This uncertainty is likely to
16 broaden the confidence intervals around effect estimates (US EPA, 2005, section 3.7). As noted
17 above, a national network is not currently in place to monitor PMio-2.5 concentrations. Ambient
18 concentrations of PMio-2.5 are generally determined by the subtraction of PMio and PM2.5
19 measurements, although methods to characterize PMio-2.5 levels in epidemiology studies vary.
20 As a result, uncertainties surround the concentration at which the observed associations occur.
21 Therefore, there is greater error in ambient exposure to PMio-2.5 compared to PM2.5. This would
22 tend to increase uncertainty and make it more difficult to detect effects of PMio-2.5 in
23 epidemiologic studies. In addition, the various differences between eastern and western U.S.
24 counties can lead to exposure misclassification, and the potential underestimation of effects in
25 western counties (US EPA, 2009a, section 2.3.3.2).
26 Because of the greater heterogeneity of PMio-2.5, exposure measurement error is more
27 likely to bias health effect estimates towards the null for epidemiologic of PMio-2.5 versus PMio
28 or PM2.s, making it more difficult to detect an effect of the coarse size fraction. In addition,
29 models that include both PMio-2.5 and PM2.5 may suffer from instability due to collinearity.
30 Further the lag structure of PMio-2.5 effects on risk of cardiovascular hospital admissions and ED
31 visits, as well as mortality has not been examined in detail (US EPA, 2009a, section 6.2.12.2).
32 3.4.5 Alternative Model Specifications
33 3.4.5.1 Time-series epidemiologic studies
34 The prior review addressed in great detail statistical modeling issues for epidemiologic
35 studies (USA EPA, 2005 p. 3-42 - 3-43). Of particular concern were the default convergence
PRELIMINARY DRAFT-September 2009 3 -86
Do Not Quote or Cite
-------
1 criteria and standard error calculations made using Generalized Additive Models (GAM) in time-
2 series studies (US EPA, 2004 section 8.4.2). A number of time-series studies were reanalyzed
3 using alternative methods, typically GAM with more stringent convergence criteria and
4 alternative models such as Generalized Linear Models (GLM) with natural smoothing splines.
5 The results of the reanalyses were compiled and reviewed (HEI, 2003), and generally did not
6 substantially change the findings of the original analyses. Generally, changes in effect estimates
7 with alternative analysis methods were much smaller than the variation in effects across studies.
8 Consequently, this evaluation led to the conclusion that PM-related effects observed in
9 epidemiologic studies were unlikely to be seriously biased by inadequate statistical modeling
10 (addref).
11 Alternative model specifications to control for potential weather effects and temporal
12 trends were also considered in the prior review (USA EPA, 2005, p. 3-42 - 3-43). The
13 magnitude of the effect estimate for PM can decrease with increasing control for weather and
14 temporal trend, though it generally stabilizes at some point. If the model does not adequately
15 address daily changes in weather-related variables, then some effects of temperature on health
16 would be erroneously ascribed to the pollution variable. Conversely, if the model over controls
17 for weather, then the result will be a much less efficient estimate of the pollutant effect (US EPA,
18 2004, p. 8-236). This would result in incorrectly ascribing some of the true pollution effect to
19 the temperature variable, and thereby limit the detection of a real but small pollution effect. The
20 prior review concluded there was no clear consensus as to what constitutes appropriate control
21 for weather and temporally related variables, while recognizing that no single approach is likely
22 to be most appropriate in all cases.
23 In summary, for quantitative assessment staff continues to consider it appropriate to use
24 results from short-term exposure studies that used either more stringent GAM or GLM analyses.
25 Moreover, an advantage to the use of GAM in time-series is that the model selects the degree of
26 smoothing or adjustment for weather and temporal covariates that best fits the data (US EPA,
27 2009a, section 7.x). The GLM approach is advantageous in allowing a more accurate calculation
28 of standard errors. However, a small number of PMio-2.5 studies have investigated the influence
29 of model specification on PMio-2.5 risk estimates, and further evaluation of PMio-2.5 risk estimates
30 with alternative model structures is needed to best characterize PMio-2.5 associated health effects
31 in conjunction with weather and other time-varying covariates (US EPA, 2009a, section 6.5.2.3).
32 3.4.5.2 Prospective cohort epidemiologic studies
33 The prior review discussed the HEI-sponsored reanalysis of data from the ACS and Six
34 Cities prospective cohort studies that evaluated a number of issues related to long-term exposure
35 studies. In particular, the reanalysis included a replication and validation study, and a sensitivity
PRELIMINARY DRAFT-September 2009 3 -87
Do Not Quote or Cite
-------
1 analysis, where alternative risk models and analytic approaches were used to test the robustness
2 of the original analyses. The replication study confirmed the original investigators' findings of
3 associations with both total and cardiorespiratory mortality (Krewski et al., 2000; US EPA,
4 2004,, p. 8-91), and the sensitivity analyses generally reported that the use of alternative models,
5 including variables that had not been used in the original analyses (e.g., physical activity, lung
6 function, marital status), did not alter the original findings. Data were also obtained for
7 additional city-level variables that were not available in the original data sets (e.g., population
8 change, measures of income, maximum temperature, number of hospital beds, water hardness)
9 and reanalysis investigators included these data in the models. The associations between fine
10 particles and mortality were generally unchanged in these new analyses, with the exception of
11 population change, which did somewhat reduce the size of the associations with fine particles or
12 sulfates (US EPA, 2004, p. 8-92). As discussed in section 3.3.2.2.5, Krewski et al.'s recent
13 analysis of ACS data with extended follow-up data assessed the robustness of the Cox
14 proportional hazard model results to alternative forms (i.e. log-log) compared to the linear
15 function for the PM-mortality relation, and adjusted for individual-level variables, and addressed
16 spatial patterns in the data (2009). Overall, the results of sensitivity analyses did not appreciably
17 change the previously reported association between long-term exposure to PM2.5 and mortality
18 among ACS participants.
19 3.4.6 Co-pollutant Confounding and Effect Modification
20 Confounding occurs when a health effect caused by one risk factor is attributed to another
21 variable that is correlated with causal risk factor and itself is a risk factor for the health effect.
22 Epidemiologic analyses attempt to adjust or control for these characteristics (i.e., potential
23 confounders) that differ between exposed and non-exposed individuals (US EPA, 2009a, section
24 1.5.3). A gaseous copollutant (e.g., 63, CO, 862 and NO2) meets the criteria for potential
25 confounding in PM-health associations if: (1) it is a potential risk factor for the health effect
26 under study; (2) it is correlated with PM; and (3) it does not act as an intermediate step in the
27 pathway between PM exposure and the health effect under study (US EPA, 2004 p. 8-10). Effect
28 modifiers include variables that may influence the health response to PM exposure (e.g., co-
29 pollutants, individual susceptibility, smoking or age). Confounding and effect modification are
30 important considerations in the evaluation of effects in a mixture of pollutants observed in
31 epidemiological studies, but for confounding, the emphasis is on controlling or adjusting for
32 potential confounders in estimating the independent effect of PM, while the emphasis for effect
33 modification is on identifying and assessing the level of modulation of PM associated health
34 effects (US EPA, 2004 p. 8-12). In addition to confounding and effect modification, animal
35 toxicologic evidence in the prior review (Table 7-13, US EPA, 2004) and gene-PM
PRELIMINARY DRAFT-September 2009 3-88
Do Not Quote or Cite
-------
1 epidemiological evidence in the current review (p.8-11, US EPA, 2009a) indicate that pollutant
2 mixtures or PM in conjunction with risk factors (e.g., genes) may act together, potentially having
3 additive or synergistic health effects.
4 3.4.6.1 Co-pollutant Confounding
5 Multivariate regression models are commonly used to assess and control for potential
6 confounding of PM associated health effects by gaseous copollutants (US EPA, 2009a, section
7 1.5.3). There are several statistical issues influencing results generated using multi-pollutant
8 models, including poorly fit models that force all pollutants to follow the same lag structure in
9 short-term PM exposure studies, adding correlated but non-causal variables, or omitting
10 important variables. Moreover, inclusion of pollutants in a multi-pollutant model that are highly
11 correlated with one another can lead to erroneous pollutant effect estimates, and, thus, incorrectly
12 identify a specific causal pollutant. Collinearity between pollutants may occur if the gaseous
13 pollutants and PM come from the same sources, if PM constituents are derived from gaseous
14 pollutants (e.g., sulfates from 802), or if meteorological conditions contribute to the formation of
15 both PM and gaseous pollutants (US EPA, 2009a section 3.3.2). For example, sources of fine
16 particle constituents include combustion of various fuels, gasoline or diesel engine exhaust, and
17 some industrial processes (US EPA, 2009a section 3.3.3); these sources also emit gaseous
18 pollutants. In addition, 862 and PM2.s are often derived from the same sources in an area (e.g.,
19 coal-fired power plants) and thus simultaneous inclusion in multivariate regression models may
20 result in instability in the results, including an under- or over-estimation of effects for one or
21 both pollutants.
22 Multi-pollutant model results for the risk of hospital admissions and ED visits for
23 cardiovascular and respiratory diseases, respectively, with PM2.5,PMi0-2.5, and gaseous co-
24 pollutants are presented in Figures 6-5 and 6-15 of the second draft ISA. Overall, for cardiac
25 hospitalizations the addition of SC>2 and Os had little influence on PM associations, although
26 substantial reduction in associations with PM2.s could be seen in some cases when NC>2 and CO
27 are added to the model (Tolbert et al., 2007). Few studies provide evidence on the effects of one
28 PM size fraction in models adjusting for the associated effect of another PM size fraction.
29 Specifically, the MCAPS study reported the effect estimate for PMio-2.s associated with
30 cardiovascular hospital admissions lost precision when a covariate for PM2.5 was included in the
31 same model (Peng et al., 2008). Studies evaluating PMi0 associated health effects using multi-
32 pollutant models observe inconsistent results after controlling for gaseous pollutants or other size
33 fractions, which the second draft ISA posits is "likely due to differences in the correlation
34 structure among pollutants as well as differing degrees of exposure measurement error" (US
35 EPA, 2009a, section 6.2.10.9).
PRELIMINARY DRAFT-September 2009 3-89
Do Not Quote or Cite
-------
1 For respiratory hospital admissions, several Canadian studies report greater effect
2 estimates for PMio-2.s compared to PM2.5 that are robust to adjustment for gaseous pollutants (US
3 EPA, 2009a, section 6.3.8.5. Chen et al. (2004) observed PM2.5 and PMi0-2.5 effects that are
4 slightly attenuated but remain positive with adjustment for NO2, whereas other studies report
5 inconsistent results with adjustment for gaseous co-pollutants (Ito et al., 2003; Moolgavkar et al.,
6 2003; Delfino et al., 1998). In multi-city analysis, the MCAPS study reported robust effects for
7 PM2.s in models adjusted for PMi0.2.5, but PMi0.2.5 effect estimates were attenuated for
8 respiratory admissions with adjustment for PM2 5 (Peng et al., 2008). PMio effect estimates for
9 respiratory diseases were robust to adjustment for gaseous co-pollutants in several recent studies
10 (Tolbert, 2007; Ulirsch et al., 2007; Anderson and Bogdan, 2007).
11 As noted in section 3.3.2.1.4, a key limitation of the multi-city U.S.-based studies is the
12 absence of evidence on gaseous co-pollutants potentially influencing the association between
13 short-term exposure to PM2 5 and mortality. However, as discussed in the prior review, multi-
14 pollutant modeling was used in the NMMAPS mortality analyses for 20 and 90 U.S. cities, in
15 which the authors added first O3, then O3 and another co-pollutant (e.g., CO, NO2 or SO2) to the
16 models (US EPA, 2004, p. 8-35). The relationship between PMio and mortality was similar in
17 models including control for 63 and other gaseous pollutants as compared to single-pollutant
18 models (US EPA, 2004, p. 8-35). The authors concluded that the PMio-mortality relationship
19 was not confounded by co-pollutant concentrations across 90 U.S. cities (Samet et al., 2000a, b;
20 Dominici, 2003).
21 In the long-term exposure studies, multi-pollutant models were tested in the reanalysis of
22 data from the ACS cohort. The reanalysis indicated associations between mortality and PM2 5 or
23 sulfates were reduced in size in co-pollutant models including SO2 but not with the other gaseous
24 pollutants. Since SO2 is a precursor for fine particle sulfates, it is inherently difficult to
25 distinguish effects from the precursor SO2 and fine particles (US EPA, 2004, p. 9-37). In the
26 current review, a case-crossover multi-city study matched PMio levels to control days that had
27 gaseous pollutant levels for SO2, NO2, Os, and CO, respectively and estimated the excess risk
28 for the average of 0- and 1-day lag PMio (Schwartz et al., 2004). Since the authors only
29 presented PMio risk estimates matched by gaseous pollutants, the influence of gaseous pollutants
30 on PMio risk estimates is unclear. However, the study results suggest that PMio is significantly
31 associated with all-cause mortality with adjustment for each of the gaseous co-pollutants
32 included in the analysis (US EPA, 2009a, section 6.X).
33 As noted in the prior review, exposure studies collected personal and ambient monitoring
34 data to evaluate whether co-pollutants can confound PM exposure assessments (US EPA, 2005,
35 p. 3-45). Investigators assessed the degree of day-to-day correlation between the different
36 measures of personal and ambient concentrations and reported that the personal and ambient
PRELIMINARY DRAFT-September 2009 3 -90
Do Not Quote or Cite
-------
1 PM2.5 measurements were correlated, as were personal exposure to PM2.5 and ambient
2 concentrations of the gaseous pollutants, suggesting that some gases may serve as a proxy for
3 exposures to other air pollutants (US EPA, 2009a, section 3.8.X). Similarly, a recent study
4 observed personal exposures to ambient PM2.5 were significantly associated with ambient
5 concentrations of PM2 5, NO2, and Os (inversely), and personal exposures to SC>4 were
6 significantly associated with ambient PM2.5 and 63 levels (Schwartz et al., 2007). In contrast,
7 Sarnat et al. (2001) reported personal and ambient concentrations of gaseous pollutants are not
8 well correlated, potentially due to levels observed below detection limits of personal monitors.
9 These findings suggest that associations reported with ambient PM2.5 are truly reflecting
10 associations with fine particles and that fine particles are unlikely to be simply acting as
11 surrogates for other gaseous pollutants (Sarnat et al., 2000, 2001; US EPA, 2004, p. 5-90,).
12 "There is evidence that associations between ambient gases and personal exposure to
13 PM2.s of ambient origin exist but are complex and vary by season and region" (US EPA, 2009a,
14 section 3.8.x).
15 In summary, where various indicators of PM and the other pollutants are correlated, it can
16 be difficult to distinguish effects of the various pollutants in multi-pollutant models. However, a
17 number of research groups have found the effects of various indicators of PM and gases to be
18 independent of one another, as illustrated in Figures 6-5 and 6-15 of the second draft ISA. In
19 addition, evidence on potential co-pollutant confounders of PM exposure assessments suggests
20 that it is unlikely that ambient PM2.5 exposure associated health effects are in actuality,
21 representing relationships with other pollutants.
22 Taking into consideration the findings of single- and multi-city studies and other
23 evaluations of potential confounding by gaseous co-pollutants described in preceding sections,
24 the ISA concludes that while research questions remain, in general, "associations for various PM
25 indices with mortality or morbidity are robust to confounding by co-pollutants." (US EPA,
26 2009a, section x.x). This indicates that effect estimates from single-pollutant models can be used
27 to represent the magnitude of a concentration-response relationship, though there will remain
28 uncertainty with regard to potential contributions from other pollutants. For quantitative
29 assessment, staff concludes that single-pollutant model results provide reasonable indicators of
30 the magnitude of PM-related effects, supported by analyses including multi-pollutant model
31 results, as available.
32 3.4.6.2 Effect Modification
33 As noted in the prior review, one approach to evaluating the potential modulating effect
34 of co-pollutants on associations reported with PM2.5 involves observing a consistent association
35 in a variety of locations with differing levels of co-pollutants (US EPA, 2005, pg 3-46,). Effect
PRELIMINARY DRAFT-September 2009 3 -91
Do Not Quote or Cite
-------
1 estimates for PMio-mortality associations plotted against concentrations of gaseous pollutants in
2 the prior review yielded no evidence that associations reported between PMio and mortality were
3 correlated with co-pollutant concentrations (US EPA, 2005 Figure 3-3). Specifically, the
4 magnitude and statistical significance of the associations reported between PM2.5 and mortality in
5 these studies did not exhibit trends with the levels of any of the four gaseous co-pollutants.
6 While not definitive, these consistent patterns indicate that it is more likely that the effect of
7 PM2.s is not appreciably modified by differing levels of the gaseous pollutants.
8 The current review examines several studies that assessed whether differences in effect
9 estimates between PM2.5 and morbidity or PMio and mortality are due to variations in PM2.5
10 species (US EPA, 2009a, section 6.6.x). One study of short-term PM2.s exposure associated
11 with morbidity effects reported statistically significant associations between the county average
12 concentrations of V, Ni, and EC (106 counties) and effect estimates for both cardiovascular and
13 respiratory hospital admissions (Bell et al., 2009). In the same study, an analysis for PMio
14 associated mortality risk estimates indicated Ni was the only PM constituent that increased the
15 risk estimate, although this result was highly influenced by New York City data. Similarly, other
16 analysis of PMio associated mortality risk observed Ni and V modified PMio mortality risk
17 estimates (Lippmann et al. 2006; Dominici et al. 2007) but were also influenced by New York
18 City data (Dominici et al. 2007).
19 In a secondary analysis of PM2 5 associated mortality, investigators found Al, As, Ni, Si,
20 and SC>4 were significant effect modifiers of PM2 5 risk estimates (Franklin et al. 2008, p. 6-277,
21 US EPA, 2009a). In Lippman et al (2006) and Franklin et al (2008) PMio and PM2.5 associated
22 mortality, respectively, were associated with 14 similar species (Figure 6-31 and Table 6-14,
23 US EPA, 2009a), and both studies reporting that Ni explained the heterogeneity in PM risk
24 estimates. In addition to Ni, sulfate, SCMs noted as a significant predictor of PM2.s risk
25 estimates (Franklin et al, 2008). However, as noted in the ISA there are methodological
26 limitations in analyses of PM chemical species as effect modifiers (pg 6-279, US EPA, 2009a).
27 Specifically, PM constituents acting as effect modifiers may be confounded by demographic
28 factors that vary across cities. Therefore, the ISA posits that direct associations between PM
29 chemical components and health effects can provide substantive evidence of effect modification
30 of exposure to PM and associated health effects by PM constituents (add ref).
31 3.5 PM-RELATED IMPACTS ON PUBLIC HEALTH
32 The following discussion draws from section 8.1 from the ISA to characterize
33 populations potentially at-risk for PM-related effects and the potential public health impacts
34 associated with exposure to ambient PM.
PRELIMINARY DRAFT-September 2009 3 -92
Do Not Quote or Cite
-------
1 3.5.1 Potentially At-Risk Populations
2 The ISA summarizes information on potentially susceptible or vulnerable groups in
3 section 8.1. Interindividual variation in human responses to air pollutants indicates that some
4 population groups are at increased risk for the detrimental effects of ambient exposure to PM.
5 The NAAQS are intended to provide an adequate margin of safety for sensitive populations, or
6 those groups potentially at increased risk for health effects in response to ambient air pollution.
7 To facilitate the identification of populations at the greatest risk for PM-related health effects,
8 studies have evaluated factors that contribute to the susceptibility or vulnerability of an
9 individual to PM. The terms susceptibility and vulnerability have been used to characterize
10 populations with a greater likelihood of an adverse outcome given a specific exposure in
11 comparison with the general population. This increased likelihood of response to PM can result
12 from a multitude of factors, including genetic or developmental factors, lifestages (i.e., childhood
13 or old age) gender differences, or preexisting disease states. In addition, new attention has been
14 paid to the concept of some population groups having increased responses to pollution-related
15 effects due to factors including socioeconomic status (SES) (e.g., reduced access to health care)
16 or particularly elevated exposure levels (such as living near a roadway or pollution source). It
17 should be noted that in some cases the factors that underlie susceptibility or vulnerability of a
18 population group to PM may intersect or overlap. For example, a population group that is
19 characterized as having low SES may have less access to healthcare resulting in the
20 manifestation of a disease, and they may also reside or work in locations that result in exposure
21 to higher concentrations of PM.
22 The terms susceptibility and vulnerability have sometimes been used interchangeably in
23 the literature, and as described above, have also been defined to represent two different
24 categories that could contribute to a population group experiencing increased risk to PM-related
25 health effects, resulting in the lack of a clear and consistent definition (US EPA, 2009a, Table 8-
26 1). The term "at-risk" has also been used to encompass these concepts more generally. This
27 policy assessment document uses the term "at-risk" to represent population groups that have a
28 greater likelihood of experiencing health effects related to PM exposure, since this term
29 implicitly encompasses both categories that contribute to increased risk.
30 To examine whether PM differentially affects certain population groups, epidemiologic
31 studies conduct stratified analyses to identify the presence or absence of effect modification. A
32 thorough evaluation of potential effect modifiers may help identify populations that are at greater
33 risk from exposure to PM. These analyses require the proper identification of confounders and
34 their subsequent adjustment in statistical models, which helps separate a spurious association
35 from a true causal association. Although the design of toxicological and controlled human
36 exposure studies does not allow for an extensive examination of effect modifiers, the use of
PRELIMINARY DRAFT-September 2009 3-93
Do Not Quote or Cite
-------
1 animal models of disease and the study of individuals with underlying disease or genetic
2 polymorphisms do allow for comparisons between groups and an elucidation of the underlying
3 mechanisms of susceptibility. Therefore, the results from these studies, combined with those
4 results obtained through stratified analyses in epidemiologic studies, contribute to the overall
5 weight of evidence for the increased susceptibility of specific population groups to PM.
6 3.5.1.1 Older Adults
7 Evidence for PM-related health effects in the older adult life stage spans epidemiologic,
8 controlled human exposure, and toxicological studies. The 2004 PM AQCD found evidence for
9 increased risk of cardiovascular effects in older adults exposed to PM. Older adulthood
10 represents a life stage that is potentially at greater risk due to the higher prevalence of pre-
11 existing cardiovascular and respiratory diseases found in this age range compared to younger age
12 groups, primarily due to the gradual decline in physiological processes as part of the aging
13 process. Therefore, some overlap exists between the life stage of older adults and the population
14 group that includes people with pre-existing diseases.
15 In recent publications, the epidemiologic evidence for cardiovascular morbidity effects in
16 older adults in response to short-term exposure to PM10-2.5 and PM2.5 is limited, but taken
17 together with evidence from studies of PM10 (i.e., Le Terte et al., 2002; Larrieu et al., 2007; and
18 Lanki et al., 2006), supports the increased risk of cardiovascular morbidity in older adults.
19 However, a clear pattern of positive associations only being observed in epidemiologic studies
20 conducted in non-US locations brings into question the influence of PM composition on health
21 effects. Although studies have not consistently found an association between short-term
22 exposure to PM and respiratory-related health effects in older adults, some studies have reported
23 an increase in respiratory hospital admissions in individuals 65 years of age and older (Fung,
24 2005; Andersen, 2007).
25 Recent epidemiologic studies have also found that individuals greater than 65 years old
26 are at greater risk of all-cause (non-accidental) mortality upon short-term exposure to both PM2.5
27 (Franklin, 2007; and Ostro, 2006) and PMi0 (Zeka, 2006; and Samoli et al. 2008), which is
28 consistent with the findings of the 2004 PM AQCD. Epidemiologic studies that examined the
29 association between mortality and long-term exposure to PM (i.e., PM2.5) have found results
30 contradictory to those obtained in the short-term exposure studies. Villeneuve et al. (2002),
31 Naess et al. (2007b), and Zeger et al. (2008) report evidence of differing PM2 5 relative risks by
32 age, where risk declines with increasing age starting at age 60 until there is no evidence of an
33 association among people >85 years of age.
34 Additional evidence for an increase in cardiovascular and respiratory effects among older
35 adults has been observed in controlled human exposure and dosimetric studies. Devlin et al.
PRELIMINARY DRAFT-September 2009 3 -94
Do Not Quote or Cite
-------
1 (2003) found that older subjects exposed to PM2.5 CAPs experienced significant decreases in
2 HRV (both in time and frequency) immediately following exposure, when they compared their
3 results to a previous study which used healthy young subjects (Ohio et al., 2000).
4 In addition, Gong et al. (2004) reported that older subjects demonstrated significant
5 decreases in HRV when exposed to PM2.5 CAPs, but this study did not compare the response in
6 older subjects to those elicited by young, healthy individuals. However, the study did find that
7 healthy older adults were more susceptibe to decreases in HRV compared to those with an
8 underlying health condition (i.e., COPD) (Gong et al. 2004). Dosimetric studies have shown a
9 depression of fine and coarse PM clearance in all regions of the respiratory tract with increasing
10 age beyond young adulthood in humans and laboratory animals. These results suggest that older
11 adults are also at greater risk of PM-related respiratory health effects (Section 4.3.4.1).
12 Animal toxicological studies have attempted to characterize the relationship between age
13 and PM-related health effects through the development of models that mimic the physiologic
14 conditions associated with older individuals. For example, Nadziejko et al. (Nadziejko, 2004}
15 observed arrhythmias in older, but not younger, rats exposed to PM2.5 CAPs. In addition, other
16 studies (Tankersley, 2004, 2007 and 2008) that used a mouse model of terminal senescence
17 observed various cardiovascular-related responses. Overall, these studies provide biological
18 plausibility for the increase in cardiovascular effects in older adults observed in the controlled
19 human exposure and epidemiologic studies.
20 The evidence from epidemiologic, controlled human exposure, and toxicological studies
21 provide coherence and biological plausbility for the association between short-term exposure to
22 PM and cardiovascular morbidity in older adults. As noted above, a clear pattern of positive
23 associations only being observed in epidemiologic studies conducted in non-U.S. locations
24 brings into question the influence of PM composition on health effects. The additional evidence
25 from epidemiologic studies that focus on mortality and respiratory morbidity in response to
26 short-term exposure to PM also indicate that older adults represent an at-risk life stage of the
27 population.
28 3.5.1.2 Children
29 Childhood represents a life stage that has generally been considered at greater risk from
30 exposure to PM due to multiple factors. The factors include the following: children spend more
31 time spent outdoors; children have greater activity levels than adults; children have exposures
32 resulting in higher doses per body weight and lung surface area; and also because of the potential
33 for irreversible effects on the developing lung (2004 PM AQCD). The 2004 PM AQCD found
34 that studies which stratify results by age typically report associations between PM and
35 respiratory-related health effects in children, specifically asthma. Of the recent epidemiologic
PRELIMINARY DRAFT-September 2009 3-95
Do Not Quote or Cite
-------
1 studies evaluated, only a few have examined the association between PMi0-2.5 and PM2.5 and
2 respiratory effects in children. Mar et al. (2004) found increased respiratory effects (e.g.,
3 wheeze, cough, lower respiratory symptoms) in children 7-12 years of age compared to
4 individuals 20-51 years of age in response to exposure to both PMio-2.5 and PM2.5 in Spokane,
5 Washington. In addition, Host et al. (2007) found an increase in respiratory-related hospital
6 admissions with short-term exposure to PMio-2.5 among children ages 0-14 years in 6 French
7 cities. Further support for these effects is provided by the results from studies that focused on
8 PM10 (Peel et al., 2005; and Mar et al., 2004). Recent toxicological studies (Mauad et al., 2008;
9 and Pinkerton et al., 2008) provide biological plausibility for the increase in PM-related
10 respiratory effects in children observed in the epidemiologic studies. Collectively, the evidence
11 from epidemiologic studies that have examined the health effects associated with all size
12 fractions of PM and toxicological studies that have examined individual PM components provide
13 additional support to the hypothesis that children are at greater risk of respiratory effects from
14 short-term exposure to PM.
15 Recent studies of long-term exposure to PM2.s have expanded the evidence for effects on
16 lung development in children including extended follow-up for the Southern California
17 Children's Health Study including evidence that respiratory morbidity effects from exposure to
18 PM2.5 persist into early adulthood, and are more robust and larger in magnitude than reported in
19 the prior review. New longitudinal studies following effects on pulmonary function over time in
20 other locations using different methods provide enhanced evidence for effects related to PM
21 exposures, indicating that children are at greater risk of potentially permanent respiratory effects
22 from long-term exposure to PM.
23 3.5.1.3 Pregnancy and Developmental Effects
24 Pregnant women represent an at-risk population group due to the potential effect of
25 environmental contaminants on the developing fetus. While the majority of the literature focuses
26 on epidemiologic studies that examine the potential health effects (e.g., low birth weight, growth
27 restriction) attributed to in utero exposure to PM (see Section 7.4), it is unclear if the health
28 effects observed are due to soluble fractions of PM that cross the placenta or physiologic
29 alterations in the pregnant woman. In the case of exposure to PM, adverse health effects in the
30 offspring could be mediated by a health response in the pregnant woman.
31 3.5.1.4 Gender
32 The 2004 PM AQCD did not find consistent evidence of a difference in health effects by
33 gender. However, there appeared to be gender differences in the localization of particles when
34 deposited in the respiratory tract and the deposition rate due to differences in body size,
35 conductive airway size, and ventilatory parameters (U.S. EPA, 2004). For example, females
PRELIMINARY DRAFT-September 2009 3 -96
Do Not Quote or Cite
-------
1 have proportionally smaller airways and slightly greater airway reactivity than males (Yunginger
2 etal. 1992).
3 Few recent epidemiologic studies have conducted gender-stratified analyses when
4 examining the association between either short- or long-term exposure to PMio-2.5 or PM2.5.
5 Similar to the studies evaluated in the 2004 PM AQCD, the current literature has not found a
6 consistent pattern of associations by gender for any health outcome. The majority of the
7 epidemiologic studies that examined the association between exposure to PM and gender focused
8 on exposure to PMio. Although most of these studies do not attribute the association to specific
9 size fractions (i.e., PMio-2.5 or PM2.5) or provide insight as to whether one size fraction may be
10 driving the observed effect, the studies of PMio provide further support that gender does not
11 appear to differentially affect PM-related health outcomes. Overall, the evidence from studies
12 that examined the association between short- and long-term exposure to PMio-2.5 and PM2.5,
13 along with the supporting evidence from PMio studies, further confirms that although differences
14 in dosimetry exist between males and females, neither consistently exhibit a higher disposition
15 for PM-related health effects.
16 3.5.1.5 Race/Ethnicity
17 The 2004 PM AQCD did not evaluate the potential susceptibility and/or vulnerability of
18 individuals of different races and ethnicities. The results from epidemiologic studies evaluated
19 in this review that examined the potential effect modification of the PM-morbidity and -mortality
20 relationships by race and ethnicity varied depending on the study location. In an analysis of the
21 PM2.5-mortality relationship, Ostro et al. (2006) stratified the association by race and ethnicity,
22 and observed a positive and significant effect for whites and Hispanics, but not for blacks, in
23 response to short-term exposure to PM2 5 ;n 9 California counties- An additional analysis performed by
24 Ostro et al. (2006) in 6 California counties using PM2.5 and various PM2.5 components, also
25 found a significant association between mortality, specifically cardiovascular mortality, and
26 Hispanic ethnicity (Ostro, 2008). Additional evidence for the potential susceptibility and
27 vulnerability of individuals by race and ethnicity were derived from studies on the health effects
28 associated with short-term exposure to PMio. Wellenius et al., (2006) observed that race (i.e.,
29 white vs. other) did not significantly modify the association between short-term exposure to
30 PMio and CHF hospital admissions. Additionally, Zeka et al., (2006) did not observe any
31 difference in mortality effect estimates when stratifying by race (i.e., black and white) upon
32 short-term exposure to PMio. Overall, the results from the studies that examined the potential
33 effect modification of PM risk estimates by race and ethnicity provide some evidence for
34 increased risk of mortality in Hispanics upon short-term exposure to PM2 5. However, it should
PRELIMINARY DRAFT-September 2009 3 -97
Do Not Quote or Cite
-------
1 be noted that both of these studies were conducted in California and additional studies in other
2 locations with race and ethnicity stratified analysis have not yet been conducted.
3 3.5.1.6 Gene-Environment Interaction
4 A consensus now exists that gene-environment interactions merit serious consideration
5 during the examination of the relationship between ambient exposures to air pollutants and the
6 development of health effects (Kauffmann, 2004; and Gilliland, 1999). These potential
7 interactions were not evaluated in the 2004 PM AQCD. Inter-individual variation in human
8 responses to air pollutants suggests that some population groups are at increased risk of
9 detrimental effects due to pollutant exposure, and it has become clear that the genetic makeup of
10 an individual can increase their susceptibility (Kleeberger, 2005). Gene-environment
11 interactions can result in health effects due to: genetic polymorphisms, which result in the lack of
12 a protein or a change that makes a functionally important protein dysfunctional; or genetic
13 damage in response to an exposure which potentially leads to a health response. In this review,
14 the majority of studies examine gene-environment interactions due to genetic polymorphisms. In
15 order to establish useful links between polymorphisms in candidate genes and adverse health
16 effects, several criteria must be satisfied: the product of the candidate gene must be significantly
17 involved in the pathogenesis of the adverse effect of interest; and polymorphisms in the gene
18 must produce a functional change in either the protein product or in the level of expression of the
19 protein (US EPA, 2008). Further, the issue of confounding by other environmental exposures
20 must be carefully considered.
21 It has been hypothesized that the cardiovascular and respiratory health effects that occur
22 in response to short-term exposure to PM are mediated by oxidative stress (US EPA, 2009a,
23 section 4.3). Research has examined this hypothesis by primarily focusing on the GST genes
24 because they have common, functionally important polymorphic alleles that significantly affect
25 antioxidant defense function in the lung. Exposure to radicals and oxidants in air pollution leads
26 to a cascade of events, which can result in a reduction in glutathione (GSH), and an increase in
27 the transcription of GSTs. Individuals with genotypes that result in reduced or absent enzymatic
28 activity are likely to have reduced antioxidant defenses and potentially increased susceptibility to
29 inhaled oxidants and radicals. However, in some cases genetic polymorphisms may actually
30 reduce an individual's susceptibility to PM-related health effects. Overall, the evidence from
31 epidemiologic and controlled human exposure studies suggests that specific genetic
32 polymorphisms can increase the susceptibility of an individual to PM exposure, but protective
33 polymorphisms also exist, which may diminish the health effects attributed to PM exposure in
34 some individuals.
PRELIMINARY DRAFT-September 2009 3-98
Do Not Quote or Cite
-------
1 3.5.1.7 Pre-Existing Disease
2 In 2004, the National Research Council (NRC) published a report that emphasized the
3 need to evaluate the effect of air pollution on at-risk population groups, including those with
4 respiratory illnesses and cardiovascular diseases (NRC, 2004). The 2004 PM AQCD included
5 epidemiologic evidence suggesting that individuals with pre-existing heart and lung diseases, as
6 well as diabetes may be at greater risk from exposure to PM. In addition, toxicological studies
7 that used animal models of cardiopulmonary diseases and heightened allergic sensitivity also
8 found evidence of increased risk. More recent epidemiologic and human clinical studies have
9 directly examined the effect of PM on individuals with pre-existing diseases and toxicological
10 studies have employed disease models to identify whether exposure to PM disproportionately
11 affects certain population groups.
12 3.5.1.7.1 Cardiovascular Diseases
13 The potential effect of underlying cardiovascular diseases on PM-related health responses
14 has been examined using epidemiologic studies that stratify effect estimates by underlying
15 conditions or secondary diagnoses, and toxicological studies that use animal models to mimic the
16 physiological conditions associated with various cardiovascular diseases (e.g., MI, ischemia, and
17 atherosclerosis). A limited number of controlled human exposure studies have also examined the
18 potential relationship between cardiovascular diseases and exposure to PM in individuals with
19 underlying cardiovascular conditions, but these studies have provided somewhat inconsistent
20 evidence for these associations.
21 The majority of the epidemiologic literature that examined associations between short-
22 term exposure to PM and cardiovascular outcomes focuses on cardiovascular-related hospital
23 admissions (HA) and emergency department (ED) visits. Hypertension is the pre-existing
24 condition that has been considered to the greatest extent when examining the association
25 between short-term exposure to PM and cardiovascular-related HAs and ED visits. The results
26 have been mixed, with several studies finding that hypertension did not modify the association
27 between IHD ED visits and PM2.5 in individuals with secondary hypertension in Utah (Pope et
28 al., 2006), or the association between PMi0 and cardiovascular-related health outcomes in 7 U.S.
29 cities (Wellenius et al., 2006) and Taipei (Lee et al., 2008). These results differ from those
30 presented by Peel et al. (2007), in Atlanta, which observed that exposure to PMio resulted in an
31 increase in ED visits for arrhythmias and CHF in individuals with underlying hypertension, and
32 Park et al. (2005) in Boston found that underlying hypertension increased associations between
33 HRV, specifically a reduction in the HF parameter, and short-term exposure to PM2.5.
34 Park et al. (2005), in the analysis mentioned above, examined other underlying
35 cardiovascular conditions and found associations between PM2.5 and HRV in individuals with
PRELIMINARY DRAFT-September 2009 3-99
Do Not Quote or Cite
-------
1 pre-existing IHD. In a toxicological study, Wellenius et al. (2003) examined the effects of PM2.5
2 CAPs exposure on induced myocardial ischemia in dogs, which mimics the effects associated
3 with IHD. The authors found that exposure to PM2 5 prior to the induced ischemia increased
4 ST-segment elevation, indicating greater ischemia than air-exposed animals. A follow-up study
5 implicated impaired myocardial blood flow in the response (Bartoli et al., 2009).
6 Additional studies examined the effects of PM on cardiac function in individuals with
7 dysrhythmia with mixed results. One study observed some evidence for an increase in ED visits
8 for IHD for individuals with secondary dysrhythmia and PMio exposure (Peel et al., 2007), while
9 another study found no evidence for effect modification of PMio exposure in individuals with
10 secondary dysrhythmia when examining CHF hospital admissions in 7 U.S. cities (Wellenius et
11 al., 2006).
12 Limited evidence is available from epidemiologic studies that examined other pre-
13 existing cardiovascular conditions, such as CHF and MI. Pope et al. (2006) observed an increase
14 in hospital admissions for acute IHD in individuals with underlying CHF upon short-term
15 exposure to PM2.5. However, Peel et al. (2007) did not find that underlying CHF contributed to
16 an increase in the association between IHD ED visits and short-term exposure to PMio, and in
17 another study underlying CHF was not found to increase MI hospital admissions for exposure to
18 PMio in the cohort of more than 300,000 hospital admissions (Zanobetti and Schwartz 2005).
19 Wellenius et al. (2006) examined the effect of previous diagnoses of acute MI, and found
20 no evidence of effect modification in the association between CHF hospital admissions and
21 short-term exposure to PMio in 7 U.S. cities. Toxiciological studies have provided some
22 additional evidence for the cardiovascular health effects associated with exposure to PM in
23 individuals with animal models of underlying MI and atherosclerosis, but the evidence is
24 inconclusive. Controlled human exposure studies that examined the effect of pre-existing
25 diseases on cardiovascular outcomes with exposure to PM are less consistent and difficult to
26 interpret in the context of the results from the epidemiologic and toxicological studies.
27 Although the epidemiologic studies did not examine potential effect modification of pre-
28 existing cardiovascular conditions on effects of long-term exposure to PM, a few toxicological
29 studies exposed animals with underlying cardiovascular conditions to PM for months. In studies
30 that focused on the cardiovascular effects following subchronic exposure to PM in ApoE"7" mice,
31 relatively consistent physiological effects were observed across studies. Mice exposed to
32 ultrafme and to PM2 5 CAPs demostrated enhanced size of early atherosclerotic lesions (ultrafme
33 CAPS: Araujo et al., 2008; PM2.5 CAPs: Chen and Nadziejko, 2005; Sun et al., 2005; Sun et al.,
34 2008). An additional long-term exposure study observed a decreasing trend in heart rate,
35 physical activity, and temperature along with biphasic responses in HRV (SDNN and rMSSD)
36 upon exposure to CAPs (Chen and Hwang, 2005).
PRELIMINARYDRAFT-September 2009 3-100
Do Not Quote or Cite
-------
1 While the majority of the literature examines the potential modification of the association
2 between PM and non-fatal cardiovascular health effects, a few new studies have also examined
3 effect modification in mortality associations. Zeka et al. (2006) found an increase in risk
4 estimates for associations between PMio and mortality in individuals with underlying stroke,
5 while Bateson et al. (2004) found evidence for effect modification of the PM-mortality
6 association in individuals with CHF.
7 Collectively, the evidence from epidemiologic and lexicological, and to a lesser extent,
8 controlled human exposure studies indicates increased susceptibility of individuals with
9 underlying cardiovascular diseases to PM exposure. Although the evidence for some outcomes
10 was inconsistent across epidemiologic and toxicological studies, this could be due to a variety of
11 issues including the PM size fraction used in the study along with the study location.
12 3.5.1.7.2 Respiratory Illnesses
13 Investigators have examined the effect of pre-existing respiratory illnesses on multiple
14 health outcomes (e.g., mortality, asthma symptoms, CHF) in response to exposure to ambient
15 levels of PM. Animal models have been developed and human clinical studies conducted to
16 examine the possible PM effects on pre-existing respiratory conditions in a controlled setting.
17 Epidemiologic studies have examined the effect of short-term exposure to PM on the
18 respiratory health of asthmatic individuals measuring a variety of respiratory outcomes.
19 Asthmatic individuals were found to have an increase in medication use (Rabinovitch et al.,
20 2006), respiratory symptoms (i.e., asthma symptoms, cough, shortness of breath, and chest
21 tightness (Gent et al., 2003), and asthma symptoms (Delfino et al., 2002; 2003) with short-term
22 exposure to PM2.5; and morning symptoms (Mortimer et al., 2002) and asthma attacks
23 (Desqueyroux et al., 2002) with short-term exposure to PMio.
24 Toxicological studies that have used ovalbumin-induced allergic airway disease models
25 provide evidence which supports the findings of the epidemiologic literature. These findings
26 suggest that individuals with allergic airways conditions are at greater risk of allergic airways
27 responses upon exposure to PM2.5, which may be partially attributed to increased pulmonary
28 deposition and localization of particles in the respiratory tract (Morishita et al., 2004). An
29 additional study (Heidenfelder et al., 2009) examined whether genes are differentially expressed
30 upon exposure to PM. They found that exposure to CAPs increased the expression of genes
31 associated with inflammation and airway remodeling in rats with allergic airway disease.
32 Although the evidence is much more limited, not all of the toxicological studies evaluated that
33 examined the effect of underlying respiratory conditions on PM-related respiratory morbidity
34 focused on allergic airways disease. These provide preliminary evidence, which suggests that
PRELIMINARYDRAFT-September 2009 3-101
Do Not Quote or Cite
-------
1 non-allergic respiratory morbidities may also increase the susceptibility of an individual to PM-
2 related respiratory effects.
3 The results from the epidemiologic and toxicological studies that focused on underlying
4 allergic airways disease is supported by a series of controlled human exposure studies which
5 have shown that exposure to DEPs increases the allergic inflammatory response in atopic
6 individuals (Bastain et al., 2003; Diaz-Sanchez et al., 1997; Nordenhall et al., 2001). However,
7 not all controlled human exposure studies have found evidence for differences between the
8 respiratory effects exhibited by healthy and asthmatic individuals. Studies by Gong et al. (2003;
9 2004; 2008) reported that healthy and asthmatic subjects exposed to coarse, fine and ultrafine
10 CAPs, exhibited similar respiratory responses. However, it should be noted that these studies
11 excluded moderate and severe asthmatics that would be expected to show increased
12 susceptibility to PM exposure.
13 In addition to examining the association between exposure to PM and respiratory effects
14 in asthmatics, some studies examined whether individuals with COPD represent a potentially at-
15 risk population group. Desqueyroux et al. (2002) did not observe an increase in the exacerbation
16 of COPD in response to short-term exposure to PM2.5. However, studies that examined the effect
17 of PM on lung function in individuals with COPD (Lagorio et al., 2006; Trenga et al., 2006)
18 observed declines in FEVi, and FEVi and FVC, respectively in response to PMio and/or PM2.5.
19 Silkoff et al. (2005) observed associations between PMio and a reduction in FEVi and PM2 5 and
20 a reduction in PEF, in those with COPD, but only during one winter of the analysis. A few
21 controlled human exposure studies examined the effects of PM on COPD subjects and found no
22 significant difference in respiratory effects between healthy and individuals with COPD upon
23 exposure to PM2.5 CAPs (Gong et al., 2004; Gong et al., 2005; and Blomberg et al., 2005). On
24 the other hand the results from dosimetric studies have shown that COPD patients have increased
25 dose rates and impaired mucociliary clearance relative to age matched healthy subjects,
26 suggesting that individuals with COPD are potentially at a greater risk of PM-related health
27 effects (Section 3.2.1).
28 A few of the epidemiologic studies examined the effect of underlying respiratory
29 illnesses on the association between short- and long-term exposure to PM and mortality. Using
30 different pre-existing respiratory illnesses, Zeka et al. (2006) and De Leon et al. (2003) found
31 that short-term exposure to PMio increased the risk of non-accidental mortality for pneumonia
32 and circulatory mortality for all respiratory illnesses, respectively. Additionally, Zanobetti et al.
33 (2008) observed an association between long-term exposure to PMio and mortality in individuals
34 that had previously been hospitalized for COPD. Although these studies do not examine
35 additional size fractions of PM, together they highlight the potential effect of underlying
36 respiratory illnesses on the PM-mortality relationship.
PRELIMINARYDRAFT-September 2009 3-102
Do Not Quote or Cite
-------
1 Overall, the epidemiologic, controlled human exposure, and toxiciological studies
2 evaluated provide biological plausbility for the increased health effects observed in
3 epidemiologic studies among asthmatic individuals in response to PM exposure. The evidence
4 from studies that examined associations between PM and health effects in individuals with
5 COPD is inconsistent.
6 3.5.1.7.3 Respiratory Contributions to Cardiovascular Effects
7 Although the majority of health effects observed in individuals with pre-existing
8 respiratory illnesses were associated with respiratory illness exacerbations, studies also examined
9 whether underlying respiratory illnesses can lead to cardiovascular effects with PM exposure.
10 Respiratory disease has not consistently been observed to affect cardiovascular response in
11 controlled human exposure or toxicological studies. The majority of epidemiologic studies that
12 examined whether underlying respiratory illnesses contributed to the mainfestation of PM-related
13 cardiovascular hospital admission or ED visits, did not report increases in effects for a variety of
14 cardiovascular outcomes (e.g., IHD, arrhythmias, CHF, MI) for individuals with underlying
15 respiratory infection (Wellenius et al., 2006), pneumonia (Zanobetti and Schwartz, 2005), or
16 COPD (Peel et al., 2007; Wellenius et al., 2005). However, Yeatts et al. (2007), in a panel study,
17 found evidence for cardiovascular effects, specifically reductions in HRV parameters, in
18 asthmatic adults upon short-term exposure to PMi0-2.5. It must be noted that most of the
19 aforementioned epidemiologic studies focused on exposure to PMio, and, therefore, it is unclear
20 how these results compare to those found in the controlled human exposure and toxicological
21 studies that focused on exposure to PM2 5 (e.g., CAPs). Thus, it is unclear if individuals with
22 underlying respiratory illnesses represent a population group that is potentially at greater risk of
23 PM-related cardiovascular effects.
24 3.5.1.7.4 Diabetes and Obesity
25 It has been hypothesized that the systemic inflammatory cascade leads to an increase in
26 cardiovascular risk (Dubowsky et al., 2006). As a result, individuals with conditions linked to
27 chronic inflammation (i.e., diabetes and obesity), have been examined to determine whether
28 diabetes or obesity facilitate the manifestation of PM-mediated health effects, and, therefore,
29 represent a potentially at-risk population group.
30 Epidemiologic studies have examined whether diabetes modifies the association between
31 cardiovascular health effects and PM exposure, but these studies have primarily focused on
32 short-term exposure to PMi0. Time-series studies have provided evidence through an
33 examination of hospital admission and ED visits and mortality, which suggests an increase in
34 health effects in diabetic individuals in response to PM exposure. Multeity studies have found
35 upwards of 75% greater risk of hospitalization for cardiac diseases in individuals with diabetes
PRELIMINARYDRAFT-September 2009 3-103
Do Not Quote or Cite
-------
1 upon to exposure to PMio (Zanobetti and Schwartz, 2002). Studies conducted in Atlanta,
2 Georgia have also found increased risk for cardiovascular-related ED visits in diabetics,
3 specifically for IHD, arrhythmias, and CHF (Peel et al., 2007). Additional studies found some
4 evidence that individuals with diabetes are at increased risk of mortality upon exposure to PMio
5 (Zeka et al., 2006) and PM2.5 (Goldberg et al., 2006). However, some studies (both multicity and
6 single-city) have not observed a modification of the risk of cardiovascular ED visits and hospital
7 admissions in response to exposure to PMio in diabetics (Pope et al., 2006; Wellenius et al.,
8 2006; Zanobetti and Schwartz, 2005).
9 Panel and cohort studies have been conducted to determine the physiological changes that
10 occur in individuals with diabetes in response to PM exposure. These studies examined both
11 changes in inflammatory markers along with specific physiological alterations in the
12 cardiovascular system. Schneider et al. (2008) in a panel study of 22 individuals with type 2
13 diabetes mellitus in Chapel Hill, NC found evidence that ambient exposure to PM2.5 enhanced the
14 reduction in various markers of endothelial function. Liu et al. (2007) observed an increase in
15 end-diastolic FMD and end-systolic FMD, and decreases in end-diastolic basal diameter and
16 end-systolic basal diameter in diabetics upon exposure to PMio. The authors also observed
17 positive associations with FMD and blood pressure in diabetic individuals. A controlled human
18 exposure study conducted by Carlsten et al. (2008) found that DE did not elicit any
19 prothrombotic effects in subjects with metabolic syndrome, which consists of physiological
20 alterations similar to those observed in both diabetic and obese individuals. An examination of
21 biomarkers found mixed results, with Liao et al. (2005) observing an increase in vWF; Liu et al.
22 (2007) observing an increase in TEARS, but not CRP or TNF-a; and Dubowsky et al. (2006)
23 observing an increase in CRP and WBCs. Overall, it is unclear how differences in each of the
24 aforementioned biomarkers contribute to the potential overall cardiovascular effect observed in
25 diabetic individuals; however, an increase in inflammation, oxidative stress, and acute phase
26 response may contribute to cardiovascular effects. A recent toxicological study, also
27 demonstrated the potential for PM-related health effects in diabetics. Sun et al. (2009) found that
28 PM2.5 CAPs exposure for 4 months can exaggerate insulin resistance, visceral adiposity, and
29 inflammation in a diet-induced obesity mouse model.
30 Overall, epidemiologic studies have reported evidence for increased effects in diabetics in
31 response to PM exposure, with preliminary evidence for physiological alterations from
32 toxicological studies. However, the limited evidence from toxicological and controlled human
33 exposure studies along with the lack of studies that examined additional PM size fractions
34 warrants additional research to confirm these associations and to identify the biological
35 pathway(s) that may result in a greater response to PM in diabetics.
PRELIMINARYDRAFT-September 2009 3-104
Do Not Quote or Cite
-------
1 In addition to diabetes, obesity has been examined as a health condition with the potential
2 to lead to an increase in PM-related health effects. Only a few recent studies have examined the
3 potential effect modification of PM risk estimates by obesity. Schwartz et al. (2005) reported a
4 change in HRV in obese (i.e., BMI > 30 kg/m2) compared to non-obese subjects, while
5 Dubowsky et al. (2006) observed an increase in inflammatory markers (i.e., CRP, IL-6, and
6 WBC) in response to short-term exposure to PM2.5 among obese individuals. Additionally,
7 Schneider et al. (2008) found some evidence for a larger reduction in FMD in individuals with a
8 BMI >30 kg/m3 in response to PM2.5 exposure. These effects could be due, in part, to a higher
9 PM dose rate in obese individuals, which has been demonstrated in children by Bennett and
10 Zeman (2004). These investigators also reported that tidal volume and resting minute ventilation
11 increased with body mass index. Although a limited amount of research has been conducted to
12 examine PM-related health effects in obese individuals there is an increasing trend of individuals
13 within the U.S. that have been defined as overweight or obese (BMI > 25.0) (56-65% between
14 NHANES III and NHANES [1999-2002]).
15 3.5.1.8 Socioeconomic Status
16 Socioeconomic status (SES) is a composite measure that usually consists of economic
17 status, measured by income; social status measured by education; and work status measured by
18 occupation (Button and Levine, 1989). Based on data from the U.S. Census Bureau in 2006,
19 from among commonly-used indicators of SES, about 12% of individuals and 11% of families
20 are below the poverty line (U.S., 2009). Although the measure of SES is composed of a
21 multitude of determinants, each of these linked factors can influence an individual's
22 susceptibility to PM-related health effects. Additionally, low SES individuals have been found to
23 have a higher prevalence of pre-existing diseases; inadequate medical treatment; and limited
24 access to fresh foods leading to a reduced intake of antioxidant polyunsaturated fatty acids and
25 vitamins, which can increase this population group's risk from to PM (Kan et al., 2008).
26 SES and individual determinants of SES, such as educational attainment, are not mutually
27 exclusive and together can influence the susceptibility of a population. Within the U. S.
28 approximately 16% of the population does not have a high school degree and only 27% have a
29 bachelor's degree or higher level of education (U.S., 2009). Educational attainment generally
30 coincides with an individual's income level, which is correlated to other surrogates of SES, such
31 as residential environment (Jerrett et al., 2004). Low SES, and surrogates of SES such as
32 educational attainment, have been shown in some studies to modify health outcomes of PM
33 exposure for a population. Franklin et al. (2008) noted an increased risk in mortality associated
34 with short-term exposure to PM2.5 and its components for individuals with low SES while
35 additional analyses stratified by education level have also observed consistent trends of increased
PRELIMINARYDRAFT-September 2009 3-105
Do Not Quote or Cite
-------
1 mortality for PM2.5 and PM2.5 species for individuals with low educational attainment (Ostro et
2 al., 2006; Ostro et al., 2008; Zeka et al., 2006). This is further supported by a reanalysis of the
3 ACS cohort (Krewski et al., 2009), which found moderate evidence for increased lung cancer
4 mortality in individuals with a high school education or less compared to individuals with more
5 than a high school education in response to long-term exposure to PM2.5.
6 Epidemiologic studies have also examined additional surrogates of SES, such as
7 residential location and nutritional status to identify their influence on the risk to a population
8 group. Jerrett et al. (2004) examined the modification of acute mortality effects due to particulate
9 air pollution exposure by residential location in Hamilton, Canada using educational attainment
10 as a surrogate for SES. The authors found that the area of the city with the highest SES
11 characteristics displayed no evidence of effect modification while the area with the lowest SES
12 characteristics had the largest health effects. Likewise, Wilson et al. (2007) examined the effect
13 of SES on the association between mortality and short-term exposure to PM in Phoenix, but used
14 educational attainment and income to represent SES. When stratifying Phoenix into central,
15 middle, and outer rings of varying urban density central Phoenix, the area with the lowest SES,
16 was found to exhibit the greatest association with PM2.5. However, the association with urban
17 density differed when examining PMio-2.5, with the greatest effect being observed for the middle
18 ring. Yanosky et al. (2008) examined whether long-term exposure to traffic-related pollutants,
19 using NC>2 as a surrogate, varied by SES at the block group level. The authors found higher
20 levels of NC>2 associated with lower SES areas, which suggests that lower SES individuals are
21 disproportionately exposed to traffic-related pollutants, which includes PM.
22 Nutritional deficiencies have been associated with increased susceptibility to a variety of
23 infectious diseases and chronic health effects. Low SES may decrease access to fresh foods, and
24 thus be related to nutritional deficiencies that could increase susceptibility to PM-related health
25 effects. Baccarelli et al. (2008) examined the association between exposure to PM2.5 and HRV in
26 individuals with polymorphisms in MTHFR and cSHMT genes, which are associated with
27 reduced enzyme activity and increased risk of CVD. The authors found that when individuals
28 with these genetic polymorphisms increased their intake (above median levels) of B6, B12, or
29 methionine no PM2.5 effect on HRV was observed.
30 3.5.2 Potential Public Health Impacts
31 Upon evaluating the association between short- and long-term exposure to PM and
32 various health outcomes, studies also attempted to identify population groups that are at greater
33 risk from PM exposure (i.e., populations that have a greater likelihood of experiencing health
34 effects related to PM exposure). These studies did so by: conducting stratified analyses;
35 examining individuals with an underlying health condition; or developing animal models that
PRELIMINARYDRAFT-September 2009 3-106
Do Not Quote or Cite
-------
1 mimic the physiological conditions associated with an adverse health effect. These studies
2 identified a multitude of factors that could potentially contribute to whether an individual is at-
3 risk from PM (US EPA, 2009a, Table 8-2). The size of the population groups affected by these
4 factors is large and many people are likely to be members of more than one at-risk population.
5 Although studies have primarily used exposures to PM2.5 or PMio, the available evidence
6 suggests that the identified factors may also enhance susceptibility to PMio-2.5- The examination
7 of population groups at-risk from PM exposure allows for the NAAQS to provide an adequate
8 margin of safety for them.
9 The evidence from epidemiologic, controlled human exposure, and toxicological studies
10 provide coherence and biological plausbility for the association between short-term exposure to
11 PM and cardiovascular morbidity and mortality in older adults. As the demographics of the U.S.
12 population shift over the next 20 years with a larger percentage of the population (i.e., 13% of
13 the population in 2011 and a projected 20% in 2030) encompassing individuals > 65 years
14 (U.S. Census Bureau, 2000}, an increase in the number of PM-related health effects (e.g.,
15 cardiovascular and respiratory morbidity, and mortality) in individuals > 65 years old could
16 occur.
17 Overall, the evidence from epidemiologic studies that have examined the health effects
18 associated with all size fractions of PM and toxicological studies that have examined individual
19 PM components provide additional support to the hypothesis that children are at greater risk of
20 respiratory effects from short-term exposure to PM. In 2008 there were approximately 73.9
21 million children in the U.S. This number is expected to reach 82 million children by 2021. In
22 2008, children comprised about 24% of the population, with approximately equal numbers of
23 children in three age groups, 0 to 5 (25 million), 6 to 11 (24 million) and 12 to 17 (25 million)
24 years of age (Federal Interagency Forum on Child and Family Statistics).
25 Collectively, the evidence from epidemiologic and toxicological, and to a lesser extent,
26 controlled human exposure studies indicates increased susceptibility of individuals with
27 underlying cardiovascular diseases to PM exposure. Although the evidence for some outcomes
28 was inconsistent across epidemiologic and toxicological studies, this could be due to a variety of
29 issues including the PM size fraction used in the study along with the study location. Even with
30 these caveats, a large proportion of the U.S. population has been diagnosed with cardiovascular
31 diseases (i.e., approximately 51.6 million people with hypertension, 24.1 million with heart
32 disease, and 14.1 million with coronary heart disease [see Table 3-2]), and therefore represents a
33 large population group that is potentially at greater risk from exposure to PM than the general
34 population.
35 Overall, the epidemiologic, controlled human exposure, and toxiciological studies
36 evaluated provide biological plausbility for the increased health effects observed in
PRELIMINARYDRAFT-September 2009 3-107
Do Not Quote or Cite
-------
1 epidemiologic studies among asthmatic individuals in response to PM exposure. Although, the
2 evidence from studies that examined associations between PM and health effects in individuals
3 with COPD is inconsistent, taken together individuals with COPD and asthma represent a large
4 percent of the U.S. population, which may be more at greater risk of PM-r elated health effects
5 (Table 3-2).
6 Table 3-2. Percent of the U.S. population with respiratory diseases, cardiovascular
7 diseases, and diabetes.
Chronic
Condition/
Disease
Age
Adults (18+)*
Number
(x 106)
%
18-44
%
45-64
%
65-74
%
75+
%
Regional
NE
%
MW
%
S
%
W
%
Respiratory Diseases
Asthma*
Asthma (<18 yrs)
24.2
6.8*
11.0
9.3*
11.5
—
10.5
—
11.7
—
9.3
11.
7
11.5
10.
5
10.
8
COPD
Chronic
bronchitis
Emphysema
9.5
4.1
4.3
1.8
2.9
0.3
5.5
2.4
5.6
5.0
6.7
6.4
3.8
1.4
4.4
2.3
4.9
1.9
3.5
1.6
Cardiovascular Diseases
All heart disease
Coronary heart
disease
Hypertension
Stroke
Diabetes
24.1
14.1
51.6
5.6
17.1
10.
9
6.4
23.
4
2.6
7.8
3.6
0.9
7.7
0.5
2.6
12.3
7.2
32.4
2.4
10.4
26.1
18.4
52.7
7.6
18.2
36.3
25.5
53.5
11.2
17.9
10.
8
6.4
22.
2
2.1
7.2
12.7
7.6
23.7
2.8
8.1
10.
9
6.6
25.
O
2.9
8.0
9.2
4.7
20.
6
2.2
7.4
* All data for adults except asthma prevalence data for children under 18 years of age, from CDC (2008). For adults
prevalence data based off adults responding to "ever told had asthma." Source: Pleis and Lethbridge-Cejku (2007);
CDC (2008) (US EPA 2009a, Table 8-3).
9 In addition, epidemiologic studies have reported evidence for increased effects in
10 diabetics in response to PM exposure, with preliminary evidence for physiological alterations
11 from toxicological studies. However, the limited evidence from toxicological and controlled
12 human exposure studies along with the lack of studies that examined additional PM size fractions
13 warrants additional research to confirm these associations and to identify the biological
14 pathway(s) that may result in a greater response to PM in diabetics. This potentially at-risk
15 population group is large, with an estimated 17.1 million diabetic individuals in the U.S.
16 (Table 3-2).
PRELIMINARYDRAFT-September 2009
Do Not Quote or Cite
3-108
-------
1 As summarized above, there are several populations groups that may be at-risk for effects
2 from exposure to PM. Estimates of the size of population groups, such as older adults, children
3 and people with cardiopulmonary diseases, including diabetes, that are the populations
4 considered to be at-risk for the effects of PM exposure are large. Combining fairly small risk
5 estimates and small changes in PM concentration with large groups of the U.S. population who
6 are potentially at-risk would result in large public health impacts.
7 3.6 SUMMARY AND CONCLUSIONS
8 (to be included in external review draft Policy Assessment}
PRELIMINARYDRAFT-September 2009 3-109
Do Not Quote or Cite
-------
1 3.7 REFERENCES
2 Agatston AS; Janowitz WR; Hildner FJ; Zusmer NR; Viamonte M Jr; Detrano R. (1990). Quantification of coronary
3 artery calcium using ultrafast computed tomography. J Am Coll Cardiol, 15: 827-832.
4 Anderson ME; Bogdan GM. (2007). Environments, indoor air quality, and children. Pediatr Clin North Am, 54:
5 295-307, viii.
6 Andersen ZJ; Wahlin P; Raaschou-Nielsen O; Scheike T; Loft S. (2007). Ambient particle source apportionment and
7 daily hospital admissions among children and elderly in Copenhagen. 625-636.
8 Araujo JA; Barajas B; Kleinman M; Wang X; Bennett BJ; Gong KW; Navab M; Harkema J; Sioutas C; Lusis AJ;
9 Nel AE. (2008). Ambient paniculate pollutants in the ultrafme range promote early atherosclerosis and
10 systemic oxidative stress. Circ Res, 102: 589-596.
11 Associations of traffic-related air pollutants with hospitalisation for first acute myocardial infarction: the HEAPSS
12 study. Occup Environ Med, 63: 844-851.
13 Atiga WL; Calkins H; Lawrence JH; Tomaselli GF; Smith JM; Berger RD. (1998). Beat-to-beat repolarization
14 lability identifies patients at risk for sudden cardiac death., 9: 899-908.
15 Auchincloss AH; Roux AV; Dvonch JT; Brown PL; Barr RG; Daviglus ML; Goff DC; Kaufman JD; O'Neill MS.
16 (2008). Associations between Recent Exposure to Ambient Fine Paniculate Matter and Blood Pressure in
17 the Multi-Ethnic Study of Atherosclerosis (MESA). Environ Health Perspect, 116: 486-491.
18 Avol EL; Gauderman WJ; Tan SM; London SJ; Peters JM. (2001). Respiratory effects of relocating to areas of
19 differing air pollution levels. Am J Respir Crit Care Med, 164: 2067-2072.
20 Baccarelli A; Zanobetti A; Martinelli I; Grille P; Hou L; Giacomini S; Bonzini M; Lanzani G; Mannucci PM;
21 Bertazzi PA; Schwartz J. (2007). Effects of exposure to air pollution on blood coagulation. J Thromb
22 Haemost, 5: 252-260.
23 Baccarelli A; Martinelli I; Zanobetti A; Grillo P; Hou LF; Bertazzi PA; Mannucci PM; Schwartz J. (2008). Exposure
24 to Particulate Air Pollution and Risk of Deep Vein Thrombosis. Arch Intern Med, 168: 920-927.
25 Baccarelli A; Martinelli I; Pegoraro V; Melly S; Grillo P; Zanobetti A; Hou L; Bertazzi PA; Mannucci PM;
26 Schwartz J. (2009). Living near Major Traffic Roads and Risk of Deep Vein Thrombosis. Circulation, add
27 cite.
28 Ballester F; Rodriguez P; Iniguez C; Saez M; Daponte A; Galan I; Taracido M; Arribas F; Bellido J; Cirarda FB;
29 Canada A; Guillen JJ; Guillen-Grima F; Lopez E; Perez-Hoyos S; Lertxundi A; Toro S. (2006). Air
30 pollution and cardiovascular admisisons association in Spain: results within the EMECAS project. J
31 Epidemiol Community Health, 60: 328-336.
32 Barnett AG; Williams GM; Schwartz J; Best TL; Neller AH; Petroeschevsky AL; Simpson RW. (2006). The effects
33 of air pollution on hospitalizations for cardiovascular disease in elderly people in Australian and New
34 Zealand cities. Environ Health Perspect, 114: 1018-1023.
3 5 Bastain TM; Gilliland FD; Li Y-F; Saxon A; Diaz-Sanchez D. (2003). Intraindividual reproducibility of nasal
36 allergic responses to diesel exhaust particles indicates a susceptible phenotype., 109: 130-136.
37 Bayer-Oglesby L; Grize L; Gassner M; Takken-Sahli K; Sennhauser FH; Neu U; Schindler C; Braun-Fahrlander C.
3 8 (2005). Decline of ambient air pollution levels and improved respiratory health in Swiss children. Environ
39 Health Perspect, 113: 1632-1637.
PRELIMINARYDRAFT-September 2009 3-110
Do Not Quote or Cite
-------
1 Becquemin MM; Bertholon JF; Bouchikhi A; Malarbet JL; Roy M. (1999). Oronasal Ventilation Partitioning in
2 Adults and Children: Effect on Aerosol Deposition in Airways. Radiat Prot Dosimetry, 81: 221-228.
3 Bell ML; Dominici F; Ebisu K; Zeger SL; Samet JM. (2007). Spatial and Temporal Variation in PM2.5 Chemical
4 Composition in the United States for Health Effects Studies. 115: 989.
5 Bell ML; Ebisu K; Peng RD; Walker J; Samet JM; Zeger SL; Dominic F. (2008). Seasonal and regional short-term
6 effects of fine particles on hospital admissions in 202 U.S. counties, 1999-2005. Am J Epidemiol, 168:
7 1301-1310.
8 Bell M; Ebisu K; Peng R; Samet J; Dominici F. (2009). Hospital Admissions and Chemical Composition of Fine
9 Particle Air Pollution. Am J Respir Crit Care Med, 179: 1115 -1120.
10 Bennett, W.D; Zeman, K.L. (2004) Effect of body size on breathing pattern and fine-particle deposition in children.
11 J Appl Physiol, 97: 821-826.
12 Bennett WD; Zeman KL; Jarabek AM. (2008). Nasal contribution to breathing and fine particle deposition in
13 children versus adults. J Toxicol Environ Health A, 71: 227-237.
14 Berger A; Zareba W; Schneider A; Ruckerl R; Ibald-Mulli A; Cyrys J; Wichmann HE; Peters A. (2006). Runs of
15 ventricular and supraventricular tachycardia triggered by air pollution in patients with coronary heart
16 disease. J Occup Environ Med, 48: 1149-58.
17 Berger RD; Kasper EK; Baughman KL; Marban E; Calkins H; Tomaselli GF. (1997). Beat-to-Beat QT Interval
18 Variability : Novel Evidence for Repolarization Lability in Ischemic and Nonischemic Dilated
19 Cardiomyopathy., 96: 1557-1565.
20 Blomberg A; Tornqvist H; Desmyter L; Deneys V; Hermans C. (2005). Exposure to diesel exhaust nanoparticles
21 does not induce blood hypercoagulability in an at-risk population. J Thromb Haemost, 3: 2103-2105.
22 Brook RD; Brook JR; Urch B; Vincent R; Rajagopalan S; Silverman F. (2002). Inhalation of fine paniculate air
23 pollution and ozone causes acute arterial vasoconstriction in healthy adults., 105: 1534-1536.
24 Brown JS; Wilson WE; Grant LD. (2005). Dosimetric comparisons of particle deposition and retention in rats and
25 humans. Inhal Toxicol, 18: 1005-1011.
26 Brunekreef B; Beelen R; Hoek G; Schouten L; Bausch-Goldbohm S; Fischer P; Armstrong B; Hughes E; Jerrett M;
27 van den Brandt P. (2009). Effects of long-term exposure to traffic-related air pollution on respiratory and
28 cardiovascular mortality in the Netherlands: The NLCS-AIR Study. Health Effects Institute. Boston, MA.
29 139.
30 Burnett RT; Cakmak S; Brook JR; KrewskiD. (1997). The role of paniculate size and chemistry in the association
31 between summertime ambient air pollution and hospitalization for cardiorespiratory diseases. Environ
32 HealthPerspect, 105: 614-620.
33 Burnett RT; Smith-Doiron M; Stieb D; Cakmak S; Brook JR. (1999). Effects of paniculate and gaseous air pollution
34 on cardiorespiratory hospitalizations. Arch Environ Occup Health, 54: 130-139.
3 5 Burnett RT; Stieb D; Brook JR; Cakmak S; Dales R; Raizenne M; Vincent R; Dann T. (2004). Associations between
36 short-term changes in nitrogen dioxide and mortality in Canadian cities. Arch Environ Occup Health, 59:
37 228-236.
3 8 Cakmak S; Dales R; Vida C. (2009). Components of paniculate air pollution and mortality in Chile. Int J Occup
39 Environ Health, 15: 152.
PRELIMINARY DRAFT-September 2009 3-111
Do Not Quote or Cite
-------
1 Carlsten C; Kaufman JD; Trenga CA; Allen J; Peretz A; Sullivan JH. (2008). Thrombotic markers in metabolic
2 syndrome subjects exposed to diesel exhaust. Inhal Toxicol, 20: 917-921.
3 CDC. (2008). Table 3-1: Current Asthma Population Estimates, in thousands by Age, United States: National Health
4 Interview Survey, 2006. Retrieved, from ADD CITE.
5 CDC. (2008). Table 4-1: Current Asthma Prevalence Percents by Age, United States: National Health Interview
6 Survey, 2006. Retrieved, from ADD CITE.
7 Chang CC; Hwang JS; Chan CC; Wang PY; Hu TH; Cheng TJ. (2005). Effects of concentrated ambient particles on
8 heart rate variability in spontaneously hypertensive rats. J Occup Health, 47: 471-480.
9 Chang J; Delfmo RJ; Gillen D; Tjoa T; Nickerson B; Cooper D. (2008). Repeated respiratory hospital encounters
10 among children with asthma and residential proximity to traffic. J Occup Environ Med, 66: 90-98.
11 Chen LH; Knutsen SF; Shavlik D; Beeson WL; Petersen F; Ghamsary M; Abbey D. (2005). The association
12 between fatal coronary heart disease and ambient paniculate air pollution: Are females at greater risk?.
13 Environ Health Perspect, 113: 1723-1729.
14 Chen J; Schwartz J. (2009). Neurobehavioral effects of ambient air pollution on cognitive performance in US adults.
15 Neurotoxicology, 30: 231-239.
16 Chen Y; Yang Q; Krewski D; Burnett RT; Shi Y; McGrail KM. (2005). The effect of coarse ambient paniculate
17 matter on first, second, and overall hospital admissions for respiratory disease among the elderly. Inhal
18 Toxicol, 17: 649-655.
19 Chen Y; Yang Q; Krewski D; Shi Y; Burnett RT; McGrail K. (2004). Influence of relatively low level of paniculate
20 air pollution on hospitalization for COPD in elderly people. Inhal Toxicol, 16: 21-25.
21 Chevalier P; Burri H; Adeleine P; Kirkorian G; Lopez M; Leizorovicz A; Andre-Fouet X; Chapon P; Rubel P;
22 Touboul P. (2003). QT dynamicity and sudden death after myocardial infarction: results of a long-term
23 follow-up study., 14: 227-233.
24 Choi JH; Xu QS; Park SY; Kim JH; Hwang SS; Lee KH; Lee HJ; Hong YC. (2007). Seasonal variation of effect of
25 air pollution on blood pressure. JEpidemiol Community Health, 61: 314-318.
26 Cruts B; van Etten L; Tornqvist H; Blomberg A; Sandstrom T; Mills NL; Borm PJ. (2008). Exposure to diesel
27 exhaust induces changes in EEG in human volunteers. Part Fibre Toxicol, 5: 4
28 Dales R; Liu L; Szyszkowicz M; Dalipaj M; Willey J; Kulka R; Ruddy TD. (2007). Paniculate air pollution and
29 vascular reactivity: the bus stop study. Int Arch Occup Environ Health, 81: 159-164.
30 Daniels MJ; Dominici F; Zeger SL; Samet JM. (2004). The national morbidity, mortality, and air pollution study
31 Part III: PM10 concentration-response curves and thresholds for the 20 largest US cities.
32 De Leon SF; Thurston GD; Ito K. (2003). Contribution of respiratory disease to nonrespiratory mortality
33 associations with air pollution. Am J Respir Crit Care Med, 167: 1117-1123.
34 Delfino RJ; Murphy-Moulton AM; Becklake MR. (1998). Emergency room visits for respiratory illnesses among
35 the elderly in Montreal: association with low level ozone exposure. Environ Res, 76: 67-77.
36 Delfino R; Brummel S; Wu J; Stern H; Ostro B; Lipsett M; Winer A; Street D; Zhang L; Tjoa T. (2009). The
3 7 relationship of respiratory and cardiovascular hospital admissions to the southern California wildfires of
38 2003. Occup EnvironMed, 66: 189.
PRELIMINARYDRAFT-September 2009 3-112
Do Not Quote or Cite
-------
1 Diaz-Sanchez D; Tsien A; Fleming J; Saxon A. (1997). Combined diesel exhaust paniculate and ragweed allergen
2 challenge markedly enhances human in vivo nasal ragweed-specific IgE and skews cytokine production to
3 a T helper cell 2-type pattern. J Immunol, 158: 2406-2413.
4 Diez Roux AV; Auchincloss AH; Franklin TG; Raghunathan T; Barr RG; Kaufman J; Astor B; Keeler J. (2008).
5 Long-term exposure to ambient paniculate matter and prevalence of subclinical atherosclerosis in the
6 Multi-Ethnic Study of Atherosclerosis. Am J Epidemiol, 167: 667-675.
7 Dockery DW; Pope CA III; Xu X; Spengler JD; Ware JH; Fay ME; Ferris BG Jr; Speizer FE. (1993). An association
8 between air pollution and mortality in six US cities., 329: 1753-1759.
9 Dockery DW; Cunningham J; Damokosh AI; Neas LM; Spengler JD; Koutrakis P; Ware JH; Raizenne M; Speizer
10 FE. (1996). Health effects of acid aerosols on North American children: respiratory symptoms.
11 Dockery DW; Luttmann-Gibson H; Rich DQ; Link MS; Mittleman MA; Gold DR; Koutrakis P; Schwartz JD;
12 VerrierRL. (2005). Association of air pollution with increased incidence of ventricular tachyarrhythmias
13 recorded by implanted cardioverter defibrillators. Environ Health Perspect, 113: 670-674.
14 Dockery DW; Luttmann-Gibson H; Rich DQ; Link MS; Schwartz JD; Gold DR; Koutrakis P; Verrier RL;
15 Mittleman MA. (2005). Paniculate air pollution and nonfatal cardiac events Part II Association of air
16 pollution with confirmed arrhythmias recorded by implanted defibrillators.
17 Dominici F; McDermott A; Daniels M; Zeger SL; Samet J. (2003). Revised Analyses of Time-Series Studies of Air
18 Pollution and Health: Mortality Among Residents of 90 Cities.
19 Dominici F; Peng RD; Bell ML; Pham L; McDermott A; Zeger SL; Samet JL. (2006). Fine paniculate air pollution
20 and hospital admission for cardiovascular and respiratory diseases. JAMA, 295: 1127-1134.
21 Dominici F; Peng RD; Ebisu K; Zeger SL; Samet JM; Bell ML. (2007). Does the effect of PM10 on mortality
22 depend on PM nickel and vanadium content? A reanalysis of the NMMAPS data. Environ Health Perspect,
23 115: 1701-3.
24 Dominici F; Peng RD; Zeger SL; White RH; Samet JM. (2007). Paniculate air pollution and mortality in the United
25 States: did the risks change from 1987 to 2000?. Am J Epidemiol, 166: 880-8.
26 Dubowsky SD; Suh H; Schwartz J; Coull BA; Gold DR. (2006). Diabetes, obesity, and hypertension may enhance
27 associations between air pollution and markers of systemic inflammation. Environ Health Perspect, 114:
28 992-998.
29 DuttonDB; Levine S. (1989). Overview, methodological critique, and reformulation. In JP Bunker; DS Gomby; BH
30 Kehrer (Ed.),Pathways to health (pp. 29-69). Menlo Park, CA: The Henry J. Kaiser Family Foundation.
31 Ebelt ST; Wilson WE; Brauer M. (2005). Exposure to ambient and nonambient components of paniculate matter: a
32 comparison of health effects. Epidemiology, 16: 396-405.
33 Eftim SE; Samet JM; Janes H; McDermott A; Dominici F. (2008). Fine Paniculate Matter and Mortality: A
34 Comparison of the Six Cities and American Cancer Society Cohorts With a Medicare Cohort. ,19: 209-
35 216.
36 Enstrom JE. (2005). Fine paniculate air pollution and total mortality among elderly Californians, 1973-2002. Inhal
37 Toxicol, 17: 803-816.
38 Erbel R; Mohlenkamp S; Kerkhoff G; Budde T; Schmermund A. (2007). Non-invasive screening for coronary artery
39 disease: calcium scoring. Heart, 93: 1620-1629.
PRELIMINARY DRAFT-September 2009 3-113
Do Not Quote or Cite
-------
1 Fairley, D. (2003). Mortality and air pollution for Santa Clara County, California, 1989-1996. In: Revised analyses
2 of time-series studies of air pollution and health. Special report. Boston, MA: Health Effects Institute; pp.
3 97-106. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf. October 18, 2004.
4 Federal Interagency Forum on Child and Family Statistics. America's Children: Key National Indicators of Well-
5 Being (2009). Washington, DC: U.S. Government Printing Office.
6 Foley DL; Craig JM; Morley R; Olsson CA; Dwyer T; Smith K; Saffery R. (2009). Prospects for epigenetic
7 epidemiology. Am J Epidemiol, 169: 389-400.
8 Franklin M; Zeka A; Schwartz J. (2007). Association between PM2.5 and all-cause and specific-cause mortality in
9 27 US communities. J Expo Sci Environ Epidemiol, 17: 279-287.
10 Franklin M; Koutrakis P; Schwartz J. (2008). The role of particle composition on the association between PM2.5 d
11 mortality. Epidemiology, 19: 680-689.
12 Fung KY; Luginaah IKMG; Webster G. (2005). Air pollution and daily hospitalization rates for cardiovascular and
13 respiratory diseases in London, Ontario. Int J Environ Stud, 62: 677-685.
14 Fung KY; Khan S; Krewski D; Chen Y. (2006). Association between air pollution and multiple respiratory
15 hospitalizations among the elderly in Vancouver, Canada. Inhal Toxicol, 18: 1005-1011.
16 Gauderman, W. J.; McConnell, R.; Gilliland, F.; London, S.; Thomas, D.; Avol, E.; Vora, H.; Berhane, K.;
17 Rappaport, E. B.; Lurmann, F.; Margolis, H. G.; Peters, J. (2000). Association between air pollution and
18 lung function growth in southern California children. Am. J. Respir. Crit. Care Med. 162: 1383-1390.
19 Gauderman, W. J.; Gilliland, G. F.; Vora, H.; Avol, E.; Stram, D.; McConnell, R.; Thomas, D.; Lurmann, F.;
20 Margolis, H. G.; Rappaport, E. B.; Berhane, K.; Peters, J. M. (2002). Association between air pollution and
21 lung function growth in southern California children: results from a second cohort. Am. J. Respir. Crit.
22 Care Med. 166: 76-84.
23 Gauderman, W.J; Avol, E.; Gilliland, F.; Vora, H.; Thomas, D.; Berhane, K.; McConnell, R.; Kuenzli, N.; Lurmann,
24 F.; Rappaport, E.; Margolis, H.; Bates, D.; Peters, 1(2004). The effect of air pollution on lung development
25 from 10 to 18 years of age. NEJM 351:1057-67.
26 Gent JF; Koutrakis P; Belanger K; Triche E; Holford TR; Bracken MB; Leaderer BP. (2009). Symptoms and
27 medication use in children with asthma and traffic-related sources of fine particle pollution. Environ Health
28 Perspect, In Press: 1-41.
29 Gold DR; Litonjua AA; Zanobetti A; Coull BA; Schwartz J; MacCallum G; Verrier RL; Nearing BD; Canner MJ;
30 Suh H; Stone PH. (2005). Air pollution and ST-segment depression in elderly subjects. Environ Health
31 Perspect, 113:883-887.
32 Goldberg MS; Burnett RT; Yale JF; Valois MF; Brook JR. (2006). Associations between ambient air pollution and
33 daily mortality among persons with diabetes and cardiovascular disease. Environ Res, 100: 255-267.
34 Gong H Jr; Linn WS; Terrell SL; Clark KW; Geller MD; Anderson KR; Cascio WE; Sioutas C. (2004). Altered
3 5 heart-rate variability in asthmatic and healthy volunteers exposed to concentrated ambient coarse particles.
36 Inhal Toxicol, 16: 335-343.
37 Gong H Jr; Linn WS; Clark KW; Anderson KR; Geller MD; Sioutas C. (2005). Respiratory responses to exposures
3 8 with fine particulates and nitrogen dioxide in the elderly with and without COPD. Inhal Toxicol, 17: 123-
39 132.
PRELIMINARYDRAFT-September 2009 3-114
Do Not Quote or Cite
-------
1 Gong H Jr; Linn WS; Clark KW; Anderson KR; Sioutas C; Alexis NE; Cascio WE; Devlin RB. (2008). Exposures
2 of healthy and asthmatic volunteers to concentrated ambient ultrafine particles in Los Angeles. Inhal
3 Toxicol, 20: 533-545.
4 Health Effects Institute (2003). Commentary on revised analyses of selected studies. In: Revised analyses of time-
5 series studies of air pollution and health. Special report. Boston, MA: Health Effects Institute; pp. 255-290.
6 Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf. October 18, 2004.
7 Heidenfelder BL; Reif DM; Harkema JR; Cohen Hubal EA; Hudgens EE; Bramble LA; Wagner JG; Morishita M;
8 Keeler GJ; Edwards SW; Gallagher JE. (2009). Comparative microarray analysis and pulmonary changes
9 in brown norway rats exposed to ovalbumin and concentrated air particulates. Toxicol Sci, 108: 207-221.
10 Henneberger A; Zareba W; Ibald-Mulli A; Ruckerl R; Cyrys J; Couderc J-P; Mykins B; Woelke G; Wichmann HE;
11 Peters A. (2005). Repolarization changes induced by air pollution in ischemic heart disease patients.
12 Environ Health Perspect, 113: 440-446.
13 Hopke PK; Ito K; Mar T; Christensen WF; Eatough DJ; Henry RC; Kim E; Laden F; Lall R; Larson TV; Liu H;
14 Neas L; Pinto J; Stolzel M; Suh H; Paatero P; Thurston GD. (2006). PM source apportionment and health
15 effects: 1 Intercomparison of source apportionment results. J Expo Sci Environ Epidemiol, 16: 275-286.
16 Host S; Larrieu S; Pascal L; Blanchard M; Declercq C; Fabre P; Jusot JF; Chardon B; Le Tertre A; Wagner V;
17 Prouvost H; Lefranc A. (2008). Short-term associations between fine and coarse particles and hospital
18 admissions for cardiorespiratory diseases in six French cities. Occup Environ Med, 65: 544-551 James DS;
19 Stidley CA; Lambert WE; Chick TW; Mermier CM; Samet JM. (1997). Oronasal distribution of ventilation
20 at different ages. Arch Environ Occup Health, 52: 118-123.
21 Host S; Larrieu S; Pascal L; Blanchard M; Declercq C; Fabre P; Jusot JF; Chardon B; Le Tertre A; Wagner V;
22 Prouvost H; Lefranc A. (2008). Short-term associations between fine and coarse particles and hospital
23 admissions for cardiorespiratory diseases in six French cities. Occup Environ Med, 65: 544-551.
24 Ibald-Mulli A; Timonen KL; Peters A; Heinrich J; Wolke G; Lanki T; Buzorius G; Kreyling WG; De Hartog J;
25 Hoek G; Ten Brink HM; Pekkanen J. (2004). Effects of paniculate air pollution on blood pressure and heart
26 rate in subjects with cardiovascular disease: a multicenter approach. Environ Health Perspect, 112: 369-
27 377.
28 Islam T; Gauderman WJ; Berhane K; McConnell R; Avol E; Peters JM; Gilliland FD. (2007). The relationship
29 between air pollution, lung function and asthma in adolescents. Thorax, 62: 957-963.
30 Ito, K. (2003). Associations of paniculate matter components with daily mortality and morbidity in Detroit,
31 Michigan. In: Revised analyses of time-series studies of air pollution and health. Special report. Boston,
32 MA: Health Effects Institute; pp. 143-156. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf.
33 May 12, 2004.
34 Ito K; Christensen WF; Eatough DJ; Henry RC; Kim E; Laden F; Lall R; Larson TV; Neas L; Hopke PK; Thurston
3 5 GD. (2006). PM source apportionment and health effects: 2 An investigation of intermethod variability in
36 associations between source-apportioned fine particle mass and daily mortality in Washington, DC. J Expo
37 Sci Environ Epidemiol, 16: 300-310.
3 8 Ito K; Thurston GD; Silverman RA.. (2007). Association between coarse particles and asthma emergency
3 9 department (ED) visits in New York City. Add cite.
40 James DS; Stidley CA; Lambert WE; Chick TW; Mermier CM; Samet JM. (1997). Oronasal distribution of
41 ventilation at different ages. Arch Environ Occup Health, 52: 118-123.
PRELIMINARY DRAFT-September 2009 3-115
Do Not Quote or Cite
-------
1 Janes H; Dominici F; Zeger SL. (2007). Trends in air pollution and mortality: an approach to the assessment of
2 unmeasured confounding., 18: 416-23.
3 Jansen KL; Larson TV; Koenig JQ; Mar TF; Fields C; Stewart J; Lippmann M. (2005). Associations between health
4 effects and paniculate matter and black carbon in subjects with respiratory disease. Environ Health
5 Perspect, 113: 1741-1746.
6 Jerrett M; Burnett RT; Brook J; Kanaroglou P; Giovis C; Finkelstein N; Hutchison B. (2004). Do socioeconomic
7 characteristics modify the short term association between air pollution and mortality? Evidence from a
8 zonal time series in Hamilton, Canada. J Epidemiol Community Health, 58: 31-40.
9 Kan H; London SJ; Chen G; Zhang Y; Song G; Zhao N; Jiang L; Chen B. (2008). Season, sex, age, and education as
10 modifiers of the effects of outdoor air pollution on daily mortality in Shanghai, China: The Public Health
11 and Air Pollution in Asia (PAPA) Study. Environ Health Perspect, 116: 1183-1188.
12 Keverne EB; Curley JP. (2008). Epigenetics, brain evolution and behaviour. Neuroendocrinology, 29: 398-412.
13 Kim CS; Hu SC. (1998). Regional deposition of inhaled particles in human lungs: comparison between men and
14 women. J Appl Physiol, 84: 1834-1844.
15 KimCS; Jaques PA. (2000). Respiratory dose of inhaled ultrafine particles in healthy adults. ADD CITE ,358:
16 2693-2705
17 Klemm, R. J.; Mason, R. (2003). Replication of reanalysis of Harvard Six-City mortality study. In: Revised
18 analyses of time-series studies of air pollution and health. Special report. Boston, MA: Health Effects
19 Institute; pp. 165-172. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf. May 12, 2004.
20 Klemm RJ; Lipfert FW; Wyzga RE; Gust C. (2004). Daily mortality and air pollution in Atlanta: two years of data
21 from ARIES. Inhal Toxicol, 16 Suppl 1: 131-141.
22 Krewski, D.; Burnett, R. T.; Goldberg, M. S.; Hoover, K.; Siemiatycki, J,; Jerrett, M.; Abrahamowicz, M.; White,
23 W. H. (2000). Reanalysis of the Harvard Six Cities Study and the American Cancer Society Study of
24 paniculate air pollution and mortality. A special report of the Institute's particle epidemiology reanalysis
25 project. Cambridge, MA: Health Effects Institute. http://pubs.healtheffects.org/view.php?id=6
26 Krewski D; Jerrett M; Burnett RT; Ma R; Hughes E; Shi Y; Turner MC; Pope AC III; Thurston G; Calle EE; Thun
27 MJ. (2009). Extended follow-up and spatial analysis of the American Cancer Society study linking
28 paniculate air pollution and mortality. Health Effects Institute. Cambridge, MA. 140.
29 Kunzli N; Jerrett M; Mack WJ; Beckerman B; LaBree L; Gilliland F; Thomas D; Peters J; Hodis HN. (2005).
30 Ambient air pollution and atherosclerosis in Los Angeles. Environ Health Perspect, 113: 201-206.
31 Laden F; Neas LM; Dockery DW; Schwartz J. (2000). Association of fine paniculate matter from different sources
32 with daily mortality in six US cities. Environ Health Perspect, 108: 941-947.
33 Laden, F.; Schwartz, J.; Speizer, F.E.; Dockery, D.W. (2006). Reduction in fine paniculate air pollution and
34 mortality: extended follow-up of the Harvard Six Cities Study. Am. J. Respir. Crit. Care. Med. 173: 667-
35 672.
36 Lagorio S; Forastiere F; Pistelli R; lavarone I; Michelozzi P; Fano V; Marconi A; Ziemacki G; Ostro BD. (2006).
3 7 Air pollution and lung function among susceptible adult subjects: a panel study., 5: 11.
3 8 Lanki T; Pekkanen J; Aalto P; Elosua R; Berglind N; D'Ippoliti D; Kulmala M; Nyberg F; Peters A; Picciotto S;
39 Salomaa V; Sunyer J; Tiittanen P; Von Klot S; Forastiere F; for the HEAPSS Study Group. (2006).
PRELIMINARYDRAFT-September 2009 3-116
Do Not Quote or Cite
-------
1 Larrieu S; Jusot J-F; Blanchard M; Prouvost H; Declercq C; Fabre P; Pascal L; Le Tertre A; Wagner V; Riviere S;
2 Chardon B; Borelli D; Cassadou S; Eilstein D; Lefranc A. (2007). Short term effects of air pollution on
3 hospitalizations for cardiovascular diseases in eight French cities: The PSAS program. Sci Total Environ,
4 387: 105-112.
5 Larrieu S; Lefranc A; Gault G; Chatignoux E; Couvy F; Jouves B; Filleul L. (2009). Are the Short-term Effects of
6 Air Pollution Restricted to Cardiorespiratory Diseases?. Am J Epidemiol, 169: 1-8.
7 Le Tertre A; Medina S; Samoli E; Forsberg B; Michelozzi P; Boumghar A; Vonk JM; Bellini A; Atkinson R; Ayres
8 JG; Sunyer J; Schwartz J; Katsouyanni K. (2002). Short term effects of paniculate air pollution on
9 cardiovascular diseases in eight European cities. J Epidemiol Community Health, 56: 773-779.
10 Le Tertre A; Medina S; Samoli E; Forsberg B; Michelozzi P; Boumghar A; Vonk JM; Bellini A; Atkinson R; Ayres
11 JG; Sunyer J; Schwartz J; Katsouyanni K. (2003). Short-term effects of paniculate air pollution on
12 cardiovascular diseases in eight European cities.
13 Liao D; Heiss G; Chinchilli VM; Duan Y; Folsom AR; Lin HM; Salomaa V. (2005). Association of criteria
14 pollutants with plasma hemostatic/inflammatory markers: a population-based study. J Expo Sci Environ
15 Epidemiol, 15:319-328.
16 Lin M; Chen Y; Burnett RT; Villeneuve PJ; Krewski D. (2002). The influence of ambient coarse paniculate matter
17 on asthma hospitalization in children: case-crossover and time-series analyses. Environ Health Perspect,
18 110:575-581.
19 Lin, X.; Metzger KB; Tolbert PE; Klein M; Peel JL; Flanders WD; Todd KH; Mulholland JA; Ryan PB; Frumkin H.
20 (2004). Ambient air pollution and cardiovascular emergency department visits. Add cite , 15: 46-56).
21 Lipfert FW; Baty JD; Miller JP; Wyzga RE. (2006). PM2.5 constituents and related air quality variables as
22 predictors of survival in a cohort of U.S. military veterans. Inhal Toxicol, 18: 645-657.
23 Lippmann, M.; Ito, K.; Nadas, A.; Burnett, R. T. (2000). Association of paniculate matter components with daily
24 mortality and morbidity in urban populations. Cambridge, MA: Health Effects Institute; research report 95.
25 Lippmann M; Ito K; Hwang JS; Maciejczyk P; Chen LC. (2006). Cardiovascular effects of nickel in ambient air.
26 Environ Health Perspect, 114: 1662-9.
27 Liu J; Ballaney M; Al-Alem U; Quan C; Jin X; Perera F; Chen LC; Miller RL. (2008). Combined Inhaled Diesel
28 Exhaust Particles and Allergen Exposure Alter Methylation of T Helper Genes and IgE Production In Vivo.
29 Toxicol Sci, 102: 76-81.
30 Liu L; Ruddy TD; Dalipaj M; Szyszkowicz M; You H; Poon R; Wheeler A; Dales R. (2007). Influence of personal
31 exposure to paniculate air pollution on cardiovascular physiology and biomarkers of inflammation and
32 oxidative stress in subjects with diabetes. J Occup Environ Med, 49: 258-265.
33 Maheswaran R; Haining RP; Brindley P; Law J; Pearson T; Fryers PR; Wise S; Campbell MJ. (2005). Outdoor air
34 pollution, mortality, and hospital admissions from coronary heart disease in Sheffield, UK: a small area
35 level ecological study. Eur Heart J, 26: 2543-2549.
36 Mar TF; Norris GA; Koenig JQ; Larson TV. (2000). Associations between air pollution and mortality in Phoenix,
37 1995-1997. Environ Health Perspect, 108: 347-353.
38 Mar, T. F.; Norris, G. A.; Larson, T. V; Wilson, W. E.; Koenig, J. Q. (2003). Air pollution and cardiovascular
39 mortality in Phoenix, 1995-1997. In: Revised analyses of time-series studies of air pollution and health.
40 Special report. Boston, MA: Health Effects Institute; pp. 177-182. Available:
41 http://www.healtheffects.org/Pubs/TimeSeries.pdf. October 18, 2004.
PRELIMINARYDRAFT-September 2009 3-117
Do Not Quote or Cite
-------
1 Mar TF; Larson TV; Stier RA; Claiborn C; Koenig JQ. (2004). An analysis of the association between respiratory
2 symptoms in subjects with asthma and daily air pollution in Spokane, Washington. Inhal Toxicol, 16: 809-
3 815.
4 Mar TF; Jansen K; Shepherd K; Lumley T; Larson TV; Koenig JQ. (2005). Exhaled nitric oxide in children with
5 asthma and short-term PM25 exposure in Seattle. Environ Health Perspect, 113: 1791-1794.
6 Mar TF; Ito K; Koenig JQ; Larson TV; Eatough DJ; Henry RC; Kim E; Laden F; Lall R; Neas L; Stolzel M; Paatero
7 P; Hopke PK; Thurston GD. (2006). PM source apportionment and health effects: 3 Investigation of inter-
8 method variations in associations between estimated source contributions of PM25 and daily mortality in
9 Phoenix, AZ. J Expo Sci Environ Epidemiol, Epidemiol. 16: 311-320.
10 McConnell R; Berhane K; Gilliland F; London SJ; Vora H; Avol E; Gauderman WJ; Margolis HG; Lurmann F;
11 Thomas DC; Peters JM. (1999). Air pollution and bronchitic symptoms in southern California children with
12 asthma. Environ Health Perspect, 107: 757-760.
13 McConnell R; Berhane K; Gilliland F; Molitor J; Thomas D; Lurmann F; Avol E; Gauderman WJ; Peters JM.
14 (2003). Prospective study of air pollution and bronchitic symptoms in children with asthma. Am J Respir
15 Crit Care Med, 168: 790-797.
16 Metzger KB; Tolbert PE; Klein M; Peel JL; Flanders WD; Todd KH; Mulholland JA; Ryan PB; Frumkin H. (2004).
17 Ambient air pollution and cardiovascular emergency department visits. ADD CITE, 15: 46-56.
18 Metzger KB; Klein M; Flanders WD; Peel JL; Mulholland JA; Langberg JJ; Tolbert PE. (2007). Ambient air
19 pollution and cardiac arrhythmias in patients with implantable defibrillators. Epidemiology, 18: 585-592.
20 Miller KA; Siscovick DS; Sheppard L; Shepherd K; Sullivan JH; Anderson GL; Kaufman JD. (2007). Long-term
21 exposure to air pollution and incidence of cardiovascular events in women. ,356: 447-458.
22 Moolgavkar, S.H. (2003). Air pollution and daily deaths and hospital admissions in Los Angeles and Cook
23 Counties. In: Revised analyses of time-series studies of air pollution and health. Special report. Boston,
24 MA: Health Effects Institute; pp. 183-198. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf.
25 October 18, 2004.
26 Morishita M; Keeler G; Wagner J; Marsik F; Timm E; Dvonch J; Harkema J. (2004). Pulmonary retention of
27 paniculate matter is associated with airway inflammation in allergic rats exposed to air pollution in urban
28 Detroit. Inhal Toxicol, 16: 663-674.
29 National Research Council (2004). Research Priorities for Airbourne Paniculate Matter: IV. Continuing Research
30 Progress. Washington, D.C.: National Academies Press.
31 NCHS. (2007). Classification of Diseases, Functioning, and Disability...Monitoring the Nation's Health. Retrieved ,
32 from #http://www.cdc.gov/nchs/icd9.htm#RTF#.
3 3 New York State Department of Health. (2006). Study of ambient air contaminants and asthma in New York City,
34 parts A and B.
3 5 Nordenhall C; Pourazar J; Ledin M-C; Levin J-O; Sandstrom T; Adelroth E. (2001). Diesel exhaust enhances airway
36 responsiveness in asthmatic subjects. Eur Respir J, 17: 909-915.
37 Nordling E; Berglind N; Melen E; Emenius G; Hallberg J; Nyberg F; Pershagen G; Svartengren M; Wickman M;
3 8 Bellander T. (2008). Traffic-related air pollution and childhood respiratory symptoms, function and
39 allergies., 19: 401-8.
PRELIMINARY DRAFT-September 2009 3-118
Do Not Quote or Cite
-------
1 Okin PM; Devereux RB; Howard BV; Fabsitz RR; Lee ET; Welty TK. (2000). Assessment of QT interval and QT
2 dispersion for prediction of all-cause and cardiovascular mortality in American Indians: The Strong Heart
3 Study., 101:61-66.
4 O'Neill MS; Hajat S; Zanobetti A; Ramirez-Aguilar M; Schwartz J. (2005). Impact of control for air pollution and
5 respiratory epidemics on the estimated associations of temperature and daily mortality. Int J Biometeorol,
6 50: 121-9.
7 Ostro B. (1995). Fine paniculate air pollution and mortality in two Southern California counties. Environ Res, 70:
8 98-104.
9 Ostro, B.; Lipsett, M.; Mann, J.; Braxton-Owens, H.; White, M. (2001). Air pollution and exacerbation of asthma in
10 African-American children in Los Angeles. Epidem. 12: 200-208.
11 Ostro, B. D.; Broadwin, R.; Lipsett, M. J. (2003). Coarse particles and daily mortality in Coachella Valley,
12 California. In: Revised analyses of time-series studies of air pollution and health. Special report. Boston,
13 MA: Health Effects Institute; pp. 199-204. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf.
14 October 18, 2004.
15 Ostro, B.D.; Broadwin, R.; Green, S.; Feng, W.; Lipsett, M.(2006). Fine particulate air pollution and mortality in
16 nine California counties: results from CALFINE. Env. Health Perspect. 114: 29-33.
17 Ostro B; Feng W-Y; Broadwin R; Green S; Lipsett M. (2007). The effects of components of fine particulate air
18 pollution on mortality in California: results from CALFINE. Environ Health Perspect, 115: 13-9.
19 Ostro BD; Feng WY; Broadwin R; Malig B J; Green RS; Lipsett MJ. (2008). The impact of components of fine
20 particulate matter on cardiovascular mortality in susceptible subpopulations. Occup Environ Med, 65: 750-
21 756.
22 Ostro B; Roth L; Malig B; Marty M. (2009). The effects of fine particle components on respiratory hospital
23 admissions in children. Environ Health Perspect, 117: 475.
24 Peel JL; Tolbert PE; Klein M; Metzger KB; Flanders WD; Knox T; Mulholland JA; Ryan PB; Frumkin H. (2005).
25 Ambient air pollution and respiratory emergency department visits. ADD CITE , 16: 164-174.
26 Peel JL; Metzger KB; Klein M; Flanders WD; Mulholland JA; Tolbert PE. (2007). Ambient air pollution and
27 cardiovascular emergency department visits in potentially sensitive groups. Am J Epidemiol, 165: 625-633.
28 Peng RD; Chang HH; Bell ML; McDermott A; Zeger SL; Samet JM; Dominici F. (2008). Coarse particulate matter
29 air pollution and hospital admissions for cardiovascular and respiratory diseases among Medicare patients.
30 JAMA, 299: 2172-2179.
31 Peng R; Bell M; Geyh A; McDermott A; Zeger S; Samet J; Dominici F. (2009). Emergency admissions for
32 cardiovascular and respiratory diseases and the chemical composition of fine particle air pollution. Environ
33 Health Perspect, 117: 957-963.
34 Penttinen P; Vallius M; Tiittanen P; Ruuskanen J; Pekkanen J. (2006). Source-specific fine particles in urban air and
35 respiratory function among adult asthmatics. Inhal Toxicol, 18: 191-198.
36 Peretz A; Kaufman JD; Trenga CA; Allen J; Carlsten C; Aulet MR; Adar SD; Sullivan JH. (2008). Effects of diesel
37 exhaust inhalation on heart rate variability in human volunteers. Environ Res, 107: 178-184.
38 Peters, A.; Dockery, D. W.; Muller, J. E.; Mittleman, M. A. (2001). Increased particulate air pollution and the
39 triggering of myocardial infarction. Circulation 103:2810-2815.
PRELIMINARY DRAFT-September 2009 3-119
Do Not Quote or Cite
-------
1 Peters A. (2005). Paniculate matter and heart disease: evidence from epidemiological studies. Toxicol Appl
2 Pharmacol, 207: S477-S482.
3 Pleis JR; Lethbridge-Cejku M. (2007). Summary health statistics for US adults: National Health Interview Survey,
4 2006. CDC. Maryland.
5 Pope, C. A., Ill; Burnett, R. T.; Thun, M. I; Calle, E. E.; Krewski, D.; Ito, K.; Thurston, G. D. (2002). Lung cancer,
6 cardiopulmonary mortality, and long-term exposure to fine paniculate air pollution. J. Am. Med. Assoc.
7 287:1132-1141.
8 Pope CA III; Burnett RT; Thurston GD; Thun MJ; Calle EE; Krewski D; Godleski JJ. (2004). Cardiovascular
9 mortality and long-term exposure to paniculate air pollution: epidemiological evidence of general
10 pathophysiological pathways of disease., 109: 71-77.
11 Pope CA III; Burnett RT. (2007). Confounding in air pollution epidemiology: the broader context., 18: 424-426.
12 Zeger S; McDermott A; Dominici F; Samet J. (2007). Mortality in the medicare population and chronic
13 exposure to fine paniculate air pollution. Environ Health Perspect, 116: 1614-1619.
14 Pope CA 3rd; Ezzati M; Dockery DW. (2009). Fine-particulate air pollution and life expectancy in the United
15 States. N Engl J Med, 360: 376-386.
16 Puett RC; Schwartz J; Hart JE; Yanosky JD; Speizer FE; Suh H; Paciorek CJ; Neas LM; Laden F. (2008). Chronic
17 paniculate exposure, mortality, and coronary heart disease in the nurses' health study. Am J Epidemiol,
18 168:1161-1168.
19 Raizenne M; Neas LM; Damokosh AI; Dockery DW; Spengler JD; Koutrakis P; Ware JH; Speizer FE. (1996).
20 Health effects of acid aerosols on North American children: pulmonary function. Environ Health Perspect,
21 104:506-514.
22 Reed MD; Barrett EG; Campen MJ; Divine KK; Gigliotti AP; McDonald JD; Seagrave JC; Mauderly JL; Seilkop
23 SK; Swenberg JA. (2008). Health effects of subchronic inhalation exposure to gasoline engine exhaust.
24 Inhal Toxicol, 20: 1125-1143.
25 Resnick HE; Lindsay RS; McDermott MM; Devereux RB; Jones KL; Fabsitz RR; Howard BV. (2004). Relationship
26 of high and low ankle brachial index to all-cause and cardiovascular disease mortality: the Strong Heart
27 Study., 109: 733-739.
28 Rich DQ; Schwartz J; Mittleman MA; Link M; Luttmann-Gibson H; Catalano PJ; Speizer FE; Dockery DW. (2005).
29 ssociation of short-term ambient air pollution concentrations and ventricular arrhythmias. Am J Epidemiol,
30 161:1123-1132.
31 Rich DQ; Freudenberger RS; Ohman-Strickland P; Cho Y; Kipen HM. (2008). Right heart pressure increases after
32 acute increases in ambient particulate concentration. Environ Health Perspect, 116: 1167-1171.
33 Rich KE; Petkau J; Vedal S; Brauer M. (2004). A case-crossover analysis of particulate air pollution and cardiac
34 arrhythmia in patients with implantable cardioverter defibrillators. Inhal Toxicol, 16: 363-372.
3 5 Rojas-Martinez R; Perez-Padilla R; Olaiz-Fernandez G; Mendoza-Alvarado L; Moreno-Macias H; Fortoul T;
36 McDonnell W; Loomis D; Romieu I. (2007). Lung function growth in children with long-term exposure to
37 air pollutants in Mexico City., 176: 377-84.
3 8 Roosli M; Kunzli N; Braun-Fahrlander C; Egger M. (2005). Years of life lost attributable to air pollution in
39 Switzerland: dynamic exposure-response model. Int J Epidemiol, 34: 1029-1035.
PRELIMINARYDRAFT-September 2009 3-120
Do Not Quote or Cite
-------
1 Rosenlund M; Berglind N; Pershagen G; Hallqvist J; Jonson T; Hollander T. (2006). Long-term exposure to urban
2 air pollution and myocardial infarction., 17: 383-390.
3 Samet JM; Dominici F; Zeger SL; Schwartz J; Dockery DW. (2000). National morbidity, mortality, and air pollution
4 study Part I: methods and methodologic issues. Add cite.
5 Sarnat JA; Marmur A; Klein M; Kim E; Russell AG; Sarnat SE; Mulholland JA; Hopke PK; Tolbert PE. (2008).
6 Fine particle sources and cardiorespiratory morbidity: an application of chemical mass balance and factor
7 analytical source-apportionment methods. Environ Health Perspect, 116: 459-66.
8 Sarnat SE; Suh HH; Coull BA; Schwartz J; Stone PH; Gold DR. (2006). Ambient paniculate air pollution and
9 cardiac arrhythmia in a panel of older adults in Steubenville, Ohio. Occup Environ Med, 63: 700-706.
10 Schindler C; Keidel D; Gerbase MW; Zemp E; Bettschart R; Brandli O; Brutsche MH; Burdet L; Karrer W; Knopfli
11 B; Pons M; Rapp R; Bayer-Oglesby L; Kunzli N; Schwartz J; Liu L-JS; Ackermann-Liebrich U; Rochat T;
12 the S APALDIA Team. (2009). Improvements in PM10 exposure and reduced rates of respiratory
13 symptoms in a cohort of Swiss adults (SAPALDIA)., 179: 579-587.
14 Schneider A; Neas L; Herbst M; Case M; Williams R; Cascio W; Hinderliter A; Holguin F; Buse J; Dungan K.
15 (2008). Endothelial dysfunction: associations with exposure to ambient fine particles in diabetic
16 individuals. Environ Health Perspect, 116: 1666.
17 Schreuder AB; Larson TV; Sheppard L; Claiborn CS. (2006). Ambient woodsmoke and associated respiratory
18 emergency department visits in Spokane, Washington. Int J Occup Environ Health, 12: 147-53.
19 Schwartz J; Morris R. (1995). Air pollution and hospital admissions for cardiovascular disease in Detroit, Michigan.
20 Am J Epidemiol, 142: 23-35.
21 Schwartz J; Coull B; Laden F; Ryan L. (2008). The effect of dose and timing of dose on the association between
22 airborne particles and survival. Environ Health Perspect, 116: 64-69.
23 Schwartz, J. (1997). Air pollution and hospital admissions for cardiovascular disease in Tucson. Epidemiology 8:
24 371-377.
25 Schwartz, J.; Neas, L. M. (2000). Fine particles are more strongly associated than coarse particles with acute
26 respiratory health effects in schoolchildren. Epidemiology 11:6-10.
27 Schwartz, J. (2003). Daily deaths associated with air pollution in six US cities and short-term mortality
28 displacement in Boston. In: Revised analyses of time-series studies of air pollution and health. Special
29 report. Boston, MA: Health Effects Institute; pp. 219-226. Available:
30 http://www.healtheffects.org/Pubs/TimeSeries.pdf. October 18, 2004.
31 Schwartz, J.; Zanobetti, A; and Bateson, T. (2003). Morbidity and mortality among elderly residents of cities with
32 daily PM measurements. In: Revised analyses of time-series studies of air pollution and health. Special
33 report. Boston, MA: Health Effects Institute; pp.25 to 58. Available:
34 http://www.healtheffects.org/Pubs/TimeSeries.pdf. May 2003.
3 5 Sheppard, L. (2003). Ambient air pollution and nonelderly asthma hospital admissions in Seattle, Washington,
36 1987-1994. In: Revised analyses of time-series studies of air pollution and health. Special report. Boston,
37 MA: Health Effects Institute; pp. 227-230. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf.
38 October 18, 2004.
39 Silkoff PE; Zhang L; Dutton S; Langmack EL; Vedal S; Murphy J; Make B. (2005). Winter air pollution and disease
40 parameters in advanced chronic obstructive pulmonary disease panels residing in Denver, Colorado. J
41 Allergy Clinlmmunol, 115: 337-344.
PRELIMINARYDRAFT-September 2009 3-121
Do Not Quote or Cite
-------
1 Slaughter JC; Kim E; Sheppard L; Sullivan JH; Larson TV; Claiborn C. (2005). Association between paniculate
2 matter and emergency room visits, hospital admissions and mortality in Spokane, Washington. J Expo Sci
3 Environ Epidemiol, 15: 153-159.
4 Sullivan JH; Schreuder AB; Trenga CA; Liu SL; Larson TV; Koenig JQ; Kaufman JD. (2005). Association between
5 short term exposure to fine paniculate matter and heart rate variability in older subjects with and without
6 heart disease. Thorax, 60: 462-6.
7 Sun Q; Yue P; Kirk RI; Wang A; Moatti D; Jin X; Lu B; Schecter AD; Lippmann M; Gordon T; Chen LC;
8 Rajagopalan S. (2008). Ambient air paniculate matter exposure and tissue factor expression in
9 atherosclerosis. Inhal Toxicol, 20: 127-137.
10 Sun Q; Yue P; Ying Z; Cardounel AJ; Brook RD; Devlin R; Hwang JS; Zweier JL; Chen LC; Rajagopalan S.
11 (2008). Air Pollution Exposure Potentiates Hypertension Through Reactive Oxygen Species-Mediated
12 Activation of Rho/ROCK. Arterioscler Thromb Vase Biol, 28: 1760-1766.
13 Suwa T; Hogg JC; Quinlan KB; Ohgami A; Vincent R; Van Eeden SF. (2002). Paniculate air pollution induces
14 progression of atherosclerosis. J Am Coll Cardiol, 39: 935-942.
15 Tabachnik E; Muller N; Toye B; Levison H. (1981). Measurement of ventilation in children using the respiratory
16 inductive plethysmograph. J Pediatr, 99: 895-899.
17 Thurston G; Ito K; Mar T; Christensen WF; Eatough DJ; Henry RC; Kim E; Laden F; Lall R; Larson TV; Liu H;
18 Neas L; Pinto J; Stolzel M; Suh H; Hopke PK. (2005). Results and implications of the workshop on the
19 source apportionment of PM health effects., 16: S134-S135.
20 Tolbert PE; Klein M; Peel JL; Sarnat SE; Sarnat JA. (2007). Multipollutant modeling issues in a study of ambient air
21 quality and emergency department visits in Atlanta. J Expo Sci Environ Epidemiol, 24: 938-945
22 Trenga CA; Sullivan JH; Schildcrout JS; Shepherd KP; Shapiro GG; Liu LJ; Kaufman JD; Koenig JQ. (2006).
23 Effect of paniculate air pollution on lung function in adult and pediatric subjects in a Seattle panel study.,
24 129: 1614-22.
25 Tsai FC; Apte MG; Daisey JM. (2000). An exploratory analysis of the relationship between mortality and the
26 chemical composition of airborne paniculate matter. Inhal Toxicol, 12: 121-135.
27 Ulirsch GV; Ball LM; Kaye W; Shy CM; Lee CV; Crawford-Brown D; Symons M; Holloway T. (2007). Effect of
28 paniculate matter air pollution on hospital admissions and medical visits for lung and heart disease in two
29 southeast Idaho cities. J Expo Sci Environ Epidemiol, 17: 478-487.
30 US EPA (1996). Air Quality Criteria for Paniculate Matter : IV Continuing Research Progress. Washington,
31 D.C.: National Academies Press.. National Center for Environmental Assessment, Office of Research and
32 Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711; report no.
33 EPA/600/P-95/001aF-cF. 3 volumes. Available:
34 http://www.epa.gOv/ttn/naaqs/standards/pm/s jmjr_cd.html
35 US EPA (2004). Air Quality Criteria for Paniculate Matter. National Center for Environmental Assessment, Office
36 of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711;
37 report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October 2004. Available:
3 8 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_cd.html
39 US EPA (2005). Review of the National Ambient Air Quality Standards for Paniculate Matter: Policy Assessment
40 of Scientific and Technical Information, O AQPS Staff Paper. Research Triangle Park, NC 27711: Office
41 of Air Quality Planning and Standards; report no. EPA EPA-452/R-05-005a. December 2005. Available:
42 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_sp.html
PRELIMINARYDRAFT-September 2009 3-122
Do Not Quote or Cite
-------
1 US EPA (2006). Responses to Significant Comments on the 2006 Proposed Rule on the National Ambient Air
2 Quality Standards for Paniculate Matter. US Environmental Protection Agency. September 2006.
3 Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_rc.html.
4 US EPA (2008a). Integrated Review Plan for the National Ambient Air Quality Standards fro Paniculate Matter.
5 National Center for Environmental Assessment and Office of Air Quality Planning and Standards, U.S.
6 Environmental Protection Agency, Research Triangle Park, NC. Report No. EPA 452/R-08-004. March
7 2008. Available at:
8 http://www.epa.gov/ttn/naaqs/standards/pm/data/2008_03_final_integrated_review_plan.pdf
9 US EPA (2008b). Ambient Air Quality Monitoring and Health Research: Summary of April 16-17, 2008 Workshop
10 to Discuss Key Issues. U.S. Environmental Protection Agency Office of Research and Development and
11 Air Quality and Assessment Division Health and Environmental Impacts Division, Office of Air Quality
12 Planning and Standards, Office of Air and Radiation, Research Triangle Park, NC. Report No. EPA 452/S-
13 08-001. December 2008. Available: http://www.epa.gov/ord/npd/pdfs/FINAL-April-2008-AQ-Health-
14 Research-Workshop-Summary-Dec-2008.pdf.
15 US EPA (2009a). Integrated Science Assessment for Paniculate Matter: Second External Review Draft. National
16 Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
17 Research Triangle Park, NC. EPA/600/R-08/139B. July 2009. Available:
18 http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_2007_isa.html.
19 Vedal S; Rich K; Brauer M; White R; Petkau J. (2004). Air pollution and cardiac arrhythmias in patients with
20 implantable cardiovascular defibrillators. Inhal Toxicol, 16: 353-362.
21 Villeneuve PJ; Burnett RT; Shi Y; Krewski D; Goldberg MS; Hertzman C; Chen Y; Brook J. (2003). A timeseries
22 study of air pollution, socioeconomic status, and mortality in Vancouver, Canada. J Expo Sci Environ
23 Epidemiol, 13: 427-435.
24 Von Klot S; Peters A; Aalto P; Bellander T; Berglind N; D'Ippoliti D; Elosua R; Hermann A; Kulmala M; Lanki T;
25 Lowel H; Pekkanen J; Picciotto S; Sunyer J; Forastiere F; Health Effects of Particles on Susceptible
26 Subpopulations (HEAPSS) Study Group. (2005). Ambient air pollution is associated with increased risk of
27 hospital cardiac readmissions of myocardial infarction survivors in five European cities. Circulation, 112:
28 3073-3079.
29 Wellenius GA; Coull B A; Godleski JJ; Koutrakis P; Okabe K; Savage ST. (2003). Inhalation of concentrated
3 0 ambient air particles exacerbates myocardial ischemia in conscious dogs. Environ Health Perspect, 111:
31 402-408.
32 Wellenius GA; Bateson TF; Mittleman MA; Schwartz J. (2005). Paniculate air pollution and the rate of
3 3 hospitalization for congestive heart failure among medicare beneficiaries in Pittsburgh, Pennsylvania. Am J
34 Epidemiol, 161: 1030-1036.
3 5 Wellenius GA; Coull BA; Batalha JRF; Diaz EA; Lawrence J; Godleski JJ. (2006). Effects of ambient particles and
3 6 carbon monoxide on supraventricular arrhythmias in a rat model of myocardial infarction. Inhal Toxicol,
37 18: 1077-1082.
3 8 Wellenius GA; Yeh GY; Coull BA; Suh HH; Phillips RS; Mittleman MA. (2007). Effects of ambient air pollution
39 on functional status in patients with chronic congestive heart failure: a repeated-measures study ADD
40 CITE.
41 Wilson WE; Mar TF; Koenig JQ. (2007). Influence of exposure error and effect modification by socioeconomic
42 status on the association of acute cardiovascular mortality with paniculate matter in Phoenix. J Expo Sci
43 Environ Epidemiol, 17: Sll.
PRELIMINARYDRAFT-September 2009 3-123
Do Not Quote or Cite
-------
1 Wright RO; Baccarelli A. (2007). Metals and neurotoxicology. J Nutr, 137: 2809-13.
2 Yang Q; Chen Y; Krewski D; Shi Y; Burnett RT; McGrail KM. (2004). Association between paniculate air
3 pollution and first hospital admission for childhood respiratory illness in Vancouver, Canada. Arch Environ
4 Occup Health, 59: 14-21.
5 Yanosky JD; Schwartz J; Suh HH. (2008). Associations between measures of socioeconomic position and chronic
6 nitrogen dioxide exposure in Worcester, Massachusetts., 71: 1593-1602.
7 Yauk C; Polyzos A; Rowan-Carroll A; Somers CM; Godschalk RW; Van Schooten FJ; Berndt ML; Pogribny IP;
8 Koturbash I; Williams A; Douglas GR; Kovalchuk O. (2008). Germ-line mutations, DNA damage, and
9 global hypermethylation in mice exposed to paniculate air pollution in an urban/industrial location., 105:
10 605-610.
11 Yeatts K; Svendsen E; Creason J; Alexis N; Herbst M; Scott J; Kupper L; Williams R; Neas L; Cascio W; Devlin
12 RB; Peden DB. (2007). Coarse paniculate matter (PM25-10) affects heart rate variability, blood lipids, and
13 circulating eosinophils in adults with asthma. Environ Health Perspect, 115: 709-714.
14 Yue W; Schneider A; Stolzel M; Ruckerl R; Cyrys J; Pan X; Zareba W; Koenig W; Wichmann HE; Peters A.
15 (2007). Ambient source-specific particles are associated with prolonged repolarization and increased levels
16 of inflammation in male coronary artery disease patients., 621: 50-60.
17 Zabel M; Klingenheben T; Franz MR; Hohnloser SH. (1998). Assessment of QT Dispersion for Prediction of
18 Mortality or Arrhythmic Events After Myocardial Infarction : Results of a Prospective, Long-term Follow-
19 up Study., 97: 2543-2550.
20 Zanobetti A; Schwartz J. (2002). Cardiovascular damage by airborne particles: are diabetics more susceptible?
21 Epidemiology, 13: 588-592.
22 Zanobetti A; Schwartz J. (2003). Airborne particles and hospital admissions for heart and lung disease. In: Revised
23 analyses of time-series studies of air pollution and health .HEI Special Report.
24 Zanobetti A; Canner MJ; Stone PH; Schwartz J; Sher D; Eagan-Bengston E; Gates KA; Hartley LH; Suh H; Gold
25 DR. (2004). Ambient pollution and blood pressure in cardiac rehabilitation patients., 110: 2184-2189.
26 Zanobetti A; Schwartz J. (2005). The effect of paniculate air pollution on emergency admissions for myocardial
27 infarction: a multicity case-crossover analysis. Environ Health Perspect, 113: 978-982.
28 Zanobetti A; Schwartz J. (2007). Paniculate air pollution, progression, and survival after myocardial infarction.
29 Environ Health Perspect, 115: 769-775.
30 Zanobetti A; Bind MAC; Schwartz J. (2008). Paniculate air pollution and survival in a COPD cohort. Environ
31 Health Perspect, 7:48.
32 Zanobetti A; Schwartz J. (2009). The effect of fine and coarse paniculate air pollution on mortality: A national
33 analysis. Environ Health Perspect, 117: 1-40.(2008).
34 Zeger S; Dominici F; McDermott A; Samet J. (2008). Mortality in the Medicare population and chronic exposure to
35 fine paniculate air pollution in urban centers (2000-2005). Environ Health Perspect, 116: 1614.
36 Zeka A; Sullivan JR; Vokonas PS; Sparrow D; Schwartz J. (2006). Inflammatory markers and paniculate air
37 pollution: characterizing the pathway to disease. Int J Epidemiol, 35: 1347-1354.
PRELIMINARYDRAFT-September 2009 3-124
Do Not Quote or Cite
-------
4 CHARACTERIZATION OF HEALTH RISKS
We have developed a preliminary annotated outline for chapter 4 of the Policy
Assessment. Chapter 4 will present a concise overview of the scope, technical approach, results,
and limitations/uncertainties of the PM NAAQS risk assessment. In discussing the risk estimates
in this chapter, focus will be placed on providing key observations that are most relevant to the
policy assessment discussion covered in chapter 5. The organization of this chapter is similar to
that of the draft Risk Assessment document (US EPA, 2009b) and we have included references
to those sections in the draft Risk Assessment that will serve as the basis for specific sections in
the outline below. Note, also, that this annotated outline is preliminary and the organization and
content may change as we begin to develop the draft chapter.
4.1 INTRODUCTION
(Briefly set up the risk assessment in terms of its place within the overall PM NAAQS review,
including goals of the assessment, overview of risk assessment from previous review, and
process for developing approach for current risk assessment - RA sections 1.1 and 1.2).
4.2 SCOPE
4.2.1 Key design elements
(Highlights the key aspects of this risk assessment and takes note of modifications to the
approach from the previous review RA section 2.4).
4.2.2 Alternative sets of standards included in the risk assessment
(Provides the set of alternative standards as well as a brief overview of the rationale behind
selection of those sets of standards for inclusion in the assessment - RA section 2.5).
4.3 METHODS USED IN THE URBAN CASE STUDY ANALYSIS
4.3.1 General approach
(Provides concise summary of the risk assessment approach - i.e., epi-based risk assessment
focused on estimating incidence of specific health endpoints related to PM2.5 exposure. Clarifies
that it is urban study area-based with exposure being characterized using ambient monitors as a
surrogate for actual exposures. Also clarifies that risk is estimated in excess of lowest measured
level (LML) or policy-relevant background (PRB) concentrations, depending on category of
endpoint. Introduces approach of providing a core set of risk estimates that receive greater focus
along with a set of additional/ supplemental risk estimates generated using alternative inputs and
sensitivity analysis methods - RA section 3.1).
PRELIMINARY DRAFT-September 2009 4-1
Do Not Quote or Cite
-------
4.3.2 Air quality inputs
(Describes composite monitor approach and rollback methods used to simulate just meeting
current and alternative sets of standards - RA section 3.2).
4.3.3 Selection of model inputs
(Provides brief overview of key aspects of the risk assessment model, including: selection of
study areas, selection of health effects endpoints, choice of epi studies to use in deriving
concentration-response functions, and specification of baseline incidence rates - RA sections 3.3
and 3.4).
4.3.4 Addressing uncertainty and variability
(Briefly describes approach used in addressing variability and uncertainty - including qualitative
and quantitative elements. Includes discussion of sensitivity analysis and use of results to help
identify key sources of uncertainty and to provide additional set of reasonable risk estimates to
use in augmenting core risk estimates - RA section 3.5).
4.4 METHODS USED IN THE REPRESENTATIVENESS ANALYSIS
(Provides overview of approach used, including limitations and use of estimates to inform
assessment of degree to which the 15 urban study areas are representative of populations in the
U.S with regard to PM exposure and potentially, risk - RA section 4.4).
4.5 METHODS USED IN THE NATIONAL-SCALE PM-RELATED MORTALITY
ANALYSIS
(Provides overview of approach used, including limitations of the analysis - RA section 5.2).
4.6 RESULTS
4.6.1 Core risk estimates
(Provides brief summary of risk estimates and key observations regarding the core risk estimates
-RA section 4.5.1).
4.6.1.1 Recent conditions
4.6.1.2 Current set of standards
4.6.1.3 Alternative sets of standards
(Divided into discussion of (a) results for set of alternative standards focusing on annual levels:
13/35 and 12/35 and (b) set of alternative standards focusing on both annual and daily levels:
13/30 and 12/25)
PRELIMINARYDRAFT-September 2009 4-2
Do Not Quote or Cite
-------
4.6.2 Sensitivity analysis results
4.6.2.1 Key sources of uncertainty
(Discusses key sources of uncertainty - including the qualitative assessment of uncertainty as
well as the single- and multi-factor sensitivity analyses completed - RA sections 4.3.1 and 4.3.2).
4.6.2.2 Additional set of reasonable risk estimates
(Provides observations regarding range of uncertainty resulting from consideration of alternative
set of reasonable risk estimates generated as part of the single- and multi-factor sensitivity
analysis. Provides observations regarding potential range of risk that might be associated with
consideration of key sources of uncertainty associated with the risk assessment - RA section
4.5.2).
4.6.3 Representativeness analysis
(Provides brief summary and observations related to the representativeness analysis of the 15
urban study areas based on consideration of national distribution of PM risk-related attributes -
RA section 4.5.3)
4.6.4 National-scale PMi.s-related mortality analysis
(Provides brief summary and observations related to the national-scale PM2.5-related mortality
analysis, including discussion of both the national-scale estimate of mortality as well as
observations regarding how well the 15 urban study areas capture the range of county-level
PM2.s-related mortality estimated for the U.S. - RA sections 4.5.4 and 5.3).
PRELIMINARYDRAFT-September 2009
Do Not Quote or Cite
-------
1 5 STAFF CONCLUSIONS ON PRIMARY PM NAAQS
2 5.1 INTRODUCTION
3 5.2 APPROACH
4 5.3 FINE PARTICLE STANDARDS
5 5.3.1 Adequacy of Current PM2.5 Standards
6 5.3.1.1 Evidence-based Considerations
7 5.3.1.2 Risk-based Considerations
8 5.3.1.3 Summary
9 5.3.2 Indicators
10 5.3.3 Averaging Times
11 5.3.4 Alternative PM2.s Standards to Address Health Effects Related to Long-term
12 Exposure
13 5.3.4.1 Evidence-based Considerations
14 5.3.4.2 Risk-based Considerations
15 5.3.4.3 Summary
16 5.3.5 Alternative PM2.5 Standards to Address Health Effects Related to Short-term
17 Exposure
18 5.3.5.1 Evidence-based Considerations
19 5.3.5.2 Risk-based Considerations
20 5.3.5.3 Summary
21 5.3.6 Alternative Forms for Annual and 24-hour PM2.s Standards
22 5.3.6.1 Form of Annual Standard
23 5.3.6.2 Form of 24-Hour Standard
24 5.3.7 Summary of Staff Conclusions on Primary PM2.5 NAAQS
PRELIMINARY DRAFT - September 2009 5-1
Do NOT Quote or Cite
-------
1 5.4 THORACIC COARSE PARTICLE STANDARD
2 5.4.1 Adequacy of Current PMi0 Standards
3 5.4.2 Indicators
4 5.4.3 Averaging Time
5 5.4.4 Alternative Standards to Address Health Effects Related to Short-term
6 Exposure
7 5.4.5 Summary of Staff Conclusions
8 5.5 SUMMARY OF KEY UNCERTAINTIES AND RESEARCH
9 RECOMMENDATIONS RELATED TO SETTING PRIMARY PM STANDARDS
10 5.6 REFERENCES
PRELIMINARY DRAFT - September 2009 5-2
Do NOT Quote or Cite
-------
1 6 POLICY-RELEVANT ASSESSMENT OF WELFARE EFFECTS
2 EVIDENCE
3 6.1 INTRODUCTION
4 This chapter assesses key policy-relevant information on known and anticipated adverse
5 effects associated with the presence of PM in the ambient air. It draws upon the most relevant
6 information contained in the second draft ISA (EPA 2009a), as well as other significant reports
7 referenced therein. The welfare effects to be considered in this review of the secondary PM
8 NAAQS include effects on visibility (section 6.2), climate (section 6.3), vegetation and
9 ecosystems (section 6.4), and materials (section 6.5). In the final PA document, this chapter will
10 include a summary of the relevant scientific information for each effects category and a staff
11 assessment of whether the available information is sufficient to be considered as the basis for
12 secondary standards distinct from primary standards for PM. Staff conclusions and
13 recommendations regarding alternative secondary PM standards for consideration by the
14 Administrator will be presented in Chapter 7.
15 It is important to note that discussions of PM-related effects on visibility, vegetation and
16 ecosystems, and climate change processes in Chapter 9 of the second draft ISA build upon and
17 include by reference extensive information from several other significant scientific reviews of
18 these topics. Though relatively little work on urban visibility effects, including the optical and
19 human perceptual aspects of atmospheric visibility, has occurred over the last decade or more,
20 these topics have been the subjects of numerous earlier investigations that have been summarized
21 in detail elsewhere (U.S. EPA, 1979; Latimer and Ireson, 1980; Middleton, 1952; Tombach and
22 McDonald, 2004; Trijonis et al., 1990; Watson et al., 2002), including past criteria documents on
23 PM, SO2 and NOX (U.S. EPA, 1982; U.S. EPA, 1993; U.S. EPA, 2004,
24 The two principal sources for material in section 6.3 on climate are Chapter 2, "Changes
25 in Atmospheric Constituents and in Radiative Forcing," (Forster et al., 2007) in the
26 comprehensive Working Group I report in the Fourth Assessment Report (AR4) from the
27 Intergovernmental Panel on Climate Change (IPCC), Climate Change 2007: The Physical
28 Science Basis (Intergovernmental Panel on Climate Change, 2007), hereafter IPCC AR4; and the
29 U.S. Climate Change Science Program (CCSP) Synthesis and Assessment Product 2.3,
30 "Atmospheric Aerosol Properties and Climate Impacts," by Chin et al. (CCSP 2009), hereafter
31 CCSP SAP2.3. The EPA is a constituent agency member of the U.S. federated CCSP along with
32 NOAA and NASA. The CCSP SAP2.3 document incorporated significant sections from EPA
33 data and reports related particularly to U.S. emissions and measurements. Sections from each of
34 these recent comprehensive reports are included here in their entirety or as amended as noted
35 where they represent the most thorough summary of the climate effects of aerosols. (In the
PRELIMINARY DRAFT - September 2009 6-1
Do Note Quote or Cite
-------
1 sections included from IPCC AR4 and CCSP SAP2.3, 'aerosols' is more frequently used than
2 "PM" and that word is retained.)
3 6.2 EFFECTS ON VISIBILITY
4 6.2.1 2006 PM NAAQS Review
5 In the review of the secondary PM NAAQS completed in 2006, EPA took into account
6 that the Regional Haze Program61, implemented under sections 169 A and 169B of the CAA, was
7 established to address all human-caused visibility impairment in Class I areas defined by the
8 Clean Air Act to include national parks and similar natural settings.
9 Recognizing that efforts were underway under that program, EPA focused the 2006 PM
10 NAAQS review on visibility impairment primarily in urban areas. The EPA evaluated the levels
11 of visibility impairment occurring in urban areas and assessed available information on public
12 preferences regarding acceptability of PM-related urban visibility impairment.
13 At that time, EPA's focus continued to remain on particle mass and EPA staff determined
14 that size-fractionated particle mass, rather than particle composition, remained the most
15 appropriate approach for addressing PM-related urban visibility effects. EPA recognized that
16 PM composition and relative humidity are important factors in the relationship between light
17 extinction (a measure of visibility) and PM mass concentration, and that a national standard
18 should provide comparable levels of visibility protection across the country. EPA's assessment
19 of PM and meteorological data from 161 cities showed that the least variation in the relationship
20 of light extinction to PM mass concentration was for afternoon periods when low relative
21 humidity conditions generally prevail (Schmidt, et al., 2005).
22 EPA proposed to revise the secondary standards by making them identical to the suite of
23 proposed primary standards for fine and coarse particles, providing protection against PM-related
24 public welfare effects including visibility impairment, effects on vegetation and ecosystems, and
25 materials damage and soiling (71 FR 2620). EPA also solicited comment on adding a new sub-
26 daily PM2.s secondary standard to address visibility impairment in urban areas. CASAC
27 provided additional advice to EPA in a letter to the Administrator requesting reconsideration of
28 CASAC's recommendations for both the primary and secondary PM2.5 standards as well as
29 standards for thoracic coarse particles (Henderson, 2006). With regard to the secondary,
30 CASAC reaffirmed "... the recommendation of Agency staff regarding a separate secondary fine
31 particle standard to protect visibility The CASAC wishes to emphasize that continuing to rely
32 on primary standards to protect against all PM-related adverse environmental and welfare effects
61 See http://www.epa.gov/air/visibilitv/program.html for more information on EPA's Regional Haze Program.
PRELIMINARY DRAFT - September 2009 6-2
Do Note Quote or Cite
-------
1 assures neglect, and will allow substantial continued degradation, of visual air quality over large
2 areas of the country" (Henderson, 2006).
3 On September 21, 2006, EPA announced its final decisions to revise the secondary
4 NAAQS for PM to provide increased protection of public welfare by making them identical to
5 the revised primary standards (71 FR 61144, October 17, 2006). Specifically, with regard to the
6 secondary standards for fine particles, EPA revised the level of the 24-hour PM2.5 standard to 35
7 ng/m3, retained the level of the annual PM2.5 annual standard at 15 |ig/m3, and revised the form
8 of the annual PM2.5 standard by narrowing the constraints on the optional use of spatial
9 averaging. With regard to the secondary standards for coarse particles, EPA retained PMio as the
10 indicator for purposes of regulating the coarse fraction of PMio (referred to as thoracic coarse
11 particles or coarse-fraction particles; generally including particles with a nominal mean
12 aerodynamic diameter greater than 2.5 jim and less than or equal to 10 jim, or PMio-2.s). EPA
13 retained the 24-hour PMio standard at 150 |ig/m3 and revoked the annual PMio standard because
14 available evidence generally did not suggest a link between long-term exposure to current
15 ambient levels of coarse particles and health or welfare effects.
16 Several parties filed petitions for review following promulgation of the revised PM
17 NAAQS in 2006. These petitions addressed a number of issues, including that of the setting of
18 the secondary PM2.5 standards identical to the primary standards. On judicial review the court
19 remanded the secondary PM2 5 NAAQS to EPA because the Agency failed to adequately explain
20 why setting the standards equal to the primary PM2.5 standards provided the required protection
21 from visibility impairment. In particular, the Agency failed to identify a target level of visibility
22 impairment that would be requisite, and failed to address regional differences in humidity-related
23 effects on visibility American Farm Bureau Federation v. EPA, 559 F. 3d 512, (D.C. Cir. 2009).
24 EPA intends to address the court remand in the context of this review.
25 6.2.2 Scope of Current Review
26 EPA staff continues to believe that a focus on urban area visibility is appropriate. In
27 articulating a rationale for this conclusion, we have reviewed the information contained in the
28 second draft ISA and find the following information compelling: 1) PM levels in urban areas are
29 often in excess of those of the surrounding region since urban haze typically includes both
30 regional and local contributions (US EPA, 2009a; sections 9.2.3.3 and 9.23.4), suggesting the
31 potential for higher levels of PM-induced visibility impairment in urban areas; 2) the existence of
32 numerous urban visibility protection programs and goals demonstrates that urban visual air
33 quality (VAQ) is noticed and an important value to urban residents (US EPA, 2009a; section
34 9.2.4), and 3) the existence of large urban populations suggests that potentially more people are
35 routinely affected by poor VAQ than in rural areas.
PRELIMINARY DRAFT - September 2009 6-3
Do Note Quote or Cite
-------
1 Therefore, the focus of the remainder of section 6.2 is on the role of ambient PM in the
2 impairment of visibility conditions in urban areas, drawing upon the most relevant information
3 contained in the second draft ISA (Chapter 9), as well as significant reports on the science of
4 visibility referenced therein, the results of assessments contained in the draft risk document titled
5 "Urban Focused Visibility Assessment" (EPA 2009c), and information presented in section 2.8
6 of this document. In particular, this section includes the following new information:
7
8 • Summary of results from a reanalysis of existing urban visibility preference studies
9 and discussion of how these results were used to inform the selection of an
10 appropriate range of candidate protection levels (CPLs) for further evaluation.
11
12 • Summary of results from analyses conducted in support of the Urban Focused
13 Visibility Assessment (UFVA) which show that urban PM light extinction varies
14 considerably during daylight hours as a result of varying PM component
15 concentrations including absorbed water due to variations in relative humidity.
16
17 Further, as described previously in the UFVA (EPA, 2009c), EPA staff is continuing to
18 focus assessments in this document in terms of an alternative indicator for PM visibility
19 impairment, i.e. PM light extinction, instead of the traditional PM2.5 mass concentration. The
20 2005 Staff Paper discussed the use of a four-hour afternoon PM2.s standard, where the underlying
21 rational was that the generally lower afternoon relative humidity tended to produce a more
22 uniform relationship between light extinction and PM2.5 mass concentration throughout the
23 country, therefore providing a more uniform level of visibility protection nationwide. However,
24 this more uniform level of visibility protection was limited to the afternoon hours of the day
25 when relative humidity and visibility impairment are typically the lowest. However, visibility
26 conditions can be the poorest when relative humidity levels are the highest. Thus, from a public
27 welfare perspective, greater protection from visibility impairment is needed during the times
28 when humidity is high. In that regard, morning relative humidity conditions, which are often
29 generally higher in the Eastern US and coastal areas than in the West, can cause the same PM
30 concentrations to produce much higher PM-related visibility impairment in those regions than in
31 areas with lower morning relative humidity. Thus, using a PM2.s mass indicator could
32 potentially result in unequal protection from visibility impairment at the national scale. Unlike
33 PM mass concentration, which is determined by removing the liquid water from the PM prior to
34 measuring it, PM light extinction can be measured at ambient humidity conditions so that it
35 includes the enhanced light extinction resulting from the liquid water that is associated with the
36 hygroscopic PM components in the atmosphere. PM light extinction, like PM mass
37 concentration, is a measurable physical characteristic of ambient PM. Thus, we believe that use
PRELIMINARY DRAFT - September 2009 6-4
Do Note Quote or Cite
-------
1 of PM light extinction as the indictor for a secondary PM NAAQS is appropriate and more
2 directly related to the visibility welfare effect.
3 6.2.3 Nature of Urban Focused PM-related Visibility Effects
4 Section 2.8 above, as well as section 9.2 of the ISA and chapter 1 of the UFVA, provides
5 useful descriptions of the known science regarding relationship between ambient PM and
6 visibility impacts and much of that discussion will not be repeated here. However, it is important
7 to discuss some aspects of this information as it has bearing on the unique considerations
8 required for defining what would be adequate protection from visibility impairment in urban
9 areas.
10 Light extinction, which is the optical characteristic of the atmosphere that best determines
11 the impact on perceived visibility, is the loss of light per unit of distance that occurs when light is
12 either scattered or absorbed. Particulate matter and gases can both scatter and absorb light.
13 Light scattering by gases (e.g. nitrogen, oxygen, etc.) that comprise the atmosphere (also known
14 as Rayleigh or clean-air scattering) is related to the density of the air, which is sufficiently
15 constant with elevation that it can be considered a known constant value for any location. Thus,
16 light extinction is approximated as the sum of PM light extinction (including both scattering and
17 absorption) plus Rayleigh light scattering, where the former characterizes the PM contribution to
18 visibility impacts and the latter is taken to be a time invariant constant depending only on
19 elevation above sea level. Therefore, for the purposes of this discussion, the term light extinction
20 includes both PM and Rayleigh sources of light extinction, while the term PM light extinction
21 represents total light extinction minus the Rayleigh light extinction, which is assumed in this
22 assessment to be 10 Mm"1.
23 While visibility impairment may be caused by either natural or manmade conditions (or
24 both), VAQ is used here to refer to the visibility effects caused solely by air quality conditions,
25 so for example it excludes the reduced visibility caused by fog. The term "urban visibility" is
26 used to refer to VAQ throughout a city or metropolitan area. Urban visibility includes the VAQ
27 conditions in all locations that people experience in their daily lives, including scenes such as
28 residential streets and neighborhood parks, commercial and industrial areas, highway and
29 commuting corridors, central downtown areas, and views from elevated locations providing a
30 broad overlook of the metropolitan area. Thus urban visibility includes VAQ conditions in major
31 cities and smaller towns and encompasses all the VAQ an individual resident sees on a regular
32 basis. Visibility conditions in urban and suburban locations are therefore distinct from visibility
33 in rural or wilderness settings such as the Class I areas. Daytime visibility impairment differs
34 from nighttime visibility impairment in a number of important ways. Light sources and ambient
35 conditions are typically five to seven orders of magnitude dimmer at night than in sunlight.
PRELIMINARY DRAFT - September 2009 6-5
Do Note Quote or Cite
-------
1 Moonlight, like sunlight, introduces light throughout an observer's sight path at a constant angle.
2 On the other hand, dim starlight emanates from all over the celestial hemisphere, while artificial
3 lights are concentrated in cities and illuminate the atmosphere from below. In addition to the
4 physical science differences between daytime and nighttime visibility, the human psychophysical
5 response at night is expected to differ, and there is much less known about the aspects of
6 nighttime visibility that are valued by the public (e.g. relative importance of views of celestial
7 objects versus those of artificially illuminated urban skyline/lights). For these reason, at this
8 time EPA is limiting its assessment to daytime visibility (EPA, 2009a, section 9.2.2).
9 If the relative composition of the PM as a function of particle size were held constant and
10 the ambient relative humidity is low (RH<60%) or held constant, there would be a direct linear
11 relationship between PM light extinction and PM mass concentration. Since the PM composition
12 and ambient relative humidity vary in both time and space, there is in general no simple one-to-
13 one correspondence between PM concentration and PM light extinction. These variations result
14 in significant uncertainty (i.e. greater than a factor of two) in predicting PM light extinction from
15 PM mass concentration. However the PM light extinction can be more accurately estimated
16 from PM composition and relative humidity data, using a simple algorithm with assumed light
17 extinction efficiencies for each of the major PM species and water growth factors for the
18 hygroscopic species. In addition, PM light extinction can be accurately determined by direct
19 measurements using an integrating nephelometer to measure particle light scattering and an
20 aethalometer or similar instrument to measure particle light absorption.
21 6.2.4 Public Welfare Impacts of Urban Visibility Impairment
22 The Clean Air Act §302(h) defines public welfare to include the effects of air pollution
23 on ".. .visibility, ... and personal comfort and wellbeing." Though good visibility conditions in
24 Class I areas (e.g., national parks and wilderness areas) have long been recognized as important
25 to the public welfare (see discussions in EPA (2004; 2005) and Chestnut and Dennis (1997),
26 more recently, visibility conditions in urban areas have also been recognized to contribute to the
27 public welfare.
28 Visibility impairment is the most noticeable effect of fine particles present in the
29 atmosphere. Visibility has direct significance to people's enjoyment of daily activities and their
30 overall wellbeing. Ambient particle pollution degrades the visual appearance and perceived
31 color of distant objects to an observer and reduces the range at which they can be distinguished
32 from the background. This aesthetic effect of ambient particle pollution is well known and
33 understood and includes the aesthetic and wellbeing benefits of better visibility, improved road
34 and air safety, and enhanced recreation in activities like hiking and bicycling. Due to the
35 subjective nature of aesthetics, people's preferences with respect to visibility are difficult to
PRELIMINARY DRAFT - September 2009 6-6
Do Note Quote or Cite
-------
1 express or quantify, but people have expressed in many different ways that they enjoy and value
2 a clear view. This desire for good daytime VAQ in urban areas has been clearly demonstrated by
3 the weight of available evidence, both on the basis of numerous studies in the social sciences
4 including quantitative preference and valuation studies, and by the existence of numerous
5 programs and goals regarding establishing and protecting urban VAQ (EPA, 2009a).
6 In addition, visibility impairment has also been shown to have an effect on human
7 psychology, creating an additional public welfare effect on "personal comfort and well being".
8 In this context, reduced VAQ is considered an environmental stressor (Campbell, 1983) that is
9 associated with heightened amounts of anxiety, tension, anger, fatigue, depression, and feelings
10 of helplessness (Evans et al., 1987; Zeidner and Shechter, 1988). Though the relationship
11 between impaired VAQ and mental health is poorly understood, studies have shown that reduced
12 VAQ affects people's behavior, including reductions in outdoor activities, increased hostility and
13 aggressive behavior (Evans et al., 1982; Cunningham, 1979; Jones and Bogat, 1978; Rotton et
14 al., 1979) and greater emergency calls associated with psychiatric disturbances (Rotton and Frey,
15 1982). There is also evidence from a recent Canadian study that some Aboriginal peoples value
16 good VAQ because many of their cultural practices and traditions depend upon their ability to
17 see certain viewscapes (Carlson, 2009); it is unclear to what extent this welfare effect might be
18 applicable for Native Americans in the U. S. Because there is insufficient information on the
19 relationship between these latter effects and ambient PM concentrations, they do not usefully
20 inform the development of policy options and selection of appropriate indictors, averaging times,
21 forms, and levels for a secondary standard at this time. However, the staff recommends that they
22 be considered qualitatively when projecting additional benefits that could accrue to the public
23 welfare from improved PM VAQ.
24 6.2.5 Reanalysis of Public Perception Studies
25 Chapter 2 of the UFVA document (EPA, 2009c) describes in detail the reanalysis that
26 was performed by EPA (with contractor support) to better characterize the available data on
27 visibility public preference and its usefulness and limitations in serving as a basis for informing
28 the selection of appropriate urban visibility candidate protection levels (i.e., CPLs). Much of the
29 material presented in this section comes directly from chapter 2 of the UFVA.
30 The reanalysis effort included urban visibility preference survey studies from three
31 western urban locations, plus a pair of smaller focus studies conducted in Washington, D.C. that
32 were designed to explore and further develop urban visibility survey instruments. With the
33 exception of the most recent Washington, DC study (Smith and Howell, 2009), all of these
34 studies were also available at the time of the 2006 review. The first urban visibility study
35 conducted was in Denver, Colorado (Ely et al., 1991), which developed the basic survey method
PRELIMINARY DRAFT - September 2009 6-7
Do Note Quote or Cite
-------
1 used in all the subsequent studies. The other western study areas included the lower Fraser River
2 valley near Vancouver, British Columbia (BC), Canada (Pryor, 1996), and Phoenix, Arizona
3 (BBC Research & Consulting, 2003). At the beginning of the 2006 review, EPA sponsored a
4 pilot focus group study in Washington, D.C. (Abt Associates Inc., 2001). In the current review,
5 in response to an EPA request for public comment on the Scope and Methods Plan (74 FR
6 11580, March 18, 2009), Dr. Anne Smith of Charles River Associates (CRA) International on
7 behalf of the Utility Air Regulatory Group, provided comments (Smith, 2009) which included a
8 discussion of the results from a new Washington, D.C. focus group study that had been
9 conducted for the purpose of testing the reproducibility and appropriateness of methods and
10 approaches employed in the EPA pilot study (Smith and Howell, 2009). In total, 852 individuals
11 participated in the studies conducted for these four urban areas.
12 As stated above, the approaches used in these studies are similar and are all derived from
13 the method first developed for the Denver urban visibility study. For example, each of the
14 studies reviewed in this assessment investigates the common question, "What level of visibility
15 degradation is acceptable?" The specific definition of acceptable is largely left to each individual
16 survey participant, allowing each to identify his/her own preferences. Further, each individual
17 responds to a series of questions answered while viewing a set of images of various urban VAQ
18 conditions.
19 The specific materials and methods used in each study vary, however, making direct
20 comparison of the study results between the different studies challenging. Key differences
21 between study methods and approaches include:
22
23 • use of WinHaze (a significant technical advance in the method of presenting VAQ
24 conditions),
25
26 • number of participants in each study,
27
28 • representativeness of participants for the general population of the relevant
29 metropolitan area, and
30
31 • specific wording used to frame the questions used in the group interview process.
32
33 Although the differences between the methods used in the urban visibility preference
34 studies are significant, it is possible to examine the results of the studies to identify overall trends
35 in the study findings. Figure 6-1 (Figure 2-14 in the UFVA) presents a graphical summary of the
36 results of the studies in the four cities. A number of decisions affecting the visual display of the
37 data were made in the interest of clarity. First, Figure 6-1 omits the 9:00 a.m. photograph results
38 from the Denver study. These photographs were determined to be inconsistently rated when
PRELIMINARY DRAFT - September 2009 6-8
Do Note Quote or Cite
-------
1 compared with results from other times of day that had similar light extinction levels (see
2 discussion in section 2.2 and Table 2-1 in UFVA). Second, the Chilliwack and Abbotsford
3 photographs appear as a single set of data for the BC study. The assumptions (process by which
4 those data were combined/ the approach taken to combine the data) that were made in order to
5 combine the data are described in section 2.3 in the UFVA). Third, the results from 2001 and
6 2009 (Test 1) studies of VAQ preferences in Washington, DC are presented as a single combined
7 set of data, because Test 1 appears to successfully replicate the 2001 study, thereby producing
8 comparable results. Tests 2 and 3 results from the 2009 Washington, DC study are not included
9 on Figure 6-1. Those tests are not considered comparable with the other studies because, unlike
10 the other studies, they used an altered range of VAQ conditions that did not represent the actual
11 range of VAQ conditions that occur in the study area.
12 Figure 6-1 contains lines at 20 dv and 30 dv that identify a range where the 50%
13 acceptance criteria occur across all four of the urban preference study areas. Out of the 114 data
14 points shown, only one photograph (or image) with a VAQ below 20 dv was rated as acceptable
15 by less than 50% of the participants who rated that photograph.62 Similarly, only one image with
16 a VAQ above 30 dv was rated acceptable by more than 50% of the participants who viewed it.63
17 This figure shows that, while there is a high degree of similarity between the preferences found
18 in each study, there may be important differences in VAQ preferences across the four cities as
19 well. For example, the Denver study identified preferences for a relatively good level of VAQ;
20 the 50% criteria occur between 17.7 dv and 20.3 dv. In Washington, DC, however, the 50%
21 criteria separation occurs at a substantially worse level of VAQ, between 27 dv and 31 dv.
62 Only 47% of the BC participants rated a 19.2 DV photograph as acceptable.
63 In the 2001 Washington, DC study, a 30.9 DV image was used as a repeated slide. The first time it was shown
56% of the participants rated it as acceptable, and 11% rated it as acceptable the second time it was shown. The
same VAQ level was rated as acceptable by 42% of the participants in the 2009 study (Test 1).
PRELIMINARY DRAFT - September 2009 6-9
Do Note Quote or Cite
-------
1 Figure 6-1. Summary of results of urban visibility studies in four North American cities,
2 showing the identified range of the 50% acceptance criteria.
ting Acceptable
n c
3 C
; &
% Participants Ra
3 C
? 5
f.
• ;***<** r. * : '
» o o **» * * 1
«, i
0 * * 1
• ** '!.
^» * \
^ * J
* **•
* 1 *
;. • • i •
\ *,\ r .
« i * * *
00 « $» 4*»«*» ° 0
5 10 15 20 25 30 35 40 4
Deciview
o Denver » Phoenix » BC
• Washington Lower Upper
3
4 There are several hypotheses that may explain why the results of these studies differ.
5 First, as mentioned earlier, the use of photographs versus WinHaze-generated images may play a
6 significant role in preference studies, perhaps introducing bias (such as suggested by the
7 responses to the 9:00 a.m. Denver photographs) as well as variability. Use of photographs from
8 different days and times of day that rely on associated ambient measurements of light extinction
9 to characterize their VAQ level introduces two types of uncertainty. The intrinsic appearance of
10 the scene can change due to the changing shadow pattern and cloud conditions, and spatial
11 variations in air quality can result in ambient light extinction measurements not being
12 representative of the sight-path-averaged light extinction. WinHaze has neither of these sources
13 of uncertainty because the same base photograph is used (i.e. no intrinsic change in scene
14 appearance between images) and the modeled haze that is displayed in the photograph is based
15 on uniform sight-path-averaged light extinction throughout the scene. This modeled uniform
16 light extinction, however, may not represent non-uniform haze conditions that may occur at a
17 study location during certain times of day or PM conditions (e.g., layered morning haze; point
18 source plumes entering site path). The impact of the application of a uniform versus non-
19 uniform haze layer to a scene has not been investigated.
20 Second, variation in the degree of representativeness of the participants and the sizes of
21 the participant samples involved may also be important factors. The small sample size and fairly
22 uniform population of respondents is a plausible explanation for the noisiness of the combined
PRELIMINARY DRAFT - September 2009 6-10
Do Note Quote or Cite
-------
1 Washington, DC, results (35 participants, including 26 from a single consulting firm and 10 of
2 those from a different city) compared with the larger and more representative population of
3 responders from Phoenix (385 participants, carefully selected to be representative
4 A third hypothesis explored by Smith and Howell (2009) is that the range of VAQ
5 images presented in the survey may influence the results. Though this hypothesis appears to be
6 borne out by Smith and Howell's results for Washington, DC, it seems an unlikely explanation
7 for the differences in results between the four urban preference studies. For example the Denver
8 study included photographs with the haziest conditions among the four studies, but resulted in
9 the lowest haze condition for the 50th percentile preference ratings among the four, not the
10 highest as might be expected if the range of haze levels were a significant factor influencing the
11 results of preference studies.
12 A fourth major hypothesis is that urban visibility preferences may differ by location, and
13 the differences may arise from inherent differences in the cityscape scene in each difference. The
14 key evidence to suggest this hypothesis is that the apparent differences between the Denver
15 results (which found the 50% acceptance criteria occurred in the best VAQ levels among the four
16 cities) and the Washington, DC results (which found the 50% acceptance criteria occurred at the
17 worst VAQ levels among the four cities). This hypothesis suggests that these results may occur
18 because the cityscape of Denver includes clearly visible snow-covered mountains in the distance,
19 while the prominent features of the Washington, DC cityscape are buildings relatively nearby
20 with only modest changes in elevation.
21 Finally, perhaps of significant importance is that the perceived sensitivity of individual
22 scenes to changes in light extinction can be quite different. As in the fourth hypothesis, this may
23 in part explain why the Denver study scene, with its long distance to the mountain backdrop,
24 resulted a preference for the best VAQ level with a 50% criteria value between 17.7 and 20.3 dv,
25 while in Washington, DC, the 50% criteria separation occurs at a substantially worse level of
26 VAQ, between 27 and 31 dv from Abt Associates Inc. (2001) and Smith and Howell (2009) Test
27 1. The distinction between the last two hypotheses are that the earlier one speaks to the
28 desirability of seeing distant mountains versus this hypothesis where its ability to perceive haze
29 at lower light extinction levels. Additional studies, including directly comparable studies using
30 similar methods in diverse cities, are necessary to gain further understanding of preferences for
31 urban visibility.
32 Based on the composite results and the effective range of 50th percentile acceptability
33 across the four urban preference studies shown in Figure 6-1, candidate protective levels (CPL)
34 have been selected in a range from 20 dv to 30 dv (74 Mm"1 to 201 Mm"1) for the purpose of
35 comparing to current and projected conditions in the assessment in chapters 3 and 4 of this
36 document. A midpoint of 25 dv (122 Mm"1) was also selected for use in the assessment. These
PRELIMINARY DRAFT - September 2009 6-11
Do Note Quote or Cite
-------
1 three values provide a low, middle, and high set of light extinction conditions that are used in
2 subsequent sections of the UFVA to provisionally define daylight hours with urban haze
3 conditions that have been judged unacceptable by the participants of these preference studies.
4 6.2.6 Considerations in Selecting Alternative Light Extinction Scenarios for
5 Assessment
6 In considering what alternative standard scenarios were appropriate to assess in the
7 context of providing urban visibility protection, EPA staff concluded that the available urban
8 public preference study information, as analyzed and characterized in the UFVA, though limited,
9 is sufficient to inform the selection of an appropriate range of light extinction CPLs. While less
10 information is available to support the selection of an appropriate averaging time and form, we
11 make the following observations regarding the nature of the public welfare impact anticipated to
12 occur in urban areas at both current and alternative levels of urban PM air quality that can
13 usefully inform their selection.
14 First, we observe that the strength of an adverse visibility impact is not necessarily linked
15 with any specific time duration since it only takes an instant for a person to see a visually
16 impaired scene. In urban areas, more so than recreational areas, a person is typically
17 experiencing visibility conditions intermittently and incidentally as a part of their daily activities
18 (e.g. during commutes to work or school), and thus, may have few opportunities to revise his/her
19 perception of the quality of that urban view again in any given day. Thus, it is possible that the
20 effect on wellbeing of impaired VAQ from a short-term exposure of limited duration may endure
21 well beyond the period of exposure. In addition, we have considered that though the occurrence
22 of multiple hours of poor VAQ in a single day would undoubtedly increase the number of
23 individuals with potential welfare impacts, because of the intermittent nature of exposure to
24 VAQ it would not necessarily increase the intensity of the impact on the personal wellbeing of
25 those already affected. As a result, though not specifically addressed in the preference studies
26 analyzed, we selected a one hour averaging time as that most closely linked to the short-term
27 nature of the daily impact as compared to longer periods, and are considering that the one hour
28 maximum value for a day is a good index of short term exposure that may have longer term
29 impacts on wellbeing.
30 Second, because the nature of the welfare effect is one of emotion (feelings of wellbeing)
31 and not directly related to a physical health outcome, we believe that it is not necessary to
32 eliminate all such exposures and that some number of hours/days with poor VAQ can reasonably
33 be accepted as being protective of public welfare. As a result, we have selected the 90th and 95th
34 percentiles per year averaged over a three year period as an appropriate range of frequencies to
35 be evaluated in conjunction with the range of PM light extinction CPLs.
PRELIMINARY DRAFT - September 2009 6-12
Do Note Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
6.2.7 Characterization of Current Urban Visibility Conditions
The following information is excerpted from Chapter 3 of the UFVA (EPA 2009c) and
the reader is referred to that document for further detail of the process and approaches used in
that assessment.
The goal of this part of the assessment was to estimate hourly PM light extinction for the
daylight hours for a reasonably representative number of days in each year of the 2005-2007
period, to allow the application of statistical forms based on three years of data. For efficiency,
the visibility assessment uses the same 15 urban study areas selected for the health risk
assessment. These areas are listed in Table 6-1, along with their 2005-2007 annual and 24-hour
PM2.5 design values and accompanying percentages of days with daily maximum 1-hour light
extinction levels greater than each of the three CPLs. These results will be discussed in
conjunction with those of Figure 6-2 below.
Table 6-1. PMi.s design values and percent of days with maximum light extinction
exceeding three candidate protection levels (CPLs) for 15 Urban study areas.
Study Area
Tacoma
Fresno
Los Angeles
Phoenix
Salt Lake City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit
Pittsburgh
Baltimore
Philadelphia
New York
Design Levels
(2005- 2007)
Annual
10.2
20.3
19.6
12.6
11.6
11.9
15.8
16.5
18.7
16.2
17.2
16.5
15.6
15.0
15.9
24-Hour
43
69
55
32
55
25
31
39
44
35
43
43
37
38
42
Number of
Assessment
Days
110
324
302
86
306
274
149
294
350
295
141
284
187
145
228
Percent of Days Exceeding
the Light Extinction CPL
74 Mm'1
68
80
92
59
61
86
89
100
96
95
91
93
88
95
91
122 Mm1
36
51
80
13
24
53
58
86
80
80
79
70
65
76
70
201 Mm1
9
24
49
3
9
14
21
55
52
34
57
43
38
46
46
16
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
6-13
-------
1 The results depicted in Table 6-1 were generated through the following steps. First,
2 CMAQ model output for 2004 was used to develop realistic diurnal variations of the major PM2.5
3 components as monthly averaged variation for each of the 15 urban monitoring sites, which were
4 then used in a later step of the process that generated hourly estimates from the 24-hour
5 speciation monitoring data. Another step in the process accounted for hourly variations in PM
6 measured by continuous monitoring at each site. Hourly relative humidity (RH) data needed for
7 the calculation of light extinction for each study area's primary monitoring site were obtained
8 from either an air monitoring station reporting such data to AQS, or a National Weather Service
9 (NWS) station. Lastly, the original IMPROVE light extinction algorithm was used to estimate
10 total hourly light extinction using the RH data and the hourly PM composition estimates. This
11 total light extinction (expressed simply as light extinction in the remainder of the document),
12 includes the term for Rayleigh (i.e. clean air) scattering. In the next draft UFVA, results will
13 also be expressed in terms of PM light extinction (i.e. light extinction resulting only from
14 ambient PM without the Rayleigh term included).
15 Figure 6-2 (Figure 3-8 of UFVA) presents box-and-whisker plots to illustrate the
16 distributions of the estimates of daylight 1-hour reconstructed light extinction levels in each area
17 across the three year (2005-2007) period. The distribution of the individual 1-hour values and
18 the daily maximum 1-hour values are both shown. The horizontal dashed lines in the plots
19 represent the low, middle and high light extinction CPLs as discussed above. These benchmarks
20 expressed as 74, 122, and 201 Mm"1, also correspond to the deciview benchmark values of 20,
21 25, and 30 dv for light extinction identified above. As seen in the comparable PM2.5 box and
22 whisker plots, the hourly light extinction values tends to be higher in the Eastern urban areas and
23 lower in the non-California Western urban areas. The distributions of maximum daily light
24 extinction values are higher, as expected, than for all hours.
25 Both Figure 6-2 and Table 6-1 indicate that all 15 urban areas have daily maximum
26 hourly light extinctions that exceed even the highest CPL some of the time. Again the non-
27 California Western urban locations have the lowest frequency of maximum hourly light
28 extinction with values in excess of the high CPL less than 10% of the time. Except for the two
29 Texas and the non-California Western urban areas, all of the other urban areas exceed the high
30 CPL about a quarter to a half of the time. Based on these estimated maximum hourly light
31 extinction values, all 15 of the urban areas exceed the low CPL for -60% to 100% of the days.
32 In the last review of the secondary PM NAAQS, the pattern of light extinction during the
33 day was of particular interest. To illustrate the distributions of 1-hour light extinction levels in
34 specific daylight hours, Figure 6-3 shows the distributions of 1-hour light extinction across the
35 entire three-year study period, individually for the study areas. These plots show that high light
36 extinction can occur during any of the daylight hours, though for most of these urban areas the
PRELIMINARY DRAFT - September 2009 6-14
Do Note Quote or Cite
-------
1
2
3
4
5
6
early morning hours have the highest light extinction. Urban areas without a prominent
proclivity for early morning high light extinction include Phoenix, AZ; Salt Lake City, UT;
Tacoma, WA; Fresno, CA; and Philadelphia, PA.
Figure 6-2. Distributions of estimated daylight 1-hour light extinction (top) and maximum
daily daylight 1-hour light extinction (bottom) in each study area
1236 3643 33S3
Hourly Extinction (Daylight Hours)
3457 3106 1652 3273 3930 3262 !587 3179 2095 1618 2515
" ' - __
t t t t
110 324 302
Daily Mawtmym Extirtctiop {Daylight Hours)
308 274 149 294 350 295 141
i J
i r
284 187 145 228
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
6-15
-------
1 Figure 6-3. Light extinction box and whisker plots by time of day for each of the 15 urban
2 areas.
Phoaafo.AZ { Pittsburgh. PA ] Salt take City, UT [__ Ss tours, II } Tadorna, _WA ]
Las Angeles CA
Bi'm)ngi*amJ AL
4 6.2.8 Projected Light Extinction When Meeting Alternative NAAQS Scenarios
5 Material for this section is taken from chapter 4 of the UFVA. Eight different alternative
6 NAAQS scenarios were evaluated. These scenarios included six light extinction cases and two
7 PM2 5 NAAQS cases. The six light extinction cases include the maximum hourly light extinction
8 compared to each of the three CPLs at two annual percentile values (90th and 95th) averaged over
9 a three year period. The two PM2.5 cases include the current PM2.5 NAAQS (i.e. 15 ug/m3 annual
10 and 35 ug/m3 24-hour) and an alternative case (12 ug/m3 annual and 25 ug/m3 24-hour), together
11 representing the highest and lowest alternative NAAQS scenarios considered in the health risk
12 assessment. Both PM2.5 NAAQS cases used the three year averaged 98th annual percentile
13 values. In order to model "what if'conditions for just meeting each alternative, two different
14 approaches were necessary: one for generating the alternate hourly light extinction values, and
15 one for the PM2 5 scenarios. A more detailed description on these two distinct approaches can be
16 found in sections 4.1.4 and 4.2.2 of the UFVA. A brief summary of each approach is provided
17 below.
18 The approach used for the light extinction scenarios involve using the estimates of
19 current hourly conditions generated for the 15 urban areas as described in the previous section,
20 and applying a process similar to the proportional rollback that was used in the health risk
21 assessment modeling of "what if conditions in several previous PM NAAQS reviews. The
PRELIMINARY DRAFT - September 2009 6-16
Do Note Quote or Cite
-------
1 Policy Relevant Background (PRB) is subtracted from the data that exceed the scenario
2 specifications, and then the percent reduction in the non-PRB portion of the data needed to scale
3 the total to meet the scenario conditions is determined. The percent reduction is applied to all of
4 the non-PRB data, and then PRB is added back to produce the values that meet the scenario
5 conditions. The light extinction scenarios use hourly daily maximum light extinction values. In
6 contrast, for the PM2.5 mass concentration scenarios approach, the reduction factors are
7 determined for the 24-hour periods (as done for the health risk assessment) but applied to each
8 hour and the non-PRB portion of each PM2.5 component, which is then used to estimate the
9 hourly light extinction using the IMPROVE algorithm and measured relative humidity.
10 The results of this process include displays of the required percent reductions in non-PRB
11 light extinction and PM2 5 values . Figure 6-4 (Figure 4-13 in the UFVA) contains two box and
12 whisker plots that show the distributions of estimated daily maximum daylight 1-hour light
13 extinction values for one of the six light extinction cases (122 Mm"1, 90th percentile) and for the
14 current PM2.5 NAAQS (i.e. 15 ug/m3 - annual and 35 ug/m3 - 24-hour) case. The principle
15 insight provided by the first plot is that the rollback process successfully generated hourly
16 maximum light extinction distribution for each urban area with the 90th percentile (i.e. the upper
17 whisker) is very near 122 Mm"1. Plots for the other light extinction scenarios (in UFVA,
18 Appendix F) are similar in demonstrating that the rollback process was successful.
19 The second plot in Figure 6-4 shows that uniformly applied PM mass concentration based
20 NAAQS provides uneven daylight 1-hour maximum light extinction values across the 15 urban
21 areas and that nearly every urban areas has a 90th percentile maximum 1-hour light extinction
22 well above the upper CPL (201Mm"1). Another way to assess the effectiveness of the PM2 5
23 based NAAQS scenarios in affecting the maximum 1-hour light extinction is to compare their
24 90th and 95th percentile design values with the range of CPLs. Table 6-2 (Table 4-11 in the
25 UFVA) contains these design values for the two PM2 5 NAAQS scenarios. As seen in the box
26 and whisker plot, the 90th percentile design values exceed the high CPL (201Mm"1) at all of the
27 Eastern urban areas and for Los Angeles while meeting the current PM2 5 NAAQS. The table
28 also shows that the lower PM2 5 NAAQS scenario results in design values that exceed the 90th
29 percentile high CPL at six or seven of the Eastern urban areas.
PRELIMINARY DRAFT - September 2009 6-17
Do Note Quote or Cite
-------
1
2
3
4
5
9
10
Figure 6-4. Distributions of daily maximum daylight 1-hour total light extinction under
two "just meeting" secondary NAAQS scenarios
(a) Secondary NAAQS based on measured total light extinction with a level of 122 Mm"1 and a
90th percentile form
ExtRollbackDailyMaxNAAQS122Pctl90
o a
X A
7 (b) Secondary NAAQS of 15 |ig/m3 for the annual average and 35 |ig/m3 for the 98th percentile
8 24-hour average
PMRollbackDailyMaxCasel NAAQS
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
6-18
-------
1 Table 6-2. Light extinction design values for "just meeting" secondary NAAQS scenarios
2 based on PM2.s mass
Annual/1-hour
PM2 5 NAAQS
City Name
Tacoma, WA
Fresno, CA
Los Angeles, CA
Phoenix, AZ
Salt Lake City, UT
Dallas, TX
Houston, TX
St. Louis, IL
Birmingham, AL
Atlanta, GA
Detroit, MI
Pittsburgh, PA
Baltimore, MD
Philadelphia, PA
New York, NY
15ug/m3/35ug/m3
90th %tile
Design Value
(Mm1)
188
183
221
117*
126
213*
224
384
355
249
364
Recalculating
419
377
377
95th %tile
Design Value
(Mm1)
228
238
311
154*
174
262*
261
477
476
271
520
Recalculating
459
403
450
12ug/m3/25ug/m3
90th %tile Design
Value (Mm1)
139
139
175
107
98
200
182
311
268
197
264
Recalculating
308
273
274
95th %tile
Design Value
(Mm1)
165
179
261
145
133
245
211
383
369
218
376
Recalculating
335
296
325
4 * Phoenix and Dallas meet 15/35 under current conditions, so these entries are the same as for current conditions.
5
6 Table 6-3 contains the percent reductions required in the non-PRB light extinction that
7 include both PM2.5 and PMio-2.5 contributions for the six light extinction NAAQS scenarios, and
8 the percent required in the non-PM2.s mass concentration to meet the two PM2.5 NAAQS
9 scenarios. A number of caveats and uncertainties associated with these approaches should be
10 noted. First, with regard to the rollback approach, there is an implied linearity between
11 emissions and concentrations and subsequent visibility effects, which is not true in general.
12 Second, there are uncertainties in each step of the assessment including the estimation of the
13 hourly PM composition and resulting light extinction, and estimation of the PRB, all of which
14 feed into the reduction calculations. In spite of these issues the reductions are thought to be a
15 useful metric for comparing the challenges of these scenarios at the 15 urban areas.
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
6-19
-------
1
2
Table 6-3. Percentage reductions required in the non-PRB light extinction or PM2.5 mass
to "just meet" the NAAQS scenarios.
Total Light
Extinction
Level (Mm1)
Percentile
Form
Annual PM2.s
Level (ug/m3)
24-hour PM2.5
Level (ug/m3)
Urban Area
Tacoma
Fresno
Los Angeles
Phoenix
Salt Lake
City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit
Pittsburgh
Baltimore
Philadelphia
New York
NAAQS Scenarios Based on Maximum Daily 1-
hour Daylight Total Light Extinction, Average
of Nth Percentile Value Over Three Years
201
90th
201
95th
122
90th
122
95th
74
90th
74
95th
Percentage Reduction Required in
Non-PRB Total Light Extinction
13
38
40
0
0
7
16
55
56
28
58
55
58
54
59
29
53
56
0
23
26
29
64
67
33
71
61
61
57
65
51
65
66
0
36
49
53
75
75
60
77
75
76
74
77
59
73
75
23
56
59
60
80
82
63
84
78
79
76
80
74
82
82
43
65
75
76
87
87
79
88
87
88
86
88
77
86
87
59
76
79
79
90
90
80
91
88
89
88
90
NAAQS Scenarios
Based on Annual and
24-hour PM2.5
15
35
12
25
Percentage Reduction
Required in Non-PRB
PM25
19
45
37
0
37
0
6
10
22
8
19
Recalc
6
8
17
43
61
55
22
56
7
27
37
45
30
43
ulating
33
35
41
4
5
6
7
The percent reductions in Table 6-3 provide another way to consider the differences
between the PM2.5 mass and PM light extinction. These are apparent when comparing the
reductions required in most of the Western urban areas to those in the East. For example
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
6-20
-------
1 comparing the least restrictive of the PM light extinction scenarios (201 Mm"1, 90th percentile) to
2 the current NAAQS (15 ug/m3- annual, 35 ug/m3- 24-hour), shows similar magnitude reductions
3 in the typically drier, less sulfate/nitrate PM dominated Western urban areas, but quite different
4 magnitudes for the more humid, higher sulfate/nitrate PM urban areas of the East. Also note that
5 the lower of the two PM2.5 scenarios (12 ug/m3- annual, 25 ug/m3- 24-hour) has less challenging
6 reduction values in the East than the highest of the PM light extinction scenarios (201, Mm"1 90th
7 percentile), implying less protection of Eastern urban visibility, yet it has much higher reduction
8 values for the Western urban areas implying a greater level of visibility protection.
9 Table 6-4 shows the percentage of days over the multiyear period (i.e. 3 years except at
10 Phoenix and Houston, which have 2 years) with maximum 1-hour daylight light extinction above
11 the three CPL values for each of the eight NAAQS scenario cases. Notice that the percentage of
12 days above the CPL for the corresponding CPL light extinction scenario is very nearly the
13 complimentary fraction to the percentile form of the scenario showing again that the rollback
14 approach is numerically consistent (e.g. the fraction of days above 74 Mm"1 for the 74 Mm"1, 90th
15 percentile scenario for Tacoma is 9% which is nearly 10%). More informative are the
16 percentages of days above each of the CPL for alternative scenarios. For example the 201 Mm"1,
17 90th percentile scenario would result in over half of the days (56%) at Tacoma having maximum
18 1-hour daily light extinction above 74 Mm"1 and a quarter of the days above 122 Mm"1.
19 Comparisons of these percentages allows a rough indication of how the two scenarios of a
20 NAAQS based on PM2.5 mass compare to the other six light extinction based scenarios in terms
21 of protecting visual air quality. Notice that even the lower PM2.5 NAAQS scenario (12 ug/m3-
22 annual, 25 ug/m3- 24-hour) would permit projected 1-hour maximum daily light extinction above
23 the least restrictive CPL (201 Mm"1) more that 10% of the time for most of the Eastern urban
24 areas (Dallas, Houston and Atlanta have values near 10%), while the percent of maximum hourly
25 days for the Western urban areas are all less than 10%.
PRELIMINARY DRAFT - September 2009 6-21
Do Note Quote or Cite
-------
1
2
4
Table 6-4. Percentage of days with a maximum 1-hour daylight light extinction above the three CPL for each of the eight
NAAQS scenario cases.
Mm-1 Level
Percentile
Form
Annual/24-
hour
Area
Tacoma
Fresno
Los
Angeles
Phoenix
Salt Lake
City
Dallas
Houston
St. Louis
Birmingham
Atlanta
Detroit
Pittsburgh
Baltimore
Philadelphia
New York
Average
Days above 74 Mm"1 (Percent)
201 201 122 122 74 74
90 95 90 95 90 95
15/35 12/25
Percentage of days
56 46 26 23 11 9 55 39
54 42 29 21 9 3 48 33
81 74 53 24 8 5 82 76
88 66 50 30 9 5 46 42
54 32 25 14 11 4 26 19
77 66 43 30 9 4 80 77
75 68 45 32 14 9 81 69
75 62 40 28 10 6 98 94
67 56 42 26 13 6 90 82
86 84 57 50 9 7 91 85
66 50 40 18 11 5 82 79
55 52 30 25 10 6 RecalcJating Recalculating
55 50 25 22 10 8 82 71
67 65 33 28 8 6 90 81
57 50 27 23 10 6 80 70
68 58 38 26 10 6 73 65
Days above 122 Mm"1 (Percent)
201 201 122 122 74 74
90 95 90 95 90 95
15/35 12/25
Percentage of days
25 18 8 6 21 23 11
27 19 10 4 11 23 14
49 20 8 521 59 28
46 27 9 5 2 1 7 6
24 13 10 4 3 2 11 6
42 26 11 4 21 46 42
40 30 11 7 11 48 31
36 25 10 6 21 79 66
36 22 11 5 21 66 52
50 42 10 7 00 68 49
36 16 11 5 10 70 59
27 22 106 0 0 Recalculating Recalculathg
23 22 10 6 00 60 46
29 26 8 6 00 70 54
26 21 11 6 11 60 45
34 23 10 5 11 49 36
Days above 201 Mm"1 (Percent)
201 201 122 122 74 74
90 95 90 95 90 95
15/35 12/25
Percentage of days
8620006 3
10 4 1 1 0 0 8 2
8 5 3 1 0 0 12 7
9532102 2
10 4 3 2 1 0 3 2
11 5 2 1 0 0 13 11
11 6 1 1 0 0 14 7
10 6 2 1 00 44 32
11 5 2 1 00 37 21
105 0 0 00 22 10
11 5 1 0 00 45 22
105 0 0 00 Recalcubting Recalculating
9 6 0 0 0 0 34 22
8 6 0 0 0 0 37 21
12 5 2 1 0 0 30 19
10 5 2 1 0 0 22 13
30
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
6-22
-------
1 6.2.9 Summary
2 Section 6.2 above focuses on the role of ambient PM in the impairment of visibility
3 conditions in urban areas, drawing upon the most relevant information contained in the second
4 draft ISA (Chapter 9), as well as significant reports on the science of visibility referenced
5 therein, the results of assessments contained in the draft UFVA (EPA, 2009c), and information
6 presented in section 2.8 of this document. The purpose of this section is to distill and describe the
7 most significant policy-relevant results from a reanalysis of urban visibility preference studies
8 conducted in four urban areas and an assessment of recent and projected urban visual air quality
9 conditions in 15 urban area case studies conducted in support of the secondary standard review
10 and described in the UFVA document (EPA, 2009c).
11 Further, as described previously in the UFVA (EPA, 2009c), EPA staff chose to include
12 an alternative indicator for PM visibility impairment, i.e. PM light extinction, in addition to the
13 traditional PM2.5 mass concentration indicator. Unlike PM mass concentration, which is
14 determined by removing the liquid water from the PM prior to measuring it, PM light extinction
15 can be measured at ambient humidity conditions so that it includes the enhanced light extinction
16 resulting from the liquid water that is associated with the hygroscopic PM components in the
17 atmosphere. PM light extinction, like PM mass concentration, is a measurable physical
18 characteristic of ambient PM. Thus, we believe that use of PM light extinction as the indictor for
19 a secondary PM NAAQS is appropriate and more directly related to the visibility welfare effect.
20 As discussed above, though the materials and methods used in the urban visibility
21 preference studies for the four urban areas vary from study to study, they are all similar to and
22 derived from the method first developed for the Denver urban visibility study. Taking into
23 account the associated caveats and uncertainties (see section 6.2.5), we thus concluded that this
24 information, though limited, can be used to explore overall findings and trends. Further, when
25 viewed together, the composite results presented in Figure 6-1 usefully inform the selection of an
26 appropriate range of VAQ CPLs for urban areas to consider in further analyses. Using the 50th
27 percentile acceptability criteria, we selected a range of CPLs from 20 dv to 30 dv (74 Mm"1 to
28 201 Mm"1) with a midpoint of 25 dv (122 Mm"1). These three values provide low, middle, and
29 high values of urban haze (light extinction) conditions that have been judged unacceptable by at
30 least 50% of the preference study participants for use in comparisons with recent and projected
31 PM air quality conditions.
32 Though not specifically evaluated in the preference studies analyzed, we also considered
33 a one hour averaging time as that most closely linked to the likely short-term nature of the daily
34 impact that is likely to occur in urban areas. In developing this view, we have considered that
35 the strength of an adverse visibility impact is not necessarily linked with any specific time
PRELIMINARY DRAFT - September 2009 6-23
Do Note Quote or Cite
-------
1 duration since it only takes an instant for a person to see a visually impaired scene. In urban
2 areas, more so than recreational areas, a person is typically experiencing visibility conditions
3 intermittently and incidentally as a part of their daily activities (e.g. during commutes to work or
4 school), so that longer-term, multiple hour exposures are less likely. As a result, any given
5 exposed individual may have few opportunities to revise his/her perception of the quality of that
6 urban view again in any given day. In such a case, the effect of a short-term exposure to
7 impaired VAQ on an individual's wellbeing may endure well beyond the period of exposure. In
8 addition, though the occurrence of multiple hours of poor VAQ in a single day would
9 undoubtedly increase the number of individuals exposed, there is no information at this time to
10 suggest that individuals exposed to multiple hours would necessarily experience an increased
11 intensity of the impact on their personal wellbeing than those exposed for shorter periods.
12 Further, in considering an appropriate range of frequencies to be evaluated in conjunction
13 with the range of light extinction CPLs, we took into account that the nature of the welfare effect
14 is one of emotion (feelings of wellbeing) and not directly related to a physical health outcome.
15 Thus, we believe that it is not necessary to eliminate all such exposures and that some number of
16 hours/days with poor VAQ can reasonably be accepted as being protective of public welfare. As
17 a result, we selected the 90th and 95th percentiles per year averaged over a three year period as
18 an appropriate range of frequencies for further consideration. By comparing these values to
19 current and projected PM2 5 and light extinction conditions, we were able to evaluate to what
20 extent these levels are being exceeded or met across the 15 urban area case studies.
21 Figure 6-2 and Table 6-1 indicate that all 15 urban areas have daily maximum hourly
22 light extinctions that exceed even the highest CPL some of the time. The non-California
23 Western urban locations have the lowest frequency of maximum hourly light extinction with
24 values in excess of the high CPL less than 10% of the time. Except for the two Texas and the
25 non-California Western urban areas, all of the other urban areas exceed that high CPL about a
26 quarter to a half of the time. Based on these estimated maximum hourly light extinction
27 estimates, all 15 of the urban areas exceed the low CPL for -60% to 100% of the days.
28 Figure 6-3 shows the distributions of 1-hour light extinction levels in specific daylight
29 hours, across the entire three-year (2005-2007) study period, individually for the study areas.
30 These plots show that high light extinction can occur during any of the daylight hours, though for
31 most of these urban areas the early morning hours have the highest light extinction. Urban areas
32 without a prominent tendency for high early morning light extinction include Phoenix, AZ; Salt
33 Lake City, UT; Tacoma, WA; Fresno, CA; and Philadelphia, PA. This pattern of light extinction
34 during the day, i.e. frequent occurrence of the worst visibility conditions during morning hours,
35 would seem to argue against selecting a standard form based on only afternoon hours that
PRELIMINARY DRAFT - September 2009 6-24
Do Note Quote or Cite
-------
1 typically have the least visibility impairment as did the 4-hour sub-daily form described in the
2 2006 review.
3 Additional analyses (described in section 6.2.8 above and in greater detail in chapter 4 of
4 the UFVA) conducted by EPA staff generated hourly light extinction values to produce PM
5 conditions that would meet eight different NAAQS scenarios: six light extinction cases and two
6 PM2.5 NAAQS cases. The six light extinction cases included the maximum hourly light
7 extinction compared to each of the three CPLs at two annual percentile values (90th and 95th)
8 averaged over a three year period. The two PM2.5 cases included the current PM2.5 NAAQS
9 (i.e. 15 ug/m3 annual and 35 ug/m3, 24-hour) and a lower case (12 ug/m3, annual and 25 ug/m3,
10 24-hour), with both PM2.5 NAAQS cases using the three year averaged 98th annual percentile
11 values. Figure 6-4 shows the distributions of estimated daily maximum daylight 1-hour light
12 extinction resulting from one of the six light extinction cases (122 Mm"1, 90th percentile) and
13 from the current PM2.5 NAAQS (i.e. 15 ug/m3 - annual and 35 ug/m3 - 24-hour). These results
14 show that even the lower of the two PM2.5 mass NAAQS scenarios would permit projected 1-
15 hour maximum daily light extinction above the least restrictive CPL (201 Mm"1) more that 10%
16 of the time for most of the Eastern urban areas (Dallas, Houston and Atlanta have values near
17 10%), while the percent of maximum hourly days for the Western urban areas are all less than
18 10%.
19 Given the results above, it is clear that urban visibility conditions across the 15 cities are
20 currently being impacted at ambient levels of PM within the range of CPLs for light extinction
21 derived from the reanalysis of available urban preference studies. The external review draft PA
22 to be released later this year will include additional analyses of how VAQ would be expected to
23 change under a scenario of just meeting the current PM2.s secondary standards and under
24 additional just meeting potential alternative standard scenarios.
25 6.3 EFFECTS ON CLIMATE
26 Information and conclusions about what is currently known about the role of PM in
27 climate is summarized in chapter 9 of the second draft ISA (US EPA, 2009a). The ISA
28 concludes; "that a causal relationship exists between PM and effects on climate, including both
29 direct effects on radiative forcing and indirect effects that involve cloud feedbacks that influence
30 precipitation formation and cloud lifetimes" (section 9.3.10). Material from the climate section
31 of the second draft ISA is principally drawn from the U.S. Climate Change Science Program
32 Synthesis and Assessment Product 2.3, "Atmospheric Aerosol Properties and Climate Impacts,"
33 by Chin et al., (CCSP SAP 2.3 2009) (CCSP 2009) and chapter 2, "Changes in Atmospheric
34 Constituents and in Radiative Forcing," (Forster et al., 2007) in the comprehensive Working
35 Group I report in the Fourth Assessment Report (AR4) from the Intergovernmental Panel on
PRELIMINARY DRAFT - September 2009 6-25
Do Note Quote or Cite
-------
1 Climate Change (IPCC), Climate Change 2007: The Physical Science Basis. This section of the
2 PA summarizes and synthesizes the policy-relevant science in the ISA for the purpose of helping
3 to inform consideration of climate aspects in the review of the secondary PM NAAQS.
4 Atmospheric PM (referred to as aerosols in the remainder of this section to be consistent
5 with the ISA) affects multiple aspects of climate. These include absorbing and scattering of
6 incoming solar radiation, alterations in terrestrial radiation, effects on the hydrological cycle, and
7 changes in cloud properties. Major aerosol components that contribute to climate processes
8 include sulfate, organic carbon (OC), black carbon (BC) nitrate and mineral dusts. There is a
9 considerable ongoing research effort focused on understanding aerosol contributions to changes
10 in global mean temperature and precipitation patterns. The Climate Change Research Initiative
11 identified research on atmospheric concentrations and effects of aerosols as a high research
12 priority (National Research Council, 2001) and the IPCC 2007 Summary for Policymakers states
13 that anthropogenic contributions to aerosols remain the dominant uncertainty in radiative forcing
14 (IPCC 2007). The current state of the science of climate alterations attributed to PM is in
15 constant flux as a result of continually updated information.
16 6.3.1 Aerosol Direct Effects
17 Aerosols have both direct and indirect effects on climate processes. The direct effects are
18 the result of the same processes responsible for visibility degradation, mainly radiative scattering
19 and absorption. However, while visibility impairment is caused by particle scattering in all
20 directions, climate effects result mainly from scattering light away from earth into space, directly
21 altering the radiative balance of the Earth-atmosphere system. This reflection of solar radiation
22 back to space decreases the transmission of visible radiation to the surface of the earth and
23 results in a decrease in the heating rate of the surface and the lower atmosphere. At the same
24 time, absorption of either incoming solar radiation or outgoing terrestrial radiation by particles,
25 primarily BC, results in an increased heating rate in the lower atmosphere.
26 The direct effect of radiative scattering by atmospheric particles exerts an overall net
27 cooling of the atmosphere, while particle absorption of solar radiation leads to warming. For
28 example, the presence of SC>42" and OC particles decrease warming from sunlight by scattering
29 shortwave radiation back into space. Such a perturbation of incoming radiation by
30 anthropogenic aerosols is designated as aerosol climate forcing, which is distinguished from the
31 aerosol radiative effect of the total aerosol (natural plus anthropogenic). Global estimates of
32 aerosol direct radiative forcing (RF) were recently summarized using a combined model-based
33 estimate (Forster et al., 2007). The overall, model-derived aerosol direct RF was estimated in the
34 IPCC AR4 as -0.5 (-0.9 to -0.1) watts per square meter (W/m2), with an overall level of scientific
35 understanding of this effect as "medium low" (Forster et al., 2007) indicating a net cooling effect
PRELIMINARY DRAFT - September 2009 6-26
Do Note Quote or Cite
-------
1 in contrast to greenhouse gases (GHGs) which have a warming effect. The contribution of
2 individual aerosol components to total aerosol directive radiative forcing is more uncertain than
3 the global average. The aerosol climate forcing and radiative effect are characterized by large
4 spatial and temporal heterogeneities due to the wide variety of aerosol sources, the spatial non-
5 uniformity and intermittency of these sources, the short atmospheric lifetime of aerosols (relative
6 to that of the greenhouse gases), and processing (chemical and microphysical) that occurs in the
7 atmosphere. For example, OC can be warming (positive forcer) when deposited on or suspended
8 over a highly reflective surface such as snow or ice but, on a global average, is a negative forcer
9 in the atmosphere.
10 6.3.2 Aerosol Indirect Effects
11 In addition to these direct radiative effects, aerosols can have a number of indirect effects
12 on climate related to their physical properties. Particulates in the atmosphere indirectly affect
13 both cloud albedo (reflectivity) and cloud lifetime by modifying the cloud amount, and
14 microphysical and radiative properties. The RF due to these indirect effects (cloud albedo effect)
15 of aerosols is estimated in the IPCC AR4 to be -0.7(-1.8 to -0.3) W/m2 with the level of scientific
16 understanding of this effect as "low" (Forster et al., 2007). Aerosols act as cloud condensation
17 nuclei (CCN) for cloud formation. Increased particulates in the atmosphere available as CCN
18 with no change in moisture content of the clouds have resulted in an increase in the number and
19 decrease in the size of cloud droplets in certain clouds that can increase the albedo of the clouds
20 (the Twomey effect). Smaller particles slow the onset of precipitation and prolong cloud
21 lifetime. This effect, coupled with changes in cloud albedo, increase the reflection of solar
22 radiation back into space. The altitude of clouds also effects cloud radiative forcing. Low
23 clouds reflect incoming sunlight back to space but do not effectively trap outgoing radiation,
24 thus, cooling the planet, while higher elevation clouds reflect some sunlight but more effectively
25 can trap outgoing radiation and act to warm the planet.
26 6.3.3 Summary and Conclusions
27 The current state of the science including an in-depth consideration of technological
28 advances in the quantification and characterization of aerosols are discussed in section 9.3 of the
29 ISA. The total negative RF due to direct and indirect effects of aerosols computed from the top
30 of the atmosphere, on a global average, is estimated at -1.3 (-2.2 to -0.5) W/m2 in contrast to the
31 positive RF of +2.9 (+3.2 to +2.6) W/m2 for anthropogenic GHGs (IPCC 2007, pg. 200).
32 Although considerable progress is being made in estimating aerosol contributions to climate
33 fluctuations, significant uncertainties remain that preclude consideration of climate effects as a
34 basis for a NAAQS secondary standard. A major impediment at this time to establishing a
35 secondary standard for PM based on climate is the lack of accurate measurement of aerosol
PRELIMINARY DRAFT - September 2009 6-27
Do Note Quote or Cite
-------
1 contributions, specifically quantification of aerosol absorption and inability to separate the
2 anthropogenic component from total aerosol forcing. Section 9.3.4 of the ISA details the current
3 limitations in aerosol measurement. Most measurement studies focus on the sum of natural and
4 anthropogenic contributions under clear sky conditions, however, this scenario is simplistic when
5 effects of cloud cover and differing reflective properties of land and ocean are considered.
6 Satellite measurements do not currently have the capability to distinguish anthropogenic from
7 natural aerosols. Due to a lack of data on the vertical distribution of aerosols, above-cloud
8 aerosols, profiles of atmospheric radiative heating are poorly understood.
9 Another limitation to including consideration of climate effects of PM in the NAAQS
10 review is the spatial and temporal heterogeneity of aerosols. In regions having high
11 concentrations of anthropogenic aerosols, aerosol forcing is greater than the global average, and
12 can exceed warming by GHGs, locally reversing the sign of the forcing (ISA section 9.3.1).
13 Emissions of carbonaceous aerosols from intermittent fires and volcanic activity can further
14 complicate regional climate forcing estimates (ISA sections 9.3.7 and 9.3.8). Individual
15 components of aerosols may either be positive or negative climate forcers. When BC is deposited
16 to the surface of ice or snow, solar absorption and heating occur at the surface. This can melt
17 additional snow or ice at the surface, altering reflective properties. Airborne PM components
18 may be directly emitted or undergo a variety of physical and chemical interactions and
19 transformations. These result in changes in particle size, structure and composition which alter
20 aerosol reflective properties. Aerosols can grow in size in the atmosphere because ambient water
21 vapor condenses on individual particles a phenomenon known as hydroscopic growth.
22 Atmospheric lifetimes of individual aerosol components vary greatly confounding tracking
23 source receptor relationships.
24 Improved representation of aerosols in climate models is essential to more accurately
25 predict the role of PM in climate forcing. Section 9.3.6.7 of the ISA details current gaps in
26 assessment of aerosol components through climate modeling. The influence of aerosols on
27 climate is not yet adequately taken into account in computer predictions although considerable
28 progress in being made in this area. For example, PM components underrepresented or missing
29 from many models include nitrate aerosols and anthropogenic secondary aerosols. The modeling
30 of aerosol indirect effects and absorption are difficult due to the high level of uncertainty
31 associated with these climate factors.
32 Most climate model simulations are based on global scale scenarios. These models may
33 fail to consider the local variations in climate forcing due to emissions sources and local
34 meteorological patterns. A series of studies available since the last review examine the role of
35 aerosols on local and regional scale climate processes (ISA section 9.3.9.3). Studies on the
36 South Coast Air Basin (SCAB) in California indicate aerosols may reduce near-surface wind
PRELIMINARY DRAFT - September 2009 6-28
Do Note Quote or Cite
-------
1 speeds, which, in turn reduce evaporation rates and increase cloud lifetimes. The overall impact
2 can be a reduction in local precipitation (Jacobson and Kaufmann, 2006). Conditions in the
3 SCAB impact ecologically sensitive areas including the Sierra Nevadas. Precipitation
4 suppression due to aerosols in California (Givati and Rosenfield, 2004) and other similar studies
5 in Utah and Colorado found that orographic precipitation decreased by 15-30% downwind of
6 pollution sources. Evidence of regional-scale impacts of aerosols on meteorological conditions
7 in other regions of the U.S. are lacking.
8 The interaction of PM with clouds remains a large source of uncertainty in climate
9 estimates. The interactions of aerosols with clouds and linkages between clouds and the overall
10 climate system are complex and limit the feasibility of conducting quanitative analysis for the
11 purpose of establishing a secondary PM standard based on welfare effects on climate processes.
12 Due to the lack of confidence in measuring relative aerosol contributions, the spatial and
13 temporal heterogeneity of PM components that contribute to climate, incomplete consideration
14 of aerosol impacts in climate modeling, insufficient data on local and regional microclimate
15 variations and heterogeneity of cloud formations it is not currently feasible to conduct a
16 quantitative analysis for the purpose of informing revisions of the current NAAQS PM standard
17 based on climate. The available information provides no basis for estimating how localized
18 changes in the temporal, spatial and composition patterns of aerosols, likely to occur as a result
19 of expected future emissions of particles and precursor gases across the U.S., would affect local,
20 regional, or nationwide changes in climate. Based on these considerations, staff concludes that
21 PM effects on climate processes can play no quantitative role in considering whether any
22 revisions of the secondary PM NAAQS are warranted for this review.
23 6.4 EFFECTS ON VEGETATION AND ECOSYSTEMS
24 An integrated assessment of the policy relevant science regarding what is currently
25 known about the effects of ambient PM on ecosystems and individual components of ecosystems
26 such as vegetation, soils, water, and wildlife are discussed in chapter 9 of the PM ISA (US EPA,
27 2009a). Effects of acidifying deposition associated with particulate N and S are covered in the
28 recent Integrated Science Assessment for Oxides of Nitrogen and Sulfur-Ecological Criteria (US
29 EPA, 2008c) and will not be considered further in this section. This section of the PM PA will
30 summarize and highlight key aspects of the policy relevant information from the ISA to help
31 inform the Administrator's judgments regarding the adequacy of the secondary NAAQS standard
32 in relation to ecological endpoints.
33 The ISA concludes there is sufficient information to infer a "likely causal" relationship
34 between particulate metals and organics and a variety of effects on individual organisms and
35 ecosystems (US EPA, 2009a, section 9.4.7). This review seeks to build upon and focus this body
PRELIMINARY DRAFT - September 2009 6-29
Do Note Quote or Cite
-------
1 of science using the concept of ecosystem services to qualitatively evaluate linkages between
2 biologically adverse effects and particulate deposition. This approach is similar to that taken in
3 the Second Draft Risk and Exposure Assessment for Review of the Secondary National Ambient
4 Air Quality Standards for Oxides of Nitrogen and Oxides of Sulfur (NOxSOx REA) (US EPA,
5 2009f) in which the relationship between air quality indicators, deposition of N and S,
6 ecologically relevant indicators and effects on sensitive receptors are linked to changes in
7 ecosystem structure and services. This approach considers the benefits received from the
8 resources and processes that are supplied by ecosystems (detailed in section 6.3.2). Ecosystem
9 components (e.g. plants, soils, water, wildlife) are impacted by anthropogenic PM air pollution,
10 which may alter the services provided by the ecosystems in question. The goals of this policy
11 assessment are to identify ecological effects associated with PM deposition that can be linked to
12 ecosystem services and qualitatively evaluate ecological endpoints when possible. Keeping
13 these goals and guidelines in mind, the information is organized into the following subsections:
14 major ecosystem stressors in PM (6.3.1); ecosystem services (6.3.2); effects of PM on ecological
15 receptors (6.3.3); and summary and conclusions (6.3.4).
16 6.4.1 Major Ecosystem Stressors in PM
17 As previously discussed, PM is not a single pollutant, but a heterogeneous mixture of
18 particles differing in size, origin and chemical composition. The heterogeneity of PM exists not
19 only within individual particles or samples from individual sites, but to even a greater extent,
20 between samples from different sites. Since vegetation and other ecosystem components are
21 affected more by particulate chemistry than size fraction, exposure to a given mass concentration
22 of airborne PM may lead to widely differing plant or ecosystems responses, depending on the
23 particular mix of deposited particles. Though the chemical constitution of individual particles
24 can be strongly correlated with size, the relationship between particle size and particle
25 composition can also be quite complex, making it difficult in most cases to use particle size as a
26 surrogate for chemistry. At this time it remains to be determined as to what extent NAAQS
27 standards focused on a given size fraction would result in reductions of the ecologically relevant
28 constituents of PM for any given area.
29 A number of different chemical species found within ambient PM that can have effects
30 on ecosystem components were discussed in Chapter 9 of the PM ISA. In particular, the ISA
31 focuses on metals and to a lesser extent on organics. Organics and some metals are regulated
32 under separate statutory authorities, e.g. section 112 of the Clean Air Act. The reminder of this
33 section will focus on the effects of PM-associated metals including Cadmium (Cd), Copper (Cu),
34 Chromium (Cr), Mercury (Hg), Nickel (Ni) and Zinc (Zn) and PM organics including persistent
35 organic pollutants (POPs), polyaromatic hydrocarbons (PAHs) and polybromiated diphenyl
PRELIMINARY DRAFT - September 2009 6-30
Do Note Quote or Cite
-------
1 ethers (PBDEs) which in turn, can affect overall ecosystem structure and function. Ecological
2 effects of Lead (Pb) are covered in the Air Quality Criteria Document for Lead (US EPA, 2006).
3 6.4.2 Ecosystem Services
4 The following material on ecosystem services is adapted from the ISA section 9.4.4.1 and
5 9.4.1.2 (US EPA, 2009a). For this assessment, "ecosystem" is defined as a functional entity
6 consisting of interacting groups of living organisms and their abiotic (chemical and physical
7 environment). Because ecosystems are diverse in biota, climate geochemistry, and hydrology,
8 response to pollutant exposures can vary greatly between ecosystems. Ecosystems cover a
9 hierarchy of spatial scales and can compromise the entire globe, biomes at the continental scale,
10 or small, well-circumscribed systems such as a small pond.
11 Ecosystems have both structure and function. Structure may refer to a variety of
12 measurements including the species richness, abundance, community composition and
13 biodiversity as well as landscape attributes. Competition among and within species and their
14 tolerance to environmental stresses are key elements of survivorship. When environmental
15 conditions shift, for example, by the presence of anthropogenic air pollution, these competitive
16 relationships may change and tolerance to stress may be exceeded. Function refers to the suite of
17 processes and interaction among the ecosystem components and their environment that involve
18 nutrient and energy flow as well as other attributes including water dynamics and the flux of
19 trace gases. Plant processes including photosynthesis, nutrient uptake, respiration and C
20 allocation, are directly related to functions of energy flow and nutrient cycling. The energy
21 accumulated and stored by vegetation (via photosynthetic C capture) is available to other
22 organisms. Energy moves from one organism to another through food webs, until it is ultimately
23 released as heat. Nutrients and water can be recycled. Air pollution alters the function of
24 ecosystems when element cycles or the energy flow is altered. This alteration can also be
25 manifested in changes in the biotic composition of ecosystems. There are at least three levels of
26 ecosystem response to pollutant deposition: (1) the individual organism and its environment, (2)
27 the population and its environment, and (3) the biological community composed of many species
28 and their environment (Billings, 1978).
29 Ecosystem structure and function may be translated into ecosystem services. Ecosystem
30 services identify the varied and numerous ways that ecosystems are important to human welfare.
31 Ecosystems provide many goods and services that are of vital importance for the functioning of
32 the biosphere and provide the basis for the delivery of tangible benefits to human society. The
33 Millennium Ecosystem Assessment (MEA) defines these to include supporting, provisioning,
34 regulating and cultural services (Hassan et al., 2005):
PRELIMINARY DRAFT - September 2009 6-31
Do Note Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Supporting services are necessary for the production of all other ecosystem services.
Some examples include biomass production, production of atmospheric 62, soil
formation and retention, nutrient cycling, water cycling, and provisioning of habitat.
Biodiversity is a supporting service that is increasingly recognized to sustain many of the
goods and services that humans enjoy from ecosystems. These provide a basis for three
higher-level categories of services.
Provisioning services, such as products (Gitay et al., 2001) i.e., food (including game,
roots, seeds, nuts, and other fruit, spices, fodder), fiber (including wood, textiles), and
medicinal and cosmetic products (including aromatic plants, pigments).
Regulating services that are of paramount importance for human society such as (a) C
sequestration, (b) climate and water regulation, (c) protection from natural hazards such
as floods, avalanches, or rock-fall, (d) water and air purification, and (e) disease and pest
regulation.
Cultural services that satisfy human spiritual and aesthetic appreciation of ecosystems
and their components.
Figure 6-5. Millennium ecosystem assessment categorization of ecosystem
services and their links to human well-being (Hassan et al., 2005).
CONSTITUENTS OF WELL-BEING
Security
PERSONAL SAFETY
SECURE RESOURCE ACCESS
SECURITY FROM DISASTERS
20
21
ECOSYSTEM
Supporting
NUTRIENT CYCLING
SOIL FORMATION
PRIMARY PRODUCTION
SERVICES
Provisioning
FOOD ^
FRESHWATER •
WOOD AND FIBER
FUEL
Regulating
CLIMATE REGULATION
FLOOD REGULATION
DISEASE REGULATION W
WATER PURIFICATION
1 Cultural
AESTHETIC '
SPIRITUAL
EDUCATIONAL
RECREATIONAL
LIFE ON EARTH -
BIODIVERSITY
Basic material
for good life
ADEQUATE LIVELIHOODS
SUFFICIENT NUTRITIOUS FOOD
SHELTER
ACCESS TO GOODS
Health
STRENGTH
FEELING WELL
ACCESS TO CLEAN AIR
AND WATER
Good social relations
SOCIAL COHESION
MUTUAL RESPECT
ABILITY TO HELP OTHERS
Freedom
of choice
and action
OPPORTUNITY TO BE
ABLE TO ACHIEVE
'WHAT AN INDIVIDUAL
VALUES DOING
AND BEING
Source: Millennium Ecosystem Assessment
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
6-32
-------
1 An important consideration in evaluating biologically adverse effects of PM and linkages
2 to ecosystem services is that many of the MEA categories overlap and any one pollutant may
3 impact multiple services (Figure 6-5). For example, deposited PM may alter the composition of
4 soil-associated microbial communities, which may affect supporting services such as nutrient
5 cycling. Changes in available soil nutrients could result in alterations to provisioning services
6 such as timber yield and regulating services such as climate regulation. If enough information is
7 available, these alterations can be quantified based upon economic approaches for estimating the
8 value of ecosystem services. Valuation may be important from a policy perspective because it
9 can be used to compare the benefits of altering versus maintaining an ecosystem.
10 6.4.3 Effects of Deposited PM on Ecosystem Receptors
11 In order for any specific chemical stressor present in ambient PM to impact ecosystems,
12 it must first be removed from the atmosphere through deposition. Deposition can occur in three
13 modes: wet (rain/frozen precipitation), dry, or occult (fog, mist or cloud). The exposure pathway
14 of PM can be direct (such as deposition to a surface) or indirect (e.g., mediated through soil or
15 water processes). Once PM has deposited on or is taken up by an ecological receptor (e.g.
16 plants, soils, water, wildlife) a response, known as an ecological effect, may occur. Common
17 anthropogenic stressors and the ecological attributes they may affect are summarized in Young
18 and Sanzone (2002). Ecological effects may include, but are not limited to, species losses,
19 changes in landscape condition, altered trophic relationships, disruption of biogeochemical
20 cycles, hydrology/geomorphology, community composition, decreased growth and/or
21 reproduction, habitat degradation, and shifts in ecosystem structure and function. Some
22 ecological effects can then be quantified through measurable ecological endpoints such as
23 presence/absence of lichens, soil root mass, species richness and abundance indices.
24 6.4.3.1 Plants
25 Ecosystem services derived from plants include all of the categories (supporting,
26 provisioning, regulating, cultural) identified in the MEA (Hassan et al., 2005). As primary
27 producers, plants play a pivotal role in energy flow through ecosystems. Vegetation supports
28 other ecosystem processes by cycling nutrients through food webs and serving as a source of
29 organic material for soil formation and enrichment. Trees and plants provide food, wood, fiber,
30 and fuel for human consumption. Flora help to regulate climate by sequestering carbon dioxide
31 (CC>2), control flooding by stabilizing soils and cycling water via uptake and evapotranspiration.
32 Plants are significant in aesthetic, spiritual and recreational aspects of human interactions.
33 Particulate matter can adversely impact plants and ecosystem services provided by plants
34 by deposition to surfaces. Particulates deposited on the surfaces of leaves and needles can block
35 light, altering the radiation received by the plant. PM deposition can obstruct stomata limiting
PRELIMINARY DRAFT - September 2009 6-33
Do Note Quote or Cite
-------
1 gas exchange, damage leaf cuticles and increase plant temperatures. This level of PM
2 accumulation is typically observed near sources of heavy deposition such as smelters and mining
3 operations. Plants growing on roadsides exhibit impact damage from near-road PM deposition,
4 having higher levels of organics and heavy metals, and accumulate salt from road de-icing
5 during winter months.
6 In addition to damage to plant surfaces, deposited PM can be taken up by plants from soil
7 or foliage. The ability of vegetation to take up heavy metals and organics is dependent upon the
8 amount, solubility and chemical composition of the deposited PM. Uptake of PM by plants from
9 soils and vegetative surfaces can disrupt photosynthesis, alter pigments and mineral content,
10 reduce plant vigor, decrease frost hardiness and impair root development. The ISA indicates that
11 there are little or no effects on foliar processes at ambient levels of PM (US EPA, 2009a,
12 sections 9.4.3 and 9.4.7) however, damage due to atmospheric pollution can occur near point-
13 sources or under conditions where plants are subjected to multiple stressors.
14 New information since the last review provides evidence of plant uptake of metals and
15 organics. An area of active study is the impact of PAHs on provisioning ecosystem services due
16 to the potential for human and other animal exposure via food consumption. The uptake of
17 PAHs depends on the plant species, site of deposition, physical and chemical properties of the
18 organic compound and prevailing environmental conditions. It has been established that most
19 bioaccumulation of PAHs by plants occurs via leaf uptake, and to a lesser extent, through roots.
20 Differences between species in uptake of PAHs confound attempts to quantify impacts to
21 ecosystem provisioning services. For example, zucchini (Cucurbitapepo) accumulated
22 significantly more PAHs than related plant species (Parrish et al., 2006).
23 Plants as ecosystem regulators can serve as passive monitors of pollution. Lichens and
24 mosses are sensitive to pollutants associated with PM and have been used with limited success to
25 show spatial and temporal patterns of atmospheric deposition of metals. For example, the
26 presence or absence of a specific species of lichen can be used as a bioindicator of metal or
27 organics contamination. PBDEs detected in moss and lichens in Antarctica indicate long-range
28 transport of PM components. In the U.S. Blue Ridge Mountains, a study linked metal
29 concentrations in mosses to elevation and tree canopy species at some sites but not with
30 concentrations of metals in the O horizon of soil (Schilling, 2002). A limitation to employing
31 mosses and lichens to detect for the presence of air pollutants is the difference in uptake
32 efficiencies of metals between species. The European Moss Biomonitoring Network has been
33 shown to be useful in Europe for estimating general trends in metal concentrations and
34 identification of some sources of trace contaminants, however, quantification of ecological
35 effects is not possible due to the variability of species responses.
PRELIMINARY DRAFT - September 2009 6-34
Do Note Quote or Cite
-------
1 An ecological endpoint (phytochelatin concentration) associated with presence of metals
2 in the environment has been correlated with the ecological effect of tree mortality (Grantz et al.,
3 2003). Metal stress may be contributing to tree injury and forest decline in the Northeastern U.S.
4 where red spruce populations are declining with increasing elevation. Quantitative assessment of
5 PM damage to forests potentially could be conducted by overlaying PM sampling data and
6 elevated phytochelatin levels. However, limited data on phytochelatin levels in other species
7 currently hinders use of this peptide as a general biomarker for PM.
8 A potentially important regulating ecosystem service of plants is their capacity to
9 sequester contaminants. Ongoing research on the application of plants to environmental
10 remediation efforts are yielding some success in removing heavy metals and organics from
11 contaminated sites (phytoremediation) with tolerant plants such as the willow tree (Salix spp.)
12 and members of the family Brassicaceae (US EPA,2009a, section 9.4.5.4). Tree canopies can be
13 used in urban locations to capture particulates and improve air quality (Freer-Smith et al., 2004).
14 Plant foliage is a sink for Hg and other metals and this regulating ecosystem service may be
15 impacted by atmospheric deposition of trace metals.
16 The presence of PM in the atmosphere affects ambient radiation as discussed in the
17 climate section (6.3) which can impact the amount of sunlight received by plants. Atmospheric
18 PM can change the radiation reaching leaf surfaces through attenuation and by converting direct
19 radiation to diffuse radiation. Diffuse radiation is more uniformly distributed in a tree canopy,
20 allowing radiation to reach lower leaves. The net effect of PM on photosynthesis depends on the
21 reduction of photosynthetically active radiation (PAR) and the increase in the diffuse fraction of
22 PAR. Decreases in crop yields (provisioning ecosystem service) have been attributed to regional
23 scale air pollution, however, global models suggest that the diffuse light fraction of PAR can
24 increase growth (US EPA, 2009a, section 9.4.4).
25 6.4.3.2 Soil and Nutrient Cycling
26 Many of the major indirect plant responses to PM deposition are chiefly soil-mediated
27 and depend on the chemical composition of individual components of deposited PM. Major
28 ecosystem services impacted by PM deposition to soils include support services such as nutrient
29 cycling, products such as crops and regulating flooding and water quality. Upon entering the soil
30 environment, PM pollutants can alter ecological processes of energy flow and nutrient cycling,
31 inhibit nutrient uptake to plants, change microbial community structure and, affect biodiversity.
32 Accumulation of heavy metals in soils depends on factors such as local soil characteristics,
33 geologic origin of parent soils, and metal bioavailability. It can be difficult to assess the extent
34 to which observed heavy metal concentrations in soil are of anthropogenic origin. Trace element
PRELIMINARY DRAFT - September 2009 6-35
Do Note Quote or Cite
-------
1 concentrations are higher in some soils that are remote from air pollution sources due to parent
2 material and local geomorphology.
3 Heavy metals such as Zn, Cu, and Cd and some pesticides can interfere with
4 microorganisms that are responsible for decomposition of soil litter, an important regulating
5 ecosystem service that serves as a source of soil nutrients. Surface litter decomposition is
6 reduced in soils having high metal concentrations. Soil communities have associated bacteria,
7 fungi, and invertebrates that are essential to soil nutrient cycling processes. Changes to the
8 relative species abundance and community composition can be quantified to measure impacts of
9 deposited PM to soil biota. A mutualistic relationship exists in the rhizophere (plant root zone)
10 between plant roots, fungi, and microbes. Fungi in association with plant roots form
11 mycorrhizae that are essential for nutrient uptake by plants. The role of mychorrizal fungi in
12 plant uptake of metals from soils and effects of deposited PM on soil microbes is discussed in
13 section 9.4.5.3 of the second draft ISA.
14 6.4.3.3 Wildlife
15 Evidence of deposited PM effects on animals is limited. Animals play a significant role
16 in ecosystem function including nutrient cycling and crop production (supporting ecosystem
17 service), and as a source of food (provisioning ecosystem service). Cultural ecosystem services
18 provided by wildlife include bird and animal watching, recreational hunting and fishing. Impacts
19 on these services are dependent upon the bioavailability of deposited metals and organics and
20 their respective toxicities to ecosystem receptors. Pathways of PM exposure to fauna include
21 ingestion, absorption and trophic transfer. Bioindicator species (known as sentinel organisms)
22 can provide evidence of contamination due to atmospheric pollutants. Use of sentinel species
23 can be of particular value because chemical constituents of deposited PM are difficult to
24 characterize and have varying bioavailability. Snails readily bioaccumulate contaminants such
25 as PAHs and trace metals. These organisms have been deployed as biomonitors for urban
26 pollution and have quantifiable biomarkers of exposure including growth inhibition, impairment
27 of reproduction, peroxidomal proliferation and induction of metal detoxifying proteins
28 (metallothioneins) (Gomet-de Vaufleury, 2000; Regoli, 2006). Earthworms have also been used
29 as sensitive indicators of soil metal contamination.
30 Trophic transfer of pollutants of atmospheric origin has been demonstrated in limited
31 studies. PM may also be transferred between aquatic and terrestrial compartments. There is
32 limited evidence for biomagnification of heavy metals up the food chain except for Hg which
33 moves readily through environmental compartments. Bioconcentration of POPs and PBDEs in
34 the Arctic and deep-water oceanic food webs indicated the global transport of particle-associated
35 organics. Salmon migrations are contributing to metal accumulation in inland aquatic systems
PRELIMINARY DRAFT - September 2009 6-36
Do Note Quote or Cite
-------
1 potentially impacting the provisioning and cultural ecosystem service of fishing. Stable isotope
2 analysis can be applied to establish linkages between PM exposure and impacts to food webs,
3 however, the use of this evaluation tool is limited for this ecological endpoint due to the
4 complexity of most trophic interactions. Foraging cattle have been used to assess atmospheric
5 deposition and subsequent bioaccumulation of Hg and trace metals and their impacts on
6 provisioning services.
7 6.4.3.4 Water
8 New limited information on impacts of deposited PM on receiving water bodies indicate
9 that the ecosystem services of primary production, provision of fresh water, regulation of climate
10 and floods, recreational fishing and water purification are adversely impacted by atmospheric
11 inputs of metals and organics. Deposition of PM to surfaces in urban settings increases the metal
12 and organic component of storm water runoff. This atmospherically-associated pollutant burden
13 can then be toxic to aquatic biota.
14 Atmospheric deposition can be the primary source of some organics and metals to
15 watersheds. The contribution of atmospherically deposited PAHs to aquatic food webs was
16 demonstrated in high elevation mountain lakes with no other anthropogenic contaminant sources.
17 Metals associated with PM deposition limit phytoplankton growth, impacting aquatic trophic
18 structure. Long-range atmospheric transport of 47 pesticides and degradation products to the
19 snowpack in seven national parks in the Western U.S. was recently quantified indicating PM-
20 associated contaminant inputs to receiving waters during spring snowmelt (Hageman et al.,
21 2006).
22 6.4.3.5 Ecosystem and Regional Responses
23 Most direct ecosystem effects associated with particulate pollution occur in severely
24 polluted areas near industrial point sources (quarries, cement kilns, metal smelting). Extensive
25 research on biota near point sources provide some of the best evidence of ecosystem function
26 impacts and demonstrates that deposited PM has the potential to alter species composition over
27 long time scales. Ecological field studies conducted in proximity to Cu-Ni smelter in Harjavalta,
28 Finland indicated ecological structure and community composition are altered in response to PM
29 and these effects decrease with increasing distance from the point source (US EPA, 2009a,
30 section 9.4.5.8). The ISA indicates at 4 km distance, species composition of vegetation, insects,
31 birds, and soil microbiota changed, and within 1 km only the most resistant organisms were
32 surviving. Heavy metal concentrations were quantified in understory plant species growing at
33 varying distance from the Harjavalta smelter (Salemaa et al., 2004). Heavy metal concentrations
34 were highest in bryophytes, followed by lichens and were lowest in vascular plants. At the
PRELIMINARY DRAFT - September 2009 6-37
Do Note Quote or Cite
-------
1 Harjavalta smelter there are clear links between PM deposition levels, ecological endpoints and
2 compromised ecosystem structure.
3 The recently completed Western Airborne Contaminants Assessment Project (WACAP)
4 is the most comprehensive database on contaminant transport and PM depositional effects on
5 sensitive ecosystems in the U.S. In this project, the transport, fate, and ecological impacts of
6 anthropogenic contaminants from atmospheric sources were assessed from 2002 to 2007 in seven
7 ecosystem components (air, snow, water, sediment, lichen, conifer needles and fish) in eight
8 core national parks (Landers et al., 2008). Collected data were analyzed to identify probable
9 local, regional and/or global sources of deposited PM components and their concurrent effects on
10 ecological receptors. Findings from this study included the observation of an elevational
11 gradient in PM deposition with greater accumulation at higher altitude areas of the parks.
12 Furthermore, specific ecological indicators were indentified in the WACAP that can be used in
13 assessing contamination on larger spatial scales. For example, quantification of concentrations
14 of selected pesticides in second-year conifer needles served as a method for regional-scale
15 comparison of pollutant distribution (Landers et al., 2008).
16 6.4.4 Summary and Conclusions
17 The above discussions identify linkages between ecological effects of deposited PM and
18 potential impacts to ecosystem services. Unfortunately, our ability to relate ambient
19 concentrations of PM to ecosystem response is hampered by a number of significant data gaps
20 and uncertainties. These limitations include the presence of multiple ecological stressors
21 confounding attempts to link specific ecosystem responses to PM deposition. These stressors
22 can be anthropogenic (e.g. habitat destruction, eutrophication, other pollutants) or natural (e.g.
23 drought, fire, disease). Deposited PM interacts with other stressors to affect ecosystem patterns
24 and processes. Futhermore, the environmental effects of deposited PM are decoupled in space
25 and time from the point of emission confounding efforts to identify ecological perturbations
26 attributed to PM deposition.
27 A second source of uncertainty lies in predicting the amount of PM deposited to sensitive
28 receptors from measured concentrations of PM in the ambient air. This makes it difficult to
29 related a given air concentration to a receptor response, an important factor in being able to set a
30 national ambient air quality standard. A multitude of factors such as the mode of deposition
31 (wet, dry and occult), wind speed, surface roughness or stickiness, elevation, particle
32 characteristics (e.g. size, shape, chemical composition), and relative humidity exert varying
33 degrees of influence on the deposition velocities for different PM components in any point in
34 time. Composition of ambient PM varies in time and space and the particulate mixture may have
35 synergistic, antagonistic or additive effects on ecological receptors depending upon the chemical
PRELIMINARY DRAFT - September 2009 6-38
Do Note Quote or Cite
-------
1 species present. Furthermore, presence of co-occurring pollutants make it difficult to attribute
2 observed effects to ecological receptors to PM alone or one component of deposited PM. Europe
3 and other countries are using the critical load approach to assess pollutant effects at the level of
4 the ecosystem. This type of assessment requires site-specific data and information on individual
5 species responses to PM. In respect to trace metals and organics, there are insufficient data for
6 the vast majority of U.S. ecosystems to calculate critical loads, however, a methodology is being
7 presented in the NOx/SOx Secondary REA (US EPA, 2009f) to calculate atmospheric
8 concentrations from deposition that may be applicable to other environmental contaminants.
9 Third, each ecosystem has developed within a context framed by the topography,
10 underlying bedrock, soils, climate, meteorology, hydrologic regime, natural and land use history,
11 and species composition that make it unique from all others. Sensitivity of ecosystem response
12 is highly variable in space and time. Because of this variety and lack of sufficient baseline data
13 on each of these features for most ecosystems, it is currently not possible to extrapolate with
14 confidence any effect from one ecosystem to another. The WACAP study represents an
15 experimental design in which ecological effects could be correlated to ambient pollutant levels.
16 There is a need for ecological modeling of PM components in different environmental
17 compartments to further elucidate links between PM and ecological indicators.
18 Ecosystem effects linked to PM are difficult to determine because the changes are often
19 subtle and may not be observed until pollutant deposition has occurred for many decades. Data
20 on ecological responses clearly linked with atmospheric PM is not abundant enough to perform a
21 quantitative analysis although the WACAP study may represent an opportunity for quantification
22 at a regional scale. Because the high levels necessary to cause injury occur only near a few
23 limited point sources and/or on a very local scale, protection against these effects alone may not
24 provide sufficient basis for recommending a separate secondary NAAQS based on the ecological
25 effects of particulate metals and organics. At this time, staff concludes that there is insufficient
26 information available to recommend for consideration an ecologically defined secondary
27 standard that is specifically targeted for protection of vegetation and ecosystems.
28 6.5 EFFECTS ON MATERIALS
29 Deposition of atmospheric pollution, including ambient PM, on materials are related to
30 both physical damage (materials damage effects) and impaired aesthetic qualities (soiling
31 effects). The deposition of PM can physically affect materials, adding to the effects of natural
32 weathering processes, by potentially promoting or accelerating the corrosion of metals, by
33 degrading paints and by deteriorating building materials such as stone, concrete and marble.
34 Particles contribute to these physical effects because of their electrolytic, hygroscopic and acidic
35 properties, and their ability to sorb corrosive gases (principally 802).
PRELIMINARY DRAFT - September 2009 6-39
Do Note Quote or Cite
-------
1 In addition, the deposition of ambient PM can reduce the aesthetic appeal of buildings
2 and objects through soiling. Particles consisting primarily of carbonaceous compounds cause
3 soiling of commonly used building materials and culturally important items such as statues and
4 works of art. Soiling is the deposition of particles on surfaces by impingement, and the
5 accumulation of particles on the surface of an exposed material results in degradation of its
6 appearance. Soiling can be remedied by cleaning or washing, and depending on the soiled
7 material, repainting.
8 Because the effects of PM are exacerbated by the presence of acidic gases and can be
9 additive or synergistic due to the complex mixture of pollutants in the air and surface
10 characteristics of the material, this discussion will also include those particles and gases that are
11 associated with the presence of ambient NOx and SOx, as well as NH3 and NHX for
12 completeness. Building upon the information presented in the last Staff Paper (US EPA, 2005),
13 and including the limited new information presented in chapter 9 of the PM ISA (US EPA,
14 2009a) and Annex E. Effects of NOy, NHx, and SOx on Structures and Materials of the
15 Integrated Science Assessment for Oxides of Nitrogen and Sulfur-Ecological Criteria (NOx/SOx
16 ISA) (US EPA, 2008c) the following sections summarize the physical damage and aesthetic
17 soiling effects of PM on materials including metal and stone.
18 6.5.1 Materials Damage Effects
19 Physical damage to materials associated with deposited particulate matter (especially
20 sulfates and nitrates) include the corrosion of metals, degradation of painted surfaces,
21 deterioration of building materials such as limestone, concrete and marble and weakening of
22 paper, plastics, elastomers and electronic components. Both wet and dry deposition contributes
23 to particulate accumulation and subsequent damage to surfaces. The presence of air pollutants
24 on a surface may increase the retention of moisture, accelerating natural weathering processes
25 and promoting damaging effects of deposited PM. Deposited particulates may also undergo
26 chemical transformations and are commonly oxidized to acids. For example, oxides of nitrogen
27 damage textiles, electronics and dyes.
28 Metals are susceptible to damage by ambient PM. Considerable research has been
29 conducted on the effects of air pollutants on metal surfaces due to the economic importance of
30 these materials, especially steel, zinc, aluminum, and copper. Chapter 9 of the PM ISA and
31 Annex E of the NOx/SOx ISA summarize the results of a number of studies on the corrosion of
32 metals (US EPA, 2009a; US EPA, 2008c). Moisture is the single greatest factor promoting metal
33 corrosion, however, deposited PM can have additive, antagonistic or synergistic effects. In
34 general SO2 is more corrosive than NOx although mixtures of NOx, SO2 and other particulate
35 matter corrode some metals at a faster rate than either pollutant alone. Information from both the
PRELIMINARY DRAFT - September 2009 6-40
Do Note Quote or Cite
-------
1 PM ISA and NOx/SOx ISA suggest that the extent of damage to metals due to ambient PM is
2 variable and dependent upon the type of metal, prevailing environmental conditions, rate of
3 natural weathering and presence or absence of other pollutants.
4 Deterioration of calcareous stone (marble, limestone, carbonated cement) is associated
5 with deposition and subsequent reaction of PM components to these surfaces. The rate of stone
6 deterioration is determined by the pollutant mix and concentration, the stone's permeability and
7 moisture content and the pollutant deposition velocity. Dry deposition of SO2 between rain
8 events has been reported to be a major causative factor in pollutant-related erosion of calcareous
9 stone. Gypsum (hydrated calcium sulfate) is the primary degradation product found on stone,
10 mortar, and concrete samples and forms as a result of a chemical reaction with calcium
11 carbonate. Gypsum deposits lead to pitting and deterioration of stone surfaces. Marble is
12 damaged by acids (H2SO4, HNOs) increasing the solubility of the stone and resulting in surface
13 recession. The second draft PM ISA and final NOx/SOx ISA summarize the results of a number
14 of studies on PM and stone surfaces. While it is clear from the available information that gaseous
15 air pollutants, in particular SO2, will promote the deterioration of some types of stones under
16 specific conditions, carbonaceous particles (non-carbonate carbon) and particles containing metal
17 oxides may help to promote the decay process.
18 A limited number of new studies available on materials damage effects of PM since the
19 last review consider the relationship between pollutants and biodeterioration of structures
20 associated with microbial communities that colonize monuments and buildings. Presence of air
21 pollutants may synergistically enhance microbial deterioration processes. The role of
22 heterotrophic bacteria, fungi and cyanobacteria in biodeterioration varied by local meterological
23 conditions and pollutant components. In a comparative study of biodeterioration processes on
24 monuments in Latin America, limestone deterioration at the Mayan site of Uxmal was enhanced
25 by biosolubilization by metabolic acids from bacteria and fungi while destruction of the
26 Cathedral of La Plata was attributed primarily to atmospheric pollutants (Herrera and Videla,
27 2006).
28 6.5.2 Soiling Effects
29 PM deposition onto surfaces such as metal, glass, stone and paint can lead to soiling.
30 Soiling results when PM accumulates on an object and alters the optical characteristics
31 (appearance). The reflectivity of a surface may be changed or presence of particulates may alter
32 light transmission. These effects can impact the aesthetic value of a structure or result in
33 reversible or irreversible damage to statues, artwork and architecturally or culturally significant
34 buildings. Due to soiling of building surfaces by PM, the frequency and duration of cleaning
35 may be increased. Soiling affects the aesthetic appeal of painted surfaces. In addition to natural
PRELIMINARY DRAFT - September 2009 6-41
Do Note Quote or Cite
-------
1 factors, exposure to PM may give painted surfaces a dirty appearance. Pigments in works of art
2 can be degraded or discolored by atmospheric pollutants, especially sulfates (US EPA, 2008c,
3 Annex E-15).
4 Formation of black crusts due to carbonaceous compounds and buildup of microbial
5 biofilms results in discoloration of surfaces. Black crust includes a carbonate component derived
6 from building material and organic carbon (OC) and elemental carbon (EC). In limited new
7 studies quantifying the OC and EC contribution to soiling by black crust, OC predominated over
8 EC at almost all locations (Bonazza et al., 2005). Limited new studies suggest that traffic is the
9 major source of carbon associated with black crust formation (Putaud, 2004) and that soiling of
10 structures in Oxford, UK showed a relationship with traffic and NO2 concentrations (Viles and
11 Gorbushina, 2003).
12 6.5.3 Summary and Conclusions
13 Airborne and deposited PM are associated with significant detrimental effects to
14 materials including decreased useable lifetime, increased maintenance frequency and loss of
15 aesthetic appeal. Costs associated with remediation of materials damaged by atmospheric
16 pollutants are difficult to estimate due to the role of natural weathering processes in degradation
17 and the uncertainties in estimating PM contribution to materials damage. The majority of
18 available new studies on materials effects of PM are from outside the U.S., however, they
19 provide limited new data for consideration of the NAAQS secondary standard.
20 Physical damage to materials, especially economically important metals and calcareous
21 stone, results in significant costs associated with damage repair and remediation. While several
22 studies in the PM ISA and NOx/SOx ISA suggest that particles can promote corrosion of metals
23 there remains insufficient evidence to relate corrosive effects to specific particulate levels or to
24 establish a quantitative relationship between ambient PM and metal degradation. With respect to
25 damage to calcareous stone, numerous studies suggest that wet or dry deposition of particles and
26 dry deposition of gypsum particles can enhance natural weathering processes.
27 Available data indicate that particle-related soiling can result in increased cleaning
28 frequency and repainting, and may reduce the useful life of the soiled materials. However, to
29 date, no quantitative relationships between particle characteristics (e.g., concentrations, particle
30 size, and chemical composition) have been established. Limited new data on the role of
31 microbial colonizers in biodeterioration processes and contributions of black crust to soiling are
32 not sufficient for quantitative analysis. Thus, staff concludes that PM effects on materials can
33 play no quantitative role in considering whether any revisions of the secondary PM NAAQS are
34 appropriate at this time.
PRELIMINARY DRAFT - September 2009 6-42
Do Note Quote or Cite
-------
1 6.6 REFERENCES
2 Abt Associates Inc., 2001. Assessing Public Opinions on Visibility Impairment due to Air Pollution: Summary
3 Report. Prepared for EPA Office of Air Quality Planning and Standards; funded under EPA Contract No.
4 68-D-98-001. Abt Associates Inc., Bethesda, MD. January. Available:
5 http://www.epa.gov/ttncaaal/tl/reports/vis rpt final.pdf. Accessed 9/16/2008.
6 Air Resource Specialists. 2003. 2003 WinHaze Air Quality Modeler, Version 2.9. Air Resource Specialists, Inc.
7 Available: http://www.air-resource.com/downloads.php. Accessed 9/16/2008.
8 Arizona Department of Environmental Quality. 2003. Recommendation for a Phoenix Area Visibility Index.
9 Visibility Index Oversight Committee. Available: http://www.phoenixvis.net/PDF/vis 031403final.pdf.
10 Accessed 9/16/2008. BBC Research & Consulting. 2002. Phoenix Area Visibility Survey. Draft Report.
11 Available: http://www.azdeq.gov/environ/air/download/vis 021903f.pdf. Accessed 9/16/2008.
12 BBC Research & Consulting. 2003. Phoenix Area Visibility Survey. Draft Report. Available:
13 http://www.azdeq.gov/environ/air/download/vis 021903f.pdf. Accessed 9/16/2008.
14 Billings WD. (1978). Plants and the ecosystem. Belmont, CA: Wadsworth Publishing Company, Inc.
15 Bonazza A; Sabbioni C; Ghedini N. (2005). Quantitative data on carbon fractions in interpretation of black crusts
16 and soiling on European built heritage. Atmos Environ, 39: 2607-2618.
17 Campbell FW. (1983). Ambient Stressors., 15: 355-380.
18 Carlson, K.T. (2009) Mountains that see, and that need to be seen: aboriginal perspectives on degraded visibility
19 associated with air pollution in the BC Lower Mainland and Fraser Valley. A Traditional Knowledge
20 Study, prepared for Environment Canada, available in EPA docket
21 CCSP (2009): Atmospheric Aerosol Properties and Climate Impacts, A Report by the U.S. Climate Change Science
22 Program and the Subcommittee on Global Change Research. [Mian Chin, Ralph A. Kahn, and Stephen E.
23 Schwartz (eds.)]. National Aeronautics and Space Administration, Washington, D.C., USA.
24 Chestnut LG; Dennis RL. (1997). Economic benefits of improvements in visibility: acid rain Provisions of the 1990
25 Clean Air Act Amendments. J Air Waste Manag Assoc, 47: 395-402.
26 Cunningham M. (1979). Weather, mood, and helping behavior: Quasi experiments with the sunshine Samaritan.,
27 37: 1947-1956.
28
29 DeBell L. (2006). Spatial and seasonal patterns and temporal variability of haze and its constituents in the United
30 States: Report IV. 1 Cooperative Institute for Research in the Atmosphere. Colorado State University Fort
31 Collins, CO.
32 Ely, D.W., J.T. Leary, T.R. Stewart, and D.M. Ross. 1991. The Establishment of the Denver Visibility Standard.
3 3 Presented at the 84th Annual Meeting & Exhibition of the Air and Waste Management Association. June
34 16-21.
3 5 Evans GW; Jacobs SV; Frager NB. (1982). Behavioral responses to air pollution. In A Baum; J Singer
36 (Ed.),Advances in Environmental Psychology (pp. 237-269). Hillsdale, NJ: Erlbaum.
37 Evans GW; Jacobs SV; Dooley D; Catalano R. (1987). The Interaction of Stressful Life Events and Chronic Strains
38 on Community Mental Health., 15: 23-24.
PRELIMINARY DRAFT - September 2009 6-43
Do Note Quote or Cite
-------
1 Forster P; Ramaswamy V; Artaxo P; Betss R; Fahey DW; Haywood J; Lean J; Lowe DC; Myhre G; Nganga J; Prinn
2 R; Raga G;Schultz M; Van Borland R. (2007). Changes in atmospheric constituents and in radiative
3 forcing. In Solomon, S.; Qin,D.; Manning, M.; Chen, Z.; Marquis, M.; Averyt, K. B.; Tignor, M.; Miller,
4 H. L. (Ed.),Climate Change 2007: the physical science basis. Contribution of Working Group I to the fourth
5 assessment report of the intergovernmental panel
6 Frank NH. (2006). Retained nitrate, hydrated sulfates, and carbonaceous mass in federal reference method fine
7 paniculate matter for six eastern U.S. cities. J Air Waste Manag Assoc, 56: 500-11
8 Freer-Smith PH; El-khatib A; Taylor G. (2004). Capture of paniculate pollution by trees: a comparison of species
9 typical of semi-arid areas (Ficus nitida and Eucalyptus globulus) with European and North American
10 species. Water Air Soil Pollut,155: 173-187.
11 Gitay H; Brown S; Easterling W; JallowB. (2001). Ecosystems and their goods and services. In Climate change 2001:
12 impacts,adaptation and vulnerability. Contribution of working group II to the third assessment report of the
13 Intergovernmental Panel on Climate Change (pp. 237-342). Cambridge, United Kingdom: Cambridge University Press.
14 Givati A; Rosenfeld D. (2004). Quantifying precipitation suppression due to air pollution., 43: 1038-1056
15 Gomot-De Vaufleury A; Pihan F. (2002). Methods for toxicity assessment of contaminated soil by oral or dermal
16 uptake in land snails: Metal bioavailability and bioaccumulation. Environ Toxicol Chem, 21: 820-827.
17 Grantz DA; Garner JHB; Johnson DW. (2003). Ecological effects of paniculate matter. Environ Int, 29: 213-239.
18 Hageman KJ; Simonich SL; Campbell DH; Wilson GR; Landers DH. (2006). Atmospheric deposition of current-use
19 and historic use pesticides in snow at national parks in the western United States. Environ Sci Techno 1, 40:
20 3174-3180.
21 Hassan R; Scholes R; Ash N. (2005). Ecosystems and human well-being: current state and trends, volume 1.
22 Henderson, R. (2006) Letter from Dr. Rogene Henderson, Chair, Clean Air Scientific Advisory Committee to the
23 Honorable Stephen L. Johnson, Administrator, US EPA. Clean Air Scientific Advisory Committee
24 Recommendations Concerning the Proposed National Ambient Air Quality Standards for Paniculate
25 Matter. March 21, 2006. Available: http://www.epa.gov/sab/pdf/casac-ltr-06-002.pdf
26 Henderson, R. (2008) Letter from Dr. Rogene Henderson, Chair, Clean Air Scientific Advisory Committee to the
27 Honorable Stephen L. Johnson, Administrator, US EPA. Clean Air Scientific Advisory Committee
28 Consultation on EPA's Draft Integrated Review Plan for the National Ambient Air Quality Standards for
29 Paniculate Matter January 3, 2008.
3 0 Herrera LK; Videla HA. (2004). The importance of atmospheric effects on biodeterioration of cultural heritage
31 constructional materials. Int Biodeterior Biodegradation, 54: 125-134. 155843
32 IPCC (2007). Summary for Policymakers. In Climate Change 2007: The Physical Science Basis. Contribution of
3 3 Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
34 (Solomon, S; Quin D, Manning M, Cehn Z; Marquis M; Averyt KB; Tignor M; and Miller HL (eds.)
3 5 Cambridge University Press, Cambridge United Kingdom and New York, NY, USA.
36 JacobsonMZ; Kaufman YJ. (2006). Wind reduction by aerosol particles. Geophys Res Lett, 33.
37 Jacques Whitford AXYS. 2007. The View Ahead; Identifying Options for a Visibility Management Framework for
3 8 British Columbia. Report for the British Columbia Ministry of Environment. Available:
39 http://www.env.gov.bc.ca/air/airqualitv/pdfs/view ahead.pdf. Accessed 8/5/2009.
40 Jones JW; Bogat GA. (1978). Air pollution and human aggression. Psychol Rep, 43: 721-722.
PRELIMINARY DRAFT - September 2009 6-44
Do Note Quote or Cite
-------
1 Landers DH; Simonich SL; Jaffe DA; Geiser LH; Campbell DH; Schwindt AR; Schreck CB; Kent ML; Hafner WD;
2 Taylor HE; Hageman KJ; Usenko S; Ackerman LK; Schrlau JE; Rose NL; Blett TF; Erway MM. (2008).
3 The Fate, Transport and Ecological Impacts of Airborne Contaminants in Western National Parks (USA).
4 U.S. Environmental Protection Agency, Office of Research and Development, NHEERL, Western Ecology
5 Division. Corvallis, Oregon.
6 Latimer DA; Ireson RG. (1980). Workbook for estimating visibility impairment. U.S. Enviornmental Protection
7 Agency. Research Triangle Park, NC. EPA-450/4-80-031
8 Middleton WEK. (1952). Vision through the atmosphere. Toronto, ON, Canada: University of Toronto Press.
9 Molenar, J. V., W.C. Malm, and C.E. Johnson. 1994. Visual air quality simulation techniques. Atmospheric
10 Environment 2&(5):W55-W63.
11 NARSTO (2004) Paniculate Matter Science for Policy Makers: A NARSTO Assessment. P. McMurry, M.
12 Shepherd, and J. Vickery, eds. Cambridge University Press, Cambridge, England. ISBN 0 52 184287 5
13 National Research Council. (2001). Climate change science: an analysis of some key questions.
14 Parrish ZD; White JC; Isleyen M; Gent MPN; lannucci-Berger W; Eitzer BD; Kelsey JW; Mattina MI. (2006).
15 Accumulation of weathered polycyclic aromatic hydrocarbons (PAHs) by plant and earthworm species.
16 Chemosphere, 64: 609-618.
17 Pitchford, M. and W. Malm. 1993. Development and applications of a standard visual index. Atmospheric
18 Environment 28(5): 1049-1054.
19 Pryor, S.C. 1996. Assessing public perception of visibility for standard setting exercises. Atmospheric Environment
20 30(15):2705-2716.
21 Putaud J-P; Raes F; Van Dengenen R; Bruggemann E; Facchini M-C; Decesari S; Fuzzi S; Gehrig R; Huglin C; Laj
22 P; Lorbeer G; Maenhaut W; Mihalopoulos N; Muller K; Querol X; Rodriguez S; Schneider J; Spindler G;
23 ten Brink H; Torseth K; Wiedensohler A. (2004). A European aerosol phenomenology~2: chemical
24 characteristics of paniculate matter at kerbside, urban, rural and background sites in Europe. Atmos
25 Environ, 38: 2579-2595. 055545
26 Regoli F; Gorbi S; Fattorini D; Tedesco S; Notti A; Machella N; Bocchetti R; Benedetti M; Piva F. (2006). Use of
27 the land snail Helix aspersa sentinel organism for monitoring ecotoxicologic effects of urban pollution: An
28 integrated approach., 114:63-69.
29 Rotton J; Barry T; Milligan M; Fitzpatrick M. (1979). The air pollution experience and interpersonal aggression., 9:
30 397-412.
31
32 Rotton J; Frey J. (1982). Atmospheric conditions, seasonal trends, and psychiatric emergencies. In Replications and
33 Extensions (pp. Unknown). Washington, DC: American Psychological Association.
34
3 5 RWDI AIR. 2008. Final Report: Establishing a Visibility Goal for Wilderness and Urban areas in British Columbia
3 6 and Canada. Report to the British Columbia Ministry of Environment. Available:
37 http://www.env.gov.bc.ca/air/airqualitv/pdfs/visibilitv goal report final.pdf. Accessed 8/5/2009.
3 8 Salemaa M; Derome J; Helmisaari HS; Nieminen T; Vanha-Majamaa I. (2004). Element accumulation in boreal
39 bryophytes,lichens and vascular plants exposed to heavy metal and sulfur deposition in Finland. Sci Total
40 Environ, 324: 141-160.
PRELIMINARY DRAFT - September 2009 6-45
Do Note Quote or Cite
-------
1 Samet, J. (2009) Letter from Dr. Jonathan M. Samet, Chair, Clean Air Scientific Advisory Committee (CASAC) to
2 The Honorable Lisa P. Jackson, Administrator, U.S. EPA. Consultation on EPA's Particulate Matter
3 National Ambient Air Quality Standards: Scope and Methods Plan for Urban Visibility Impact Assessment.
4 May 21,2009. Available: http://yosemite.epa.gov/sab/sabpeople.nsfAVebCommittees/CASAC.
5 Schilling JS; Lehman ME. (2002). Bioindication of atmospheric heavy metal deposition in the Southeastern US
6 using the moss Thuidium delicatulum. Atmos Environ, 36: 1611-1618.
7 Schmidt. M; Frank, N.; Mintz, D.; Rao, T.; McCluney, L. (2005). Analyses of paniculate matter (PM) data for the
8 PM NAAQS review. Memorandum to PMNAAQS review docket EPA-HQ-OAR-2001-0017. June 30,
9 2005. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_td.html
10
11 Smith, A. 2009. Comments on the First External Review Draft of EPA's "Integrated Science Assessment for
12 Particulate Matter." CRA International, Washington, DC. March 30. Prepared for the Utility Air
13 Regulatory Group. Submitted as public comment to the public meeting EPA Clean Air Science Advisory
14 Council. April 2.
15 Smith, A.E. and S. Howell. 2009. An Assessment of the Robustness of Visual Air Quality Preference Study Results.
16 CRA International, Washington, DC. March 30. Prepared for the Utility Air Regulatory Group. Submitted
17 as supplemental material to presentation by Anne Smith to the public meeting of the EPA Clean Air
18 Science Advisory Council. April 2.
19 Statistics Canada. 2009a. Population and Dwelling Count Highlight Tables, 2006 Census. Available:
20 http://wwwl2.statcan.gc.ca/census-recensement/2006/dp-pd/hlt/97-550/Index.cfm7Page = INDX&LANG
21 = Eng. Accessed 7/13/2009.
22 Statistics Canada. 2009b. Population: Chilliwack, British Columbia (Census Agglomeration). Available:
23 http://wwwl2.statcan.ca/census-recensement/2006/dp-pd/prof/92-591/details/page.cfm7Lang = E&Geol =
24 CMA&Codel = 930_&Geo2 = PR&Code2 = 59&Data = Count&SearchText = Chilliwack&SearchType
25 = Begins&SearchPR = 01&B1 = Population&Custom. Accessed 7/13/2009.
26 Tombach I; McDonald K. (2004). Visibility and Radiative Balance Effects. In P.H. McMurry MFS, and J.S. Vickery
27 (Ed.),Particulate Matter Science for Policy Makers: A NARSTO Assessment: Cambridge University Press.
28
29 Trijonis JC; Malm WC; Pitchford M; White WH; Charlson R. (1990). Acidic deposition: State of science and
30 technology. Report 24. Visibility: Existing and historical conditions-causes and effects. Final report.
31
32 U.S. EPA, 1979 U.S. EPA. (1979). Protecting Visibility, anEPA Report to Congress. U.S. Environmental Protection
33 Agency. Washington, D.C..
34
35 U.S. EPA. (1982). Air quality criteria for paniculate matter and sulfur oxides. U.S. Environmental Protection
36 Agency. Washington, D.C..
37
38 U.S. EPA. (1993). Air quality criteria for oxides of nitrogen. U.S. Environmental Protection Agency. Washington,
39 D.C.
40 US EPA (1996a). Air Quality Criteria for Particulate Matter. Research Triangle Park, NC: National Center for
41 Environmental Assessment-RTF Office; report no. EPA/600/P-95/001aF-cF. 3v. Available:
42 http://www.epa.gov/ttn/naaqs/standards/prn/sjmjr cd.html.
43 US EPA (1996b). Review of the National Ambient Air Quality Standards for Particulate Matter: Policy Assessment
44 of Scientific and Technical Information, O AQPS Staff Paper. Research Triangle Park, NC 27711: Office
45 of Air Quality Planning and Standards; report no. EPA-452/R-96-013. Available:
46 http://www.epa.gOv/ttn/naaas/standards/pm/s pm pr sp.html.
PRELIMINARY DRAFT - September 2009 6-46
Do Note Quote or Cite
-------
1 US EPA (2004). Air Quality Criteria for Paniculate Matter. National Center for Environmental Assessment, Office
2 of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711;
3 report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October 2004. Available:
4 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr cd.html
5 US EPA (2005). Review of the National Ambient Air Quality Standards for Paniculate Matter: Policy Assessment
6 of Scientific and Technical Information, OAQPS Staff Paper. Research Triangle Park, NC 27711: Office
7 of Air Quality Planning and Standards; report no. EPA EPA-452/R-05-005a. December 2005. Available:
8 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr sp.html
9 U.S. EPA. (2006) Air Quality Criteria for Lead Final Report. U.S. Environmental ProtectionAgency, Washington,
10 DC, EPA/600/R-05/144aF-bF, 2006.
11 US EPA (2008a). Integrated Review Plan for the National Ambient Air Quality Standards for Paniculate Matter.
12 National Center for Environmental Assessment and Office of Air Quality Planning and Standards, U.S.
13 Environmental Protection Agency, Research Triangle Park, NC. Report No. EPA 452/R-08-004. March
14 2008. Available at:
15 http://www.epa.gov/ttn/naaqs/standards/pm/data/2008 03 final integrated review_plan.pdf
16 U.S. EPA (2008b). Integrated Science Assessment for Paniculate Matter: First External Review Draft. National
17 Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
18 Research Triangle Park, NC. EPA/600/R-08/139 and 139A. December 2008. Available:
19 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_isa.html.
20 US EPA (2008c) U.S. EPA. Integrated Science Assessment (ISA) for Oxides of Nitrogen and Sulfur Ecological
21 Criteria (Final Report). U.S. Environmental Protection Agency, Washington, D.C., EPA/600/R-08/082F,
22 2008.
23 US EPA (2009a). Integrated Science Assessment for Paniculate Matter: Second External Review Draft. National
24 Center for Environmental Assessment-RTF Division, Office of Air Quality Planning and Standards,
25 Research Triangle Park, NC. EPA/600/R-08/139B. July 2009. Available:
26 http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_2007_isa.html.
27 US EPA (2009c) Particulate Matter Urban-Focused Visibility Assessment. External Review Draft. Office of Air
28 Quality Planning and Standards U.S. Environmental Protection Agency, Research Triangle Park, NC. EPA-
29 450/P-09-005. September 2009. Available
3 0 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_pd.html
31 US EPA (2009e). Particulate Matter National Ambient Air Quality Standards: Scope and Methods Plan for Urban
32 Visibility Assessment. Office of Air Quality Planning and Standards, U.S. Environmental Protection
33 Agency, Research Triangle Park, NC. EPA-452/P-09-001. February 2009. Available:
34 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_pd.html.
35 US EPA (20091) The Second Draft Risk and Exposure Assessment for Review of the Secondary National Ambient
3 6 Air Quality Standards for Oxides of Nitrogen and Oxides of Sulfur.
37 Viles HA; Gorbushina AA. (2003). Soiling and microbial colonisation on urban roadside limestone: A three year
38 study in Oxford, England., 38: 1217-1224. 156138
39 Young, T.F. and S. Sanzone (editors).(2002). A framework for assessing and reporting on ecological condition.
40 Prepared by the Ecological Reporting Panel, Ecological Processes and Effects Committee. EPA Science
41 Advisory Board. Washington, DC. 142 p.
42 Watson JG; Zhu T; Chow JC; Engelbrecht J; Fujita EM; Wilson WE. (2002). Receptor modeling application
43 framework for particle source apportionment. Chemosphere, 49: 1093-1136.
PRELIMINARY DRAFT - September 2009 6-47
Do Note Quote or Cite
-------
1 Zeidner M; Shechter M. (1988). Psycological responses to air polution: Some personality and demographic
2 correlates., 8: 191-208.
PRELIMINARY DRAFT - September 2009 6-48
Do Note Quote or Cite
-------
1 7 STAFF CONCLUSIONS ON SECONDARY PM NAAQS
2 7.1 INTRODUCTION
3 7.2 APPROACH
4 7.3 STANDARDS TO ADDRESS VISIBILITY IMPAIRMENT
5 7.3.1 Adequacy of Current PM2.s Standards
6 7.3.2 Indicators
7 7.3.3 Averaging Times
8 7.3.4 Alternative Standards to Address Visibility Impairment
9 7.3.5 Summary of Staff Conclusions
10 7.4 STANDARDS TO ADDRESS OTHER PM-RELATED WELFARE EFFECTS
11 7.4.1 Climate Change
12 7.4.2 Vegetation and Ecosystems
13 7.4.3 Materials
14 7.4.4 Summary of Staff Conclusions
15 7.5 SUMMARY OF KEY UNCERTAINTIES AND RESEARCH
16 RECOMMENDATIONS RELATED TO SETTING SECONDARY STANDARDS
17 7.6 REFERENCES
PRELIMINARY DRAFT - September 2009
Do Note Quote or Cite
7-1
-------
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-452/P-09-007
Environmental Protection Health and Environmental Impacts Division September 2009
Agency Research Triangle Park, NC
------- |