ro
Z
Quantitative Health Risk Assessment for
Particulate Matter
Second External Review Draft
-------
DISCLAIMER
This draft document has been prepared by staff from the 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 to obtain review and comment
from the Clean Air Scientific Advisory Committee (CASAC) and the general public.
Comments on this draft document should be addressed to Dr. Zachary Pekar, U.S.
Environmental Protection Agency, Office of Air Quality Planning and Standards, C504-
06, Research Triangle Park, North Carolina 27711 (email: pekar.zachary@epa.gov).
Elements of this report (see Acknowledgements below) have been provided to the
U.S. Environmental Protection Agency (EPA) by Abt Associates, Inc. in partial
fulfillment of Contract No. EP-D-08-100, Work Assignment 0-02. Any opinions,
findings, conclusions, or recommendations are those of the authors and do not necessarily
reflect the views of the EPA or Abt Associates, Inc. Any analyses, interpretations, or
conclusions presented in this report based on emergency department, hospitalization and
mortality baseline incidence data obtained from outside sources, are credited to the
authors and not the institutions providing the raw data.
ACKNOWLEDGEMENTS
In addition to EPA staff, personnel from Abt Associates, Inc. contributed to the
writing of this document. Specific chapters and appendices where Abt Associates, Inc.
made contributions include: chapter 3 (Overview of Methods Used in Estimating Risk for
the Urban Case Studies), and appendices A (Air Quality Assessment), C
(Epidemiological - Epi - Study Specific Information On PM^.s), E (Risk Analysis - Core
Analysis), and F (Sensitivity Analysis Results).
-------
EPA-452/P-10-001
February 2010
Quantitative Health Risk Assessment for
Particulate Matter
Second External Review Draft
US Environmental Protection Agency
Office of Air and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
-------
i Table of Contents
2
3 List of Tables iv
4 List of Figures vi
5 List of Acronyms/Abbreviations viii
6 1 Introduction 1-1
7 1.1 Background 1-3
8 1.2 Current risk Assessment: Goals and Planned Approach 1-5
9 1.3 Organization of Document 1-6
10 2 Scope 2-1
11 2.1 Overview of Risk Assessment from Last Review 2-2
12 2.2 Original Assessment Plan 2-3
13 2.2.1 Risk Assessment 2-3
14 2.2.2 Population Exposure Analysis 2-5
15 2.3 Current Scope and Key Design Elements 2-6
16 2.4 Alternative Suites of PM2.5 Standards Evalutated 2-9
17 3 Urban Case Study Analysis Methods 3-1
18 3.1 General Approach 3-1
19 3.1.1 Basic Structure of the Risk Assessment 3-1
20 3.1.2 Calculating PM2.5-Related Health Effects Incidence 3-7
21 3.1.2.1 Short-term vs. Long-term Exposure 3-9
22 3.1.2.2 Calculating Annual Incidence 3-10
23 3.2 Air Quality Inputs 3-11
24 3.2.1 Characterizing Recent Conditions 3-11
25 3.2.2 Estimating Policy Relevant Background 3-13
26 3.2.3 Simulating Air Quality to Just Meet Current and Alternative Standards 3-14
27 3.2.3.1 Proportional Rollback Method 3-15
28 3.2.3.2 Hybrid Rollback Method 3-19
29 3.2.3.3 Peak Shaving Rollback Method 3-19
30 3.3 Selection of Model Inputs 3-20
31 3.3.1 Health Endpoints 3-20
32 3.3.2 Selection and Delineation of Urban Study Areas 3-22
33 3.3.3 Selection of Epidemiological Studies and Concentration-response (C-R)
34 Functions within those Studies 3-26
35 3.3.4 Summary of Selected Health Endpoints, Urban Areas, Studies, and C-R
36 Functions 3-32
-------
1 3.4 Baseline Health Effects Incidence Data 3-47
2 3.4.1 Data Sources 3-47
3 3.4.1.1 Mortality 3-47
4 3.4.1.2 Hospital Admission and Emergency Department Visits 3-47
5 3.4.1.3 Populations 3-49
6 3.4.2 Calculation of Baseline Incidence Rates 3-52
7 3.5 Addressing Uncertainty and Variability 3-56
8 3.5.1 Overview 3-56
9 3.5.2 Treatment of Key Sources Of Variability 3-59
10 3.5.3 Qualitative Assessment of Uncertainty 3-62
11 3.5.4 Single and Multi-Factor Sensitivity Analyses 3-72
12 3.5.4.1 Sensitivity Analyses for Long-Term Exposure-Related Mortality 3-72
13 3.5.4.2 Sensitivity Analyses for Short-Term Exposure-Related Mortality and
14 Morbidity 3-76
15 3.5.4.3 Multi-factor Sensitivity Analyses 3-77
16 3.5.5 Summary of Approach to Addressing Variability and Uncertainty 3-78
17 4 Urban Case Study Results 4-1
18 4.1 Assessment of Health Risk Associated with Recent Conditions (core analysis) ... 4-16
19 4.2 Assessment of Health Risk Associated with Just Meeting the Current and
20 Alternative Suites of Standards (core analysis) 4-19
21 4.2.1 Core Risk Estimates for Just Meeting the Current Suite of Standards 4-21
22 4.2.2 Core Risk Estimates for Just Meeting Alternative Suites of Standards 4-22
23 4.3 Sensitivity Analysis Results 4-28
24 4.3.1 Sensitivity Analysis Results to Identify Potentially Important Sources of
25 Uncertainty and Variability 4-28
26 4.3.1.1 Single-factor Sensitivity Analysis 4-35
27 4.3.1.2 Multi-factor Sensitivity Analysis Results 4-41
28 4.3.2 Additional Set of Reasonable Risk Estimates to Inform Consideration of
29 Uncertainty in Core Risk Estimates 4-43
30 4.4 Evaluating the representativeness of the Urban Study Areas in the National
31 Context 4-48
32 4.5 Consideration of Design Values and Patterns of PM2.5 Monitoring Data in
33 Intrepreting Core Risk Estimates 4-64
34 4.5.1 Design Values 4-65
35 4.5.2 Patterns in PM2.5 Monitoring Data 4-71
February 2010 ii Draft - Do Not Quote or Cite
-------
1 5 National-Scale Assessment of Long-Term Mortality Related to PM2.s
2 Exposure 5-1
3 5.1 Overview 5-1
4 5.2 Methods 5-3
5 5.2.1 Population Estimates 5-3
6 5.2.2 Population Exposure 5-3
7 5.2.3 Premature Mortality Estimates 5-5
8 5.3 Results 5-5
9 6 Integrative Discussion of Urban Case Study Analysis of PM2.s-related Risks 6-1
10 6.1 Key Analytical Elements in this Risk Assessment 6-2
11 6.2 Interpreation of Urban Study Area Results 6-4
12 6.2.1 Simulation of Just Meeting the Current Suite of PM2.5 Standards 6-7
13 6.2.2 Simulation of Just Meeting Alternative Annual Standards 6-10
14 6.2.3 Simulation of Just Meeting Alternative Suites of Annual and 24-hour
15 Standards 6-12
16 6.3 National Perspective on PM2.s-related Risks 6-16
17 6.4 Key Observations 6-18
18 7 REFERENCES 7-1
19
20 Appendix A. Air Quality Assessment
21 Appendix B. Hybrid (Non-Proportional) Rollback Approach
22 Appendix C. Epidemiological Study Specific Information for PM Risk Assessment
23 Appendix D. Supplement to Representativeness of the 15 Urban Study Areas
24 Appendix E. Risk Estimates (Core Analysis)
25 Appendix F. Sensitivity Analysis Results
26 Appendix G. Supplement to the National-Scale Assessment of Long-Term Mortality Related to
27 PM2.5 Exposure
28 Appendix H. Consideration off Risk Associated with Exposure to Thoracic Coarse PM
29 (PMio.2.5)
February 2010 iii Draft - Do Not Quote or Cite
-------
i List of Tables
2
3 Table 3-1. Numbers of Monitors in Risk Assessment Locations From Which Composite
4 Monitor Values Were Calculated* 3-13
5 Table 3-2 Regional Policy-Relevant Background Estimates Used in the Risk Assessment. ..3-14
6 Table 3-3. EPA Design Values for Annual and \24-hour PM2.5 Standards for the Period 2005-
7 2007.* 3-16
8 Table 3-4. Urban Study Areas Selected for the Risk Assessment 3-24
9 Table 3-5. Locations, Health Endpoints, and Short-Term Exposure Studies Included in the PM2.5
10 Risk Assessment* 3-34
11 Table 3-6. Locations, Health Endpoints, and Long-Term Exposure Studies Included in the PM2.s
12 Risk Assessment* 3-35
13 Table 3-7. Summary of Locations, Health Endpoints, Studies and Concentration-Response
14 Functions Included in the Core Analysis.* 3-36
15 Table 3-8. Summary of Locations, Health Endpoints, Studies and Concentration-Response
16 Functions Included in Sensitivity Analyses 3-44
17 Table 3-9. Sources of Hospital Admissions (HA) and Emergency Department (ED) Visit Data.
18 3-48
19 Table 3-10.Relevant Population Sizes for PM Risk Assessment Locations 3-50
20 Table 3-11. Baseline Mortality Rates (Deaths per 100,000 Relevant Population per Year) for
21 2006 for PM Risk Assessment Locations.* 3-53
22 Table 3-12.Baseline Hospital Admission (HA) and Emergency Department (ED) Rates
23 (Admissions/Visits per 100,000 Relevant Population per Year) for 2007 for PM Risk
24 Assessment Locations.* 3-55
25 Table 3-13.Summary of Qualitative Uncertainty Analysis of Key Modeling Elements in the PM
26 NAAQSRisk Assessment 3-64
27 Table 4-1. Estimated Annual Incidence of Selected Mortality and Morbidity Endpoints
28 Associated with Long- and Short-Term Exposure to Ambient PM2.5 Concentrations
29 that Just Meet the Current Standards, Based on Adjusting 2007 PM2.5 Concentrations.
30 u 4-6
31 Table 4-2 Estimated Percent of Total Annual Incidence of Selected Mortality and Morbidity
32 Endpoints Associated with Long- and Short-Term Exposure to Ambient PM2.5
33 Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2.5
34 Concentrations. 1,2 4-7
35 Table 4-3 Overview of Sensitivity Analysis Results 4-31
36 Table 4-4 Derivation of a set of reasonable alternative risk estimates to supplement the core
37 risk estimates (Los Angeles and Philadelphia, current standards, for long-term IHD
38 mortality) 4-45
39 Table 4-5 Data Sources for PM NAAQS Risk Assessment Risk Distribution Analysis 4-50
40 Table 4-6 Summary Statistics for Selected PM Risk Attributes 4-52
41 Table 4-7 Results of Kolomogrov-Smirnoff Tests for Equality Between National and Urban
42 Study Area Distributions for Selected National Risk Characteristic Variables.. 4-56
43 Table 4-8 Identification of controlling standard (24-hour or annual) for alternative suites of
44 standard levels 4-70
February 2010
IV
Draft - Do Not Quote or Cite
-------
1 Table 5-1 Estimated PM2.5-related premature mortality associated with incremental air
2 quality differences between 2005 ambient mean PM2 5 levels and lowest measured
3 level from the epidemiology studies or policy relevant background (90th percentile
4 confidence interval) 5-6
5
6
February 2010 v Draft - Do Not Quote or Cite
-------
i List of Figures
2
3 Figure 3-1. Major components of particulate matter health risk assessment 3-2
4 Figure 3-2. Flow diagram of risk assessment for short-term exposure studies 3-5
5 Figure 3-3 .Flow diagram of risk assessment for long-term exposure studies 3-6
6 Figure 3-4 15 urban study areas included in the risk assessment (including seven PM regions
7 used to guide selection of study areas) 3-25
8 Figure 4-1 Percent reduction in long-term exposure-related mortality risk (alternative standards
9 and recent conditions relative to the current standards) (Note: inset shows PM2.5
10 related incidence and percent of total incidence for IHD mortality under the current
11 suite of standards) 4-8
12 Figure 4-2 Percent reduction in long-term exposure-related mortality risk (recent conditions
13 relative to the current standards) (Note: inset shows PM2.5 related incidence and
14 percent of total incidence for IHD mortality under the current suite of standards) ..4-9
15 Figure 4-3 Percent reduction in long-term exposure-related mortality risk (alternative standards
16 relative to the current standards) (Note: inset shows PM2.5 related incidence and
17 percent of total incidence for IHD mortality under the current suite of standards) 4-10
18 Figure 4-4 Percent reduction in short-term exposure-related mortality and morbidity risk
19 (alternative standards and recent conditions relative to the current standards) (Note:
20 inset shows PM2.5 related incidence and percent of total incidence for CV under the
21 current suite of standards) 4-11
22 Figure 4-5 Percent reduction in short-term exposure-related mortality and morbidity risk (recent
23 conditions relative to the current standards) (Note: inset shows PM2.5 related
24 incidence and percent of total incidence for CV under the current suite of standards)
25 4-12
26 Figure 4-6 Percent reduction in short-term exposure-related mortality and morbidity risk
27 (alternative standards relative to the current standards) (Note: inset shows PM2.5
28 related incidence and percent of total incidence for CV under the current suite of
29 standards) 4-13
30 Figure 4-7 Comparison of core risk estimates with reasonable alternative set of risk estimates for
31 Los Angeles and Philadelphia (IFID mortality) 4-46
32 Figure 4-8 Comparison of core risk estimates with reasonable alternative set of risk estimates for
33 Los Angeles and Philadelphia (all cause mortality) 4-46
34 Figure 4-9 Comparison of distributions for key elements of the risk equation: total population. 4-
35 57
36 Figure 4-10 Comparison of distributions for key elements of the risk equation: 98th percentile 24-
37 hour average PM2.5 4-58
38 Figure 4-11 Comparison of distributions for key elements of the risk equation: all use mortality
39 rate. 4-59
40 Figure 4-12Comparison of distributions for key elements of the risk equation: Mortality risk
41 effect estimate from Zanobetti and Schwartz (2008) 4-60
42 Figure 4-13Comparison of distributions for selected variables expected to influence the relative
43 risk from PM2.s: long term average July temperature 4-61
44 Figure 4-14Comparison of distributions for selected variables expected to influence the relative
45 risk from PM2.s: percent of population 65 and older 4-62
February 2010 vi Draft - Do Not Quote or Cite
-------
1 Figure 4-1 SComparison of distributions for selected variables expected to influence the relative
2 risk from PM2.s: per capita annual personal income 4-63
3 Figure 4-16Comparison of distributions for selected variables expected to influence the relative
4 risk from PM2.s: per capita annual personal income 4-64
5 Figure 4-17Design values in 15 urban study areas and broader set of U.S. urban areas relative to
6 the current suite of standards (15/35) 4-67
7 Figure 4-18Design values in 15 urban study areas and broader set of U.S. urban areas relative to
8 the 12/35 alternative suite of standards 4-68
9 Figure 4-19Design values in 15 urban study areas and broader set of U.S. urban areas relative to
10 the 12/25 alternative suite of standards) 4-69
11 Figure 4-20Annual and 24-hour design values (for individual monitors and at the study-area
12 level) for the 15 urban study areas (with the presentation of values scaled to reflect
13 current standard of 15/35) 4-73
14 Figure 4-21 Annual and 24-hour design values (for individual monitors and at the study-area
15 level) for the 15 urban study areas (with the presentation of values scaled to reflect
16 current standard of 12/25) 4-74
17 Figure 5-1 Conceptual diagram of data inputs and outputs for national long-term mortality risk
18 assessment 5-3
19 Figure 5-2 2005 fused surface baseline PM2.5 concentrations 5-4
20 Figure 5-3 Percentage of premature mortality attributable to PM2.5 exposure at various 2005
21 annual average PM2.5 levels* 5-7
22 Figure 5-4 Cumulative distribution of county-level percentage of total mortality attributable to
23 PM2.5 for the U.S. with markers identifying where along that distribution the urban
24 case study area analysis fall* 5-8
25
February 2010 vii Draft - Do Not Quote or Cite
-------
1
2
3
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
A/C
ACS
Act
AMI
AQS
P
BenMAP
BMI
BRFSS
CASAC
CAA
CBS A
CDC
CDF
CFR
CHD
CMAQ
CO
COPD
CPD
C-R
CSA
CV
CVD
df
ED
ER
EPA
FACA
FIPS
GAM
GEOS-CHEM
GLMs
List of Acronyms/Abbreviations
Air conditioning
American Cancer Society
Clean Air Act
Acute Myocardial Infarction
EPA's Air Quality System
Slope coefficient
Benefits Mapping Analysis Program
Body Mass Index
Behavioral Risk Factor Surveillance System
Clean Air Scientific Advisory Committee
Clean Air Act
Core-based Statistical Area
Centers for Disease Control
Cumulative Distribution Function
Code of Federal Regulations
Coronary Heart Disease
Community Multiscale Air Quality
Carbon Monoxide
Chronic Obstructive Pulmonary Disease
Cardio-pulmonary Disease
Concentration-response
Consolidated Statistical Area
Cardiovascular
Cardiovascular Disease
Degrees of freedom
Emergency Department
Emergency Room
United States Environmental Protection Agency
Federal Advisory Committee Act
Federal Information Processing System
Generalized additive model
Goddard Earth Observing System-Chemical Model
Generalized linear model
February 2010
Vlll
Draft - 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
35
36
37
38
39
HA
HCUP
HEI
HS
ICD
IHD
INF
IRP
ISA
KB
km
K-S
LML
MCAPS
MSA
NA
NAAQS
NCEA
NEI
NCHS
NMMAPS
NOx
03
OAQPS
PA
PM
PMX
PMio
Hospital Admissions
Healthcare Cost and Utilization Project
Health Effects Institute
High School
International Classification of Diseases
ischemic heart disease
Influence of uncertainty on risk estimates
Integrated Review Plan
Integrated Science Assessment Document
Knowledge Base
Kilometer
Kolmogorov-Smirnov
Lowest Measured Level
Medicare Air Pollution Study
Metropolitan Statistical Area
Not Applicable
National Ambient Air Quality Standards
National Center for Environmental Assessment
National Emissions Inventory
National Center for Health Statistics
National Morbidity, Mortality, and Air Pollution Study
Nitrogen oxides
Ozone
Office of Air Quality Planning and Standards
Policy Assessment Document
Particulate Matter
The legal definition for PMx, as defined in the Code of Federal
Regulations, includes both a 50% cut-point and a penetration
curve. A 50% cut-point of X um diameter means that 50% of
particles with aerodynamic diameter of X are removed by the inlet
and 50% pass through the inlet and are collected on the filter.
Depending on the specific penetration curve specified, particles
larger than X um aerodynamic diameter are collected with an
efficiently than decreases rapidly for particles larger than X while
the collection efficiency for particles smaller than X increases
rapidly with decreasing size until 100 % efficiency is reached.
Particles with a 50% upper cut-point of 10± 0.5 um aerodynamic
diameter and a penetration curve as specified in the Code of
Federal Regulations.
February 2010
IX
Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
PM2.5
PMiQ-2.5
PRB
RA
RR
REA
SAB
SEDD
SID
S02
sox
SES
TRIM
TRIM.Risk
UFP
USDA
VNA
WHI
WHO
ZCA
Particles with a 50% upper cut-point of 2.5 um aerodynamic
diameter and a penetration curve as specified in the Code of
Federal Regulations.
Particles with a 50% upper cut-point of 10 um aerodynamic
diameter and a lower 50% cut-point of 2.5 um aerodynamic
diameter.
Policy-Relevant Background
Risk Assessment Document
Relative risk
Risk and Exposure Assessment
Science Advisory Board
State Emergency Department Databases
State Inpatient Database
Sulfur Dioxide
Sulfur Oxides
Socio-economic Status
Total Risk Integrated Methodology
Total Risk Integrated Methodology - Risk Assessment component
Ultrafine Particles
U.S. Department of Agriculture
Voronoi Neighbor Averaging
Women's Health Initiative
World Health Organization
Zip Code Area
February 2010
Draft - Do Not Quote or Cite
-------
1 1 INTRODUCTION
2 The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
3 the national ambient air quality standards (NAAQS) for particulate matter (PM). Sections 108
4 and 109 of the Clean Air Act (Act) govern the establishment and periodic review of the NAAQS.
5 These standards are established for pollutants that may reasonably be anticipated to endanger
6 public health and welfare, and whose presence in the ambient air results from numerous or
7 diverse mobile or stationary sources. The NAAQS are to be based on air quality criteria, which
8 are to accurately reflect the latest scientific knowledge useful in indicating the kind and extent of
9 identifiable effects on public health or welfare that may be expected from the presence of the
10 pollutant in ambient air. The EPA Administrator is to promulgate and periodically review, at
11 five-year intervals, "primary" (health-based) and "secondary" (welfare-based) NAAQS for such
12 pollutants. Based on periodic reviews of the air quality criteria and standards, the Administrator
13 is to make revisions in the criteria and standards, and promulgate any new standards, as may be
14 appropriate. The Act also requires that an independent scientific review committee advise the
15 Administrator as part of this NAAQS review process, a function performed by the Clean Air
16 Scientific Advisory Committee (CASAC).1
17 The current NAAQS for PM include a suite of standards to provide protection for
18 exposures to fine and coarse particles using PM2.5 and PMio, as indicators, respectively (71 FR
19 61144, October 17, 2006). With regard to the primary and secondary standards for fine particles,
20 in 2006 EPA revised the level of the 24-hour PM2.5 standard to 35 ug/m3 (calculated as a 3-year
21 average of the 98th percentile of 24-hour concentrations at each population-oriented monitor),
22 retained the level of the annual PM2.5 annual standard at 15 ug/m3 (calculated as the 3-year
23 average of the weighted annual mean PM2.5 concentrations from single or multiple community -
24 oriented monitors), and revised the form of the annual PM2 5 standard by narrowing the
25 constraints on the optional use of spatial averaging.2 With regard to the primary and secondary
26 standards for PMio, EPA retained the 24-hour PMio standard at 150 ug/m3 (not to be exceeded
1 The Clean Air Scientific Advisory Committee (CAS AC) was established under section 109(d)(2) of the Clean Air
Act (CAA or Act) (42 U.S.C. 7409) as an independent scientific advisory committee. CASAC provides advice,
information and recommendations on the scientific and technical aspects of air quality criteria and NAAQS under
sections 108 and 109 of the CAA. The CASAC is a Federal advisory committee chartered under the Federal
Advisory Committee Act (FACA). See
http://vosemite.epa.gov/sab/sabpeople.nsfAVebComniitteesSubcomniittees/CASAC%20Particulate%20Matter%20R
eview%20Panel for a list of the CASAC PM Panel members and current advisory activities.
In the revisions to the PM NAAQS finalized in 2006, EPA tightened 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).
February 2010 1-1 Draft - Do Not Quote or Cite
-------
1 more than once per year on average over 3 years) and revoked the annual standard because
2 available evidence generally did not suggest a link between long-term exposure to current
3 ambient levels of coarse particles and health or welfare effects. These standards were based
4 primarily on a large body of epidemiological evidence relating ambient PM concentrations to
5 various adverse health endpoints. Secondary standards for PM2.5 and PMio were revised to be
6 identical to the primary standards.
7 The next periodic review of the PM NAAQS is now underway.3 The review process
8 includes four key phases: planning, science assessment, risk assessment, and policy
9 assessment/rulemaking. A planning document, the Integrated Review Plan for the National
10 Ambient Air Quality Standards for Paniculate Matter (IRP; EPA, 2008a), outlined the science-
11 policy questions that frame this review, the process and schedule for the review, and descriptions
12 of the purpose, contents, and approach for developing the other key documents for this review.4
13 The science assessment document, the Integrated Science Assessment for Paniculate Matter
14 (ISA; EPA, 2009a and b), includes an evaluation of the scientific evidence on the health effects
15 of PM, including information on exposure, physiological mechanisms by which PM might
16 damage human health, and an evaluation of the epidemiological evidence including information
17 on reported concentration-response (C-R) relationships for PM-related morbidity and mortality
18 associations, including consideration of effects on at-risk populations.5
19 This second draft quantitative health risk assessment (RA) presents the quantitative
20 assessments of PM-related risks to public health being conducted by staff in EPA's Office of Air
21 Quality Planning and Standards (OAQPS) to support the review of the primary PM standards.
22 The development of this document is described below in chapter 2. This draft RA is being
23 released for review by the CASAC PM Panel and the public at a public meeting to be held on
24 March 10-11, 2010. Comments received on this draft will be taken into consideration in
25 preparing a final quantitative health RA for PM, which is scheduled to be completed in April
26 2010.
27 The final ISA and final quantitative health RA will inform the policy assessment and
28 rulemaking steps that will lead to final decisions on the primary PM NAAQS. A policy
29 assessment (PA) is now being prepared by OAQPS staff to provide a staff analysis of the
30 scientific basis for alternative policy options for consideration by senior EPA management prior
3 See http://www.epa.gov/ttn/naaqs/standards/pm/sjm index.html for more information on the current and
previous PM NAAQS reviews.
4 On November 30, 2007, EPA held a public consultation with the CASAC PM Panel on the draft IRP. The final
IRP took into consideration comments received from CASAC and the public on the draft plan as well as input from
senior Agency managers.
5 On October 5-6, 2009, the CASAC PM Panel met to review the second draft ISA (EPA, 2009a). The final ISA
took into consideration CASAC and public comments received on that draft.
February 2010 1-2 Draft - Do Not Quote or Cite
-------
1 to rulemaking. The PA is intended to help "bridge the gap" between the Agency's scientific
2 assessments, presented in the ISA and RA, and the judgments required of the Administrator in
3 determining whether it is appropriate to retain or revise the standards. The PA will integrate and
4 interpret information from the ISA and the RA to frame policy options and to facilitate
5 CASAC's advice to the Agency and recommendations on any new standards or revisions to
6 existing standards as may be appropriate, as provided for in the Clean Air Act. The first draft PA
7 is planned for release around the end of February 2010 for review by the CASAC PM Panel and
8 the public during a public teleconference being planned for late March. Proposed and final
9 rulemaking notices are now scheduled for November 2010 and July 2011, respectively.
10 1.1 BACKGROUND
11 As part of the last PM NAAQS review completed in 2006, EPA's OAQPS conducted a
12 quantitative risk assessment to estimate risks of various health effects associated with exposure
13 to ambient PM2.5 and PMio-2.5 in a number of urban study areas selected to illustrate the public
14 health impacts of these pollutants (U.S. EPA, 2005, chapter 4; Abt Associates, 2005). The
15 assessment scope and methodology were developed with considerable input from the CASAC
16 Review Panel and the public, with CASAC concluding that the general assessment methodology
17 and framework were appropriate (Hopke, 2002). The final quantitative risk assessment took into
18 consideration CASAC advice (Hopke, 2004; Henderson, 2005) and public comments on two
19 drafts of the risk assessment.
20 The extensive quantitative assessment conducted for fine particles in the last review
21 included estimates of risks of mortality (total non-accidental, cardiovascular, and respiratory),
22 morbidity (hospital admissions for cardiovascular and respiratory causes), and respiratory
23 symptoms (not requiring hospitalization) associated with recent short-term (daily) ambient PM2.5
24 levels and risks of total, cardiopulmonary, and lung cancer mortality associated with long-term
25 exposure to PM2 5 in nine urban study areas. The quantitative risk assessment included estimates
26 of: (1) risks of mortality, morbidity, and symptoms associated with recent ambient PM2.5 levels;
27 (2) risk reductions and remaining risks associated with just meeting the existing suite of PM2 5
28 NAAQS (1997 standards); and (3) risk reductions and remaining risks associated with just
29 meeting various alternative PM2.5 standards.
30 The quantitative risk assessment conducted in the last review for thoracic coarse particles
31 was much more limited than the analyses conducted for fine particles. The PMio-2.s risk
32 assessment included risk estimates for just three urban areas for two categories of health
33 endpoints related to short-term exposure to PMio-2.s: hospital admissions for cardiovascular and
34 respiratory causes and respiratory symptoms. While one of the goals of the PMio-2.s risk
35 assessment was to provide estimates of the risk reductions associated with just meeting
February 2010 1-3 Draft - Do Not Quote or Cite
-------
1 alternative PMio-2.s standards, OAQPS staff concluded that the nature and magnitude of the
2 uncertainties and concerns associated with this portion of the risk assessment weighed against
3 use of these risk estimates as a basis for recommending specific standard levels (U.S. EPA, 2005,
4 p. 5-69).
5 Prior to the issuance of a proposed rulemaking in the last review, CASAC presented
6 recommendations to the Administrator supporting revisions of the PM2.5 primary standards.
7 These recommendations placed substantial reliance on the results of the quantitative risk
8 assessment (Henderson, 2005, pp 6-7). In a letter to the Administrator following the 2006
9 proposed rule (71 FR 12592, January 17, 2006), CASAC requested reconsideration of the
10 Agency's proposed decisions and reiterated and elaborated on the scientific bases for its earlier
11 recommendations which included placing greater weight on the result of the Agency's risk
12 assessment. With regard to the quantitative risk assessment, CASAC concluded, "While the risk
13 assessment is subject to uncertainties, most of the PM Panel found EPA's risk assessment to be
14 of sufficient quality to inform its recommendations." (Henderson, 2006a, p. 3).
15 In the 2006 final rule, the EPA Administrator recognized that the quantitative risk
16 assessment for fine particles was based upon a more extensive body of data and was more
17 comprehensive in scope than the previous assessment conducted for the review completed in
18 1997. However, as presented in the final rulemaking notice, the Administrator was mindful of
19 significant uncertainties associated with the risk estimates for fine particles. More specifically,
20
21 Such uncertainties generally related to a lack of clear understanding of a number of
22 important factors, including, for example, the shape of the concentration-response
23 functions, particularly when, as here, effect thresholds can neither be discerned nor
24 determined not to exist; issues related to selection of appropriate statistical models for the
25 analysis of the epidemiologic data; the role of potentially confounding and modifying
26 factors in the concentration-response relationships; issues related to simulating how PM2.5
27 air quality distributions will likely change in any given area upon attaining a particular
28 standard, since strategies to reduce emissions are not yet defined; and whether there
29 would be differential reductions in the many components within PM2 5 and, if so, whether
30 this would result in differential reductions in risk. In the case of fine particles, the
31 Administrator recognized that for purposes of developing quantitative risk estimates,
32 such uncertainties are likely to [be] amplified by the complexity in the composition of the
33 mix of fine particles generally present in the ambient air. (72 FR 61168, October 17,
34 2006).
35
36 As a result, the Administrator viewed that the quantitative risk assessment provided supporting
37 evidence for the conclusion that there was a need to revise the PM2.5 primary standards, but he
38 judged that the assessment did not provide an appropriate basis to determine the level of the
39 standards (72 FR 61168, October 17, 2006).
February 2010 1-4 Draft - Do Not Quote or Cite
-------
1 In a letter to the EPA Administrator following the issuance of the final rule, CASAC
2 expressed "serious scientific concerns" regarding the final PM standards. In particular, CASAC
3 was concerned that the Agency "did not accept our finding that the annual PM2 5 standard was
4 not protective of human health and did not follow our recommendation for a change in that
5 standard" (Henderson et al, 2006b, p. 1). With respect to the use of the risk assessment to inform
6 EPA's decision on the primary PM2.5 standard, CASAC stated, "While there is uncertainty
7 associated with the risk assessment for the PM2 5 standard, this very uncertainty suggests a need
8 for a prudent approach to providing an adequate margin of safety" (Henderson et al., 2006b, p.2)
9 Several parties filed petitions for review following promulgation of the revised PM
10 NAAQS in 2006. These petitions for review addressed the following issues with regard to the
11 primary PM NAAQS: (1) selecting the level of the annual primary PM2 5 standard, (2) retaining
12 PMio as the indicator for coarse particles and retaining the level and form of the 24-hour PMio
13 standard, and (3) revoking the PMio annual standard. On judicial review, the D.C. Circuit
14 remanded the annual primary PM2.5 NAAQS to EPA because the Agency failed to adequately
15 explain why the standard provided the requisite protection from both short- and long-term
16 exposures to fine particles including protection for at-risk populations. The court upheld the
17 Agency's use of the quantitative risk assessment to inform the decision to revise the PM2.5
18 standards but not to inform the selection of level.6 The court also upheld the decision to retain
19 the 24-hour PMio standard and revoke the annual PMio standard. American Farm Bureau
20 Federation v. EPA, 559 F. 3d 512, (D.C. Cir. 2009).
21 1.2 CURRENT RISK ASSESSMENT: GOALS AND PLANNED APPROACH
22 The goals of the current quantitative health risk assessment remain largely the same as
23 those articulated in the risk assessment conducted in the last review. These goals include: (a) to
24 provide estimates of the potential magnitude of premature mortality and/or selected morbidity
25 effects in the population associated with recent ambient levels of PM and with just meeting the
26 current and alternative suites of PM standards considered in selected urban study areas,
27 including, where data are available, consideration of impacts on at-risk populations; (b) to
28 develop a better understanding of the influence of various inputs and assumptions on the risk
29 estimates to more clearly differentiate among alternative suites of standards, including potential
30 impacts on various at-risk populations; and (c) to gain insights into the distribution of risks and
31 patterns of risk reductions and the variability and uncertainties in those risk estimates. In
6 One petition for review addressed the issue of setting the secondary PM2 5 standards identical to the primary
standards. On judicial review, the court remanded the secondary PM2 5 NAAQS to EPA because the Agency failed
to adequately explain why the standards provided the required protection from visibility impairment. American
Farm Bureau Federation v. EPA, 559 F. 3d 512, (D.C. Cir. 2009).
February 2010 1-5 Draft - Do Not Quote or Cite
-------
1 addition, this assessment includes nationwide estimates of the potential magnitude of premature
2 mortality associated with long-term exposure to recent levels of ambient PM2.5 to more broadly
3 characterize this risk on a national scale and to support the interpretation of the more detailed
4 risk estimates generated for selected urban study areas. The overall scope and design of this
5 quantitative risk assessment, discussed below in chapters 2 and 3, reflect efforts to achieve these
6 goals.
7 This current quantitative risk assessment builds on the approach used and lessons learned
8 in the last PM risk assessment and attempts to reduce and better characterize overall uncertainly
9 associated with the analysis by incorporating a number of enhancements, in terms of both the
10 methods and data used in the analyses. This assessment covers a variety of health endpoints for
11 which, in staff s judgment, there is adequate information to develop quantitative risk estimates
12 that can meaningfully inform the review of the primary PM NAAQS. Evidence of relationships
13 between PM and other health endpoints for which, in staff s judgment, there currently is
14 insufficient information to develop meaningful quantitative risk estimates will be more generally
15 considered in the PA as part of the evidence-based considerations that inform staffs assessment
16 of policy options.
17 1.3 ORGANIZATION OF DOCUMENT
18 The remainder of this document is organized as follows. Chapter 2 provides an overview
19 of the scope of the quantitative risk assessment, including a summary of the previous risk
20 assessment, the original planned approach and the key design elements reflected in this second
21 draft assessment, and the rationale for the alternative standard levels evaluated in this
22 assessment. Chapter 3 describes the analytical approach, methods, and data used in conducting
23 the risk assessment, including the approach used to generate risk estimates for the set of urban
24 case studies included in this analysis and the approaches used in addressing variability and
25 uncertainty (Appendices A, B, and C provide supplemental information regarding the data and
26 methods used). Chapter 4 presents selected risk estimates generated for the urban case studies,
27 including the results of single- and multi-factor sensitivity analyses and a national-scale analysis
28 of the representativeness of relevant risk-related factors (Appendix D provides supplemental
29 information on risk-related factors; Appendices E and F provide detailed risk estimates and
30 sensitivity analysis results, respectively). Chapter 5 presents the approach used and results of a
31 national-scale assessment of PM2.5-related long-term mortality risks associated with recent air
32 quality (Appendix G provides supplemental information to the national-scale mortality analysis).
33 Chapter 6 provides an integrative discussion of the various risk estimates generated in these
34 assessments that draws on the results of the urban area case studies, the uncertainty/variability
February 2010 1-6 Draft - Do Not Quote or Cite
-------
1 characterization, and the national-scale analyses to inform our quantitative characterization of
2 PM-related risks to public health.
February 2010 1-7 Draft - Do Not Quote or Cite
-------
1 2 SCOPE
2 This chapter provides an overview of the scope and key design elements of this
3 quantitative health risk assessment. The design of this assessment began with a review of the
4 risk assessment completed during the last PM NAAQS review (Abt Associates, 2005; EPA,
5 2005, chapter 4), with an emphasis on considering key limitations and sources of uncertainty
6 recognized in that analysis.
7 As an initial step in the this PM NAAQS review, EPA invited outside experts,
8 representing a broad range of expertise (e.g., epidemiology, human and animal toxicology,
9 statistics, risk/exposure analysis, atmospheric science) to participate in a workshop with EPA
10 staff to help inform EPA's plan for the review. The participants discussed key policy-relevant
11 issues that would frame the review and the most relevant new science that would be available to
12 inform our understanding of these issues. One workshop session focused on planning for
13 quantitative risk/exposure assessments, taking into consideration what new research and/or
14 improved methodologies would be available to inform the design of a quantitative health risk
15 assessment and whether, and if so how, it might be appropriate to conduct a quantitative
16 exposure assessment. These workshop discussions informed the preparation of the IRP, which
17 included initial plans for quantitative risk and exposure assessments.
18 As a next step in the design of these quantitative assessments, OAQPS staff developed a
19 more detailed planning document, Particulate Matter National Ambient Air Quality Standards:
20 Scope and Methods Plan for Health Risk and Exposure Assessment (Scope and Methods Plan;
21 EPA, 2009b). This Scope and Methods Plan was the subject of a consultation with the CAS AC
22 PM Panel and public review on April 1-2, 2009 (at which the first draft ISA was also reviewed).
23 Based on consideration of CASAC and public comments on the Scope and Methods Plan and
24 information in the first draft ISA, we modified the scope and design of the risk assessment and
25 completed initial analyses that were presented in an initial draft of this RA (first draft RA; EPA,
26 2009e). The CASAC PM Panel met on October 5-6, 2009 to review the first draft RA (as well as
27 the second draft ISA).7 Based on consideration of CASAC (Samet, 2009) and public comments
28 on the first draft RA, together with ongoing refinement of elements of the risk assessment
29 approach informed by the second draft ISA, we have prepared this second draft RA.
30 In presenting the scope and key design elements of the current risk assessment, this
31 chapter first provides a brief overview of the risk assessment completed for the previous PM
32 NAAQS review in section 2.1, including key limitations and uncertainties associated with that
7 A public teleconference was held on November 12, 2009, during which CASAC reviewed the draft comment letter
prepared by the CASAC PM Panel.
February 2010 2-1 Draft - Do Not Quote or Cite
-------
1 analysis. Section 2.2 provides a summary of the initial design of the risk assessment as outlined
2 in the Scope and Methods Plan. Section 2.3 provides an overview of key design elements
3 reflected in this second draft risk assessment that reflect consideration of previous CASAC and
4 public comments. Section 2.4 provides a summary of the alternative air quality scenarios
5 simulated in this assessment, including recent air quality and the current and alternative suites of
6 PM2.5 24-hour and annual standards.
7 2.1 OVERVIEW OF RISK ASSESSMENT FROM LAST REVIEW
8 The quantitative risk assessment from the last review included a broad assessment of
9 PM2.5-related risk and a much more limited treatment of PMi0-2.5-related risk. That assessment
10 included estimates of risks of mortality (total non-accidental, cardiovascular, and respiratory),
11 morbidity (hospital admissions for cardiovascular and respiratory causes), and respiratory
12 symptoms (not requiring hospitalization) associated with short-term (24-hour) exposure to
13 ambient PM2.5 and risks of total, cardiopulmonary, and lung cancer mortality associated with
14 long-term exposure to PM2.5 in selected urban areas. Nine urban areas were selected across the
15 U.S.: Boston, MA; Detroit, MI; Los Angeles, CA; Philadelphia, PA; Phoenix, AZ; Pittsburgh,
16 PA; San Jose, CA; Seattle, WA; and St. Louis, MO.
17 The EPA recognized that there were many sources of uncertainly and variability inherent
18 in the inputs to the assessment and that there was a high degree of uncertainty in the resulting
19 PM2.5 risk estimates. Such uncertainties generally related to a number of important factors,
20 including: (a) the shape of the concentration-response (C-R) function (and whether or not a
21 population threshold exists); (b) issues related to the selection of appropriate statistical models
22 for the analysis of epidemiological data; (c) the role of potentially confounding and modifying
23 factors in the C-R relationships; (d) methods for simulating how daily PM2.s ambient
24 concentrations would likely change in any given area upon meeting a particular suite of
25 standards; and (e) the potential for differences in the relative toxicity of the components within
26 the mix of ambient PM2.5.
27 While some of these uncertainties were addressed quantitatively in the form of estimated
28 confidence ranges around central risk estimates, other uncertainties and the variability in key
29 inputs were not reflected in these confidence ranges, but rather were addressed through separate
30 sensitivity analyses or characterized qualitatively (EPA, 2005, chapter 4; Abt Associates, 2005).
31 The C-R relationships used in the quantitative risk assessment were based on findings from
32 human epidemiological studies that relied on fixed-site, population oriented, ambient monitors as
33 a surrogate for actual ambient PM2.s exposures. The assessment included a series of base case
34 estimates that, for example, included various cutpoints intended as surrogates for alternative
35 potential population thresholds. Other uncertainties were addressed in various sensitivity
February 2010 2-2 Draft - Do Not Quote or Cite
-------
1 analyses (e.g., the use of single- versus multi-pollutant models, use of single versus multi-city
2 models, use of a distributed lag model) and had a more moderate and often variable impact on
3 the risk estimates in some or all of the cities.
4 These same sources of uncertainty and variability were also applicable to the quantitative
5 risk assessment conducted for PMio-2.5 in the last review. However, the scope of the risk
6 assessment for PMio-2.5 was much more limited than that for PM2.5 reflecting the much more
7 limited body of epidemiological evidence and air quality information available for PMio-2.5. The
8 PMio-2.5 risk assessment included risk estimates for just three urban areas for two categories of
9 health endpoints related to short-term exposure to PMio-2.5: hospital admissions for
10 cardiovascular and respiratory causes and respiratory symptoms. While one of the goals of the
11 PMio-2.5 risk assessment was to provide estimates of the risk reductions associated with just
12 meeting alternative PMio-2.5 standards, EPA staff concluded that the nature and magnitude of the
13 uncertainties and concerns associated with this portion of the risk assessment weighed against
14 use of these risk estimates as a basis for recommending specific standard levels (EPA, 2005, see
15 p. 5-69). These uncertainties and concerns were summarized in the proposal notice (see FR 71
16 2662, January 17, 2006) and discussed more fully in the Staff Paper (EPA, 2005, chapter 4) and
17 associated technical support document (Abt Associates Inc., 2005).
18 2.2 ORIGINAL ASSESSMENT PLAN
19 The Scope and Methods Plan outlined a planned approach for conducting the current
20 quantitative PM risk assessment, including broad design issues as well as more detailed aspects
21 of the analyses. That document also outlined plans for a population exposure analysis based on
22 micro-environmental exposure modeling. The planned approaches for conducting both analyses
23 are briefly summarized below.
24 2.2.1 Risk Assessment
25 Key design elements for the quantitative risk assessment, as presented in the Scope and
26 Methods Plan, included:
27 PM size fractions: We planned to focus primarily on estimating risk associated with
28 exposure to PM2.s with a much more limited assessment of PMio-2.5- Regarding PM
29 components and ultrafine particles, we concluded that, based on review of evidence in
30 the first draft ISA, there was insufficient data to support quantitative risk assessment
31 at this time.
32 Selection of health effects categories (PM2.s): We planned to focus primarily on
33 categories for which the evidence supports a judgment that there is at least a likely
34 causal relationship. We also planned to consider including additional categories for
35 which evidence supports a judgment that there is a suggestive causal relationship
February 2010 2-3 Draft - Do Not Quote or Cite
-------
1 (e.g., reproductive, developmental outcomes), if sufficient information was available
2 to develop meaningful risk estimates for these additional categories.
3 Selection of health effect categories (PMi0-2.5): We planned to build on the limited
4 risk assessment conducted in the last review (EPA, 2005) with a focus on health
5 effect categories that staff judged to be sufficiently suggestive of a causal relationship
6 with short-term exposure to warrant analysis.
7 Selection of urban study areas: We planned to expand the number of urban study
8 areas to between 15 and 20, with selection of these study areas being based on
9 consideration of a number of factors (e.g., availability of location-specific C-R
10 functions and baseline incidence data, coverage for geographic heterogeneity in PM
11 risk-related attributes, coverage for areas with more vulnerable populations). We also
12 discussed the possibility of including more refined risk assessments for locations
13 where more detailed exposure studies had been completed (e.g., L.A., where a zip
14 code level analysis of long-term PM2.-exposure related mortality was presented in
15 Krewski et al., 2009).
16 Simulation of air quality levels that just meet current or alternative suites of
17 standards: We planned to consider the use of non-proportional air quality
18 adjustment methods in addition to the proportional approach that has been used
19 previously. These non-proportional adjustment methods could be based on (a)
20 historical patterns of reductions in urban areas, if these result in support for non-
21 proportional reductions across monitors and/or (b) model-based (e.g., CMAQ)
22 rollback designed to more realistically reflect patterns of PM reductions across
23 monitors in an urban area.
24 Characterization of policy relevant background (PRB): We planned to use
25 modeling (combination of the global-scale circulation model, GEOS-Chem, with the
26 regional scale air quality model, CMAQ) as presented in the first draft ISA, rather
27 than empirical data to characterize PRB levels for use in the risk assessment model.
28 Selection of epidemiological studies to provide C-R functions: We planned to
29 include both multi- and single-city studies (given advantages associated with both
30 designs) as well as multi- and single-pollutant studies, placing greater weight on the
31 use of C-R functions reflecting adjusted single-city estimates obtained from multi-city
32 studies.
33 Shape of the functional form of the risk model: We planned to emphasize non-
34 threshold C-R functions in the risk assessment model, based on the first draft ISA
35 conclusion that there was little support in the literature for population thresholds for
36 mortality effects associated with either long-term or short-term PM2.5 ambient
37 concentrations.8 We also stated that we may consider population thresholds as part of
38 the sensitivity analysis.
8 In discussing short-term exposure mortality studies, the first draft ISA (U.S. EPA, 2009a) indicated support for no-
threshold log-linear models, while acknowledging that the possible influence of exposure error and heterogeneity of
shapes across cities remains to be resolved.
February 2010 2-4 Draft - Do Not Quote or Cite
-------
1 Modeling of risk down to PRB versus lowest measured level (LML): We planned
2 to model risk down to LML for estimating risk associated with long-term PM2.5
3 exposures and down to PRB for estimating risks associated with short-term PM2.s
4 exposures.
5 Characterization of uncertainty and variability: We planned to include a
6 discussion in the risk assessment report on the degree to which the risk assessment
7 covers key sources of variability related to PM risk. For uncertainty, we planned to
8 include a qualitative discussion of key sources of uncertainty and provide ratings
9 (low, medium and high) in terms of their potential impact on risk estimates. We also
10 described the use of sensitivity analysis methods planned both to characterize the
11 potential impact of sources of uncertainty on risk estimates and to provide an
12 alternative set of reasonable estimates to supplement the main ("core") set of risk
13 estimates generated for the urban study areas.
14 National-scale assessment: We planned to conduct a limited national-scale
15 assessment of mortality associated with long-term exposure to recent ambient PM2.5
16 levels.
17 Representativeness analysis for the urban study areas: We planned to conduct an
18 analysis to evaluate the representativeness of the selected urban study areas against
19 national distributions for key PM risk-related attributes to determine whether they are
20 nationally representative or more focused on a particular portion of the distribution
21 for a given attribute.
22 2.2.2 Population Exposure Analysis
23 The Scope and Methods Plan also described a population exposure analysis based on
24 micro-environmental exposure modeling using the Air Pollution Exposure Model (APEX). The
25 planned analysis would have focused on PM2.5 and have involved a subset of the urban study
26 areas included in the risk assessment. The results of this analysis were planned to focus on
27 providing insights on population exposure with respect to informing the interpretation of
28 available epidemiological studies.
29 Following release of the Scope and Methods Plan, we continued development of the
30 approach for conducting a population exposure analysis, with the goal of completing the analysis
31 as part of the current PM review. However, this additional design work highlighted the need to
32 more clearly define the intended purpose of the analysis, including specific ways in which the
33 results would be used to interpret the estimates generated from the risk assessment (e.g.,
34 potentially identifying sources of exposure measurement error associated with the
35 epidemiological studies from which C-R functions were drawn for the risk assessment and the
36 magnitude of the impact of those sources of error on risk estimates). Taking CASAC comments
37 into consideration, which emphasized the same point regarding the importance of more clearly
38 defining how the exposure assessment results would be used, as well as the complexities
39 associated with designing and conducting such an assessment, we decided to continue methods
February 2010 2-5 Draft - Do Not Quote or Cite
-------
1 development work rather than attempt to complete a preliminary population exposure analysis as
2 part of this review. Development of the population exposure analysis methodology is ongoing,
3 and we anticipate that such an assessment could be conducted as part of the next PM NAAQS
4 review.
5 2.3 CURRENT SCOPE AND KEY DESIGN ELEMENTS
6 An overview of the scope and key design elements that are the basis for this second draft
7 RA are presented below, focusing on those aspects of the risk assessment approach which differ
8 from the originally planned approach.
9 PM size fractions: This quantitative risk assessment characterizes risk associated
10 with PM2.5-related exposures only. With regard to PMio-2.5, we have concluded that
11 continued limitations in data available for characterizing PMio-2.5 exposure and risk
12 would introduce significant uncertainty into a PMio-2.5 risk assessment such that the
13 risk estimates generated would be of limited utility in informing review of the
14 standard. This conclusion was reached by reviewing the set of limitations cited in the
15 last PM NAAQS risk assessment for not using the PMio-2.5 risk estimates in
16 recommending specific standard levels. We then considered whether health effects
17 data released since the last review (as summarized in the final PM ISA) as well as any
18 enhancements to the PMio-2.5 monitoring network would fundamentally address these
19 limitations. We concluded that significant limitations in both health effects data and
20 the PMio-2.5 monitoring network continue to exist such that a quantitative risk
21 assessment for PMio-2.5 is not supported at this time (a more in-depth discussion of the
22 rationale behind the decision not to conduct a quantitative risk assessment for PMio-2.5
23 is presented in Appendix H). Furthermore, based on the final PM ISA, we continue to
24 conclude that available data are too limited to support a quantitative risk assessment
25 for any specific PM components or for ultrafine particles (UFPs). We note, however,
26 that the evidence for health effects associated with thoracic coarse particles, PM
27 components, and UFPs will be included in the evidence-based considerations that will
28 be presented in the draft PA..
29 Selection of health effects categories (PM2.s): A multi-factor decision framework
30 was used to select the final set of health effects categories included in the risk
31 assessment for PM2.5 (section 3.3.1). This set of endpoints is consistent with those
32 outlined in the Scope and Methods Plan for PM2.5 (i.e., all of the selected endpoints
33 are from categories classified in the ISA as having a causal or likely causal
34 relationship with PM2.s exposure), although selecting endpoints limited to these two
35 classifications is a consequence of applying our multi-factor decision framework and
36 not the sole determining factor. A number of health effect categories classified as
37 suggestive of a casual relationship in the ISA (e.g., reproductive effects) were
38 considered, but were not selected for inclusion due in part to limited information
39 available to support selection of C-R functions for specific endpoints within these
40 health effect categories and/or lack of available baseline incidence data. Inn addition,
41 CAS AC members expressed differing views as to the appropriateness of including
42 these categories.
February 2010 2-6 Draft - Do Not Quote or Cite
-------
1 Selection of urban study areas: We have included 15 urban study areas in the risk
2 assessment, with the selection of these areas being based on a number of criteria
3 including: (a) consideration of urban study areas evaluated in the last PM risk
4 assessment; (b) consideration of locations evaluated in key epidemiological studies;
5 (c) preference for locations with relatively elevated 24-hour and/or annual PM2 5
6 monitored levels so that the assessment can provide potential insights into the degree
7 of risk reduction associated with just meeting the current and alternative suites of
8 standards; and (d) preference to include locations in different regions across the
9 country, reflecting potential differences in PM sources, composition, and potentially
10 other factors which might impact PM-related risk (section 3.3.2). Due in part to time
11 and resource limitations, we have not included a specialized analysis of risk based on
12 epidemiology studies using more highly-refined exposure analysis (e.g., the study of
13 L.A. involving zip code-level effect estimates, as presented in Krewski et al., 2009).
14 We have included consideration of studies with more refined surrogate measures of
15 exposure in our discussion of uncertainty related to long-term mortality, since they
16 can inform our interpretation of the degree of potential bias associated with the effect
17 estimates used to model risks (section 3.5.3).
18 Method used to develop composite monitor values: Ongoing methods
19 development has resulted in revisions to the methods used to derive composite
20 monitor values for both the annual and 24-hour distributions (section 3.2.1). The
21 revised methods ensure that monitors contributing to a composite calculation in a
22 particular study area are given equal weight, in contrast to the approach used in the
23 first draft RA, which effectively weighted monitors by their sampling frequency,
24 potentially leading to estimates that were biased high.
25 Simulation of air quality levels that just meet current or alternative suites of
26 standards: In addition to applying the proportional rollback approach used in the
27 first draft RA (and in the last risk assessment) to simulate PM2.5 ambient levels that
28 would "just meet" the current and alternative suites of standards, we have developed
29 and applied two alternative approaches (hybrid and peak-shaving) to help characterize
30 the uncertainty associated with this aspect of the assessment (section 3.2.3). We have
31 also refined our rollback approach for the Pittsburgh study area, using a dual-zone
32 approach to take into account monitor locations and the related topography in that
33 area (section 3.2.3).
34 Characterization of PRB: Consistent with the planned approach, we have used
35 regional PRB estimates generated using a combination of GEOS-Chem and CMAQ
36 modeling as presented in the ISA (section 3.2.2).
37 Selection of epidemiological studies to provide C-R functions: In modeling risk
38 associated with both short-term and long-term PM2.5 exposures, we have focused on
39 larger multi-city studies based on our conclusion that these studies provided more
40 defensible effect estimates. In modeling short-term exposure-related mortality and
41 morbidity, we obtained more spatially-refined effect estimates at the city- and
42 regional-levels, respectively (in both cases, these effects estimates are based on
43 application of Bayesian methods). We also included C-R functions selected from
44 several single city studies to provide coverage for additional health effect endpoints
February 2010 2-7 Draft - Do Not Quote or Cite
-------
1 associated with short-term PM2.5 exposures (e.g., emergency department visits).
2 Modeling of long-term exposure-related mortality focused on the latest reanalysis of
3 the ACS dataset (Krewski et al., 2009). This study expands upon previous
4 publications presenting evaluations of the ACS long-term cohort study and in
5 particular includes rigorous examination of different model forms for estimating
6 effects estimates (in addition to including updated and expanded datasets on
7 incidence and exposure). Our rationale for selecting the specific studies used in the
8 assessment, as well as our rationale for not selecting alternative studies, is discussed
9 below in section 3.3.3.
10 Characterization of uncertainty and variability: Our approach to characterizing
11 uncertainty and variability is based on application of the WHO Guidance on
12 Characterizing and Communicating Uncertainty In Exposure Assessment (WHO,
13 2008). This guidance provides a four-tiered approach for characterizing uncertainty
14 (and to a lesser extent variability) in the context of a risk assessment, with tiers
15 ranging from qualitative characterization (Tier 1) to use of full-probabilistic Monte
16 Carlo-based simulation (Tier 3). Sensitivity analysis methods, which are used in the
17 RA to assess sources of uncertainty and variability, represent a Tier 2 approach. The
18 application of single- and multi-factor sensitivity analysis methods in the RA serves
19 two purposes: (a) to characterize the potential magnitude of impact that a source(s) of
20 uncertainty and/or variability can have on risk estimates and (b) to provide an
21 additional set of reasonable risk estimates to supplement the "core" risk estimates in
22 characterizing the potential magnitude of uncertainty in the risk estimates. The
23 "core" risk estimates produced in this assessment refer to those generated using the
24 combination of modeling elements and input datasets in which we had the highest
25 confidence relative to other modeling choices (section 3.5.1 and 3.5.4).
26 National-scale assessment: As planned, we have conducted a limited national-scale
27 assessment of (chapter 5). This analysis provides estimates of mortality associated
28 with long-term exposure to recent ambient PM2 5 levels at the national scale, which
29 provides some context for considering the risks estimated for the urban study areas.
30 We continue to conclude that any expansion of this assessment (e.g., to include
31 additional health endpoints or additional air quality scenarios that simulate just
32 meeting alternative suites of standards), as suggested by some CAS AC Panel
33 members, was beyond the scope of what was needed or could reasonably be done
34 within the time and resources available for this review (section 5.1).
35 Representativeness analysis for the urban study areas: As planned, we have
36 conducted an analysis to evaluate the representativeness of the selected urban study
37 areas against national distributions for key PM risk-related attributes to determine
38 whether they are nationally representative or more focused on a particular portion of
39 the distribution for a given attribute (section 4.4).
40 Consideration of patterns in design values and ambient PMi.s monitoring data
41 across urban areas: We have included in this second draft assessment an
42 examination of how 24-hour and annual design values, together with patterns in PM2.5
43 monitoring data within an area, can influence the degree of risk reduction estimated to
44 occur upon just meeting the current or alternative suites of standards. This analysis
February 2010 2-8 Draft - Do Not Quote or Cite
-------
1 has resulted in a better understanding of the factors behind specific patterns of risk
2 reduction. We have also compared patterns of design values for the urban study areas
3 with patterns across the broader set of urban areas in the U.S. in order to help place
4 core risk estimates generated for the set of urban study areas in a broader national
5 context.
6 Integrated discussion of results and key observations: To enhance the utility of the
7 risk estimates generated for the 15 urban study areas in supporting the review of the
8 PM NAAQS, we have added a new chapter 6: Integrative Discussion of PM2.5-
9 Related Risks. This chapter integrates the core risk estimates generated for the 15
10 urban study areas with information from the sensitivity analyses and the qualitative
11 analysis of uncertainty, analyses of representativeness and patterns of design values,
12 and the national-scale mortality analysis.
13 2.4 ALTERNATIVE SUITES OF PM2.5 STANDARDS EVALUTATED
14 In developing estimates of risks associated with just meeting alternative suites of PM2.5
15 standards, we selected alternative levels for the annual and 24-hour PM2.5 standards during the
16 development of the first draft RA that we judged to be appropriate, drawing from the information
17 available to us at that time from the second draft ISA. In defining alternative suites of standards
18 to be evaluated, we identified alternative standard levels in conjunction with the averaging times
19 (24-hour and annual) and forms for the current suite of standards.9 We note that all of the basic
20 elements of the standards (e.g., indicator, averaging time, level, and form) will be discussed in a
21 forthcoming draft Policy Assessment which will present staff conclusions based on both
22 evidence-based and risk-based considerations to inform judgments that the EPA Administrator
23 must make in deciding whether to retain or revise the existing suite of PM standards.
24 In selecting alternative levels for the annual and 24-hour PM2.5 standards for the purpose
25 of evaluation in the quantitative risk assessment, we considered ambient air quality levels
26 associated with health effects in epidemiological studies of long- and short-term exposure to
27 PM2.5, as assessed in the second draft ISA. As discussed further below (section 3.3.3), in
28 selecting alternative levels for consideration in the risk assessment, we placed emphasis on air
29 quality information from multi-city studies because these studies have a number of advantages
30 compared to single-city studies including: (1) multi-city studies reflect ambient PM2.5 levels and
31 potential health impacts across a range of diverse locations; (2) multi-city studies "clearly do not
32 suffer from potential omission of negative analyses due to 'publication bias'" (EPA, 2004a, p. 8-
33 30); and (3) multi-city studies generally have higher statistical power.
9 The "form" of a standard defines the air quality statistic that is compared to the level of the standard in determining
whether an area attains the standard. The form of the 24-hour PM25 standard is the 98th percentile of the distribution
of 24-hour PM2 5 concentrations at each population-oriented monitor within an area, averaged over 3 years. The
form of the annual PM2 5 standard is an annual arithmetic mean, averaged over 3 years, from single or multiple
community-oriented monitors.
February 2010 2-9 Draft - Do Not Quote or Cite
-------
1 Specifically, regarding alternative levels for the annual PM2.5 standard to be evaluated in
2 this risk assessment, we first considered long-term average PM2.5 concentrations associated with
3 health effects observed in long-term epidemiological studies, as summarized in Figure 2-2 of the
4 second draft ISA. The second draft ISA concluded that the association between increased risk of
5 mortality and long-term PM2.5 exposure becomes more precise and consistently positive in
6 locations with mean PM2.5 concentrations of 13.5 |ig/m3 and above. (EPA, 2009a, section
7 2.3.1.2). The second draft ISA also concluded that the strongest evidence for cardiovascular-
8 related effects related to long-term PM2.5 exposures has been reported in large, multi-city U.S.-
9 based studies and, specifically, one of these studies, the Women's Health Initiative (WHI) Study,
10 reports associations between PM2.5 and cardiovascular effects among post-menopausal women
11 with a mean annual average PM25 concentration of 13.5 |ig/m3 (EPA, 2009a, section 2.3.1.2). In
12 addition, we evaluated long-term average PM2.5 concentrations in short-term exposure studies
13 that reported statistically significant effects. More specifically, as reported in the second draft
14 ISA, both cardiovascular and respiratory morbidity effects (e.g., emergency department visits,
15 hospital admissions) have been observed and become more precise and consistently positive in
16 locations with mean PM2s concentrations of 13 |ig/m3 and above (EPA, 2009a, section 2.3.1;
17 also see Figure 2-1).10
18 Based on the available epidemiological evidence indicating effects associated with a
19 range of annual averaged PM2 5 concentrations, as briefly described above, we selected levels of
20 12 and 13 |ig/m3 as the alternative annual standard levels to be evaluated in the quantitative risk
21 assessment. We have added 14 |ig/m3to the set of annual levels evaluated in this second draft
22 RA to provide fuller coverage for the range of values between the current annual standard level
23 of 15 |ig/m3 and the lowest level evaluated.
24 In identifying alternative levels for the 24-hour PM2 5 standard to be evaluated in this risk
25 assessment, we considered the ambient PM2.5 levels associated with mortality and morbidity
26 effects as reported in key short-term epidemiological studies. We focused on the 98th percentile
27 PM2.s ambient levels reported in two multi-city studies that provided C-R functions used in the
28 core risk assessment, Zanobetti and Schwartz (2009) and Bell et al. (2008). The focus on the
29 98th percentile of the 24-hour PM2 5 concentrations observed in the epidemiological studies is
30 consistent with the approach used in the prior PM NAAQS review and is consistent with the
31 current form of the 24-hour PM2 5 standard.
10 We note that the association between long-term mean ambient PM25 levels and statistically-significant health
effects reported in short-term exposure studies would be dependent on the specific relationship between day-to-day
variation in the 24-hour PM2 5 levels (in the underlying study counties) and the associated long-term mean PM2 5
levels (i.e., the association between mean PM2 5 levels and short-term health effects, would not hold for counties
with notably different relationships between short-term day-to-day variation and longer-term mean PM2 5 levels).
February 2010 2-10 Draft - Do Not Quote or Cite
-------
1 The second draft ISA presented 98th percentile 24-hour PM2.5 values for each of the 112
2 urban areas included in the Zanobetti and Schwartz (2009) short-term mortality study (EPA,
3 2009a, Figure 6-22). We evaluated the trend in these county-level 98th percentile 24-hour PM2 5
4 levels in conjunction with the statistical significance of the associated county-level effect
5 estimates. If we had found an association between the air quality levels and statistically
6 significant effect estimates (i.e., higher 98th percentile PM2.5 levels were consistently associated
7 with statistically significant effect estimates), then it would have been reasonable to consider the
8 lowest 98th percentile PM2.5 level associated with the set of counties for which a statistically
9 significant effect estimates was observed as the basis for selecting an alternative standard level
10 for evaluation in this risk assessment. However, no such association was observed. Rather, we
11 observed mixed results with no clear correlation between 98th percentile air quality levels and
12 statistically significant effect estimates. Therefore, we focused on the overall range of 98th
13 percentile values across the entire set of counties and considered the lower quartile of that
14 distribution as representative of a reasonably precautionary approach for identifying alternative
15 levels for consideration in the risk assessment. The 10th and 25th percentiles values were 25.5
16 and 29.8 |ig/m3, respectively (Zanobetti, 2009). We note that the overall 98th percentile value
17 across the entire set of urban areas analyzed in Zanobetti and Schwartz. (2009) was 34.3 |ig/m3
18 (EPA, 2009a, Figure 2-1; Zanobetti and Schwartz, 2009)
19 We also completed a similar analysis of the county-level ambient air quality data (Bell,
20 2009) for the 202 counties associated with the Bell et al. (2008) study. Analysis of the overall
21 distribution of 98th percentile values across the entire dataset resulted in identifying 10th and 25th
22 percentile values of about 24.4 and 29.3 |ig/m3, respectively. We note that the overall 98th
23 percentile value across the entire set of counties analyzed in Bell et al. (2008)) was 34.2 |ig/m3
24 (EPA, 2009a, Table 6-11; Bell, 2009).
25 Based on the available epidemiological evidence indicating effects associated with a
26 range of 98th percentile 24-hour PM2.5 concentrations, as briefly described above, we selected
27 levels of 25 and 30 |ig/m3 as the alternative 24-hour standard levels to be evaluated in this
28 quantitative risk assessment.
29 Once alternative levels were identified for the annual and 24-hour PM standards, we then
30 identified specific combinations of these standard levels to be considered in evaluating suites of
31 alternative standards in the risk assessment. In selecting the pairing of annual and 24-hour
32 standard levels, we considered which standard was likely to be controlling across the set of 15
33 urban study areas (either the annual or 24-hour standard will be the "controlling standard" at a
February 2010 2-11 Draft - Do Not Quote or Cite
-------
1 given location, depending on the design value associated with that location).11 For this risk
2 assessment, the goal was to select combinations of annual and 24-hour levels that would result in
3 a mixture of behavior in terms of which standards would control across the various urban study
4 areas. For example, with the 12/35 combination (i.e., an annual standard level of 12 |ig/m3 and a
5 24-hour standard level of 35 |ig/m3), the annual level of 12 |ig/m3 is the controlling standard for
6 all 15 urban study areas, while with the 12/25 combination, the annual standard is the controlling
7 standard at some locations and the 24-hour standard is the controlling standard at other locations.
8 Consideration of these factors resulted in a set of five alternative combinations of annual and 24-
9 hour standards being identified for inclusion in the risk assessment.
10 The full set of air quality scenarios included in the risk assessment, including the recent
11 conditions air quality scenario and current standards scenario along with the five alternative sets
12 of standards are as follows:
13 Recent conditions (risk estimates based on ambient PM2.5 monitoring data for the
14 analysis period - 2005 to 2007)
15 Current PM2.5 NAAQS: annual 15 |ig/m3; 24-hour 35 |ig/m3
16 Alternative PM2.5 standards: annual 14 |ig/m3; 24-hour 35 |ig/m3
17 Alternative PM2 5 standards: annual 13 |ig/m3; 24-hour 35 |ig/m3
18 Alternative PM2 5 standards: annual 12 |ig/m3; 24-hour 35 |ig/m3
19 Alternative PM2.s standards: annual 13 |ig/m3; 24-hour 30 |ig/m3
20 Alternative PM2.5 standards: annual 12 |ig/m3; 24-hour 25 |ig/m3.
1: The controlling standard is the standard which requires the greatest percentage reduction to get the design value
monitor to meet that standard - see section 3.3.3 for additional detail on the issue of controlling standards.
February 2010 2-12 Draft - Do Not Quote or Cite
-------
1 3 URBAN CASE STUDY ANALYSIS METHODS
2 This chapter provides an overview of the methods used in the risk assessment. Section
3 3.1 discusses the basic structure of the risk assessment, identifying the modeling elements and
4 related sources of input data needed for the analysis. Section 3.2 discusses air quality
5 considerations. Section 3.3 discusses the selection of health endpoints, urban study areas and C-
6 R functions from key epidemiological studies used in modeling those endpoints. Section 3.4
7 discusses baseline health effects incidence rates. Finally, section 3.5 describes how uncertainty
8 and variability are addressed in the risk assessment.
9 3.1 GENERAL APPROACH
10 3.1.1 Basic Structure of the Risk Assessment
11 The general approach used in both the prior and the current PM risk assessment relies
12 upon C-R functions which have been estimated in epidemiological studies. Since these studies
13 estimate C-R functions using ambient air quality data from fixed-site, population-oriented
14 monitors, the appropriate application of these functions in a PM risk assessment similarly
15 requires the use of ambient air quality data at fixed-site, population-oriented monitors.
16 The general PM health risk model, illustrated in Figure 3-1, combines information about
17 PM air quality for specific urban areas with C-R functions derived from epidemiological studies,
18 baseline health incidence data for specific health endpoints, and population estimates to derive
19 estimates of the annual incidence of specified health effects attributable to ambient PM
20 concentrations under different air quality scenarios. This assessment was implemented within
21 TRIM.Risk, the component of EPA's Total Risk Integrated Methodology (TRIM) model that
22 estimates human health risks.12
23 The analyses conducted for this review focused on estimating risks associated with recent
24 PM2.5 air quality and estimating changes in these risks associated with air quality simulated to
25 reflect just meeting the current suite of PM2 5 ambient standards, as well as any additional
26 reductions in incidence estimated to occur upon just meeting alternative suites of PM2.5
27 standards.
28 Consistent with past risk assessments for NAAQS reviews, this risk assessment is
29 intended to estimate risks attributable to anthropogenic sources and activities only. Therefore, for
30 all health endpoints associated with short-term exposure to PM2.5, the risk assessment considers
31 only the incidence of health effects associated with PM2.5 concentrations in excess
12 For more detailed information about TRIM.Risk, see: http://www.epa.gov/ttn/fera/trim_risk.html
February 2010 3-1 Draft - Do Not Quote or Cite
-------
Figure 3-1. Major components of particulate matter health risk assessment.
Air Quality
Ambient Population-
Oriented Monitoring and
Estimated Policy
Relevant Background
Levels for Selected Cities
Air Quality Adjustment
Procedures
Alternative Proposed
Standards
Concentration-Response
Human Epidemiological
Studies (various health
endpoints)
Estimates of City-specific
Baseline Health Effects
Incidence Rates
(various health
endpoints) and
Population Data
Concentration
Response
Relationships
Risk Estimates:
Recent Air
Quality
Alternative
Scenarios
February 2010
3-2
Draft - Do Not Quote or Cite
-------
of policy relevant background (PRB) levels. In the studies estimating a relationship between
mortality and long-term exposure to PM2.5, however, the lowest measured levels (LMLs)
reported in the epidemiological studies were substantially above PRB. Thus, estimating risk
down to PRB would have required substantial extrapolation of the estimated C-R functions
below the range of the data on which they were estimated. Therefore, we estimated risk only
down to the LML to avoid introducing additional uncertainty related to this extrapolation into
this analysis. To provide consistency for the different C-R functions selected from the long-term
exposure studies, and, in particular, to avoid the choice of LML unduly influencing the results of
the risk assessment, we selected a single LML - 5.8 |ig/m3 from the later exposure period
evaluated in Krewski et al. (2009) to be used in estimating risks associated with long-term
PM2 5 exposures.
For each health effect that has been associated with PM2.5, the risk assessment may be
viewed as assessing the incidence of the health effect associated with PM2.5 concentrations under
a given air quality scenario (e.g., a scenario in which PM2.5 concentrations just meet a specified
suite of standards) above PRB or the LML . Equivalently, the risk assessment may be viewed as
assessing the change in incidence of each health effect associated with a change in PM2.5
concentrations from some higher level (e.g., PM2.5 concentrations that just meet a specified suite
of standards) to specified lower levels (PRB levels or the LML).
The risk assessment procedures described in more detail below are diagramed in Figure
3-2 for analyses based on short-term exposure studies and in Figure 3-3 for analyses based on
long-term exposure studies. To estimate the change in incidence of a given health effect
resulting from a given change in ambient PM2 5 concentrations in an assessment location, the
following analysis inputs are necessary:
Air quality information including: (1) PM2.5 air quality data from one or more recent
years from population-oriented monitors in the assessment location, (2) estimates of
PM2.s PRB concentrations appropriate to this location, and (3) a method for adjusting the
air quality data to reflect patterns of air quality changes to simulate just meeting the
current or alternative suite of PM2.5 standards. (These air quality inputs are discussed in
more detail in section 3.2).
C-R function(s) which provide an estimate of the relationship between the health
endpoint of interest and PM2 5 concentrations (preferably derived in the assessment
location, although functions estimated in other locations can be used at the cost of
increased uncertainty - see section 3.5.3). For PM2.5, C-R functions are available from
epidemiological studies that assessed PM2.s-related health effects associated with either
short- or long-term exposures. (Section 3.1.2 describes the role of C-R functions in
estimating health risks associated with PM2.5).
Baseline health effects incidence rate and population The baseline incidence rate
provides an estimate of the incidence rate (number of cases of the health effect per year,
February 2010 3-3 Draft - Do Not Quote or Cite
-------
usually per 10,000 or 100,000 general population) in the assessment location
corresponding to recent ambient PM2 5 levels in that location. To derive the total baseline
incidence per year, this rate must be multiplied by the corresponding population number
(e.g., if the baseline incidence rate is number of cases per year per 100,000 population, it
must be multiplied by the number of 100,000s in the population). (Section 3.4
summarizes considerations related to the baseline incidence rate and population data
inputs to the risk assessment).
February 2010 3-4 Draft - Do Not Quote or Cite
-------
Figure 3-2. Flow diagram of risk assessment for short-term exposure studies.
Air Quality Data
Concentration-Response Functions
Identity
location-
specific
studies
Identify
Relative Risk
or slope
coefficients (B)
i Convert RR |
> (if necessary j
Identify
functional form
| Speciyrolback
method I
(for certain I
! analyses) i
Baseline Health Incidence
Compute
annual
n
associated
with
change in PM
Estimate of
percent change in
total incidence
Estimate of
PM-associated
incidence
February 2010
3-5
Draft - Do Not Quote or Cite
-------
Figure 3-3. Flow diagram of risk assessment for long-term exposure studies.
Air Quality Data
Concentration-Response Functions
Identify
studies
Identifyfunctional
form
Identify
Relative Risk
(RR) or slope
coefficients (B)
i
r
1 Convert RR
* * toB
{if necessary
Baseline Health Incidence
I Specify roltisck
method
(for certain I
analyses) I
Estimate of
percent change in
total incidence
Compute #
associated
with
change in PM
Estimate of
PM-associated
incidence
February 2010
3-6
Draft - Do Not Quote or Cite
-------
1 The risk assessment was carried out using three years of recent air quality data from
2 2005, 2006, and 2007 (see section 3.2.1). We matched the population data used in the risk
3 assessment to the year of the air quality data. For example, when we used 2005 air quality data,
4 we used 2005 population estimates. It was not possible to obtain the necessary data to calculate
5 baseline incidence rates separately for each of the three years for each of the risk assessment
6 locations, therefore, we calculated these rates for a single year, under the assumption that these
7 rates are unlikely to have changed significantly from 2005 to 2007. The calculation of baseline
8 incidence rates is described in detail in section 3.4.
9 For this risk assessment, we developed a core (primary) set of risk results based on the
10 application of modeling element choices (e.g., C-R functions, lag periods) that we believe have
11 the greatest overall support in the literature (hereafter referred to as the "core" results). While it
12 is not possible at this time to assign quantitative levels of confidence to these core risk estimates,
13 we do believe these estimates are generally based on inputs having higher overall levels of
14 confidence relative to risk estimates that could have been generated using other inputs identified
15 in the literature.
16 In addition, as discussed above in section 2.1 and later in section 3.5, we have also used
17 single-element and multi-element sensitivity analysis techniques to generate a set of reasonable
18 alternative risk estimates based on the application of alternative modeling element choices that,
19 while not having as much support in the literature as those used in the core analysis, do still
20 represent plausible inputs. The results of these sensitivity analyses allow us to gain insights into
21 which sources of uncertainty and variability may have the greatest impact on risk estimates when
22 acting alone, or in combination with other sources of uncertainty. The sensitivity analysis-based
23 risk estimates also provide us with an additional set of reasonable risk results that allow us to
24 place the results of the core analysis in context with regard to uncertainty. A number of
25 modeling elements were used in differentiating core analyses from sensitivity analyses (e.g., C-R
26 function shape, alternative effect estimates, alternative lag structures, different methods used to
27 rollback air quality to simulate attainment to current or alternative standard levels, application of
28 PRB versus LML). Specific choices made in relation to individual modeling elements in
29 differentiating the core analysis from sensitivity analyses are described, as appropriate, in the
30 sections that follow, which cover specific aspects of the risk assessment design. The potential
31 utility of the sensitivity analysis-based risk estimates in informing consideration of uncertainty
32 and variability in the core results is discussed in section 4.5.2.
33 3.1.2 Calculating PM2.5-Related Health Effects Incidence
34 The C-R functions used in the risk assessment are empirically estimated relations
35 between average ambient concentrations of PM2.5 and the health endpoints of interest (e.g.,
February 2010 3-7 Draft - Do Not Quote or Cite
-------
1 mortality or hospital admissions reported by epidemiological studies for specific locations). This
2 section describes the basic method used to estimate changes in the incidence of a health endpoint
3 associated with changes in PM2 5, using a "generic" C-R function of the most common functional
4 form.
5 Although some epidemiological studies have estimated linear C-R functions and some
6 have estimated logistic functions, most of the studies used a method referred to as "Poisson
7 regression" to estimate exponential (or log-linear) C-R functions in which the natural logarithm
8 of the health endpoint is a linear function of PM2 5:
9
10 y = Beftc (1)
11
12 where x is the ambient PM2.5 level, y is the incidence of the health endpoint of interest at
13 PM2.s level x, p is the coefficient of ambient PM2.5 concentration, and B is the incidence at x=0,
14 i.e., when there is no ambient PM2.5. The relationship between a specified ambient PM2.5 level,
15 XQ, for example, and the incidence of a given health endpoint associated with that level (denoted
16 asj/o)isthen
17
18 y0=Beflc" (2)
19
20 Because the log-linear form of a C-R function (equation (1)) is by far the most common
21 form, we use this form to illustrate the "health impact function" used in the PM2.5 risk
22 assessment.
23 If we let x0 denote the baseline (upper) PM2 5 level, and x} denote the lower PM2 5 level,
24 and yo and yi denote the corresponding incidences of the health effect, we can derive the
25 following relationship between the change in x, Ax= (XQ- xj), and the corresponding change my,
26 Ay, from equation (I).13
27 Aj = (j0-j1) = Jo[l-e-^]. (3)
28
29 Alternatively, the difference in health effects incidence can be calculated indirectly using
30 relative risk. Relative risk (RR) is a measure commonly used by epidemiologists to characterize
31 the comparative health effects associated with a particular air quality comparison. The risk of
32 mortality at ambient PM2.5 level x0 relative to the risk of mortality at ambient PM2.5 level xj, for
13 If Ax < 0 - i.e., if Ax = (xr x0) - then the relationship between Ax and Ay can be shown to be
Ay = (yl - y0) = y0 [efA* - 1]. If Ax < 0, Ay will similarly be negative. However, the magnitude of Ay will be the
same whether Ax>OorAx<0- i.e., the absolute value of Ay does not depend on which equation is used.
February 2010 3-8 Draft - Do Not Quote or Cite
-------
1 example, may be characterized by the ratio of the two mortality rates: the mortality rate among
2 individuals when the ambient PM2 5 level is XQ and the mortality rate among (otherwise identical)
3 individuals when the ambient PM2 5 level is x}. This is the RR for mortality associated with the
4 difference between the two ambient PM2 5 levels, XQ and xj. Given a C-R function of the form
5 shown in equation (1) and a particular difference in ambient PM2 5 levels, Ax, the RR associated
6 with that difference in ambient PM2.5, denoted as RR-Ax, is equal to epAx. The difference in health
7 effects incidence, Ay, corresponding to a given difference in ambient PM2.5 levels, Ax, can then
8 be calculated based on this RRAx as:
9
10 4y = (.y0 -y^ = y0[!-(!/RR^)]- (4)
11
12 Equations (3) and (4) are simply alternative ways of expressing the relationship between
13 a given difference in ambient PM2 5 levels, Ax > 0, and the corresponding difference in health
14 effects incidence, Ay. These health impact equations are the key equations that combine air
15 quality information, C-R function information, and baseline health effects incidence information
16 to estimate ambient PM2 5 health risk.
17 3.1.2.1 Short-term vs. Long-term Exposure
18 Concentration-response (C-R) functions that use as input annual average PM2 5 levels (or
19 some function of these, such as the average over a period of several years) relate these to the
20 annual incidence of the health endpoint - i.e., in such studies x in equation (1) above is the
21 average PM2 5 concentration over a period of one or more years, meant to represent long-term
22 exposure, and_y is the annual incidence of the health effect associated with that long-term
23 exposure.
24 Concentration-response (C-R) functions that use as input 24-hour average PM2 5 levels (or
25 some function of these, such as the average over one or more days) relate these to the daily
26 incidence of the health endpoint - i.e., in such studies x in equation (1) above is the average
27 PM2 5 concentration over a period of one or a few days (short-term exposure), andy is the daily
28 incidence of the health effect associated with that short-term exposure.
29 There are several variants of the short-term (daily) C-R function. Some C-R functions
30 were estimated by using moving averages of ambient PM2.5 to predict daily health effects
31 incidence. Such a function might, for example, relate the incidence of the health effect on day t
32 to the average of PM2 5 concentrations on days t and (/-I). Some C-R functions consider the
33 relationship between daily incidence and daily average PM2.5 lagged a certain number of days.
34 For example, a study might estimate the C-R relationship between mortality on day t and average
35 PM2 5 on a prior day (t-l). A few studies have estimated distributed lag models, in which health
February 2010 3-9 Draft - Do Not Quote or Cite
-------
1 effect incidence is a function of PM2.5 concentrations on several prior days - that is, the incidence
2 of the health endpoint on day Hs a function of the PM2.5 concentration on day t, day (/-I), day (t-
3 2), and so forth. Such models can be reconfigured so that the sum of the coefficients of the
4 different PM2.5 lags in the model can be used to predict the changes in incidence on several days.
5 For example, corresponding to a change in PM on day tin a distributed lag model with 0-day, 1-
6 day, and 2- day lags considered, the sum of the coefficients of the 0-day, 1-day, and 2-day lagged
7 PM2.s concentrations can be used to predict the sum of incidence changes on days t, (M-l) and
8 0+2).
9 Most daily time-series epidemiological studies estimated C-R functions in which the PM-
10 related incidence on a given day depends only on same-day PM concentration(i.e. lag 0), the
11 previous-day PM concentration (i.e. lag 1), or some variant of those, such as a two-day average
12 concentration (e.g. lag 0-1). Such models necessarily assume that the longer pattern of PM
13 levels preceding the PM concentration on a given day does not affect mortality or morbidity on
14 that day. To the extent that PM-related mortality on a given day is affected by PM concentrations
15 over a longer period of time, then these models would be mis-specified, and this mis-
16 specification would affect the predictions of daily incidence based on the model.
17 The extent to which time-series studies using single-day PM2.5 concentrations may under
18 or over-estimate the relationship between short-term PM2 5 exposure and risk of mortality is
19 unknown. However, there is some evidence, based on analyses of PMi0 data, that mortality or
20 morbidity on a given day is influenced by prior PM exposures up to more than a month before
21 the date of death (Schwartz, 2000). The extent to which short-term exposure studies (including
22 those that consider distributed lags) may not capture the full impact of long-term exposures to
23 PM2.5 is similarly not adequately understood, although the current evidence (e.g., Krewski et al.,
24 2009; Krewski et al., 2000) suggests that there is a substantial impact of long-term exposures on
25 health effects that is not picked up in the short-term exposure studies.
26 3.1.2.2 Calculating Annual Incidence
27 The risk assessment estimated health effects incidence, and changes in incidence, on an
28 annual basis, for 2005, 2006, and 2007. For mortality, both short-term and long-term exposure
29 studies have reported estimated C-R functions. As noted above, most short-term exposure C-R
30 functions estimated by daily time-series epidemiological studies relate daily mortality to same-
31 day PM2 5 concentration or previous-day PM2 5 concentration (or some variant of those).
32 To estimate the daily health impacts of 24-hour average ambient PM2.5 levels above PRB,
33 C-R functions from short-term exposure studies were used together with estimated changes in
34 24-hour ambient PM2 5 concentrations to calculate the daily changes in the incidence of the
February 2010 3-10 Draft - Do Not Quote or Cite
-------
1 health endpoint. After daily changes in health effects were calculated, an annual change was
2 calculated by summing the daily changes.
3 The mortality associated with long-term exposure is likely to include mortality related to
4 short-term exposures as well as mortality related to longer-term exposures. As discussed
5 previously, estimates of daily mortality based on the time-series studies also are likely influenced
6 by prior PM exposures. Therefore, the estimated annual incidences of mortality calculated based
7 on the short- and long-term exposure studies are not likely to be completely independent and
8 should not be added together. While we can characterize the statistical uncertainty surrounding
9 the estimated PM2.5 coefficient in a reported C-R function, there are other sources of uncertainty
10 associated with the C-R functions used in the risk assessment that are addressed via sensitivity
11 analyses and/or qualitatively discussed in section 3.5.3.
12 3.2 AIR QUALITY INPUTS
13 3.2.1 Characterizing Recent Conditions
14 As noted earlier, a major input to the PM2.5 risk assessment is ambient PM2.5 air quality
15 data for each assessment location. Twenty-four hour PM2.5 air quality data for 2005, 2006, and
16 2007 were obtained for each of the urban study areas from monitors in EPA's Air Quality
17 System (AQS). To characterize PM2.5 air quality in each risk assessment location as accurately
18 as possible, we used only those monitors that were located within the county or counties that
19 were analyzed in the epidemiological studies used to select C-R functions. In a few cases, an
20 urban area was delineated differently by two or more epidemiological studies used in the risk
21 assessment. For example, Birmingham, AL was defined as Blount, Jefferson, Shelby, St. Clair,
22 and Walker Counties in one study and as only Jefferson County in another study. In such cases,
23 we matched our delineation of the urban study area to that used in each study, resulting in two or
24 more different delineations of the urban study area and identified them as, for example,
25 Birmingham 1 and Birmingham 2. The counties and the number of air quality monitors included
26 within each urban area are given in Table 3-1.
27 In order to be consistent with the approach generally used in the epidemiological studies
28 that estimated PM2 5 C-R functions, the average ambient PM2 5 concentration on each day for
29 which measured data were available was deemed most appropriate for use in the risk assessment
30 (i.e., we created a composite monitor average). Consistent with the approach used in the prior
31 PM risk assessment, a composite monitor data set was created for each assessment location
32 based on a composite of all monitors located within each urban study area. For this risk
33 assessment, we have used an approach for creating composite monitors (see description below)
34 that reflects equal weighting of monitors in computing both 24-hour and annual composite
35 monitor values. (This reflects a change from the approach used in the first draft RA which
February 2010 3-11 Draft - Do Not Quote or Cite
-------
1 weighted monitors by sampling frequency - an approach which could result in bias being
2 introduced into the analysis.)
3 To calculate daily averages at the composite monitor for a location, we first checked the
4 number of observations at each monitor at that location. If a monitor had fewer than 11
5 observations in a quarter of the year (three months, the first quarter being January, February, and
6 March), we left the days in that quarter without observations as missing. If a monitor had at least
7 11 observations in a quarter, we filled in the missing days at that monitor in that quarter as
8 follows: For each series of seven or fewer consecutive days with missing values, we took the
9 average of the closest day with a reported value before the missing days and the closest day with
10 a reported value after the missing days, and we assigned that average to all days in the series of
11 missing days. If a series of consecutive missing days was greater than seven, we did not fill
12 them in. After the missing days at monitors had been filled in as described, we calculated the
13 composite monitor value for a given day as the average of values across all monitors for that day.
14 If there were any days for which the composite monitor value was missing, we filled them in
15 with 7-day moving averages (i.e., an average of the 3 days before and the 3 days after the
16 missing day). Given the approach for interpolating missing days at individual monitors (just
17 described), the incidence of missing days at composite monitors was very low. The numbers of
18 monitors in the risk assessment locations are given in Table 3-1.
19 To calculate annual averages at the composite monitor for a location, we first checked the
20 number of observations in each quarter of each year at each monitor at the location. If a monitor
21 had fewer than 11 observations in a quarter of the year, we set the quarterly average at that
22 monitor to "missing." If the monitor had at least 11 observations in a quarter, we calculated the
23 quarterly average at the monitor as the average of the reported observations at the monitor in that
24 quarter. For each quarter of the year, we then calculated the composite monitor quarterly
25 average as the average of the monitor-specific quarterly averages. The annual average at the
26 composite monitor was then calculated as the average of the four composite monitor quarterly
27 averages.14
28
14 Pittsburgh was treated somewhat differently from the other locations because there are effectively two attainment
areas in Pittsburgh - one containing ten of the monitors we're using in the risk assessment ("Pittsburgh-1"), and the
other containing the remaining 2 monitors ("Pittsburgh-2"). We treated each of these two sets of monitors as a
separate "location," and calculated both daily and annual composite monitor values in each "location." We then
calculated composite monitor values for Pittsburgh as weighted averages of the composite monitor values for
"Pittsburgh-1" and "Pittsburgh-2", where the weights were the proportion of the monitors in each (i.e., 10/12 and
2/12).
February 2010 3-12 Draft - Do Not Quote or Cite
-------
1
2
Table 3-1. Numbers of Monitors in Risk Assessment Locations From Which Composite
Monitor Values Were Calculated*
Risk Assessment
Location
Atlanta, GA - 1
Atlanta, GA - 2
Atlanta, GA - 3
Baltimore, MD
Birmingham, AL - 1
Birmingham, AL - 2
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY- 1***
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO - 1
St. Louis, MO - 2
Tacoma, WA
Counties
Cobb, De Kalb, Fulton, Gwinnett
Cobb, De Kalb, Fulton
20-County MSA**
Baltimore city, Baltimore county
Blount, Jefferson, Shelby, St. Clair, Walker
Jefferson
Dallas
Wayne
Fresno
Harris
Los Angeles
Kings, New York City (Manhattan), Queens, Richmond, Bronx
Philadelphia
Maricopa
Allegheny
Salt Lake
Jefferson, Madison (IL), St. Louis, St. Louis City, St. Clair (IL)
Madison (IL), St. Louis, St. Louis City, St. Clair (IL)
Pierce
Number of Monitors
8
7
10
8
10
8
6
9
3
6
10
12
7
5
12
7
15
14
1
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
* Calculation of composite monitor values is described in the text above.
** Barrow, Bartow, Carroll, Cherokee, Clayton, Cobb, Coweta, DeKalb, Douglas, Fayette, Forsyth, Fulton, Gwinett,
Henry, Newton, Paulding, Pickens, Rockdale, Spalding, and Walton.
*** The sets of monitors for New York (Manhattan) have l-in-3 day sampling, with sampling schedules synced
across monitors. This means that for the three year simulation period, roughly 2/3 of the days (i.e., 731) had no
monitor coverage for the New York urban study area, resulting in a need to interpolate estimates for these days (for
the composite monitor) using the approach described above. Similarly, with Tacoma, the single monitor at that
location also has 1 in 3 day sampling, resulting again, in 2/3 of the days not having data with interpolation being
used to derive estimates for those days (for the composite monitor).
Appendix A summarizes the PM2.s air quality data that were used in each of the
assessment locations, including quarterly and annual counts, quarterly and annual averages, and
the 98th percentile of the daily (24-hour) averages.
3.2.2 Estimating Policy Relevant Background
Policy-relevant background estimates used in the risk assessment model (see Table 3-2
below) were obtained from the ISA (Table 3-23, final ISA, EPA, 2009d). These values were
generated based on a combination of Community Multiscale Air Quality model (CMAQ) and
Goddard Earth Observing System (GEOS)-Chem modeling as described in the draft ISA (see
section 3.7.1.2). Annual values presented in Table 3-2 were used in modeling health endpoints
associated with long-term exposure (in those sensitivity analysis scenarios where risk was
modeled down to PRB - see section 3.5.4). For health endpoints associated with short-term
exposure (which involved modeling down to PRB, exclusively), quarterly values presented in
Table 3-2 were used to represent the appropriate block of days within a simulated year.
February 2010
3-13
Draft - Do Not Quote or Cite
-------
1
2
Table 3-2 Regional Policy-Relevant Background Estimates Used in the Risk
Assessment.
U.S. Region
Northeast
Southeast
Industrial Midwest
Upper Midwest
Southwest
Northwest
Southern California
Annual
0.74
1.72
0.86
0.84
0.62
1.01
0.84
January-
March
0.85
2.43
0.89
0.79
0.61
0.48
0.54
April-June
0.78
1.41
0.89
0.93
0.76
0.81
0.92
July-
September
0.67
1.41
0.94
0.99
0.70
1.42
1.21
October-
December
0.68
1.64
0.73
0.66
0.40
1.32
0.67
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
3.2.3 Simulating Air Quality to Just Meet Current and Alternative Standards
This section describes the methodologies used to simulate ambient PM2 5 levels in an area
that would just meet specified PM2.5 standards. The form of the current PM2.5 standards requires
that the 3-year average (rounded to the nearest 0.1 |ig/m3) of the annual means from each single
monitor or the average of multiple monitors must be at or below the level of the annual standard
and the 3-year average (rounded to the nearest 1 |ig/m3) of the ninety-eighth percentile values at
each monitor cannot exceed the level of the 24-hour standard. In determining attainment of the
annual average standard, an area may choose to use either the spatially averaged concentrations
across all population-oriented monitors, subject to meeting certain criteria detailed in Part 50,
Appendix N, of the CFR, or it may use the highest 3-year average based on individual monitors.
The most realistic simulation of just meeting both the annual and the 24-hour PM2.5 standards in
a location would require changing the distribution of 24-hour PM2.5 concentrations at each
monitor separately, based on the specific mix of local and regional controls impacting that
particular location. This would require extensive analysis and assumptions about the nature of
future control strategies that is beyond the scope of quantitative risk assessments done as part of
the review of the NAAQS.15
In the last PM risk assessment, just meeting the current or alternative PM2.5 standards was
simulated by changing 24-hour PM2.5 concentrations at a "composite monitor," which
represented the average of the monitors in a location. In the current PM risk assessment, just
meeting the current or alternative PM2.5 standards was simulated by changing 24-hour PM2.5
concentrations at each monitor separately. This change was made because the current PM risk
assessment considers three alternative approaches to simulating PM2 5 concentrations that just
15 Such modeling analyses are done by States in developing state implementation plans that demonstrate how areas
will come into attainment with standards that have been promulgated.
February 2010
3-14
Draft - Do Not Quote or Cite
-------
1 meet a given suite of standards (i.e., proportional, hybrid and peak-shaving - see below), and two
2 of these methods (hybrid and peak-shaving) involve making monitor-specific changes of 24-hour
3 PM2 5 concentrations to simulate just meeting standards. All three of these methods start with
4 monitor-specific series of PM2.5 concentrations in which missing days have been filled in as
5 described above.
6 In simulating ambient PM2.5 levels that would just meet current and alternative suites of
7 standards, we have applied the following approaches to rolling back air quality levels: (a)
8 proportional rollback, in which the same proportional adjustment is applied to all monitors in a
9 study area, has traditionally been used in the NAAQS risk assessments since it generally reflects
10 historical patterns in how air quality has changed over time, (b) hybrid rollback, which involves
11 an initial localized reduction to bring higher monitors down to the range of their neighbors,
12 followed by proportional reduction, if needed, to just meet a given suite of standards; and (c)
13 peak-shaving rollback, in where each monitor that exceeds the 24-hour standard is simulated to
14 just meet the 24-hour standard through proportional reduction of its annual 24-hour PM2.5
15 distribution (with no impact on monitors that are meeting the 24-hour standard). The
16 proportional rollback approach is applied to each of the 15 urban study areas, while the other two
17 rollback approaches are applied to a subset of areas as appropriate (e.g., the peak-shaving
18 approach is only used for those study areas where the 24-hour standard is both controlling and
19 being exceeded by one or more monitors).
20 The proportional rollback approach was used in generating the core risk estimates in light
21 of its use in past risk assessments, while the other two rollback approaches (hybrid and peak
22 shaving) were considered in sensitivity analyses to characterize potential variability in the way
23 urban areas may respond to suites of current or alternative standards. As described below, the
24 proportional rollback reflects a regional pattern of ambient PM2.5 reduction, the hybrid approach
25 reflects a combination of local and regional patterns in ambient PM2.5 reduction, and the peak
26 shaving approach reflects a localized pattern of ambient PM2.5 reduction. We have not ascribed
27 greater confidence to the proportional approach, since we have no basis for predicting which
28 approach would likely be most reflective of future patterns of ambient PM2.5 reductions in each
29 study area.
30 3.2.3.1 Proportional Rollback Method
31 The proportional approach, which reflects a regional pattern of reductions in ambient
32 PM2.5 concentrations, was used in previous PM2.5 risk assessments. This approach involves
33 proportional adjustments to monitor levels, in which PM2.5 concentrations are reduced ("rolled
34 back") by the same percentage each day. When this approach is used, it does not matter whether
35 (1) PM2.5 concentrations are first rolled back by the same percentage each day at each monitor,
February 2010 3-15 Draft - Do Not Quote or Cite
-------
1 and then the composite monitor values are calculated from these monitor-specific values or (2)
2 first the composite monitor values are calculated and then these are rolled back by the same
3 percentage each day - the results will be the same.
4 The percent reduction of 24-hour PM2.5 concentrations in the proportional rollback
5 approach (and in the second step of the hybrid rollback approach, described below) at each
6 monitor each day to simulate just meeting current and alternative set of standard levels is
7 determined by the PM2.5 annual and 24-hour design values. The annual design value (in |ig/m3)
8 was calculated as follows:
9 At each monitor, the annual average PM2.5 concentration was calculated for each of the
10 years 2005, 2006, and 2007, and these three annual average concentrations were then
11 averaged.
12 The maximum of these monitor-specific 3-year averages of annual averages is the annual
13 design value, denoted dvannual,
14 The 24-hour design value (in |ig/m3) was similarly calculated as follows:
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
At each monitor, the 98th percentile 24-hour PM2.5 concentration was calculated for each
of the years 2005, 2006, and 2007, and these three 98th percentile concentrations were
then averaged.
The maximum of these monitor-specific 3-year averages of 98th percentile concentrations
is the 24-hour design value, denoted dvdaily 93 (note, we will refer to the 98th percentile
design value as the 24-hour design value throughout the rest of the document).
The annual and 24-hour design values used in assessing the current and alternative
standards for PM2.5 are given in Table 3-3. Note that monitors that were closed in 2005 (and
therefore, did not include monitoring data for the majority of the three year simulation period), or
which were missing an entire year's worth of monitoring data during any of the three simulation
years (2005, 2006 or 2007) were excluded from consideration as design value monitors, although
these monitors were still used to construct composite monitors for purposes of estimating risks.
Table 3-3. EPA Design Values for Annual and \24-hour PM2.s Standards for the Period
2005-2007.*
Location
Atlanta
Baltimore
Birmingham
Dallas
Annual
(ug/m3)
16.2
15.6
18.7
12.8
24-hour
(us/m3)
35
37
44
26
February 2010
3-16
Draft - Do Not Quote or Cite
-------
Location
Detroit
Fresno
Houston
Los Angeles
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
St. Louis
Tacoma
Annual
(ug/m3)
17.2
17.4
15.8
19.6
15.9
15.0
12.6
19.8
11.6
16.5
10.2
24-hour
(us/m3)
43
63
31
55
42
38
32
60
55
39
43
1 *The calculation of design values is explained in the text above.
2
3 The percent reduction required to meet a standard (annual or 24-hour) was determined by
4 comparing the design value for that standard with the level of the standard. Because pollution
5 abatement methods are applied largely to anthropogenic sources of PM2.5, rollbacks were applied
6 only to PM2.5 above estimated PRB levels. The percent reduction was determined by the
7 controlling standard. For example, suppose both annual and 24-hour PM2 5 standards are being
8 simulated. Suppose^ is the percent reduction required to just meet the annual standard (i.e., the
9 percent reduction of daily PM2 5 above background necessary to get the annual design value
10 down to the current or alternative annual standard). Suppose/^ is the percent reduction required
11 to just meet the 24-hour standard (i .e., the percent reduction of daily PM2.5 above background
12 necessary to get the 24-hour PM2 5 design value down to the 24-hour standard). Ifpd is greater
13 than pa, then all 24-hour average PM2 5 concentrations above background are reduced by pd
14 percent. If pa is greater thanp^ then all 24-hour average PM2.5 concentrations are reduced by pa
15 percent. The method of rollbacks to meet a set of annual and 24-hour PM2 5 standards is
16 summarized as follows:
17 1. The percent by which the above-PRB portion of all daily PM2 5 concentrations (at the
18 composite monitor) would have to be reduced to just meet the annual standard (denoted
19 stdo) is
W.-PRB^)
February 2010
3-17
Draft - Do Not Quote or Cite
-------
1
2 where PRBavg is the average of the daily PRB concentrations.16
O
4
5 2. The percent by which the above-PRB portion of all 24-hour PM2.5 concentrations (at the
6 composite monitor) would have to be reduced to just meet the current or alternative 24-
7 hour standard (denoted std^s) is:
8
(stdd9,-PRBmg}
9 p , = 1
10
1 1 Letpmax = maximum of (maximum of pa andp^s) and zero. 17
12
13
14 3. Then if PM0 denotes the original PM value on a given day (at the composite monitor), the
1 5 rolled back PM value on that day, denoted PMrb, is:
16
17 PMrb = PRB +(PM0 - PRB)*(\-pmax ).
18
19 Results of the simulations done in each urban study area using the proportional rollback
20 approach, as well as the hybrid and peak shaving approaches discussed below, are presented in
21 Appendix F, Tables F-49 and F-50. For each urban study area and suite of standards, two sets of
22 values are presented in each table based on application of each rollback approach including: (a)
23 the maximum monitor-specific three-year (2005-2007) annual average (i.e., "Max. M-S" in both
24 tables) and (b) the composite monitor value for 2007 (i.e., "2007 CM" in both tables). The first
25 estimate (Max M-S) allows us to see how the design value changes in just meeting each suite of
26 standards based on application of the different rollback methods, while the second estimate
27 (2007 CM) is the surrogate for long-term exposure-related mortality, as described below in
28 section 3.5.4. The tables differ in terms of the information presented in the last set of columns,
29 with Table F-49 showing the percent reduction in the composite monitor values given
30 application of a particular rollback approach (allows comparison of the pattern of risk reduction
16 In the previous PM risk assessment, a constant PRB level was assumed for all days, and that constant PRB level
was used in the formulas to calculate percent rollbacks necessary to just meet a standard. It can be shown that, if
PRB levels vary from day to day, the average PRB level takes the place of the constant PRB level in the previous
formula, as shown in the above equation.
17 If the percent rollback necessary to just meet the annual standard and the percent rollback necessary to just meet
the 24-hour standard were both negative ~ i.e., if both standards were already met ~ then the percent rollback
applied in the risk assessment was zero. That is, PM values were never increased, or "rolled up."
February 2010 3-18 Draft - Do Not Quote or Cite
-------
1 across standard levels generated using each rollback approach), and Table F-50 showing the
2 percent difference in the composite monitor values in comparing the hybrid and peak shaving
3 results to that obtained with the proportional rollback approach for a given standard level (allows
4 comparison of residual risk estimates generated using the different rollback approaches for each
5 standard level). The information in the last set of columns in each table is considered below in
6 the sensitivity analysis (section 3.5.4).
7 3.2.3.2 Hybrid Rollback Method
8 The hybrid rollback approach reflects a combination of first localized and then regional
9 patterns of reductions in ambient PM2.5 concentrations. In comparison to the proportional
10 rollback approach, this approach has two steps: (1) first PM2 5 concentrations are reduced at a
11 specific monitor location within an urban study area and then additional monitors within that
12 urban study area are adjusted to a lesser extent (with the magnitude of adjustment based on a
13 distance-decay function); then (2) a proportional rollback of the adjusted PM2.5 concentrations at
14 all of the different monitors is carried out, as described in Section 3.2.3.1 above. Because the
15 initial step reflecting localized controls is non-proportional, this needs to be completed on the
16 monitor datasets (associated with a particular study area) prior to construction of the composite
17 monitor. However, once those non-proportional reductions have been implemented, a composite
18 monitor can then be constructed (as described earlier) and the second step of conducting
19 proportional adjustment to simulate the current or alternative suites of standards can be
20 calculated for the composite monitor. New design values are calculated for the hybrid rollback
21 approach based on the PM2 5 concentrations that have been adjusted in the first step of the two-
22 step process.18 The hybrid approach is described in more details in Appendix B.
23 3.2.3.3 Peak Shaving Rollback Method
24 The peak shaving approach reflects localized patterns of reduction in ambient PM2.5
25 concentrations and has only been applied in cases where the 24-hour standard is controlling (i.e.,
26 the percent rollback necessary to meet the daily standard is greater than the percent rollback
27 necessary to meet the annual standard in that location). This approach was used to calculate
28 annual averages for 2005, 2006, and 2007 at composite monitors for comparison with the
29 composite monitor annual averages calculated using the proportional and hybrid rollback
18 As with the composite monitor values representing recent air quality, "rolled back" composite monitor values in
Pittsburgh, for both the proportional rollback and the hybrid rollback methods, were calculated based on the division
of monitors into the 10 in "Pittsburgh-1" and the remaining 2 in "Pittsburgh-2" (see footnote in Section 3.2.1).
Daily and annual composite monitor values in "Pittsburgh-1" and "Pittsburgh-2" were rolled back as described in
Sections 3.2.3.1 and 3.2.3.2; rolled back composite monitor values for Pittsburgh were calculated as weighted
averages of the rolled back composite monitor values for "Pittsburgh-1" and "Pittsburgh-2", where the weights were
the proportion of the monitors in each (i.e., 10/12 and 2/12).
February 2010 3-19 Draft - Do Not Quote or Cite
-------
1 approaches. Because of time constraints, we did not calculate health risks with the application of
2 the peak shaving rollback approach. Because the C-R functions used in the risk assessment are
3 almost linear, a comparison of annual averages at composite monitors using the three different
4 approaches for simulating just meeting alternative standards provides a good surrogate for
5 estimates of health risks when alternative standards are just met (see Section 3.5.4 for additional
6 detail on the composite monitor-based comparison of the three rollback strategies completed as
7 part of the sensitivity analysis).
8 As with the proportional and hybrid rollback approaches, the peak shaving approach for
9 calculating annual averages at composite monitors starts with monitor-specific quarterly
10 averages that have been calculated as described above in Section 3.2.1. In contrast to the
11 proportional and hybrid rollback approaches, the peak shaving method uses monitor-specific
12 design values. For each monitor, we compared the monitor-specific 24-hour design value to the
13 level of the 24-hour standard and calculated the percent rollback necessary to reduce the
14 concentration at each monitor to the standard level (using a formula that is analogous to the
15 proportional rollback formula given above in Section 3.2.3.1). We then rolled back each
16 quarterly average at the monitor by this percent rollback. We calculated the average quarterly
17 average across all monitors in the location, for each quarter. Finally, we calculated the annual
18 average at the composite monitor under the standard by averaging the four quarterly averages
19 calculated on the previous step.19
20 3.3 SELECTION OF MODEL INPUTS
21 3.3.1 Health Endpoints
22 The selection of health effect endpoints reflects consideration for a number of factors.
23 The specific set of factors considered in selecting health effects endpoints to model in this
24 assessment included:
25 The overall weight of evidence from the collective body of epidemiological, controlled
26 human exposure, and toxicological studies and the determination made in the final ISA
27 regarding the strength of the causal relationship between PM2.5 and the more general
28 health effect category;
19 As with the rolled back composite monitor values in Pittsburgh using both the proportional and hybrid rollback
methods, rolled back composite monitor values in Pittsburgh using the peak shaving method were calculated based
on the division of monitors into the 10 in "Pittsburgh-1" and the remaining 2 in "Pittsburgh-2" (as explained in the
footnote in Section 3.2.3.2). However, unlike in the other locations, if the annual standard was controlling in one of
the Pittsburgh attainment areas (i.e., in "Pittsburgh-1" or "Pittsburgh-2"), monitor-specific quarterly averages in that
attainment area were rolled back by the percent rollback necessary to just meet the annual standard there. Once
monitors in "Pittsburgh-1" and "Pittsburgh-2" were rolled back, the procedure to calculate annual composite
monitor values in Pittsburgh was the same as in the other risk assessment locations.
February 2010 3-20 Draft - Do Not Quote or Cite
-------
1 The extent to which particular health effect endpoints within these broader health effect
2 categories are considered significant from a public health standpoint;
3 The availability of well-conducted epidemiological studies providing C-R functions for
4 specific health effect endpoints;
5 The availability of sufficient air quality monitoring data in areas that were evaluated in
6 the epidemiological studies;
7 The availability of baseline incidence data to support population risk (incidence)
8 modeling; and
9 The anticipated value of developing quantitative risk estimates for the health effect
10 endpoint(s) to inform decision-making in the context of the PM NAAQS review.
11
12 In selecting the set of health effect endpoint categories (and associated endpoints and
13 related at-risk populations) to include in the PM2.5 risk assessment, we considered the health
14 effects evidence presented in the final ISA (EPA, 2009d), as well as CAS AC (Samet, 2009a) and
15 public comments received on the Scope and Methods Plan and CASAC (Samet, 2009b) and
16 public comments received on the first draft RA. In reviewing the final ISA in relation to PM2.5,
17 we focused on the following sections: (a) section 2.3.1.1 (Effects of Short-Term Exposure to
18 PM2.5), (b) section 2.3.1.2 (Effects of Long-Term Exposure to PM2.5), (c) section 2.3.2
19 (Integration of PM2.5 Health Effects), and (d) subsections in Chapter 6 and 7 of the final ISA
20 providing summaries of causal determination (for both morbidity and mortality endpoints)
21 related to short-term and long-term exposure, respectively. We also considered information in
22 the ISA on at-risk populations, which identified the life stages of children and older adults,
23 people with pre-existing cardiovascular and respiratory diseases, and people with lower
24 socioeconomic status as populations at increased risk for PM-related health effects.
25 Based on the evidence presented in the ISA and application of the above criteria, we
26 identified the following health effects endpoints for inclusion in the risk assessment:
27 Health effects associated with short-term PM^exposure:
28 Mortality (causal relationship)
29 o non-accidental,
30 o cardiovascular-related
31 o respiratory-related,
32 Cardiovascular effects (causal relationship)
33 o cardiovascular-related hospital admissions
34 Respiratory effects (likely causal relationship)
35 o respiratory-related hospital admissions
February 2010 3-21 Draft - Do Not Quote or Cite
-------
1 o asthma-related emergency department visits
2 Health effects associated with long-term PM^exposure:
3 Mortality (causal relationship)
4 o all-cause
5 o ischemic heart disease (IHD)-related
6 o cardiopulmonary-related
7 o lung cancer
8 While we selected specific health effect endpoints that were all within broad health effect
9 categories classified in the ISA as having a "causal" or "likely causal" association with PM2.5
10 exposure, our selection is a based on applying the multi-factor approach described above.
11 The evidence available for these selected health effect endpoints generally focused on
12 the entire population, although some information was available that allowed us to consider
13 differences in estimated risk for the at-risk populations of older adults and people with pre-
14 existing cardiovascular and respiratory diseases. While evidence of effects in other important at-
15 risk populations, including children and people with lower socioeconomic status, was not judged
16 to be sufficient to support quantitative risk assessment, this evidence will be part of the evidence-
17 based considerations to be discussed in the policy assessment document currently being
18 developed.
19 3.3.2 Selection and Delineation of Urban Study Areas
20 This section describes the approach used in selecting the 15 urban study areas included in
21 this risk assessment (see Table 3-3 for a listing of the urban study areas). This approach builds
22 upon and expands the approach for selecting urban study areas from the prior risk assessment
23 (EPA, 2005, section 3.2, p. 37).
24 Criteria used in the prior risk assessment and updated in this analysis include:
25 Availability of sufficient air quality data: Sufficient air quality data was
26 identified as having at least 11 observations per quarter for a one year period and
27 at least 122 observations per year. We assessed prospective study areas by
28 insuring that there was at least one PM2.5 monitor within the boundaries of the
29 prospective study area that met these completeness criteria for the period 2005 to
30 2007 with additional preference given to locations with more than one PM2 5
31 monitor meeting completeness criteria, since this provided a better
32 characterization of ambient air levels for that urban location.
33
34 Inclusion in epidemiology study: Coverage of the location within one of the key
35 epidemiology studies included in the risk assessment (at or close to the location
36 where at least one C-R function for one of the recommended health endpoints has
37 been estimated by a study satisfying the selection criteria used in the risk
38 assessment). In this review, because the current risk assessment primarily utilizes
February 2010 3-22 Draft - Do Not Quote or Cite
-------
1 multi-city studies to evaluate risk for short-term and long-term PM2.5 exposures
2 (whereas the prior risk assessment used city-specific studies in modeling
3 endpoints associated with short-term exposures), this criterion no longer applies
4 for most prospective areas.
5
6 Availability of city-specific baseline incidence data: Regarding sufficiency of
7 baseline health effects incidence data, an ongoing effort by EPA to collect county -
8 level hospital and emergency department admissions data from states to support
9 this risk assessment (see section 3.5) has resulted in enhanced health effects
10 baseline incidence data, largely addressing this criterion (i.e., most urban areas in
11 the U.S. now have coverage with the updated baseline health effects incidence
12 data).
13
14 Two additional factors considered in selecting locations to model in the current
15 assessment included:
16 Potential for risk reductions using alternative standard levels: Specifically,
17 we focused on those urban areas with PM2.5 monitoring levels suggesting the
18 potential for risk reduction under alternative (24-hour or annual) standards under
19 consideration, particularly focusing on urban locations with at least one monitor
20 having an annual average above 12 |ig/m3 and/or a 24-hour value above 25 |ig/m3.
21 Furthermore, locations with ambient PM2.5 level significantly higher than these
22 levels were favored (with several urban study areas selected having both annual
23 and 24-hour design values exceeding the current standards - Table 3-4).
24
25 Regional representation: The second criterion we added for study area selection
26 focused on providing coverage for factors believed to play a role in influencing
27 risk heterogeneity at the national-level (e.g., PM2 5 source characteristics and
28 composition, demographics, SES status, air conditioner use). Building on the 7
29 regions originally identified in the 1996 PM Criteria Document (EPA, 1996,
30 section 6.4) (i.e., PM regions), we evaluated several urban locations from each of
31 these PM regions with the goal to identify one or more candidate urban study
32 areas in each region. Ultimately, consideration of the criteria described here
33 resulted in an urban study area not being identified for one of the PM regions (the
34 Upper Midwest), however, the remaining six PM regions each included at least
35 one urban study areas evaluated in the risk assessment. While the PM regions
36 were originally defined focusing primarily on differences in PM composition, size
37 and seasonality, by selecting urban study areas from regions across the continental
38 U.S., we recognize the potential for covering regional differences in other factors
39 related to risk heterogeneity as well (e.g., demographics, SES). The
40 representativeness analysis (section 4.4) specifically assesses the degree to which
41 the 15 urban study areas provide coverage for national trends in key risk-related
42 factors such as those listed here.
43
44 Based on consideration of the above criteria, 15 study areas were selected for inclusion in
45 this risk assessment. Table 3-4 presents the 15 urban study areas including (a) whether the urban
February 2010 3-23 Draft - Do Not Quote or Cite
-------
1
2
3
4
5
study area was included in the prior risk assessment, (b) which PM region the urban study area is
located in, and (c) the 24-hour and annual design values using 2005-2007 air quality data. Figure
3-4 identifies each of the 15 urban study areas in relation to the 7 regions used to guide the
selection of the urban study areas.
Table 3-4. Urban Study Areas Selected for the Risk Assessment.
Urban study
area
Atlanta
Baltimore
Birmingham
Dallas
Detroit
Fresno
Houston
LA
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
St. Louis
Tacoma
State
GA
MD
AL
TX
MI
CA
TX
CA
NY
PA
AZ
PA
UT
MO
WA
Modeled in last
NAAQS review
X
X
X
X
X
X
X
PM
region*
SE
NE
SE
SE
IM
SCA
SE
SCA
NE
NE
SW
IM
NW
IM
NW
Annual design
value (ug/m3)
16.2
15.6
18.7
12.8
17.2
17.4
15.8
19.6
15.9
15.0
12.6
19.8
11.6
16.5
10.2
24-hour design
value (ug/m3)
35
37
44
26
43
63
31
55
42
38
32
60
55
39
43
7
8
9
10
* SE (Southeast), IM (industrial Midwest), SCA (Southern California), NE (Northeast), NW (Northwest), SW
(Southwest) (See, EPA, 1996, section 6.4 for description of these regions).
February 2010
3-24
Draft - Do Not Quote or Cite
-------
2 Figure 3-4 15 urban study areas included in the risk assessment (including seven PM
3 regions used to guide selection of study areas).
4
5 Once the 15 urban study areas were selected, the next step was to identify the spatial
6 template to use in defining each study area (i.e., the geographical area associated with each study
7 area that would be used in identifying which counties and PM2.5 monitors were associated with a
8 particular study area). For 12 of the 15 urban study areas, we either used a combined statistical
9 area (CSA) as the basis for the spatial template, or if that was not available, we used a core-based
10 statistical area (CBSA). The three remaining urban study areas were special cases and were
11 handled as follows:
12 Baltimore: We used counties in the Baltimore CBS A only and did not consider the
13 larger Baltimore-DC CSA since we felt it unlikely that the entire larger CSA would
14 behave similarly with regard to PM2.5 emissions reduction strategies;
15 Philadelphia: We used the Philadelphia CSA, but excluded Berks County (Reading),
16 and
17 Tacoma: we only used Pierce County (since we felt it unlikely that efforts to reduce
18 emissions at the "elevated" monitor in Pierce County, would significantly impact
19 monitors in Seattle).
February 2010
3-25
Draft - Do Not Quote or Cite
-------
1 As noted above, in a few instances, two or more epidemiological studies used different
2 geographic boundaries for determining which populations were included in their studies. For
3 example, in one study conducted in Birmingham, AL populations from Blount, Jefferson,
4 Shelby, St. Clair, and Walker Counties were included, while another study included the
5 population residing in only Jefferson County. In such cases, we matched our delineation of the
6 urban area to that of each study, resulting in two or more different delineations of the urban area.
7 As we discuss below, two of the studies on which we rely for our core analysis -
8 Zanobetti and Schwartz (2009) and Bell et al. (2008) - are multi-location studies. Zanobetti and
9 Schwartz (2009) specified the county or counties included in each of the urban areas they
10 included in their analysis. Bell et al. (2008), however, did not focus on urban areas, but instead
11 focused on counties with populations above a specified threshold number. To limit the number
12 of different "versions" of a risk assessment location, wherever possible we specified the counties
13 in a risk assessment location for Bell et al. (2008) to match the set specified for Zanobetti and
14 Schwartz (2009). This was possible in those cases in which Zanobetti and Schwartz (2009)
15 identified an urban area as a single county, and that county was also included in Bell et al.
16 (2008). This was the case for several of the risk assessment locations. In some cases, however,
17 Zanobetti and Schwartz (2009) used a multi-county delineation of an urban area where at least
18 one of the counties was not among those included in Bell et al. (2008). In those cases, we had to
19 delineate two definitions of the urban area - one corresponding to Zanobetti and Schwartz (2009)
20 and the other corresponding to Bell et al. (2008). This was the case for Atlanta, Birmingham,
21 and St. Louis. In both Atlanta and New York, other delineations by other studies forced
22 additional delineation of these urban areas, as shown in Table 3-1 above.
23 Finally, we applied the studies of mortality associated with long-term exposure to PM2.5
24 to the urban areas as defined by the short-term exposure mortality study, Zanobetti and Schwartz
25 (2009), to enable meaningful comparisons between estimates of premature morality associated
26 with short-term and long-term exposure to PM2 5.
27 3.3.3 Selection of Epidemiological Studies and Concentration-response (C-R) Functions
28 within those Studies
29 As discussed above, we included in the PM2.5 risk assessment only those health effect
30 endpoint categories (and specific health effects) that met the set of criteria reflected in the multi-
31 factor approach we developed for selecting health effect endpoints (see section 3.3.1). One of
32 these factors was the strength of evidence supporting a causal association between PM2.5
33 exposure and the endpoint of interest. Thus, in cases where the majority of the available studies
34 did not report a statistically significant relationship, the effect endpoint was not included. Once
35 it had been determined that a health endpoint would be included in the analysis, however,
36 inclusion of a study on that health endpoint was not based on statistical significance alone, but
February 2010 3-26 Draft - Do Not Quote or Cite
-------
1 considered other factors (e.g., overall design of the study including degree of control for
2 confounders, method used to characterize exposure to PM2.5 within the risk assessment).
3 A significant change since the previous PM risk assessment is the addition to the relevant
4 epidemiological literature of several multi-city studies. This type of study has several
5 advantages over single-city studies. First, multi-city studies use the same study design in each of
6 the cities included in the study, so that city-specific results are readily comparable. Second,
7 when they are estimating a single C-R function based on several cities, multi-city studies also
8 tend to have more statistical power and provide effect estimates with relatively greater precision
9 than single city studies due to larger sample sizes, reducing the uncertainty around the estimated
10 coefficient. Moreover, in a multi-city study the statistical power to detect an effect in any given
11 city can be supplemented by drawing statistical power from data across all the cities included in
12 the study (or all the cities in the same region) to adjust city-specific estimates towards the mean
13 across all cities included in the analysis (or in the same region). This is particularly useful in
14 those instances, where a city has relatively less data resulting in a larger standard error for the
15 effect estimate. In this situation, the information on the C-R relationship in all the other cities
16 included in a multi-city study can be used to help inform an assessment of the C-R relationship
17 in the city in question. Finally, multi-city studies tend to avoid the often-noted problem of
18 publication bias that single-city studies confront (in which studies with statistically insignificant
19 or negative results are less likely to get published than those with positive and/or statistically
20 significant results).
21 For this risk assessment, we selected what we considered to be the best study to assess
22 the C-R relationship between PM2 5 and a given health endpoint, and we included other studies
23 for that health endpoint only if they were judged to contribute something above and beyond what
24 we could learn from the primary study selected.
25 A primary study for a given health endpoint had to satisfy the study selection criteria that
26 we have used in past PM (and other) risk assessments. In particular:
27 It had to be a published, peer-reviewed study that has been evaluated in the PM ISA and
28 judged adequate by EPA staff for purposes of inclusion in this risk assessment based on
29 that evaluation.
30 It had to directly measure, rather than estimate, PM2.5 on a reasonable proportion of the
31 days in the study.
February 2010 3-27 Draft - Do Not Quote or Cite
-------
1 It had to either not rely on Generalized Additive Models (GAMs) using the S-Plus
2 software to estimate C-R functions or to appropriately have re-estimated these functions using
3 revised methods.20
4 Because of the advantages noted above, we selected multi-city studies as our primary
5 studies for assessing the risks of premature non-accidental, cardiovascular, and respiratory
6 mortality (Zanobetti and Schwartz, 2009) and cardiovascular and respiratory hospital admissions
7 (Bell et al., 2008) associated with short-term exposure to PM2.5 in our core analysis. In each of
8 these studies, the 15 urban areas selected for the PM risk assessment were among the locations
9 included in their analysis. These two multi-city studies are based on more recent air quality and
10 health effects incidence data for short-term exposure-related mortality and morbidity and
11 therefore represent the best studies to use in deriving C-R functions for this risk assessment.
12 Dominici et al. (2007) was considered as an alternative study in identifying C-R functions for
13 modeling short-term exposure-related mortality, however its study period and the underlying air
14 quality data and disease incidence data (1987-2000) are not as current as that of Zanobetti and
15 Schwartz et al., 2009 (study period of 2001-2005), and therefore, we decided to focus on
16 Zanobetti and Schwartz et al. (2009) as the source of C-R functions for modeling short-term
17 exposure-related mortality.
18 Studies often report more than one estimated C-R function for the same location and
19 health endpoint. Sometimes models including different sets of co-pollutants are estimated in a
20 study; sometimes different lag structures are used. Sometimes different modeling approaches are
21 used to fit weather and temporal variables in the model. Once a study has been selected, the next
22 step is to select one or more C-R functions from among those reported in the study.
23 Zanobetti and Schwartz (2009) divided the United States into six regions, based on the
24 Koppen climate classification (Kottek 2006; Kottek et al. 2006)(http://koeppen-
25 eiger.vuwien.ac.at/).21 They estimated the coefficient of PM2.5 in single-pollutant log-linear
26 models using Poisson regression for each of 112 cities, as well as in two-pollutant models with
27 coarse PM. They estimated annual models (which assume that the relationship between
28 mortality and PM2.5 is the same through the year), as well as four seasonal models per location.
29 They then used a random effects meta-analysis to combine the city-specific results (Berkey et al.
20 The GAM S-Plus problem was discovered prior to the recent final PM risk assessment carried out as part of the
PM NAAQS review completed in 2006. It is discussed in the 2004 PM Criteria Document (EPA, 2004), PM Staff
Paper (EPA, 2005c), and PM Health Risk Assessment Technical Support Document (Abt Associates, 2005).
21 Zanobetti and Schwartz delineate regions as follows: "region 1: humid subtropical climates and maritime
temperate climates (Cfa, Cfb), which includes FL, LA TX, GA, AL, MS, AR, OK, KS, MO, TN, SC, NC, VA, WV,
KY; region 2: warm summer continental climates (Dfb), including ND, MN, WI, MI, PA, NY, CT, RI, MA, VT,
NH, ME; region 3: hot summer continental climates (Dfa) with SD, NE, IA, IL, IN, OH; region 4: dry climates
(BSk) (MM, AZ, NV); region 5: dry climates together with continental climate (Dfc, BSk) with MT, ID, WY, UT,
CO; region 6: Mediterranean climates which includes CA, OR, WA (Csa, Csb)" (p. 10).
February 2010 3-28 Draft - Do Not Quote or Cite
-------
1 1998). Pooling of city-specific results was done at the national level as well as at the regional
2 level, and separately for each season as well as for the annual functions.
3 With respect to the multi-city study for short-term exposure mortality, at the request of
4 EPA, the authors produced Empirical Bayes "shrunken" city-specific estimates, adjusted towards
5 the appropriate regional mean, using the approach described in Le Tertre et al. (2005). This was
6 done for the annual estimates as well as for each season-specific estimate.22 The annual city-
7 specific "shrunken" estimates were used in our core analysis.23 The seasonal estimates were
8 used in a sensitivity analysis. City-specific estimates have the advantage of relying on city-
9 specific data; however, as noted above, such estimates can have large standard errors (and thus
10 be unreliable); "shrinking" city-specific estimates towards the regional mean estimate is a more
11 efficient use of the data.24 Such "shrinking" can be thought of as combining the advantages of a
12 single-city study (in which the estimation of a city-specific coefficient is not influenced by data
13 from other locations) with the advantages of a multi-city study (in which there is much greater
14 statistical power to detect small effects).
15 In Zanobetti and Schwartz (2009) all PM2.5 models used the same lag structure (i.e., an
16 average of same-day and the previous day's PM^.s). The study did, however, examine both
17 single-pollutant and two-pollutant models (with coarse PM). We selected the single-pollutant
18 models, in part to avoid collinearity problems, and in part to be consistent with most of the other
19 studies used in the risk assessment, which were single-pollutant studies.
20 Bell et al. (2008) estimated log-linear models relating short-term exposure to PM2.5 and
21 hospital admissions for cardiovascular and respiratory illnesses among people 65 and older,
22 using a 2-stage Bayesian hierarchical model, for each of 202 counties in the United States. They
23 reported both annual and season-specific results, nationally and regionally (for four regions:
24 Northeast, Southeast, Northwest, and Southwest), but not at the local (city-specific) level. All
25 cardiovascular hospital admissions models were single-pollutant, 0-day lag models; for
26 respiratory hospital admissions, both single-pollutant 0-day models and single-pollutant 2-day
27 models were estimated. We used the regional, annual C-R functions in our core analysis
28 (identifying the appropriate region for each of our 15 risk assessment locations).25 For
22 These city-specific "shrunken" estimates were provided to EPA (see Zanobetti, 2009).
23 One reason we selected the annual functions over the season-specific functions for the core analysis is that, while
we can sum the season-specific mortality estimates across the four seasons, we cannot do the same for the upper and
lower bounds of 95% confidence intervals around those estimates. To produce correct confidence bounds around
annual mortality estimates based on seasonal functions, we would need the covariance matrix of the season-specific
estimates, separately for each location, which we do not have.
24 The degree to which a city-specific estimate is "shrunken" towards the regional mean depends on the size of the
standard error of the city-specific estimate relative to that of the regional mean estimate. The larger the city-specific
estimate relative to the regional mean estimate, the less shrinkage toward the regional mean.
25 The region into which each of the 202 counties in Bell et al. (2008) falls is given at:
http://www.biostat.jhsph.edu/MCAPS/estimates-full.html.
February 2010 3-29 Draft - Do Not Quote or Cite
-------
1 respiratory hospital admissions (for the core analysis), we selected the 2-day lag models, based
2 on evidence that for respiratory effects the strongest associations with PM exposure may be
3 associated with longer lag periods (on the order of 2 days or more).26 We used the regional
4 season-specific functions in a sensitivity analysis.
5 We identified two studies that estimated C-R relationships between short-term exposure
6 to PM2.5 and emergency department (ED) visits for cardiovascular and/or respiratory illnesses.
7 (There were no multi-city studies for this category of health endpoint.) Tolbert et al. (2007)
8 examined both cardiovascular and respiratory ED visits in Atlanta, GA, using single-pollutant
9 log-linear models with a 3-day moving average (0-day, 1-day, and 2-day lags) of PM2.5. Ito et al.
10 (2007) estimated the relationship between short-term exposure to PM2.5 and ED visits for asthma
11 in New York City (Manhattan). They estimated two single-pollutant models, one for the whole
12 year and one for the period from April through August; in addition, they estimated several two-
13 pollutant models for the period from April through August. We selected the single-pollutant
14 model for the whole year for the core analysis, and we explored the impacts of using the annual
15 versus the April-through-August model, as well as the single- versus multi-pollutant models in
16 sensitivity analyses.
17 For the purpose of conducting a sensitivity analysis to show the impact of different lag
18 structures, different modeling approaches, and single- versus two-pollutant models on estimates
19 of the risks of premature mortality and hospital admissions associated with short-term exposure
20 to PM2.5, we selected Moolgavkar (2003). This study reported results for premature non-
21 accidental, cardiovascular, and respiratory mortality and for cardiovascular and respiratory
22 hospital admissions associated with short-term exposures to PM2 5 in Los Angeles, using several
23 different lag structures and several different approaches to modeling the effects of weather and
24 temporal variables.
25 In modeling premature mortality associated with long-term exposure to PM2.5 in our core
26 analysis, we selected Krewski et al. (2009) as our primary study. This study is an extension of
27 the ACS prospective cohort study (Pope et al., 2002), used in the previous PM risk assessment,.
28 The Krewski et al., 2009 study (and the underlying ACS dataset) has a number of advantages
29 which informed our selection of this study as the basis for C-R functions used in the core
30 analysis, including: (a) extended air quality analysis incorporating data from 1989 to 2000
31 (extending the period of observation to eighteen years: 1982-2000), which increases the power of
32 the study and allows the study authors to examine the important issue of exposure time windows,
26 The ISA states that, "Generally, recent studies of respiratory HAs that evaluate multiple lags, have found effect
sizes to be larger when using longer moving averages or distributed lag models. For example, when examining HAs
for all respiratory diseases among older adults, the strongest associations where observed when using PM
concentrations 2 days prior to the HA." (EPA, 2009d, section 2.4.2.2).
February 2010 3-30 Draft - Do Not Quote or Cite
-------
1 (b) rigorous examination of a range of model forms and effect estimates, including consideration
2 for such factors as spatial autocorrelation in specifying response functions, (c) coverage for a
3 range of ecological variables (social, economic and demographic) which allows for consideration
4 for whether these confound or modify the relationship between PM2.5 exposure and mortality, (d)
5 inclusion of a related analysis (focusing on Los Angeles), which allowed for consideration of
6 spatial gradients in PM2.5 and whether they effect response models (by addressing effect
7 modification, for example) and (e) large overall dataset with over 1.2 million individuals and 156
8 MS As. To provide coverage for one of the other larger datasets used in prospective cohort
9 analyses of long-term mortality (the six-cites dataset), we selected the Krewski et al. (2000)
10 study to provide C-R functions that were used in the sensitivity analysis completed for this risk
11 assessment.
12 A number of other studies were considered as candidates for use in modeling long-term
13 exposure-related mortality in this analysis. For purposes of transparency, we have included a
14 brief summary here of our rationale for not selecting a number of the more high-profile studies
15 for use in the core analysis. The Laden et al. (2006) study (which focused on the six-cities
16 dataset) was not selected because it used visibility data to estimate ambient PM2.5 levels. The
17 Goss et al. (2004) study (based on the cystic fibrosis data), while addressing an at-risk population
18 of concern, was not selected because of a lack of baseline incidence data for this population
19 which prevents quantitative modeling of mortality incidence. The Miller et al. (2007) study
20 (focusing on the Women's Health Initiative dataset) while providing coverage for a population of
21 particular interest, was not used, again due to an absence of baseline incidence data (which is
22 particularly important for this population which is typically healthier than the general
23 population). And finally, the Eftim et al. (2008) study (focusing on the Medicare population)
24 was not included because this study did not include representative confounder control for
25 smoking, which introduces uncertainty into C-R functions obtained from the study.
26 Krewski et al. (2009) (the study selected as the basis for C-R functions used in the core
27 analysis) considered mortality from all causes, as well as cardiopulmonary mortality, mortality
28 from ischemic heart disease, and lung cancer mortality. The study presents a variety of C-R
29 functions, in an effort to show how the results vary with various changes to the method/model
30 used. It was not readily apparent from review of the HEI report, that the authors of the study
31 recommended any one of these as clearly superior to the others. Therefore, we corresponded
32 with the authors of the Krewski et al. (2009) study to obtain additional clarification regarding
33 specific aspects of the study and associated results as presented in the HEI report (Krewski et al.,
34 2009). In response to the our question of whether the study authors had a preference for a
35 particular model (in the context of using that model and its hazard ratio(s) in risk assessment),
36 the authors stated that they had "refrained from expressing a preference among the results for
February 2010 3-31 Draft - Do Not Quote or Cite
-------
1 their use in quantitative risk assessment," preferring to "explore several plausible statistical
2 models that we have fit to the available data." However, the authors go on to state that "...if one
3 had to choose a model for use in practical applications involved in air quality management, one
4 could argue that a random effects model (which accounts for apparent spatial autocorrelation in
5 the data) might be preferable. A model that included ecological covariates, which has the effect
6 of reducing the residual variation in mortality, might also be of interest. If forced to pick a single
7 model for risk assessment applications in air quality management, our random effects model with
8 ecological covariates might be selected" (Krewski, 2009).
9 In addition to these statements from the study authors regarding the model form to use,
10 EPA staff also considered the results of an analysis presented in the study examining the
11 importance of exposure time windows in deriving C-R functions. This analysis suggested that
12 models developed using both exposure time windows considered in the analysis (1979-1983 and
13 1999-2000) were equally effective at representing the relationship between PM2.5 exposure and
14 long-term exposure-related mortality. Therefore, we concluded that C-R functions used in the
15 core analysis should include functions fitted to both exposure time windows. However, the study
16 does not provide random effects models with ecological covariates for both exposure time
17 windows (this form of model is only provided with a fit to the latter exposure window).
18 Therefore, for the core analysis, we decided to use the Cox proportional hazard model with 44
19 individual and 7 ecological variables fitted to both exposure time windows (note, that if the
20 Krewski et al. (2009) study had provided a random effects model with ecological covariates (for
21 both PM monitoring periods - 1979-1983 and 1999-2000), then we would have used those
22 models in our core analysis).
23 In specifying effect estimates for each set of models, the relative risks for a 10 |ig/m3
24 change in PM2.5 were back-calculated from Table 33 of Krewski et al. (2009). We selected
25 several additional C-R functions from Krewski et al. (2009) to use in sensitivity analyses carried
26 out in two risk assessment locations (Los Angeles and Philadelphia), including the random
27 effects form (section 3.5.4), as described below. In addition, as mentioned earlier, we used C-R
28 functions obtained from Krewski et al. (2000) [reanalysis of the Six Cities Study] in the
29 sensitivity analysis.
30 3.3.4 Summary of Selected Health Endpoints, Urban Areas, Studies, and C-R Functions
31 A summary of the selected health endpoints, urban areas, and epidemiological studies used
32 in the risk assessment is given below in Tables 3-5 and 3-6 for short-term and long-term
33 exposure studies, respectively. A more detailed overview of the locations, health endpoints,
34 studies, and C-R functions included in the core analysis is given in Table 3-7. An overview of
February 2010 3-32 Draft - Do Not Quote or Cite
-------
1 the locations, health endpoints, studies, and C-R functions included in sensitivity analyses is
2 given in Table 3-8.
February 2010 3-33 Draft - Do Not Quote or Cite
-------
Table 3-5. Locations, Health Endpoints, and Short-Term Exposure Studies Included in the PM2.s Risk
Assessment*
Urban Area
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City,
UT
St. Louis, MO
Tacoma, WA
Premature Mortality
Non-Accidental
Zanobetti and
Schwartz (2009)
Moolgavkar
(2003)
Zanobetti and
Schwartz (2009)
Cardiovascular
Zanobetti and
Schwartz (2009)
Moolgavkar
(2003)
Zanobetti and
Schwartz (2009)
Respiratory
Zanobetti and
Schwartz (2009)
Zanobetti and
Schwartz (2009)
Hospital Admissions
Cardiovascular
Bell et al. (2008)
Moolgavkar
(2003)
Bell et al. (2008)
Respiratory
Bell et al. (2008)
Bell et al. (2008)
ED Visits
Cardiovascular
Tolbert et al.
(2007)
Respiratory
Tolbert et al.
(2007)
Ito et al. (2007)
* Studies in italics are used only in sensitivity analyses.
February 2010
Draft - Do Not Quote or Cite
-------
Table 3-6. Locations, Health Endpoints, and Long-Term Exposure Studies Included in the PM2.s Risk
Assessment*
Urban Area
Premature Mortality
All-Cause
Cardiopulmonary
Ischemic Heart Disease
Lung Cancer
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Krewski et al. (2009) [extension
of the ACS study]
Krewski et al. (2009) [extension
of the ACS study]
New York, NY
Krewski et al. (2009) [extension
of the ACS study]
Krewski et al. (2009) [extension
of the ACS study]
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Los Angeles, CA
Philadelphia, PA
Krewski et al. (2009) [extension
of the ACS study]
Krewski et al. (2000) [reanalysis
of the Six Cities Study]
Krewski et al. (2009) [extension
of the ACS study]
Krewski et al. (2000) [reanalysis
of the Six Cities Study]
Krewski et al. (2009) [extension
of the ACS study]
Krewski et al. (2000) [reanalysis
of the Six Cities Study]
* Studies in italics are used only in sensitivity analyses.
February 2010
Draft - Do Not Quote or Cite
-------
Table 3-7. Summary of Locations, Health Endpoints, Studies and Concentration-Response Functions Included in
the Core Analysis.*
Risk
Assessment
Location
Atlanta
Baltimore
Counties
Cobb, De Kalb, Fulton,
Gwinnett
Cobb, DeKalb, Fulton,
Barrow, Bartow,
Carroll, Cherokee,
Clayton, Cobb, Coweta,
DeKalb, Douglas,
Fayette, Forsyth,
Fulton, Gwinett, Henry,
Newton, Paulding,
Pickens, Rockdale,
Spalding, Walton
Baltimore city,
Baltimore county
Study/C-R Function
Zanobetti and Schwartz (2009)1
Zanobetti and Schwartz (2009) l
Zanobetti and Schwartz (2009) l
Krewski et al. (2009) 2
Krewski et al. (2009) 2
Krewski et al. (2009) 2
Krewski et al. (2009) 2
Bell et al. (2008)3
Bell et al. (2008)3
Tolbert et al. (2007)
Tolbert et al. (2007)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Health Endpoint
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure Emergency room (ED) visits,
cardiovascular
Short-term exposure Emergency room (ED) visits,
respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Lag Structure
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-,1-day, and
2-day lags
Avg. of 0-,1-day, and
2-day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
February 2010
Draft - Do Not Quote or Cite
-------
Risk
Assessment
Location
Birmingham
Dallas
Counties
Blount, Jefferson,
Shelby, St. Clair,
Walker
Jefferson
Dallas
Study/C-R Function
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Health Endpoint
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Lag Structure
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
February 2010
Draft - Do Not Quote or Cite
-------
Risk
Assessment
Location
Detroit
Fresno
Counties
Wayne
Fresno
Study/C-R Function
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Health Endpoint
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Lag Structure
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
February 2010
Draft - Do Not Quote or Cite
-------
Risk
Assessment
Location
Houston
Los Angeles
Counties
Harris
Los Angeles
Study/C-R Function
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Health Endpoint
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Lag Structure
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
February 2010
Draft - Do Not Quote or Cite
-------
Risk
Assessment
Location
New York
Philadelphia
Phoenix
Counties
Kings, New York City
(Manhattan), Queens,
Richmond, Bronx
New York City
(Manhattan)
Philadelphia
Maricopa
Study/C-R Function
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Ito et al. (2007)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Health Endpoint
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure Emergency room (ED) visits,
asthma
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Lag Structure
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day andl-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
February 2010
5-40
Draft - Do Not Quote or Cite
-------
Risk
Assessment
Location
Pittsburgh
Salt Lake City
Counties
Allegheny
Salt Lake
Study/C-R Function
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Health Endpoint
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Lag Structure
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
February 2010
5-41
Draft - Do Not Quote or Cite
-------
Risk
Assessment
Location
St. Louis
Tacoma
Counties
Jeffferson, Madison
(IL), St. Louis, St.
Louis city, St. Clair (IL)
Madison (IL), St. Louis,
St. Louis city, St. Clair
(IL)
Pierce
Study/C-R Function
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Health Endpoint
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Lag Structure
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
* All C-R functions in the core analysis are single-pollutant, log-linear models; all are for a full year. The exposure metric for all short-term exposure C-R
functions is the 24-hour average; the exposure metric for all long-term exposure C-R functions is the annual average.
1 This is a multi-city study; city-specific estimates "shrunken" towards the mean across all cities in a region were supplied to EPA (Zanobetti, 2009).
2 Two C-R functions were used for the core analysis - one corresponding to the earlier exposure period, from 1979 - 1983, and the other corresponding to the
later exposure period, from 1999 - 2000. Both C-R functions were based on follow-up of the cohort through 2000. Both used the standard Cox proportional
February 2010
5-42
Draft - Do Not Quote or Cite
-------
hazards model, with 44 individual and 7 ecologic covariates. The relative risks for a 10 ug/m3 change in PM2 5 from which the PM2 5 coefficients were back-
calculated were taken from Table 33 of Krewski et al. (2009).
3 This study estimated four regional C-R functions - for the Northeast, Southeast, Northwest, and Southwest - for each health endpoint. For each risk
assessment location, we used the regional C-R function for the region containing the risk assessment location. The designation of counties to each of these
four regions can be found at http://www.biostat.jhsph.edu/MCAPS/estimates-full.html.
February 2010 3-43 Draft - Do Not Quote or Cite
-------
Table 3-8. Summary of Locations, Health Endpoints, Studies and Concentration-Response Functions Included in
Sensitivity Analyses.
Sensitivity Analysis
Study/C-R Function
Health Endpoint**
Risk Assessment
Location(s)
Single-Factor Sensitivity Analyses:
Impact of using different model choices - fixed effects
log-linear vs. random effects log-linear vs. random
effects log-log C-R function*
Impact of using copollutant models in modeling long-
term exposure-related mortality
Impact of estimating risks down to PRB rather than
down to LML
Impact of C-R function from alternative long-term
exposure study
Impact of using alternative hybrid rollback approach
(note, that as discussed in section 3.2.3, in addition to
the hybrid rollback approach, we have also included a
peak-shaving rollback approach as an alternative to the
proportional rollback approach).27
Impact of using season-specific C-R functions (vs. an
annual C-R function)
random effects log-linear:
Krewski et al. (2009) [Table 9,
"Autocorrelation at MSA and
ZCA levels" group - "MSA &
Diff ' row]
random effects log-log:
Krewski et al. (2009) [Table 11,
"MSA and DIFF" rows]
Krewski et al., 2000 (reanalysis
of ACS) - provides 2-pollutant
models combining PM2 5 with
CO, NO2
,O3orSO2.
Krewski et al. (2009) - C-R
functions for each of two
exposure periods
Krewski et al. (2000) [reanalysis
of the Harvard Six Cities study]
Krewski et al. (2009)
Zanobetti and Schwartz (2009) -
seasonal functions vs. annual
function
All-cause, cardiopulmonary, ischemic
heart disease, and lung cancer mortality
associated with long-term exposure
All-cause mortality associated with
long-term exposure
Long-term exposure all-cause mortality
All-cause, cardiovascular, respiratory,
lung cancer mortality associated with
long-term exposure
All-cause mortality associated with
long-term exposure
Non-accidental mortality,
cardiovascular mortality, respiratory
mortality associated with short-term
Los Angeles and
Philadelphia
Los Angeles and
Philadelphia
All 15 urban areas
Los Angeles and
Philadelphia
Baltimore, Birmingham,
Detroit, Los Angeles, New
York, Pittsburgh, and St.
Louis
All 15 urban areas
27 However, as noted in section 3.2.3 and in section 3.5.4, quantitative risk estimates were not generated using the peak-shaving approach and
instead, composite monitor values (acting as surrogates for long-term exposure-related risk) were used as the basis for the sensitivity analysis involving the
peak-shaving rollback approach.
February 2010
5-44
Draft - Do Not Quote or Cite
-------
Sensitivity Analysis
Impact of using season-specific C-R functions (vs. an
annual C-R function)
Impact of using an annual C-R function (applied to the
whole year) vs. a seasonal function for April through
August (applied only to that period) (using a single
pollutant model).
Impact of model selection (e.g., log-linear GAM with 30
df; log-linear GAM with 100 df; and log-linear GLM
with 100 df)
Impact of lag structure (0-day, 1-day, 2-day)
Impact of single- vs. multi-pollutant models (PM2 5
with CO)
Impact of using alternative hybrid rollback approach
Impact of lag structure (0-day, 1-day, 2-day)
Study/C-R Function
Bell et al. (2008) - seasonal
functions vs. annual function
Ito et al. (2007)
Moolgavkar (2003)
Moolgavkar (2003)
Moolgavkar (2003)
Zanobetti and Schwartz (2009)
Bell et al., 2008
Health Endpoint**
exposure
HA (unscheduled), cardiovascular and
respiratory, associated with short-term
exposure
Asthma ED visits
Non-accidental and cardiovascular
mortality; and cardiovascular and
COPD+ HA associated with short-term
exposure
Non-accidental and cardiovascular and
COPD+ HA associated with short-term
exposure
Non-accidental and cardiovascular
mortality; and cardiovascular and
COPD+ HA associated with short-term
exposure
Non-accidental mortality associated
with short-term exposure
Cardiovascular and respiratory hospital
admissions associated with short-term
exposure
Risk Assessment
Location(s)
All 15 urban areas
New York
Los Angeles
Los Angeles
Los Angeles
Baltimore, Birmingham,
Detroit, Los Angeles, New
York, Pittsburgh, and St.
Louis
Los Angeles and
Philadelphia
Multi-Factor Sensitivity Analyses:
Impact of using a fixed effects log-linear vs. a random
effects log-log model, estimating incidence down to the
lowest measured level (LML) in the study vs. down to
PRB, and using a proportional vs. hybrid rollback to
estimate incidence associated with long-term exposure
to PM2 .5 concentrations that just meet the current
standards
Impact of using season-specific vs. all-year C-R
functions and proportional vs. hybrid rollbacks to
estimate incidence associated with short-term exposure
to PM2 5 concentrations that just meet the current
Zanobetti and Schwartz (2009)
All-cause and ischemic heart disease
mortality associated with long-term
exposure
Non-accidental mortality associated
with short-term exposure
Los Angeles and
Philadelphia
Baltimore, Birmingham,
Detroit, Los Angeles, New
York, Pittsburgh, and St.
Louis
February 2010
5-45
Draft - Do Not Quote or Cite
-------
Sensitivity Analysis
standards
Study/C-R Function
Health Endpoint**
Risk Assessment
Location(s)
*This "single-factor" sensitivity analysis is actually two factors - first the change from a fixed effects log-linear model to a random effects log linear model,
and then the change from a random effects log-linear model to a random effects log-log model. These were combined into a single sensitivity analysis
because Krewski et al. (2009) did not present the results of a fixed effects log-log model (to compare to the core analysis fixed effects log-linear model).
**"HA" = hospital admissions, "ED" = emergency department visits, "COPD+" = chronic obstructive pulmonary disease.
February 2010
5-46
Draft - Do Not Quote or Cite
-------
1 3.4 BASELINE HEALTH EFFECTS INCIDENCE DATA
2 As noted in section 3.2.1 above, the form of C-R function most commonly used in
3 epidemiological studies on PM, shown in equation (1), is log-linear. To estimate the change in
4 incidence of a health endpoint associated with a given change in PM2.5 concentrations using this
5 form of C-R function requires the baseline incidence (often calculated as the baseline incidence
6 rate times the population) of the health endpoint, that is, the number of cases per unit time (e.g.,
7 per year) in the location before a change in PM2.5 air quality (denoted yo in equations 3 and 4).
8 Incidence rates express the occurrence of a disease or event (e.g., asthma episode, death,
9 hospital admission) in a specific period of time, usually per year. Rates are expressed either as a
10 value per population group (e.g., the number of cases in Philadelphia County) or a value per
11 number of people (e.g., the number of cases per 10,000 residents in Philadelphia County), and
12 may be age- and sex-specific. Incidence rates vary among geographic areas due to differences in
13 population characteristics (e.g., age distribution) and factors promoting illness (e.g., smoking, air
14 pollution levels).
15 3.4.1 Data Sources
16 3.4.1.1 Mortality
17 We obtained individual-level mortality data for 2006 for the whole United States from
18 the Centers for Disease Control (CDC), National Center for Health Statistics (NCHS). The data
19 are compressed into a CD-ROM, which contains death information for each decedent, including
20 residence county FIPS, age at death, month of death, and underlying causes (ICD-10 codes).
21 The detailed mortality data allow us to generate cause-specific death counts at the county level
22 for selected age groups. Below we describe how we generated the county-level death counts.
23 3.4.1.2 Hospital Admission and Emergency Department Visits
24 For hospital admissions (HA) and emergency department (ED) visits, there are multiple
25 data sources:
26 Healthcare Cost and Utilization Project (HCUP) Central Distributor HCUP is a
27 family of health care databases developed through a Federal-State-Industry partnership
28 and sponsored by the Agency for Healthcare Research and Quality (AHRQ). The HCUP
29 databases are based on the data collection efforts of data organizations in participating
30 states. We used two HCUP databases: the State Inpatient Database (SID) and the State
31 Emergency Department Database (SEDD) respectively. SID/SEDD include detailed
32 HA/ED information for each discharge, including patient county FIPS, age, admission
33 type (e.g., emergent, urgent), admission/discharge season, and principle diagnosis (ICD-9
34 codes). The HCUP databases can be purchased from the HCUP Central Distributor,
35 although not all participant states release the data to the Central Distributor.
February 2010 3-47 Draft - Do Not Quote or Cite
-------
1 HCUP State Partners. For those HCUP participating states that don't release their data
2 to the Central Distributor, we contacted the HCUP state partners to obtain the HA and/or
3 ED data.
4 Communication with the author(s) of selected epidemiological studies. The ED data
5 for Atlanta in 2004 were sent to EPA by one of the authors of Tolbert et al. (2007).
6 Table 3-9 shows the states for which we obtained data from the HCUP Central
7 Distributor and the HCUP State Partners. The data are at the discharge level if not otherwise
8 noted, and the data year is 2007 for all the states in the table. The column "PM RA Location"
9 indicates the selected risk assessment location(s) where the incidence rate is applied.
10 The necessary baseline incidence data were not available for Atlanta, Birmingham,
11 Philadelphia, Pittsburgh and St. Louis. Therefore, for each of these five risk assessment
12 locations EPA instead used the baseline incidence rate for a designated surrogate location.
13 Surrogate locations were chosen if they were deemed to be sufficiently similar to the urban area
14 whose baseline incidence data were not available. Surrogate locations are noted in Table 3-9.
15
16 Table 3-9. Sources of Hospital Admissions (HA) and Emergency Department (ED)
17 Visit Data.
States
Arizona
California
Illinois
Maryland
Michigan
New York
North Carolina
Texas
Utah
Washington
HCUP
Central
Distributor
HA data
NA*
NA
HA data
HA data
NA
HA data
NA
HA data
HA data
HCUP State
Partner
-
HA data
HA data
-
-
HA and ED data
-
HA data
-
-
PMRA
Location
Phoenix
Fresno, Los
Angeles
St. Louis
Baltimore,
Philadelphia
Detroit
New York,
Pittsburgh
Atlanta and
Birmingham
Dallas,
Houston
Salt Lake
City
Tacoma
Notes
Due to privacy concerns, CA state
agency provided county level data.
1 . Due to privacy concerns, IL state
agency provided county level data.
2. Two IL counties (Madison and St.
Clair) serve as the surrogate for the
St. Louis metropolitan region.
Baltimore serves as the surrogate for
Philadelphia.
Buffalo, NY serves as the surrogate
for Pittsburgh.
Charlotte, NC serves as the surrogate
for both Atlanta and Birmingham.
February 2010
3-48
Draft - Do Not Quote or Cite
-------
1 *NA denotes "not available, or not available with all variables required for our analysis. If data were not available
2 from the HCUP Central Distributor, we contacted the HCUP State Partner.
3
4 3.4.1.3 Populations
5 To calculate baseline incidence rate, in addition to the health baseline incidence data we
6 also need the corresponding population. We obtained population data from the U.S. Census
7 Bureau (http://www.census.gov/popest/counties/asrh/). These data, released on May 14, 2009,
8 are the population estimates of the resident populations by selected age groups and sex for
9 counties in each U.S. state from 2000 to 2008. We used 2007 populations for calculating most
10 incidence rates except for the ED visit rate in Atlanta. Because the ED visit data obtained from
11 the authors of Tolbert et al. (2007) are for 2004, we used 2004 population estimates for the 20-
12 county Metropolitan area used in the Tolbert et al. study for the Atlanta area to calculate the ED
13 incidence rates to be applied when using that study in the risk assessment; we then applied the
14 2004 rates to the 2007 population, assuming the ED incidence rates in Atlanta did not change
15 significantly from 2004 to 2007. The sizes of the populations in the assessment locations that are
16 relevant are shown below in Table 3-10.
February 2010 3-49 Draft - Do Not Quote or Cite
-------
Table 3-10. Relevant Population Sizes for PM Risk Assessment Locations.
City
Atlanta, GA - 1
Atlanta, GA - 2
Atlanta, GA - 3
Baltimore, MD
Birmingham, AL - 1
Birmingham, AL - 2
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY - 1
New York, NY - 2
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO - 1
Counties
Cobb, De Kalb, Fulton, Gwinnett
Cobb, De Kalb, Fulton
20-County MSA**
Baltimore city, Baltimore county
Blount, Jefferson, Shelby, St. Clair,
Walker
Jefferson
Dallas
Wayne
Fresno
Harris
Los Angeles
Kings, New York City (Manhattan),
Queens, Richmond, Bronx
New York city (Manhattan)
Philadelphia
Maricopa
Allegheny
Salt Lake
Jefferson, Madison (IL), St. Louis, St.
Louis city, St. Clair (IL)
Population (Year 2006 and 2007)*
All Ages
2006
3,126,000
2,376,000,
4,975,000
1,429,000
1,037,000
660,000
2,338,000
2,012,000
886,000
3,876,000
9,881,000
8,251,000
1,613,000
833,000
3,779,000
1,225,000
991,000
2,093,000
2007
3,198,000
2,421,000
5,123,000
1,426,000
1,044,000
659,000
2,367,000
1,985,000
899,000
3,936,000
9,879,000
8,275,000
1,621,000
1,450,000
3,880,000
1,219,000
1,010,000
2,091,000
Ages ^30
2006
1,817,000
1,400,000
2,831,000
849,000
619,000
397,000
1,285,000
1,176,000
444,000
2,097,000
5,544,000
4,940,000
1,061,000
833,000
2,103,000
790,000
504,000
1,259,000
2007
1,865,000
1,433,000
2,918,000
848,000
625,000
397,000
1,308,000
1,168,000
452,000
2,139,000
5,579,000
4,975,000
1,074,000
833,000
2,167,000
786,000
517,000
1,261,000
Ages £ 65
2006
236,000
191,000
391,000
190,000
131,000
88,000
195,000
236,000
86,000
299,000
1,011,000
1,004,000
201,000
189,000
417,000
208,000
83,000
274,000
2007
245,000
198,000
408,000
189,000
133,000
88,000
199,000
234,000
87,000
307,000
1,030,000
1,013,000
204,000
187,000
432,000
206,000
86,000
275,000
February 2010
5-50
Draft - Do Not Quote or Cite
-------
City
St. Louis, MO - 2
Tacoma, WA
Counties
Madison (IL), St. Louis, St. Louis city,
St. Clair (IL)
Pierce
Population (Year 2006 and 2007)*
All Ages
2006
1,879,000
764,000
2007
1,875,000
773,000
Ages ^30
2006
1,134,000
437,000
2007
1,134,000
444,000
Ages £ 65
2006
253,000
79,000
2007
252,000
81,000
* Not all populations listed in the table were used for calculating the incidence rates. As noted above, the population year needs to match the year of the health
data and the population age group needs to match what is used in the epidemiological studies. In addition, 2004 population (all ages) is used for ED visits in
Atlanta-3, which is 4,663,946. Populations in this table are rounded to the nearest 1,000.
** The 20 counties are Barrow, Bartow, Carroll, Cherokee, Clayton, Cobb, Coweta, DeKalb, Douglas, Fayette, Forsyth, Fulton, Gwinett, Henry, Newton, Paulding, Pickens,
Rockdale, Spalding, and Walton.
February 2010
5-51
Draft - Do Not Quote or Cite
-------
1 3.4.2 Calculation of Baseline Incidence Rates
2 To calculate a baseline incidence rate to be used with a C-R function from a given study,
3 we matched the counties, age groupings, and ICD codes used in that study. For example, Bell et
4 al. (2008) designated Dallas, TX as Dallas County and estimated a C-R function for ICD-9 codes
5 490-492, 464-466, and 480-487 (respiratory HA) among ages 65 and up; we therefore selected
6 only those HA records that had corresponding ICD codes for ages 65 and up in Dallas County
7 and also selected the population for the same age group in the same county. The incidence rate
8 is simply the ratio of the selected HA count to the population. The same procedure was used to
9 calculate baseline incidence rates for all of the risk assessment locations.28
10 If a C-R function was estimated for a specific season, we selected only those HA records
11 within that season. The season definitions are: winter (December, January, and February), spring
12 (March, April, and May), summer (June, July, and August) and fall (September, October, and
13 November). Note that the HA data for some states didn't include information about admission
14 season but only discharge season or discharge quarter. The admission season was then
15 approximated using discharge season or discharge quarter.29
16 Some studies (e.g., Bell et al., 2008) look at the unscheduled HAs only, so we excluded
17 scheduled admissions from the analyses to match the study. A HA is unscheduled if the
18 admission type is emergency or urgent.
19 The baseline mortality rates are given in Table 3-11. The baseline HA and ED visit rates
20 are given in Table 3-12.
28 For Atlanta, Birmingham, Philadelphia, Pittsburgh and St. Louis, the HA data are not available. We calculated the
hospital admission rates for the surrogate cities. These cities are listed in Table 3-7.
29 Based on communication with the HCUP state partner in Texas, patients are normally admitted and discharged in
the same season.
February 2010 3-52 Draft - Do Not Quote or Cite
-------
Table 3-11. Baseline Mortality Rates (Deaths per 100,000 Relevant Population per Year) for 2006 for PM Risk Assessment
Locations.*
City
Atlanta, GA - 1
Atlanta, GA - 1
Atlanta, GA - 2
Atlanta, GA - 3
Baltimore, MD
Baltimore, MD
Birmingham, AL - 1
Birmingham, AL - 1
Birmingham, AL - 2
Dallas, TX
Dallas, TX
Detroit, MI
Detroit, MI
Fresno, CA
Fresno, CA
Houston, TX
Houston, TX
Los Angeles, CA
Los Angeles, CA
New York, NY - 1
Age Group
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
Type of Mortality (ICD-10 or ICD-9 Codes)
All-Cause
860
1,700
1,600
1,020
1,500
1,300
920
1,030
Non-accidental
(AOO-R99)
480
950
920
540
850
620
480
560
630
Cardiovascular
(101-159)
120
270
260
150
300
190
130
190
270
Respiratory
(JOOJ99)
41
85
85
48
67
67
37
57
52
Cardio-
pulmonary
(401-440, 460-
519)
330
690
680
420
700
590
370
510
Ischemic
Heart
Disease
(410-414)
89
300
190
170
360
260
150
250
Lung
Cancer
(162)
51
110
104
66
107
66
57
55
COPD
(490-496)
29
February 2010
5-53
Draft - Do Not Quote or Cite
-------
City
New York, NY - 1
New York, NY - 2
Philadelphia, PA
Philadelphia, PA
Phoenix, AZ
Phoenix, AZ
Pittsburgh, PA
Pittsburgh, PA
Salt Lake City, UT
Salt Lake City, UT
St. Louis, MO - 1
St. Louis, MO - 1
St. Louis, MO - 2
Tacoma, WA
Tacoma, WA
National
National
Age Group
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
Type of Mortality (ICD-10 or ICD-9 Codes)
All-Cause
1,0800
1,700
1,100
1,800
980
1,500
1,200
810
1,300
Non-accidental
(AOO-R99)
970
600
1,090
480
870
660
750
1,300
Cardiovascular
(101-159)
280
160
330
110
270
190
220
370
Respiratory
(JOO-J99)
83
67
96
45
83
66
76
130
Cardio-
pulmonary
(401-440, 460-
519)
580
720
470
770
350
680
510
340
580
Ischemic
Heart
Disease
(410-414)
380
300
220
350
101
320
240
140
240
Lung
Cancer
(162)
56
120
68
120
37
106
88
53
90
COPD
(490-496)
42
71
: Figures in this table are rounded to a two-integer level of precision.
February 2010
5-54
Draft - Do Not Quote or Cite
-------
Table 3-12. Baseline Hospital Admission (HA) and Emergency Department (ED) Rates (Admissions/Visits per 100,000
Relevant Population per Year) for 2007 for PM Risk Assessment Locations.*
City
Atlanta, GA - 1
Atlanta, GA - 2
Atlanta, GA - 3
Baltimore, MD
Birmingham, AL - 1
Birmingham, AL - 2
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Los Angeles, CA
Los Angeles, CA
New York, NY - 1
New York, NY - 2
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO - 1
St. Louis, MO - 2
Tacoma, WA
Age Group
>65
All ages
>65
>65
>65
>65
>65
>65
All ages
>65
>65
All ages
>65
>65
>65
>65
>65
>65
Health Endpoints (ICD-9 Codes)
HA, cardio-
vascular (390-
429)
5,500
HA (unscheduled),
cardiovascular(426
-429, 430-438,
410-414, 440-449)
5,700
8,600
5,700
5,000
8,800
5,600
5,900
5,500
6,400
8,600
5,020
6,100
3,030
5,600
4,500
HA, COPD
(490-496)
223
HA (unscheduled),
respiratory (490-492,
464-466, 480-487)
2,020
2,600
2,020
2,000
3,000
2,100
2,200
2,000
2,030
2,600
1,600
1,900
1,200
2,600
1,600
ED visits,
cardiovascular (410-
414, 427, 428, 433-
437, 440, 443^145,
451-453)
690**
ED visits, respiratory
(460-465, 466.1, 466.11,
466.19, 477, 480-486, 491-
493, 496, 786.07, 786.09)
2600**
ED visits,
asthma
(493)
1,100
* Figures in this table are rounded to a two-integer level of precision.
** These are 2004 incidence rates because Tolbert et al. (2007) provided 2004 ED visit data in a 20-county delineation of Atlanta. However, the 2004 rates were
applied to the appropriate year population in the risk assessment.
February 2010
5-55
Draft - Do Not Quote or Cite
-------
1 3.5 ADDRESSING UNCERTAINTY AND VARIABILITY
2 3.5.1 Overview
3 An important component of a population health risk assessment is the characterization of
4 both uncertainty and variability. Variability refers to the heterogeneity of a variable of interest
5 within a population or across different populations. For example, populations in different
6 regions of the country may have different behavior and activity patterns (e.g., air conditioning
7 use, time spent indoors) that affect their exposure to ambient PM and thus the population health
8 response. The composition of populations in different regions of the country may vary in ways
9 that can affect the population response to exposure to PM - e.g., two populations exposed to the
10 same levels of PM might respond differently if one population is older than the other. In
11 addition, the composition of the PM to which different populations are exposed may differ, with
12 different levels of toxicity and thus different population responses. Variability is inherent and
13 cannot be reduced through further research. Refinements in the design of a population risk
14 assessment are often focused on more completely characterizing variability in key factors
15 affecting population risk - e.g., factors affecting population exposure or response - in order to
16 produce risk estimates whose distribution adequately characterizes the distribution in the
17 underlying population(s).
18 Uncertainty refers to the lack of knowledge regarding the actual values of inputs to an
19 analysis. Models are typically used in analyses, and there is uncertainty about the true values of
20 the parameters of the model (parameter uncertainty) - e.g., the value of the coefficient for PM2 5
21 in a C-R function. There is also uncertainty about the extent to which the model is an accurate
22 representation of the underlying physical systems or relationships being modeled (model
23 uncertainty) - e.g., the shapes of C-R functions. In addition, there may be some uncertainty
24 surrounding other inputs to an analysis due to possible measurement errore.g., the values of
25 daily PM2.5 concentrations in a risk assessment location, or the value of the baseline incidence
26 rate for a health effect in a population. In any risk assessment, uncertainty is, ideally, reduced to
27 the maximum extent possible through improved measurement of key variables and ongoing
28 model refinement. However, significant uncertainty often remains, and emphasis is then placed
29 on characterizing the nature of that uncertainty and its impact on risk estimates. The
30 characterization of uncertainty can be both qualitative and, if a sufficient knowledgebase is
31 available, quantitative.
32 The selection of urban study areas for the PM2.s risk assessment was designed to cover
33 the range of PM2.s-related risk experienced by the U.S. population and, in general, to adequately
34 reflect the inherent variability in those factors affecting the public health impact of PM2.5
35 exposure. Sources of variability reflected in the risk assessment design are discussed in section
February 2010 3-56 Draft - Do Not Quote or Cite
-------
1 3.5.2, along with a discussion of those sources of variability which are not fully reflected in the
2 risk assessment and consequently introduce uncertainty into the analysis.
3 The characterization of uncertainty associated with risk assessment is often addressed in
4 the regulatory context using a tiered approach in which progressively more sophisticated
5 methods are used to evaluate and characterize sources of uncertainty depending on the overall
6 complexity of the risk assessment (WHO, 2008). Guidance documents developed by EPA for
7 assessing air toxics-related risk and Superfund Site risks (USEPA, 2004b and 2001, respectively)
8 as well as recent guidance from the World Health Organization (WHO, 2008) specify multi-
9 tiered approaches for addressing uncertainty.
10 The WHO guidance presents a four-tiered approach, where the decision to proceed to the
11 next tier is based on the outcome of the previous tier's assessment. The four tiers described in the
12 WHO guidance include:
13 Tier 0 - recommended for routine screening assessments, uses default uncertainty factors
14 (rather than developing site-specific uncertainty characterizations);
15 Tier 1 - the lowest level of site-specific uncertainty characterization, involves qualitative
16 characterization of sources of uncertainty (e.g., a qualitative assessment of the general
17 magnitude and direction of the effect on risk results);
18 Tier 2 - site-specific deterministic quantitative analysis involving sensitivity analysis,
19 interval-based assessment, and possibly probability bound (high- and low-end)
20 assessment; and
21 Tier 3 - uses probabilistic methods to characterize the effects on risk estimates of sources
22 of uncertainty, individually and combined.
23 With this four-tiered approach, the WHO framework provides a means for systematically
24 linking the characterization of uncertainty to the sophistication of the underlying risk assessment.
25 Ultimately, the decision as to which tier of uncertainty characterization to include in a risk
26 assessment will depend both on the overall sophistication of the risk assessment and the
27 availability of information for characterizing the various sources of uncertainty. EPA staff has
28 used the WHO guidance as a framework for developing the approach used for characterizing
29 uncertainty in this risk assessment.
30 The overall analysis in the PM NAAQS risk assessment is relatively complex, thereby
31 warranting consideration of a full probabilistic (WHO Tier 3) uncertainty analysis. However,
32 limitations in available information prevent this level of analysis from being completed at this
33 time. In particular, the incorporation of uncertainty related to key elements of C-R functions
34 (e.g., competing lag structures, alternative functional forms, etc.) into a full probabilistic WHO
35 Tier 3 analysis would require that probabilities be assigned to each competing specification of a
36 given model element (with each probability reflecting a subjective assessment of the probability
February 2010 3-57 Draft - Do Not Quote or Cite
-------
1 that the given specification is the "correct" description of reality). However, for many model
2 elements there is insufficient information on which to base these probabilities. One approach that
3 has been taken in such cases is expert elicitation; however, this approach is resource- and time-
4 intensive and consequently, it was not feasible to use this technique in the current PM NAAQS
5 review to support a WHO Tier 3 analysis.30
6 For most elements of this risk assessment, rather than conducting a full probabilistic
7 uncertainty analysis, we have included qualitative discussions of the potential impact of
8 uncertainty on risk results (WHO Tierl) and/or completed sensitivity analyses assessing the
9 potential impact of sources of uncertainty on risk results (WHO Tier 2). Note, however, that in
10 conducting sensitivity analyses, we have used both single- and multi-factor approaches (to look
11 at the individual and combined impacts of sources of uncertainty on risk estimates). Also, as
12 discussed below in section 3.5.4, in conducting sensitivity analyses, we used only those
13 alternative specifications for input parameters or modeling approaches that were deemed to have
14 scientific support in the literature (and so represent alternative reasonable input parameter values
15 or modeling options). This means that the alternative risk results generated in the sensitivity
16 analyses represent reasonable risk estimates that can be used to provide a context, with regard to
17 uncertainty, within which to assess the set of core (base case) risk results (see section 4.5.3).
18 The sensitivity analysis also includes coverage for potential variability in the pattern of
19 reductions in ambient PM2 5 concentrations associated with simulations of just meeting the
20 current and alternative suites of standards. Specifically, as discussed above in section 3.2.3, we
21 have included three alternative rollback methods (proportional, hybrid and peak shaving) to
22 provide coverage for variability in this potentially important factor influencing risk estimates.
23 In addition to the qualitative and quantitative treatment of uncertainty and variability
24 which are described here, we have also completed two additional analyses intended to place the
25 risk results generated for the 15 urban study areas in a broader national context. The first is a
26 representativeness analysis (described in section 4.4) which evaluates the set of urban study areas
27 against national-distributions of key PM risk-related attributes (with the goal of determining the
28 degree to which the study areas are representative of national trends in these parameters). The
29 second is a national-scale assessment of long-term mortality related to PM2.5 exposures
30 (discussed in chapter 5). In addition to providing an estimate of the national impact of PM2 5 on
31 long-term mortality, this analysis also evaluates whether the set of 15 urban study areas generally
32 represents the broader distribution of risk across the U.S., or a more focused portion of the
33 national risk distribution (e.g., the higher-end).
30 Note, that while a full probabilistic uncertainty analysis was not completed for this risk assessment, we were able
to use confidence intervals associated with effects estimates (obtained from epidemiological studies) to incorporate
statistical uncertainty associated with sample size considerations in the presentation of risk estimates.
February 2010 3-58 Draft - Do Not Quote or Cite
-------
1 The remainder of this section is organized as follows. Key sources of variability which
2 are reflected in the design of the risk assessment, along with sources excluded from the design,
3 are discussed in section 3.5.2. A qualitative discussion of key sources of uncertainty associated
4 with the risk assessment (including the potential direction, magnitude and degree of confidence
5 associated with our understanding of the source of uncertainty - the knowledge base) is
6 presented in section 3.5.3. The methods and results of the single- and multi -factor sensitivity
7 analyses completed for the risk assessment are presented in section 3.5.4. An overall summary
8 of the methods used to address uncertainty and variability for the 15 urban study areas (including
9 the two assessments intended to place the urban study areas in a broader national context) is
10 presented in section 3.5.5.
11 3.5.2 Treatment of Key Sources Of Variability
12 The risk assessment was designed to cover the key sources of variability related to
13 population exposure and exposure response, to the extent supported by available data.31
14 However, as with all risk assessments, there are sources of variability which have not been fully
15 reflected in the design of the risk assessment and consequently introduce a degree of uncertainty
16 into the risk estimates. While different sources of variability were captured in the risk
17 assessment, it was generally not possible to separate out the impact of each factor on population
18 risk estimates, since many of the sources of variability are reflected collectively in a specific
19 aspect of the risk model. For example, inclusion of urban study areas from different PM regions
20 likely provides some degree of coverage for a variety of factors associated with PM2.5 risk (e.g.,
21 air conditioner use, PM2 5 composition, differences in population commuting and exercise
22 patterns, weather). However, the model is not sufficiently precise or disaggregated to allow the
23 individual impacts of any one of these sources of variability on the risk estimates to be
24 characterized.
25 Key sources of potential variability that are likely to affect population risks are discussed
26 below, including the degree to which they are (or are not) fully captured in the design of the risk
27 assessment:
31 The term "key sources of variability" refers to those sources that the EPA staff believes have the potential to play
an important role in impacting population incidence estimates generated for this risk assessment. Specifically, EPA
staff has concluded that these sources of uncertainty, if fully addressed and integrated into the analysis, could result
in adjustments to the core risk estimates which might be relevant from the standpoint of interpreting the risk
estimates in the context of the PM NAAQS review. The identification of sources of variability as "key" reflects
consideration for sensitivity analyses conducted for previous PM NAAQS risk assessments, which have provided
insights into which sources of variability (reflected in different elements of those earlier sensitivity analyses) can
influence risk estimates, as well as information presented in the final PM ISA. For example, chapter 2 of the final
PM ISA addresses such issues as: ambient PM variability and correlations (section 2.1.1), trends and temporal
variability (section 2.1.2), correlations between pollutants (section 2.1.4), and source contributions to PM (section
2.1.6). These discussions were carefully considered by staff in identifying key sources of variability to address both
in the risk assessment and in the qualitative discussion of variability presented in this section.
February 2010 3-59 Draft - Do Not Quote or Cite
-------
1 PM2.s composition: While information was not available to support modeling risk
2 associated with different components of PM2.5, the assessment did use effect estimates
3 (for a number of the short-term exposure-related health endpoints) differentiated by
4 region of the country, or differentiated for specific urban locations (sections 3.3.3 and
5 3.3.4). While many factors may contribute to differences in effect estimates (for the
6 same health endpoint) across different locations, compositional differences in PM2.5
7 may be partially responsible. Therefore, while the analysis did not explicitly address
8 compositional differences in generating risk estimates, potential differences in PM2.5
9 composition may be reflected in those effect estimates that are differentiated by region
10 and/or urban study area. The effect estimates for mortality associated with long-term
11 exposure to PM2 5 are not regionally differentiated and instead, a single national-scale
12 estimate is used. This means that any differences in risks of mortality associated with
13 long-term exposure to PM2 5 that are linked to differences in PM2 5 composition (or to
14 any other differences across regions or locations) would not be discernable, since a
15 single national-scale risk estimate is generated for each mortality category. In addition
16 to using region- or location-specific effect estimates for health effects associated with
17 short-term exposures, the selection of urban areas to include in the risk assessment was
18 designed in part to ensure that areas in different regions of the country, with different
19 PM2.5 composition, were included.
20 Intra-urban variability in ambient PMi.s levels: Several recent studies (e.g., Jerrett
21 et al., 2005) have addressed the issue of heterogeneity of PM concentrations within
22 urban areas and its potential impact on the estimation of premature mortality associated
23 with long-term exposure to PM2 5. Most recently, the HEI Reanalysis II (Krewski et
24 al., 2009), focusing on the ACS dataset, discusses epidemiological analyses completed
25 for Los Angeles and New York City which included more highly-refined (zip code
26 level) characterizations of spatial gradients in population exposure within each urban
27 area based on land-use regression methods and/or kriging. While both analyses
28 provide insights into the issue of intra-urban heterogeneity in PM2.5 concentrations and
29 its potential implications for epidemiology-based health assessments, due to the time
30 and resource necessary to integrate them into the risk assessment, we were not able to
31 incorporate these studies quantitatively. The implications of these studies for
32 interpretation of long-term mortality C-R functions and potential exposure error
33 associated with those functions is discussed below in section 3.5.3.
34 Variability in the patterns of ambient PM2.s reduction as urban areas: In
35 simulating just meeting the current or alternative suites of standards, there can be
36 considerable variability in the patterns of ambient PM2 5 reductions that result from
37 different simulation approaches (i.e., they can be more localized, more regional, or
38 some combination thereof). To address this issue in the risk assessment, we have
39 included three rollback approaches as part of the sensitivity analysis including:
40 proportional (reflecting regional patterns of reduction), hybrid (reflecting a
41 combination of localized and regional patterns of reduction), and peak shaving
42 (reflecting localized patterns of reduction) (see section 3.2.3 for additional detail on
43 these rollback methods and section 3.5.4 for a description of how this factor is
44 addressed in the sensitivity analysis).
February 2010 3-60 Draft - Do Not Quote or Cite
-------
1 Copollutant concentrations: Inclusion of copollutant models in short-term exposure-
2 related time series studies has produced mixed results in terms of the degree of
3 attenuation of the PM2.s signal that results from inclusion of other pollutants (see final
4 PM ISA, sections 6.2.10.9 and 6.3.8.5). The PM ISA (section 6.2.10.9) suggests that
5 these inconsistent findings associated with controlling for gaseous pollutants are likely
6 due to differences in the correlation structure among pollutants as well as differing
7 degrees of exposure measurement error related to the copollutants. Further, the PM
8 ISA (section 2.1.3) notes that correlations between PM and copollutants (including CO,
9 Os, SC>2 and NO2) can vary both seasonally and spatially. Therefore, it is possible that
10 the degree of attenuation of PM2.s-related risk by copollutants may differ across study
11 areas. However, because the multi-city studies used in the core risk assessment
12 (Zanobetti and Schwartz., 2009; Bell et al., 2008; and Krewski et al., 2009) provide
13 single pollutant models, our analysis does not directly address the issue of copollutant
14 confounding (see section 3.5.3 for additional discussion of uncertainty introduced into
15 the analysis as a result of not including copollutant models in the core risk assessment).
16 We did explore the issue of copollutant modeling in the context of modeling long-term
17 exposure-related mortality as part of the sensitivity analysis (section 3.5.4). In
18 addition, the potential impact of copollutant confounding on short-term exposure-
19 related mortality and morbidity was explored in the Moolgavkar et al., 2003 study, as
20 discussed below in section 4.3.1.1 (although they have limited applicability to the core
21 risk estimates generated in this RA).
22 Demographics and socioeconomic-status (SES)-related factors: Variability in
23 population density particularly in relation to elevated levels of PM2.5 has the potential
24 to influence population risk. In addition, other aspects of demographics such as age of
25 housing stock (which can influence rates of air conditioner use thereby impacting rates
26 of infiltration of PM indoors) can impact exposure and therefore risk (discussed in PM
27 ISA - sections 2.2.1 and 2.3.2). While risk modeling completed for this analysis is
28 based on concentrations measured at central-site monitors used as surrogates for
29 population exposure and does not explicitly consider more detailed patterns of PM
30 exposure by different subpopulations, potential differences in exposure to PM2.s
31 reflecting demographic and SES-related factors is covered to some degree by the use of
32 urban study area-differentiated effects estimates (for short-term exposure-related
33 mortality) and regionally-differentiated effects estimates (in the case of short-term
34 exposure-related morbidity). In the case of long-term exposure-related mortality, while
35 the modeling for this group of endpoints does not utilize location-specific or
36 regionally-differentiated effects estimates, the national-scale effects estimates that are
37 used do reflect differences in exposure and health response across urban study areas
38 (which will reflect, to some extent, differences in demographics and SES-related
39 factors to the extent that these factors influence the relationship between PM2.5
40 exposure and mortality response, as detected by the underling cohort studies).
41 Behavior affecting exposure to PM2.s: We have incorporated, where available,
42 region- and/or city-specific effect estimates in order to capture behavioral differences
43 across locations that could affect population exposures to PM2.s (e.g., time spent
44 outdoors, air conditioning use). However, while these location-specific effect
45 estimates may be capturing differences in behavior, they may also be capturing other
February 2010 3-61 Draft - Do Not Quote or Cite
-------
1 differences (e.g., differences in the composition of PM2.5 to which populations are
2 exposed). As noted above, it was not possible to separate out the impact of these
3 different factors, which may vary across locations and populations, on effect estimates.
4 Baseline incidence of disease: We collected baseline health effects incidence data (for
5 mortality and morbidity endpoints) from a number of different sources (see section
6 3.4). Often the data were available at the county-level, providing a relatively high
7 degree of spatial refinement in characterizing baseline incidence given the overall level
8 of spatial refinement reflected in the risk assessment as a whole. Otherwise, for urban
9 study areas without county-level data, either (a) a surrogate urban study area (with its
10 baseline incidence rates) was used, or (b) less refined state-level incidence rate data
11 were used.
12 Longer-term temporal variability in ambient PM2.s levels (reflecting meteorological
13 trends, as well as future changes in the mix of PM2.5 sources and regulations impacting
14 PM2 5): Risk estimates for the PM2 5 NAAQS review have been generated using recent
15 years of air quality data. In other words, efforts have not been made to simulate
16 potential future changes in either the concentrations or composition of ambient PM2 5
17 in the risk assessment locations based on possible changes in economic activity,
18 demographics or meteorology. Actual risk levels potentially experienced in the future
19 as a result of implementing alternative standard levels may differ from those presented
20 in this report due, in part, to potential changes in these factors related to ambient PM2 5.
21 3.5.3 Qualitative Assessment of Uncertainty
22 As noted in section 3.5.1, we have based the design of the uncertainty analysis carried out
23 for this risk assessment on the framework outlined in the WHO guidance document (WHO,
24 2008). That guidance calls for the completion of a Tier 1 qualitative uncertainty analysis,
25 provided the initial Tier 0 screening analysis suggests there is concern that uncertainty associated
26 with the analysis is sufficient to significantly impact risk results (i.e., to potentially affect
27 decision making based on those risk results). Given previous sensitivity analyses completed for
28 prior PM NAAQS reviews, which have shown various sources of uncertainty to have a
29 potentially significant impact on risk results, we believe that there is justification for conducting
30 a Tier 1 analysis. In fact, as argued earlier, given the complexity of the overall risk assessment, a
31 full Tier 3 uncertainty analysis is warranted for consideration under the WHO guidelines
32 (although as discussed later, limitations in available data preclude completion of this level of
33 more-refined uncertainty analysis at this time).
34 For the qualitative uncertainty analysis, we have described each source of uncertainty and
35 qualitatively assessed its potential impact (including both the magnitude and direction of the
36 impact) on risk results, as specified in the WHO guidance. As shown in Table 3-13, for each
37 source of uncertainty, we have (a) provided a description, (b) estimated the direction of influence
38 {over, under, both, or unknown) and magnitude (low, medium, high) of the potential impact of
39 each source of uncertainty on the risk estimates, (c) assessed the degree of uncertainty (low,
February 2010 3-62 Draft - Do Not Quote or Cite
-------
1 medium, or high) associated with the knowledge-base (i.e., assessed how well we understand
2 each source of uncertainty), and (d) provided comments further clarifying the qualitative
3 assessment presented. Table 3-13 includes all key sources of uncertainty identified for the PM2 5
4 NAAQS risk assessment. A subset of these sources has been included in the Tier 2 quantitative
5 assessment discussed in section 3.5.4.
6 The categories used in describing the potential magnitude of impact for specific sources
7 of uncertainty on risk estimates (i.e., low, medium, or high) reflect EPA staff consensus on the
8 degree to which a particular source could produce a sufficient impact on risk estimates to
9 influence the interpretation of those estimates in the context of the PM NAAQS review.32
10 Sources classified as having a "low" impact would not be expected to impact the interpretation
11 of risk estimates in the context of the PM NAAQS review; sources classified as having a
12 "medium" impact have the potential to change the interpretation; and sources classified as "high"
13 are likely to influence the interpretation of risk in the context of the PM NAAQS review (if those
14 sources of uncertainty are reduced or more fully characterized). Because this classification of
15 the potential magnitude of impact of sources of uncertainty is qualitative and not informed
16 directly by any type of analytical results, it is not possible to place a quantitative level of impact
17 on each of the categories.33 Therefore, the results of the qualitative analysis of uncertainty have
18 limited utility in informing consideration of overall confidence in the core risk estimates and,
19 instead, serve primarily as a means for guiding future research to reduce uncertainty related to
20 PM2.5 risk assessment.
21 As with the qualitative discussion of sources of variability included in the last section, the
22 characterization and relative ranking of sources of uncertainty addressed here is based on
23 consideration by EPA staff of information provided in previous PM NAAQS risk assessments
24 (particularly sensitivity analyses), the results of the sensitivity analyses completed for the current
25 PM NAAQS risk assessment and information provided in the final PM ISA as well as earlier PM
26 Criteria Documents. Where appropriate, in Table 3-13, we have included references to specific
27 sources of information considered in arriving at a ranking and classification for a particular
28 source of uncertainty.
32 For example, if a particular source of uncertainty were more fully characterized (or if that source was reduced,
potentially reducing bias in a core risk estimate), would the estimate of incremental risk reduction in going from the
current to an alternative standard level change sufficiently to produce a different conclusion regarding the magnitude
of that risk reduction in the context of the PM NAAQS review?
33 Thematically, the categories used in the qualitative uncertainty analysis are similar to the categories used in
categorizing the results of the single- and multi-factor sensitivity analyses completed for this analysis (section 4.3).
However, in the context of the sensitivity analysis results, because we do have quantitative estimates of the impact
of individual modeling elements, it is possible to categorize the modeling elements included in the sensitivity
analysis based on magnitude of impact on risk estimates. This is not possible for the qualitative uncertainty analysis
described in this section.
February 2010 3-63 Draft - Do Not Quote or Cite
-------
1
2
Table 3-13. Summary of Qualitative Uncertainty Analysis of Key Modeling Elements in the PM NAAQS Risk
Assessment.
Source
Description
Potential influence of
uncertainty on risk
estimates
Direction
Magnitude
Knowledge-
Base
uncertainty*
Comments
(KB: knowledge base, INF: influence of uncertainty on risk
estimates)
A.
Characterizing
ambient PM2 5
levels for study
populations
using the
existing
ambient
monitoring
network
If the set of monitors used in a
particular urban study area to
characterize population
exposure as part of an ongoing
risk assessment do not match
the ambient monitoring data
used in the original
epidemiological study, then
uncertainty can be introduced
into the risk estimates.
Both
Low-
medium
Low-medium
KB and INF: In modeling risk, we focus on those counties that
were included in the epidemiological studies supplying the
underlying C-R functions. This means that, particularly for those
endpoints modeled using C-R functions obtained from more recent
studies, there is likely a close association between the monitoring
network used in the risk assessment and the network used in the
study supplying the C-R function(s). Note, however, that in those
instances where the networks are different (e.g., when older
studies are used, resulting in an increased potential for networks to
have changed), uncertainty may be introduced into the risk
assessment and it is challenging to evaluate the nature and
magnitude of the impact that that uncertainty would have on risk
estimates, given the complex interplay of factors associated with
mismatched monitoring networks (i.e., differences in the set of
monitors used in modeling risk and those used in the underlying
epidemiological study).
B.
Characterizing
policy-relevant
background
(PRB)
For this analysis, we have used
modeling to estimate PRB
levels for each urban study
area. Depending on the nature
of errors reflected in that
modeling, uncertainty (in both
directions) may be introduced
into the analysis.
Both
Low
Low
INF: Given that the risk assessment focuses primarily on the
reduction in risk associated with moving from the current NAAQS
to alternative standard levels, the impact of uncertainty in PRB
levels on the risk estimates is expected to be low. In addition, for
long-term exposure related mortality, we have based the core
analysis on modeling risk down to LML rather than PRB, which
reduces the significance of the PRB issue in the context of
modeling long-term exposure-related mortality.
C.
Characterizing
intra-urban
population
exposure in the
context of
epidemiology
studies linking
Exposure misclassification
within communities that is
associated with the use of
generalized population
monitors (which may miss
important patterns of exposure
within urban study areas)
introduces uncertainty into the
Under
(generally)
Medium-
high
High
KB and INF: Recent analyses in Los Angeles and New York City
based on ACS data (as reported in Krewski et al., 2009)
demonstrate the relatively significant effect that this source of
uncertainty can have on effect estimates (and therefore on risk
results). These analyses also illustrate the complexity and site-
specific nature of this source of uncertainty. The results of the Los
Angeles analysis suggest that exposure error may result in effects
estimates that are biased low and therefore result in the
February 2010
5-64
Draft - Do Not Quote or Cite
-------
Source
Description
Potential influence of
uncertainty on risk
estimates
Direction Magnitude
Knowledge-
Base
uncertainty*
Comments
(KB: knowledge base, INF: influence of uncertainty on risk
estimates)
PM2 5 to
specific health
effects
effect estimates obtained from
epidemiology studies.
underestimation of risk. Specifically in relation to the zip-code
level analysis based on ACS data conducted in Los Angeles
(Jerrett et al, 2005), the final ISA states that, "This [the refined
exposure analysis reported in the Jerrett study] resulted in both
improved exposure assessment and an increased focus on local
sources of fine particle pollution. Significant associations between
PM2 5 and mortality from all causes and cardiopulmonary diseases
were reported with the magnitude of the relative risks being
greater than those reported in previous assessments. In general,
the associations for PM2 5 and mortality using these two methods
[kriging and land-use regression] for exposure assessment were
similar, though the use of land use regression resulted in
somewhat smaller hazard ratios and tighter confidence intervals
(see Table 7-9). This indicates that city-to-city confounding was
not the cause of the associations found in the earlier ACS Cohort
studies. This provides evidence that reducing exposure error can
result in stronger associations between PM2 5 and mortality than
generally observed in broader studies having less exposure detail"
(final ISA, section 7.6.3, p. 7-90).
D. Statistical fit
ofthe C-R
functions
Exposure measurement error
combined with other factors
(e.g., size of the effect itself,
sample size, control for
confounders) can effect the
overall level of confidence
associated with the fitting of
statistical effect-response
models in epidemiological
studies.
Both
Low-
medium
(long-term
health
endpoints)
Medium
(short-term
health
endpoints)
Medium
INF: Long-term mortality studies benefit from (a) having larger
sample sizes (given that large national datasets are typically used
in deriving national-scale models), (b) the fact that the form of the
models used appears to be subject to relatively low uncertainty
(see next row below) and (c) our not attempting to derive location-
specific effects estimates (but instead, relying on national-scale
estimates). These factors combine to produce effects estimates that
tend to be statistically robust (as reflected in results presented in
Krewski et al., 2009). In addition, while concerns remain
regarding exposure misclassification and potential confounding,
generally we do not believe that the effects estimates are
consistently biased in a particular direction. In the case of short-
term mortality and morbidity health endpoints, there is greater
uncertainty associated with the fit of models given the smaller
sample sizes often involved, difficulty in identifying the
etiologically relevant time period for short-term PM exposure, and
the fact that models tend to be fitted to individual counties or
February 2010
5-65
Draft - Do Not Quote or Cite
-------
Source
Description
Potential influence of
uncertainty on risk
estimates
Direction Magnitude
Knowledge-
Base
uncertainty*
Comments
(KB: knowledge base, INF: influence of uncertainty on risk
estimates)
urban areas (which introduces the potential for varying degrees of
confounding and effects modification across the locations). In
contrast to the long-term mortality studies, the short-term
mortality and morbidity endpoints occasionally have effects
estimates that are not statistically significant. Note, however that
for this risk assessment, in modeling both short-term mortality and
morbidity endpoints, we are not relying on location-specific
models. In the case of short-term mortality, we are using city-
specific effects estimates derived using Bayesian techniques (these
combine national-scale models with local-scale models) (personal
communication with Zanobetti, 2009). For short-term morbidity,
we are using regional effects estimates (Bell et al., 2008). In both
cases, while effects estimates are at times non-statistically
significant, these models do benefit from larger sample sizes
compared to city-specific models.
E. Shape of the
C-R functions
Uncertainty in predicting the
shape of the C-R function,
particularly in the lower
exposure regions which are
often the focus in PM NAAQS
regulatory reviews.
Both
Medium
Low-medium
INF: Regarding long-term mortality, the ISA suggests that a log-
linear non-threshold model is best supported in the literature for
modeling both short-term and long-term health endpoints.
Although consideration for alternative model forms (Krewski et
al., 2009) does suggest that different models can impact risk
estimates to a certain extent, generally this appears to be a
moderate source of overall uncertainty. Particularly if, as is the
case in this risk assessment, we are not extrapolating below the
lowest measured levels found in the underlying epidemiological
studies. With regard to long-term mortality, the final ISA
concludes that, "In addition to examining the concentration-
response relationship between short-term exposure to PM and
mortality, Schwartz et al. (2008, 156963) conducted an analysis of
the shape of the concentration-response relationship associated
with long-term exposure to PM. Using a variety of statistical
methods, the concentration-response curve was found to be
indistinguishable from linear, and, therefore, little evidence was
observed to suggest that a threshold exists in the association
between long-term exposure to PM2 5 and the risk of death
(Section 7.6)." (section 2.4.3, p. 2-26). Regarding short-term
morbidity, the final ISA states that, "Overall, the studies evaluated
February 2010
5-66
Draft - Do Not Quote or Cite
-------
Source
Description
Potential influence of
uncertainty on risk
estimates
Direction
Magnitude
Knowledge-
Base
uncertainty*
Comments
(KB: knowledge base, INF: influence of uncertainty on risk
estimates)
further support the use of a no-threshold log-linear model, but
additional issues such as the influence of heterogeneity in
estimates between cities, and the effect of seasonal and regional
differences in PM on the concentration-response relationship still
require further investigation." (section 2.4.3, p. 2-25).
F. Addressing
co-pollutants
The inclusion or exclusion of
co-pollutants which may
confound, or in other ways,
affect the PM effect, introduces
uncertainty into the analysis.
Both
Low-
medium
Medium
INF: With regard to long-term health endpoints, the final ISA
states that, "Given similar sources for multiple pollutants (e.g.,
traffic), disentangling the health responses of co-pollutants is a
challenge in the study of ambient air pollution." (ISA, section
7.5.1, p. 7-57). The final ISA also notes that in some instances,
consideration of copollutants can have a significant impact on risk
estimates. For example, the more refined study of mortality in LA
as reported in Krewski et al., 2009 suggested that inclusion of
ozone in the model along with PM2 5 results in statistically non-
significant results for long-cancer mortality, while IHD-associated
mortality remained statistically significant (Krewski et al., 2009 -
Table 23). With regard to short-term mortality and morbidity, the
final ISA generally concludes that observed associations are fairly
robust to the inclusion of copollutants in the predictive models
(see ISA, sections 6.3.8, 6.3.9, and 6.3.10). The mixed impact of
considering multi-pollutant models in assessing PM2 5-associated
risk for short-term and long-term exposure related endpoints, leads
us to conclude that the potential impact of this source of
uncertainty is low-medium (depending on the specific endpoints
under consideration). The epidemiological studies used as the
basis for selecting C-R functions for the core risk assessment did
not include multi-pollutant models (with the exception of PM10.25
and PM2 5 combined models in Zanobetti and Schwartz, 2009).
However, we have included copollutant models in the sensitivity
analysis (see Section 4.3).
G. Potential
variation in
effects
estimates
reflecting
compositional
The composition of PM can
differ across study areas
reflecting underlying
differences in primary and
secondary PM2 5 sources (both
natural and anthropogenic). If
Both
Medium-
High
Medium-High
KB and INF: Epidemiology studies examining regional
differences in PM2 5-related health effects have found differences
in the magnitude of those effects (see sections 2.3.1.1 and 2.3.2 in
the draft ISA). While these may be the result of factors other than
composition (e.g., different degrees of exposure misclassification),
composition remains one potential explanatory factor. For short-
February2010
5-67
Draft - Do Not Quote or Cite
-------
Source
Description
Potential influence of
uncertainty on risk
estimates
Direction
Magnitude
Knowledge-
Base
uncertainty*
Comments
(KB: knowledge base, INF: influence of uncertainty on risk
estimates)
differences for
PM
these compositional differences
in fact translate into significant
differences in public health
impact (per unit concentration
in ambient air) for PM2 5 then
significant uncertainty may be
introduced into risk
assessments if these
compositional differences are
not explicitly addressed.
term exposure morbidity and mortality effects, the inclusion of
city-specific and/or regional-specific effect estimates in the risk
assessment may well reflect differences in PM composition and,
thus consideration of differences in risk due to city-specific
differences in composition may already be incorporated in the risk
estimates for these endpoints to some extent.
H. Specifying
lag structure
(short-term
exposure
studies)
Different lags may have
varying degrees of association
with a particular health
endpoint and it may be difficult
to clearly identify a specific lag
as producing the majority of a
PM-related effect (recently,
distributed lags have been
recommended since they allow
for a distribution of the impact
across multiple days of PM
exposure prior to the health
outcome). A lack of clarity
regarding the specific lag(s)
associated with a particular
health endpoint adds
uncertainty into risk estimates
generated for that endpoint.
Both
Medium
Medium
KB and INF: With regard to lag periods, the ISA states, "An
attempt has been made to identify whether certain lag periods are
more strongly associated with specific health outcomes. The
epidemiologic evidence evaluated in the 2004 PM AQCD
supported the use of lags of 0-1 days for cardiovascular effects and
longer moving averages or distributed lags for respiratory diseases
(U.S. EPA, 2004a). However, currently, little consensus exists as
to the most appropriate a priori lag times to use when examining
morbidity and mortality outcomes." (final ISA, section 2.4.2, p. 2-
24). This suggests that uncertainty remains concerning the
identification of appropriate lags, and thus the etiologically
relevant time period for exposure to PM for specific health
endpoints.
I.
Transferability
ofC-R
functions from
study locations
to urban study
area locations
The use of effects estimates
based on data collected in a
particular location(s) as part of
the underlying epidemiological
study in different locations (the
focus of the risk assessment)
introduces uncertainty into the
analysis.
Both
Medium (for
long-term
exposure
mortality)
Not
applicable
(for short-
Medium (for
long-term
exposure
mortality)
Low (for short-
term exposure
mortality)
INF: This issue has been ameliorated to a great extent in this risk
assessment since we are now using multi-city studies for key
short-term endpoints with effects estimates generally being
applied only to urban study areas matching locations used in the
underlying epidemiological study. In the case of long-term
exposure mortality studies, these are designed to capture a more
generalized national signal and therefore, concerns over the
transferability of functions between locations is of greater concern.
February 2010
5-68
Draft - Do Not Quote or Cite
-------
Source
Description
Potential influence of
uncertainty on risk
estimates
Direction
Magnitude
Knowledge-
Base
uncertainty*
Comments
(KB: knowledge base, INF: influence of uncertainty on risk
estimates)
term
exposure
health effect
risk
estimates)
J. Use of
single-city
versus multi-
city studies in
the derivation
ofC-R
functions
Often both single-city and
multi-city studies are available
(for a given health effect
endpoint) for the derivation of
C-R functions. Each of these
study designs has advantages
and disadvantages which
should be considered in the
context of assessing
uncertainty in a risk assessment
(Note, that generally this issue
applies more to the modeling
of short-term exposure-related
endpoints then to the modeling
of long-term exposure related
endpoints, since the latter is
typically based on multi-city
prospective cohort studies).
Both
Medium
High
KB: Because many health endpoints have been evaluated using
both single-city and multi-city studies, we have a relative large
selection of single city studies and a few large multi-city studies to
consider in examining this issue.
INF: For reasons presented in section 3.3.3, we have decided to
focus on multi-city studies as a source of C-R functions for the
core risk assessment, reflecting advantages that these studies offer
(e.g., they tend to have more statistical power and provide effect
estimates with relatively greater precision than single city studies
due to larger sample sizes, reducing the uncertainty around the
estimated coefficient, and reducing publication bias). While the
choice of multi-city studies is well-supported, this decision does
introduce uncertainty since single city studies can provide a wider
range of C-R functions (and associated effects estimates)
reflecting greater variation in study design, differences in
composition, human behavior, and copollutants, and differences in
the input datasets used (e.g., ambient air monitors and disease
baseline incidence data). Even if there is greater confidence in C-
R functions obtained from multi-city studies, overall uncertainty in
those C-R functions may be reflected to some extent in the range
of C-R functions seen across single-city studies.
K. Impact of
historical air
quality on
estimates of
health risk
from long-term
PM25
exposures
Long-term studies of mortality
suggest that different time
periods of PM exposure can
produce significantly different
effects estimates, raising the
issue of uncertainty in relation
to determining which exposure
window is most strongly
associated with mortality.
Both
Medium
Medium
INF: The latest HEI Reanalysis II study (HEI, 2009) which looked
at exposure windows (1979-1983 and 1999-2000) for long-term
exposure in relation to mortality, did not draw any conclusions as
to which window was more strongly associated with mortality.
However, the study did suggest that moderately different effects
estimates are associated with the different exposure periods (with
the more recent period having larger estimates). Overall, the
evidence for determining the window over which the mortality
effects of long-term pollution exposures occur suggests a latency
February 2010
5-69
Draft - Do Not Quote or Cite
-------
Source
Description
Potential influence of
uncertainty on risk
estimates
Direction Magnitude
Knowledge-
Base
uncertainty*
Comments
(KB: knowledge base, INF: influence of uncertainty on risk
estimates)
period of up to five years, with the strongest results observed in
the first few years after intervention (final ISA, section 7.6.4. p. 7-
95).
L.
Characterizing
baseline
incidence rates
Uncertainty can be introduced
into the characterization of
baseline incidence in a number
of different ways (e.g., error in
reporting incidence for specific
endpoints, mismatch between
the spatial scale in which the
baseline data were captured
and the level of the risk
assessment).
Both
Low-
medium
Low
INF: The degree of influence of this source of uncertainty on the
risk estimates likely varies with the health endpoint category under
consideration. There is no reason to believe that there are any
systematic biases in estimates of the baseline incidence data. The
influence on risk estimates that are expressed as incremental risk
reductions between alternative standards should be relatively
unaffected by this source of uncertainty.
KB: The county level baseline incidence and population estimates
at the county level were obtained from data bases where the
relative degree of uncertainty is low.
1 * Refers to the degree of uncertainty associated with our understanding of the phenomenon,
2 (specifically in the context of modeling PM risk)
in the context of assessing and characterizing its uncertainty
February 2010
5-70
Draft - Do Not Quote or Cite
-------
1 The results presented in Table 3-13 consider only the potential impact of each source of
2 uncertainty when acting in isolation to impact core risk estimates. However, it is likely that a
3 number of these sources of uncertainty could act in concert to impact risk estimates and
4 furthermore, that these combined effects could be more than additive in certain circumstances.
5 EPA staff has identified several combinations of sources of uncertainty addressed in Table 3-13
6 that should be highlighted due to their potential to produce significant impacts on core risk
7 estimates when acting in concert. These are briefly described below:
8 Uncertainty source D (statistical fit of the C-R functions), Source E (shape of the
9 C-R functions), Source F (addressing copollutants), and Source J (use of single-
10 city versus multi-city studies in the derivation of C-R functions): Consideration of
11 uncertainty associated with the shape of C-R functions needs to be considered in light
12 of overall confidence (uncertainty) associated with a particular model. A number of
13 factors contribute to an interpretation of confidence in a model including: statistical fit
14 of the model, degree to which potential confounding by copollutants is considered, and
15 other aspects of study design including single- versus multi-city study design. While
16 choice of a particular model (e.g., threshold model, or log-log model) may produce a
17 significant impact on risk estimates relative to alternative model forms, the overall
18 scientific support for that particular model form (informed by consideration of the
19 factors listed above) is an important consideration in assessing overall uncertainty both
20 from a qualitative and quantitative standpoint.
21 In addition, there is the potential for sources of uncertainty discussed in Table 3-13 to
22 interact with sources of variability covered in section 3.5.2 in impacting core risk estimates. One
23 such interaction is discussed below:
24 Uncertainty source A (characterizing ambient PM2.s levels for study populations
25 using the existing ambient monitoring network) and variability related to the
26 pattern of ambient PM2.s reductions at urban study areas (see section 3.5.2): The
27 estimation of a composite monitor value to use in modeling risk for a study area under
28 an alternative suite of standards is dependent both on the specification of the
29 monitoring network and the approach used in adjusting the concentrations for the
30 monitors in that network (i.e., the rollback approach used to simulate the pattern of
31 ambient PM2 5 reductions associated with just meeting the current or alternative suites
32 of standards). As we have seen in modeling risk for Pittsburgh, refinements in the
33 approach used to simulate air quality just meeting alternative suites of standards (in the
34 case of Pittsburgh transit!oning from a single study area to two distinct study areas
35 each with different design values and separate assessments of rollback) produced
36 significant differences in composite monitor values for the study area. Therefore, both
37 of these factors (the definition of the monitoring network and rollback approach) can
38 work in concert to impact ambient PM2.5 levels and hence risk estimates.
39
February 2010 3-71 Draft - Do Not Quote or Cite
-------
1 3.5.4 Single and Multi-Factor Sensitivity Analyses
2 We quantitatively examined the impact of several inputs to the risk assessment in a series
3 of single-factor sensitivity analyses summarized above in Table 3-8. A number of these sources
4 of uncertainty were also examined in-concert to assess their combined impact on core risk
5 estimates through the multi-factor sensitivity analysis. In addition, the sensitivity analysis
6 considered variability in the pattern of reductions in ambient PM2 5 associated with just meeting
7 the current and alternative suites of standards (i.e., consideration of variability in the simulation
8 of rollback). This section focuses on providing additional detail on the sources of alternative
9 model specifications and input datasets used in the sensitivity analysis (as alternative to the core
10 modeling approach).
11 Rather than present results for each sensitivity analysis for all of the air quality scenarios
12 considered in the core analysis, we selected a single air quality scenario - PM2.5 concentrations
13 that just meet the current standards - to use for the sensitivity analyses. The one exception to
14 this was the sensitivity analyses examining the impact of alternative approaches to simulating
15 just meeting alternative standards (the hybrid and peak-shaving rollback methods).34
16 In discussing the approach used in conducting the sensitivity analysis, we focus first on
17 methods used in assessing long-term exposure related health endpoints followed by the methods
18 used in assessing short-term exposure related health endpoints. We then discuss multi-factor
19 sensitivity analyses completed for both short-term and long-term exposure-related health
20 endpoints. Note, that the results of the sensitivity analyses (including both single- and multi -
21 factor analyses) are presented and discussed in section 4.3.
22 3.5.4.1 Sensitivity Analyses for Long-Term Exposure-Related Mortality
23 Because Krewski et al. (2009) presented results based on alternative model specifications
24 only for the later exposure period (1999 - 2000), our sensitivity analyses focusing on the
25 estimates of health effects incidence associated with long-term exposure to PM2.5 similarly used
26 the C-R functions based on this later exposure period. Krewski et al. (2009) considered several
27 alternative modeling approaches to estimate the relationship between mortality (both all cause
28 and cause-specific) and long-term exposure to PM2.5, providing us the opportunity to examine
29 the impact of alternative modeling approaches on the estimate of mortality risk associated with
30 long-term exposure. In particular, we examined the impact of using a random effects log-linear
31 model and of using a random effects log-log model35 (rather than the standard fixed effects log-
32 linear model used in the core analysis) to estimate the risks of all cause mortality,
34 Sensitivity analyses focusing on the hybrid and peak-shaving rollback approach (relative to the proportional
rollback approach used in the core analysis) involved the full set of alternative standard levels, in order to assess
potential differences in risk across the range of standard levels.
35 In the log-log model, the natural logarithm of mortality is a linear function of the natural logarithm of PM25.
February 2010 3-72 Draft - Do Not Quote or Cite
-------
1 cardiopulmonary mortality, ischemic heart disease mortality, and lung cancer mortality
2 associated with long-term exposure in Los Angeles and Philadelphia.36 The coefficient of PM2.5
3 in the random effects log-linear model was back-calculated from the relative risk reported in
4 Table 9 ("Autocorrelation at MSA and ZCA levels" group - "MSA & DIFF" row) of Krewski et
5 al. (2009). The coefficient of PM2.5 in the random effects log-log model was back-calculated
6 from the relative risks reported in Table 11 ("MSA and DIFF" rows) of Krewski et al. (2009).
7 As noted above, for all health endpoints associated with long-term exposure to PM2 5 we
8 estimated risk associated with PM2.5 concentrations above 5.8 |ig/m3 (the LML for the later
9 exposure period used in Krewski et al., 2009). In a sensitivity analysis we examined the impact
10 of that limitation by comparing those mortality risk estimates to the mortality risk estimates
11 obtained when we estimated risk associated with PM2 5 concentrations above estimated PRB
12 levels. This sensitivity analysis was carried out for all cause mortality in all 15 risk assessment
13 urban areas.
14 In addition, we compared the impact of using the primary C-R functions used in the risk
15 assessment, taken from Table 33 of Krewski et al. (2009), versus C-R functions for mortality
16 associated with long-term exposure reported in another study, Krewski et al. (2000), which was
17 based on a reanalysis of the Harvard Six Cities Study. The C-R functions estimated in Krewski
18 et al. (2000) from the Harvard Six Cities cohort were estimated for ages 25 and up, while the C-
19 R functions estimated in Krewski et al. (2009) from the ACS cohort were for ages 30 and up.
20 For purposes of consistency in the comparison, however, we applied the C-R functions from
21 Krewski et al. (2000) to ages 30 and up (and used the baseline incidence rates for that age group
22 as well).37 This sensitivity analysis was carried out for all cause mortality, cardiopulmonary
23 mortality, and lung cancer mortality in Los Angeles and Philadelphia.
24 We also considered the impact of using multi-pollutant models in estimating long-term
25 exposure-related mortality. Specifically, we obtained 2-pollutant models (considering CO, NO2,
26 Os or SO2 together with PM2 5) from Krewski et al., 2000, which is an earlier reanalysis of the
27 ACS dataset and used them in generating alternative estimates of all-cause mortality to contrast
28 with the core estimates generated using Krewski et al., 2009.
29 For all of the sensitivity analyses involving alternative C-R functions, in addition to
30 calculating the incidence of the health effect when an alternative approach is taken, we
36As noted in Table 3-8, we combined both of these alternative modeling approaches in a single sensitivity analysis.
In changing from a fixed effects log-linear model to a random effects log-log model, two changes are actually being
made - the change from a fixed effects log-linear model to a random effects log-linear model, and the change from a
random effects log-linear model to a random effects log-log model. However, because Krewski et al. (2009) did not
present results for a fixed effects log-log model, it was not possible to compare the impact of making the single
change from a fixed effects log-linear model (our core analysis selection) to a fixed effects log-log model. We thus
instead present a two-stage sensitivity analysis incorporating both of the changes.
37 The baseline incidence rates for ages 25 and up and ages 30 and up are likely to be very similar.
February 2010 3-73 Draft - Do Not Quote or Cite
-------
1 calculated the percent difference in estimates from the core analysis resulting from the change in
2 analysis input. So for example, when we calculated the incidence of all cause mortality
3 associated with long-term exposure to PM2 5 using a random effects log-log model (instead of the
4 fixed effects log-linear model used in the core analysis), we calculated the percent difference in
5 the result as (incidence estimated using a random effects log-log model - incidence estimated
6 using a fixed effects log-linear model)/( incidence estimated using a fixed effects log-linear
7 model).
8 Finally, we also examined the issue of variability in estimating the pattern of reductions
9 in ambient PM2.5 levels under the current and alternative standard levels (i.e., conducting
10 rollback). For the first draft RA, we considered the impact of using a hybrid rollback approach in
11 addition to the proportional rollback approach which has been more traditionally used in PM
12 NAAQS risk assessment (this sensitivity analysis was implemented including the generation of
13 quantitative risk estimates for a full suite of long-term exposure-related mortality categories).
14 For this second draft, as discussed above in sections 2.6, and 3.2.3, we have included
15 consideration of a peak shaving rollback approach in addition to the hybrid as non-proportional
16 methods to contrast with proportional rollback. As discussed in Section 3.2.3, for the second
17 draft risk assessment, rather than generating quantitative risk estimates, we have calculated
18 composite monitor estimates using the different rollback methods (proportional, hybrid and peak
19 shaving). The composite monitor values are surrogates for long-term exposure-related mortality.
20 Therefore, by comparing composite monitor values generated for the same study area/standard
21 level combination (using different rollback methods), we can obtain insights into the potential
22 impact of the rollback method used on long-term exposure-related mortality. Specifically, for
23 this sensitivity analysis, we compared composite monitor values in two ways:
24 Potential difference in composite monitor values at the current or alternative standard
25 level (for the same study area) given application of alternative rollback methods: We
26 compared the absolute magnitude of composite monitors values produced using different
27 rollback methods for the same study area/standard level combination to provide insights
28 into differences in the magnitude of residual risk for a given suite of standards in a study
29 area using different rollback methods (Appendix F, Table F-50).38 For example, in Table
30 F-50, for Los Angeles, we see that for the current standard suite of standards, use of
31 proportional rollback and peak shaving rollback methods results in composite monitor
32 values of 9.5 |ig/m3 and 12.0 |ig/m3, respectively, with the peak shaving value being 40%
33 higher than the value derived using proportional rollback. Given that the composite
34 monitor values are surrogates for long-term exposure-related mortality, we conclude that
35 for this combination of urban study area and suite of standards, use of the peak shaving
36 rollback method could produce PM2.5-attributable long-term mortality risk estimates that
37 are approximately 40% higher than use of the proportional rollback method.
38 This calculation reflects the fact that we model long-term exposure-related mortality down to LML.
February 2010 3-74 Draft - Do Not Quote or Cite
-------
1 Potential difference in the pattern of reduction in composite monitor values across
2 alternative standards: We compared differences in the percent reduction in composite
3 monitor values across alternative suites of standards for the same study area using
4 different rollback methods to provide insights into differences in incremental risk
5 reduction resulting from the use of different rollback approaches (Appendix F, Table F-
6 49).39 For example, in Table F-49, for Baltimore, we see that the proportional rollback
7 and hybrid rollback approaches resulted in composite monitor values for the 13/35
8 alternative suite of standards of 11.6 |ig/m3 and 11.8 |ig/m3, respectively, with these
9 translating into a percent reduction (compared with their respective values under the
10 current suite of standards) of 21% and 16%, respectively. Given that the composite
11 monitor values are surrogates for long-term exposure-related mortality, we conclude that
12 use of the two rollback methods (in the case of Baltimore for these two suites of
13 standards) does not appear to produce notably different patterns of risk reduction (in
14 terms of percent reduction), although residual risk could differ using the two approaches.
15 The peak-shaving and hybrid rollback approaches were not applied to all study areas,
16 since they are primarily applicable in certain situations.40 The sensitivity analysis results
17 described above (presented in Appendix F, Tables F-49 and F-50) form the basis for summary
18 information related to rollback approaches presented in Table 4-3.
19 In addition to the above insights regarding potential impacts on residual risk and the
20 degree of risk reduction across standard levels, inclusion of multiple rollback approaches also
21 allowed us to more fully examine the degree to which alternative 24-hour standards can produce
22 reductions in annual-average PM2.5 concentrations, thereby producing reductions in long-term
23 exposure-related mortality. As discussed below in section 6.2, alternative 24-hour standards,
24 when controlling, can result in reductions in annual average PM2.5 concentrations, particularly if
25 proportional rollback is used. In this case, the assumption of more regional patterns of PM2.5
26 reduction in reducing PM2.5 concentrations to just meet alternative 24-hour standards results in
27 an equivalent magnitude of reduction in the annual average. However, in simulating more
28 localized patterns of PM2.5 reductions to just meet alternative 24-hour standards, the PM2.5
29 reductions can be more limited to the monitor(s) (and areas) exceeding the 24-hour standard, and
30 other monitors may not be effected, resulting in a smaller impact on the annual average.
31 Inclusion of rollback approaches reflecting more localized patterns of ambient PM2 5 reduction
32 (i.e., the hybrid and particularly the peak shaving methods) allows us to assess the degree to
33 which alternative 24-hour standards (when controlling) produce appreciable reductions in
39 We note that this analysis also reflects calculation of long-term exposure-related mortality down to LML.
40 For the hybrid rollback approach, only select study areas had the mix of local sources in proximity to monitor
with elevated levels necessary to support consideration of a hybrid local/regional attainment strategy (i.e.,
application of the hybrid rollback) (i.e., Baltimore, Birmingham, Detroit, Los Angeles, New York, St. Louis). In the
case of the peak sharing approach, only those locations where the 24-hour standard was controlling were considered
for this sensitivity analysis (i.e., Atlanta, Baltimore, Birmingham, Detroit, Fresno, Los Angeles, New York,
Philadelphia, Phoenix, Pittsburgh, St. Louis, Tacoma).
February 2010 3-75 Draft - Do Not Quote or Cite
-------
1 annual-average PM2.5 concentrations and consequently in long-term exposure-related mortality.
2 This issue is revisited in discussing the results of the sensitivity analysis (section 4.3.1.1) and in
3 the integrative discussion of the core risk estimates (section 6.2).
4 3.5.4.2 Sensitivity Analyses for Short-Term Exposure-Related Mortality and
5 Morbidity
6 The scope of the sensitivity analysis completed for short-term exposure-related mortality
7 and morbidity is more limited than that completed for long-term exposure-related mortality.
8 This reflects, in part, the much greater magnitude of long-term exposure-related mortality. An
9 additional factor is that while there has been considerable research in the area of short-term
10 exposure-related mortality and morbidity which sheds light on uncertainty in such factors as C-R
11 function specification, this information is not directly applicable in a sensitivity analysis. In
12 order to complete a quantitative sensitivity analysis, we need alternative C-R function
13 specifications that produce risk estimates that can be directly compared to the core risk estimates.
14 Ideally, this is done by identifying alternative model forms in the epidemiological study used in
15 the core risk model. However, in the case of short-term exposure-related mortality, the studies
16 providing our core risk models (Zanobetti and Schwartz et al., 2009 and Bell et al., 2008), only
17 provide limited alternative model specifications, as described below. Further, alternative
18 epidemiological studies, such as Moolgavkar et al., 2003, while providing useful insights into
19 which factors can impact risk estimates (e.g., lag, multipollutant forms), cannot generate
20 alternative risk estimates that can be readily compared with the core risk estimates given
21 differences in the underlying study designs and datasets employed.
22 The primary studies selected to assess mortality risk and risk of hospitalization associated
23 with short-term exposure to PM2.5 (Zanobetti and Schwartz, 2009, and Bell et al., 2008,
24 respectively) both provided all-year C-R functions as well as season-specific C-R functions. We
25 examined the impact of using season-specific functions by applying these functions to each
26 season, as defined by the study authors,41 and summing the estimated season-specific incidences
27 of mortality and hospitalizations. We compared these estimates to the estimates obtained by
28 applying the corresponding all-year C-R functions to a year of air quality data.42 This sensitivity
29 analysis was carried out for all 15 of the risk assessment urban areas.
41 Both studies defined each season as three months, beginning with winter defined as December, January, and
February. In applying a season-specific function to a year of air quality data, we chose to keep a calendar year
together, so that, for example, winter 2005 was defined as December 2005, January 2005, and February 2005.
42 The mean season-specific incidence estimates can be summed to produce an all-year estimate of incidence.
However, the 2.5th and 97.5th percentile season-specific estimates cannot be summed. To calculate the 2.5th and
97.5th percentile estimates of all-year incidence from the season-specific estimates would require the variance-
covariance matrix of the season-specific coefficient estimators, which was not available. Therefore our comparison
of all-year estimates based on summed season-specific estimates versus estimates based on an all-year C-R function
was carried out only using the mean estimates.
February 2010 3-76 Draft - Do Not Quote or Cite
-------
1 In addition, Ito et al. (2007) estimated an annual C-R function as well as a seasonal
2 function for April through August for asthma ED visits in New York City. We compared the
3 results of applying the annual C-R function to a whole year of air quality data to the results of
4 applying the seasonal function to only those months (April through August) for which it was
5 estimated.
6 Moolgavkar (2003) estimated C-R functions for several health endpoints - non-accidental
7 and cardiovascular mortality; and cardiovascular and respiratory HAs - associated with short-
8 term exposures to PM2.5 in Los Angeles using different lag structures, different modeling
9 approaches to incorporating weather and temporal variables, and single-pollutant versus multi-
10 pollutant models. This study thus provided an opportunity to show the impact of lag structure,
11 modeling approach, and single- vs. multi-pollutant models, individually, for several health
12 endpoints associated with short-term exposures, although it is difficult to generalize to other
13 locations since the study was only conducted in a single urban area. As noted earlier, differences
14 in study design and the underlying datasets used prevent the results based on application of
15 models from Moolgavkar et al., 2003 from being compared directly to the core risk estimates.
16 Finally, as with estimates of long-term exposure-related mortality, we also considered the
17 impact of variability related to simulating ambient PM2.5 levels under the suite of current
18 standard levels (i.e., variability in conducting rollback) on estimates of non-accidental mortality
19 associated with short-term exposures to PM2 5 (using Zanobetti and Schwartz, 2009). However,
20 in this case, we only considered the hybrid model (consideration of peak shaving focused on the
21 impact on long-term exposure-related mortality). We note however, that sensitivity analysis
22 findings based on consideration for peak shaving generally will hold for short-term exposure-
23 related mortality and morbidity since both categories of health endpoints are also driven primary
24 by annual-average PM2.5 levels (see section 6.2). .
25 In all cases except the ED visits sensitivity analysis, in addition to calculating the
26 incidence of the health effect when an alternative approach is taken, we calculated the percent
27 difference in estimates from the core analysis resulting from the change in analysis input.43
28 3.5.4.3 Multi-factor Sensitivity Analyses
29 Each single-element sensitivity analysis shows how the estimates of PM2 5-related health
30 effects incidence change as we change a single element of the analysis (such as the form of the
31 C-R function or the way we simulate just meeting a set of standards). Because each of the
32 alternative modeling choices is considered to be a reasonable choice, the results of these single-
43 We did not calculate percent different for the ED visits sensitivity analysis because the two different C-R
functions (all-year in the core analysis vs. April through August in the sensitivity analysis) are also being applied to
different portions of the year (all year vs. April through August), so it is something of an "apple to oranges"
comparison.
February 2010 3-77 Draft - Do Not Quote or Cite
-------
1 element sensitivity analyses provide a set of reasonable alternative estimates that may similarly
2 be considered plausible (see section 4.3). The results of the single-element sensitivity analysis
3 are presented and discussed in section 4.3.1.
4 The single-element sensitivity analyses provide insight into which sources of uncertainty
5 may have the greatest impact on risk estimates when acting alone. However, there are several
6 sources of uncertainty in estimating PM2.5-related health effects. To provide a more complete
7 picture of the uncertainty surrounding estimates of PM2.5-related health effects incidence - and
8 to expand the set of reasonable alternative estimates - we next carried out multi-element
9 sensitivity analyses. The results of the multi-factor sensitivity analysis are presented and
10 discussed in section 4.3.1.2.
11 The choice of uncertain analysis elements to include in the multi-element sensitivity
12 analyses was guided by the single-element sensitivity analyses. In particular, we selected those
13 modeling choices that had the greatest impacts on the estimates of health effects incidence in the
14 single-element sensitivity analyses to provide insight into the scope of possible estimates that,
15 while perhaps not based on our first choice of analysis elements, are nevertheless plausible
16 alternative estimates.
17 We identified three analysis elements that substantially affected the estimates of mortality
18 associated with long-term exposure to PM2.5 the model choice (fixed effects log linear vs.
19 random effects log-log), whether effects are estimated associated with PM2 5 concentrations
20 down to the LML in the study (5.8 |ig/m3) or down to PRB, and whether a proportional or a
21 hybrid rollback is used to simulate PM2.5 concentrations that just meet a given set of standards.
22 This resulted in 2x2x2 = 8 different estimates of mortality, all of which could be considered
23 plausible, based on the fact that the underlying model choices are all considered reasonable.
24 We identified two analysis elements that substantially affected the estimates of mortality
25 associated with short-term exposure to PM2.5 - whether season-specific or all-year C-R functions
26 were used and whether a proportional or a hybrid rollback approach was used to simulate just
27 meeting the current and alternative standards.
28 3.5.5 Summary of Approach to Addressing Variability and Uncertainty
29 The characterization of uncertainty and variability associated with the risk assessment
30 includes a number of elements, which have been discussed in detail above. These include:
31 Identification of key sources of variability associated with PM2.5-related population
32 exposure and hazard response and the degree to which they are captured in the risk
33 assessment (see section 3.5.2). When important sources of variability in exposure
34 and/or hazard response are not reflected in a risk assessment, significant uncertainty
35 can be introduced into the risk estimates that are generated. While not explicitly
36 referenced in the WHO guidance, this assessment (focused on coverage for key sources
February 2010 3-78 Draft - Do Not Quote or Cite
-------
1 of variability) could be considered part of a Tier 1 analysis (i.e., the qualitative
2 characterization of sources of uncertainty).
3 Qualitative assessment of uncertainty, including both an assessment of the magnitude
4 of potential impact of each source on risk estimates (along with the potential direction
5 of that impact) as well as an assessment of overall confidence associated with our
6 understanding of that source of uncertainty (see section 3.5.3). This represents a WHO
7 Tier 1 analysis.
8 Single-factor sensitivity analysis intended to evaluate the impact of individual sources
9 of uncertainty and variability on risk estimates (see section 3.5.4). The goal of this
10 assessment is to evaluate the relative importance of these sources of uncertainty and
11 variability in impacting core risk estimates. The single-factor sensitivity analysis
12 represents a WHO Tier 2 analysis. In conducting these assessments, we have used
13 alternative representations of modeling elements that have support in the literature to
14 ensure that the risk estimates that are generated represent reasonable alternate estimates
15 that can supplement the core risk estimates generated in the analysis (see section 4.5.3).
16 Multi-factor sensitivity analysis intended to assess the combined impact of multiple
17 sources of uncertainty and variability on risk estimates (see section 3.5.4). By
18 considering the combined effect of multiple sources of uncertainty and variability, this
19 analysis has the potential to identify any non-linearities which can magnify the impact
20 of uncertainty and variability on risk estimates, especially if several non-linear factors
21 act in concert. This also represents a WHO Tier 2 analysis. As with the single-factor
22 sensitivity analysis results, these risk estimates are also generated using modeling
23 inputs which have support in the literature and consequently, they also represent
24 reasonable alternate estimates that supplement the core risk estimates (see section
25 4.5.2).
26 As noted above, since information was not available to characterize overall levels of
27 confidence in alternative model inputs, the uncertainty characterization completed for this risk
28 assessment did not include a full probabilistic assessment of uncertainty and its impact on core
29 risk estimates (i.e., a WHO Tier 3 analysis was not completed). Further, the risk estimates
30 generated using the single- and multi-factor sensitivity analyses do not represent uncertainty
31 distributions, but rather additional plausible point estimates of risk (i.e., we do not know whether
32 they represent risk estimates near the upper or lower bounds of a true but undefined uncertainty
33 distribution and we do not know the actual population percentiles that they represent). The
34 appropriate use for these reasonable alternate risk estimates in informing consideration of
35 uncertainty in the core risk estimates is discussed in section 4.5.3.
36 In addition to the specific analyses discussed above, we have also completed two
37 additional analyses intended to place the 15 urban study areas in a broader national context with
38 regard to risk. These include a representativeness analysis which evaluates the way the 15 urban
39 study areas compare to national distributions for key PM-related risk attributes (discussed in
40 section 4.4). We have also completed a national-scale assessment of long-term mortality related
February 2010 3-79 Draft - Do Not Quote or Cite
-------
1 to PM2.5 exposures (chapter 5), which, in addition to providing an estimate of the national impact
2 of PM2.5 on long-term mortality, also evaluates whether the set of 15 urban study areas generally
3 represents the broader distribution of risk across the U.S., or a more focused portion of the
4 national risk distribution (e.g., the higher-end).
5 A third set of analyses that has been added to this second draft RA focuses on evaluating
6 patterns in the design values (including both 24-hour and annual) and underlying PM2.5
7 monitoring data for the 15 urban study areas (see Section 4.5). The goal of this analysis is to use
8 this information to enhance our understanding of patterns in risk reduction seen under both the
9 current and alternative suites of standards across the urban study areas. The interplay of design
10 values and underlying PM2.5 monitoring data play a key role in determining whether a location
11 will experience risk reductions when just meeting any given suite of standards is simulated and,
12 if so, the magnitude of those reduction. As part of this analysis, we contrast patterns in design
13 values for the 15 urban study areas with patterns seen more broadly across urban areas in the
14 U.S. with the goal of placing the urban study areas in a national context with regard to this key
15 factor influencing risk.
February 2010 3-80 Draft - Do Not Quote or Cite
-------
1 4 URBAN CASE STUDY RESULTS
2 For this risk assessment, we have developed a core set of risk estimates supplemented by
3 an alternative set of risk results generated using single-factor and multi-factor sensitivity
4 analysis. The core set of risk estimates was developed using model inputs that staff judge to
5 have a greater degree of support in the literature relative to inputs used in the sensitivity analyses
6 (the rationale for selection of specific epidemiological studies and associated C-R functions for
7 the core analysis is discussed above in section 3.3.3). This chapter presents and discusses the
8 core set of risk estimates generated for the urban case study area, and also discusses the results of
9 the sensitivity analyses which serve to augment the core risk estimates. The results of the
10 sensitivity analyses allow us to evaluate and rank the potential impact of key sources of
11 uncertainty on the core risk estimates. In addition, because the sensitivity analyses were
12 conducted using alternative modeling inputs having some degree of support in the literature, the
13 results of the sensitivity analysis also represent a set of reasonable alternatives to the core set of
14 risk estimates that can be used to inform characterization of uncertainty in the core results (see
15 section 4.3 below).
16 As discussed above in section 2.2 and 3.2, this risk assessment includes consideration of
17 the following air quality scenarios:
18 Recent conditions: based on PM2.5 concentrations characterized through monitoring for
19 the period 2005-2007 at each urban case study location;
20 Current NAAQS: based on rolling back PM2.5 concentrations to just meet the current
21 suite of standards in each urban study area (annual standard of 15 |ig/m3 and a 24-hour
22 standard of 3 5 |ig/m3, denoted 15/3 5);
23 Alternative NAAQS: based on rolling back PM2.5 concentrations to just meet alternative
24 suites of standards in each urban study area:
25 o annual standard of 14 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted 13/35);
26 o annual standard of 13 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted 13/35);
27 o annual standard of 12 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted 12/35);
28 o annual standard of 13 |ig/m3 and a 24-hour standard of 30 |ig/m3 (denoted 13/30);
29 o annual standard of 12 |ig/m3 and a 24-hour standard of 25 |ig/m3 (denoted 12/25).
30 In simulating both current and alternative suites of standards, for the core analysis, we
31 used a proportional roll-back approach (see section 3.2.3), while a hybrid roll-back approach
32 reflecting the potential for local source control was used for a subset of urban study areas as part
33 of the sensitivity analysis conducted for this assessment (see section 3.2.3). In addition, we have
34 considered the peak-shaving approach as a further alternative to proportional rollback in
February 2010 4-1 Draft - Do Not Quote or Cite
-------
1 simulating just meeting the current and alternative suites of standards. While we did not generate
2 risk estimates based on application of the peak-shaving approach, we did generate composite
3 monitor-based annual average PM2 5 levels which allow us to assess how long-term exposure-
4 related risk could vary if this alternative roll-back method was used (see Section 4.3).
5 As described in section 2.1 and 3.3.2, we assessed risk for 15 urban study areas chosen to
6 provide coverage for the diversity of urban settings across the U.S. that reflect areas with
7 elevated annual and/or daily PM2.5 concentrations. At a minimum, all areas selected had recent
8 air quality levels at or above the lowest annual and/or 24-hour standards analyzed. In addition,
9 our goal was to select areas reflecting the heterogeneity in PM risk-related attributes such as
10 sources, composition, demographics, and population behavior.
11 Risk estimates were generated for the following health effects endpoints: (a) long-term
12 exposure-related mortality (all-cause, cardiopulmonary disease-related (CPD), ischemic heart
13 disease-related (IHD) and lung cancer-related), (b) short-term exposure-related mortality (non-
14 accidental, cardiovascular disease-related (CVD), respiratory), and (c) short-term exposure-
15 related morbidity (hospital admissions (HA) for CVD and respiratory illness and emergency
16 department (ED) visits). Risk estimates are presented separately for each of these 15 study areas,
17 although in certain circumstances, risk estimates may be restricted to a subset of these locations
18 if, for example, an endpoint is modeled using a C-R function derived from an epidemiological
19 study that was conducted only in a subset of the urban areas. For the core analysis, long-term
20 exposure mortality risk was modeled down to lowest measured level (LML), because the LML
21 was higher than estimated PRB and because there is substantial uncertainty as to the shape of the
22 concentration-response (C-R) function at concentrations below the LML. For long-term
23 exposure mortality a sensitivity analysis was conducted that estimated risk down to policy-
24 relevant background (PRB). In contrast, all short-term exposure health effects endpoints were
25 modeled down to PRB, since this was higher than the LML across all studies and for purposes of
26 NAAQS decision making, EPA is focused on risks associated with PM2 5 levels that are due to
27 anthropogenic sources that can be controlled by U.S. regulations (or through international
28 agreements with neighboring countries).
29 In modeling long-term exposure mortality, for the core analysis, we have based estimates
30 on the latest reanalysis of the American Cancer Society (ACS) dataset, with two sets of risk
31 estimates being generated; one using a C-R function derived by fitting PM2.5 monitoring data
32 from 1979-1983 and a second set based on fitting PM2.5 monitoring data from 1999-2000
33 (Krewski et al., 2009) (see section 3.3.3). In presenting core risk estimates for long-term
34 mortality, both sets of estimates are given equal weight.
35 In modeling short-term exposure mortality and morbidity for the core analysis, we have
36 used the latest multi-city studies (Zanobetti and Schwartz, 2009; Bell et al., 2008) (see section
February 2010 4-2 Draft - Do Not Quote or Cite
-------
1 3.3.3). In the case of short-term exposure mortality, we obtained and used city-specific effects
2 estimates derived using empirical Bayes methods from the study authors (Zanobetti, 2009).
3 Multi-city studies were favored for the core analysis, since these studies are not subject to
4 publication bias and because they reflect a diverse set of locations with regard to the observed
5 relationship between short-term PM2.5 exposure and health affect response in the population.
6 Additional detail on the specific C-R functions and related modeling elements such as effects
7 estimates and lag periods used in the core analysis relative to the sensitivity analysis are
8 presented above in sections 3.3 and 3.4 and called out where appropriate below as specific risk
9 estimates are discussed.
10 The pattern of mortality incidence across the urban study areas is markedly different for
11 short-term exposure-related mortality compared with long-term exposure-related mortality
12 reflecting a number of factors including: (a) differences in patterns of daily PM2.5 levels versus
13 annual average values across the urban study areas and (b) the fact that urban study area-specific
14 effect estimates are used in modeling short-term exposure-related mortality, while a single effect
15 estimate is used for all study areas for long-term exposure-related mortality (for a particular
16 mortality category). Further, effect estimates for short-term exposure-related mortality can be
17 notably small for some study areas (e.g., the effect estimates for non-accidental mortality for Los
18 Angeles is significantly smaller than effect estimates for the other study areas, thereby
19 accounting for the relatively small total incidence estimate for this study area - see Appendix C,
20 Table C-l).
21 Because the recent conditions air quality scenario spans three years (2005-2007), risk
22 estimates are generated for each of these years, reflecting the underlying air quality data for a
23 particular year. Risk metrics generated for the above health effects endpoints include:
24 Annual incidence of the endpoint due to PM2.s exposure (annual incidence).
25 Generated for the population associated with a given urban study area (for a given
26 simulation year), in most cases, these risk estimates include both a point estimate as well
27 as a 95th percentile confidence interval, the latter reflecting sampling error as
28 characterized in the underlying epidemiological study.
29 Percent of total annual incidence for the health endpoint due to PM2.s exposure
30 (percent of total incidence attributable to PM2.s): Again, generated for the population
31 associated with a given urban study area (and simulation year), this metric characterizes
32 the fraction of total incidence that is associated with PM2.5 exposure. As with the
33 underlying PM-related incidence estimates, this risk metric also typically includes a 95th
34 percentile confidence interval reflecting sampling error associated with the effects
35 estimate. Compared with the annual incidence metric which reflects underlying
36 population size for each study area, this risk metric has the advantage of not being
37 dependent on the size of the underlying population, thereby allowing direct comparison
38 of the potential impact of PM2.5 for the health effect endpoint of interest across urban
39 study area locations. For this reason, in discussing risk estimates in this section, the
February 2010 4-3 Draft - Do Not Quote or Cite
-------
1 percent of total incidence attributable to PM2.s risk metric is given greater emphasis than
2 the absolute measure of annual incidence attributable to PM2.s.
3 Percent reduction in PM2.s-related health effect incidence for an alternative set of
4 standards or the recent conditions scenario, relative to the current standards
5 (percent change from the current set of standards). Also estimated separately for each
6 urban study area and simulation year, this metric characterizes the degree of risk
7 reduction (for alternative standard levels) or increased risk (for the recent conditions
8 scenario) relative to the current NAAQS. For this metric, a negative value represents an
9 increase in risk (this is the case for the recent conditions scenario, where risks are higher
10 than those associated with just meeting the current suite of standards). This metric is
11 positive, or zero, for alternative suites of standards since they either produce no risk
12 reduction (if ambient air levels under recent conditions are already at or below that
13 alternative standard levels), or a positive risk reduction for alternative standards resulting
14 in a reductions in ambient PM2.5 concentrations. Because this metric is incremental, it
15 was not possible to generate the 95th percentile confidence intervals included with the
16 other two "absolute" risk metrics described above. As with the previous risk metric, this
17 metric is not dependent on the underlying population size and therefore, allows direct
18 comparison across urban study areas.
19 In addition to presenting the central-tendency (highest confidence) estimates for each of
20 these metrics, we also include 95th percentile confidence intervals, reflecting statistical
21 uncertainty surrounding the estimated coefficients in the reported C-R functions used in deriving
22 the risk estimates (note, that these confidence intervals only capture this statistical fit uncertainty
23 - other sources of uncertainty including shape and form of the function, are addressed separately
24 as part of the sensitivity analysis - see Section 4.3.1.1 and the qualitative analysis of uncertainty
25 - see Section 3.5.3).
26 Detailed tables presenting estimates for these risk metrics for the complete set of air
27 quality scenarios (for all 15 urban study areas) are included in Appendix E and referenced as
28 needed in the discussion of risk estimates presented in the following sections. To support the
29 discussion of risk estimates presented in this chapter, we have included a subset of tables and
30 summary figures including:
31 Tables summarizing risk for the current standard levels: Two tables are included
32 which summarize both long-term and short-term exposure-related risk for the 15 urban
33 study areas associated with just meeting the current suite of standards. Both tables
34 include a subset of the health endpoints believed to have the greatest support in the
35 literature including IHD mortality for long-term exposure, cardiovascular mortality and
36 hospital admissions for short-term exposure. Table 4-1 presents total incidence
37 attributable to PM2.5 exposure for the endpoints and Table 4-2 presents percent of total
38 incidence attributable to PM2.5 exposure for these endpoints. Together, these tables
39 inform consideration for the magnitude of public health impact (related to both long-term
40 and short-term exposure to PM2 5) associated with just meeting the current suite of
41 standards in the 15 urban study areas.
February 2010 4-4 Draft - Do Not Quote or Cite
-------
1 Figures illustrating the percent reduction in long-term and short-term exposure-
2 related risk for the alternative standard levels relative to the current standard (as
3 well as increases in risk under recent conditions relative to the current standard):
4 Figures 4-1 and 4-4 provide a snapshot of trends in risk reduction for long-term exposure-
5 related risk (Figure 4-1) and short-term exposure-related risk (Figure 4-4) across
6 alternative standard levels relative to the risk under the current standard. These figures
7 include plots for each of the 15 urban study areas, thereby allowing trends in risk
8 reduction across standard levels (and urban study areas) to be assessed simultaneously.44
9 Each of these figures is presented in additional detail by splitting each into (a)
10 comparison of the recent conditions risk against the current standard level and (b)
11 comparison of risk under alternative standard level against the current standard, in order
12 to allow a more detailed look at patterns in risk reduction for individual urban study areas
13 (splitting Figures 4-1 and 4-4 in this fashion allows greater resolution in tracing the linear
14 risk plots for each study area). Specifically, Figures 4-2 and 4-3 provide these higher-
15 resolution plots for long-term exposure-related risk and Figures 4-5 and 4-6 provide
16 higher-resolution plots for short-term exposure related risk.
17 Although risk estimates were generated for all three simulation years, in this chapter core risk
18 estimates primarily from 2007 are presented and discussed for both the recent conditions air
19 quality scenario and just meeting current and alternative suites of standards. This reflects the
20 observation that in generally 2007 represents a reasonable central year (in terms of the magnitude
21 of risk generated for the three simulated years), when considering results for all modeled health
22 effect endpoints across the 15 study areas. In addition, 2007 is the most recent year of the three
23 simulated. We note, however, that while we do focus on 2007 in presenting and discussing risk
24 estimates, we include an assessment of general trends across the three simulation years to gain
25 perspective on year-to-year variation in PM2.5-related risk estimates as assessed here.
26
27
28
29
30
31
32
33
44 Note, that importantly, patterns of risk reduction across standard levels (in terms of percent change relative to risk
for the current standard level) are similar for all health endpoints modeled for a particular exposure duration (i.e.,
patterns of percent risk reduction will be similar for long-term exposure related all-cause, IHD and cardiopulmonary
mortality). This reflects the fact that the C-R functions used in this risk assessment are close to linear across the
range of ambient air levels evaluated. This allows us to present these figures plotting changes in risk more generally
for short-term exposure-related endpoints and long-term exposure related endpoints without having to provide
figures for each specific endpoint category.
February 2010 4-5 Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
Table 4-1. Estimated Annual Incidence of Selected Mortality and Morbidity Endpoints
Associated with Long- and Short-Term Exposure to Ambient PM2.s
Concentrations that Just Meet the Current Standards, Based on Adjusting
2007 PM2.5 Concentrations. u
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease
Mortality Associated with Long-term
Exposure to PM2.53
Exposure Period:
1979-1983
220
(180-258)
297
(243 - 349)
131
(107- 154)
195
(159-230)
377
(308 - 445)
77
(63 - 92)
344
(281 -405)
860
(701 - 1018)
1755
(1435-2070)
261
(214-308)
317
(258 - 374)
256
(209-302)
15
(12- 18)
446
(365 - 525)
38
(31 - 46)
Exposure Period:
1999-2000
277
(227-324)
374
(307-440)
165
(135- 194)
247
(202-291)
478
(390-563)
98
(80- 116)
434
(355-511)
1094
(890 - 1296)
2222
(1814-2620)
330
(270-389)
402
(327-476)
324
(264-382)
19
(16-23)
563
(461-662)
49
(40-58)
Incidence of
Cardiovascular
Mortality Associated
with Short-term
Exposure to PM2.54
32
(-33-95)
62
(-4 - 126)
-1
(-42-40)
29
(-19-76)
60
(-8 - 127)
12
(-9 - 33)
46
(-31 - 122)
-30
(-132-72)
473
(276 - 668)
84
(22 - 145)
84
(-4 - 170)
43
(-9 - 93)
9
(-2 - 20)
106
(24 - 187)
1 1
(-6 - 27)
Incidence of
Cardiovascular
Hospitalizations
Associated with Short
term Exposure to
PM2.55
41
(-27- 109)
216
(159-273)
16
(-1 1 - 43)
28
(-18-73)
233
(171 -295)
23
(0 - 46)
56
(-37- 149)
258
(3-511)
752
(552-951)
203
(149-257)
108
(1 -215)
140
(103- 177)
9
(0 - 18)
178
(131-225)
19
(-46 - 82)
1The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible interval based on statistbal
uncertainty surrounding the PM coefficient.
SEstimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1979-1983 and from 1999-2000 respectively.
4Basedon location-specific single pollutant concentration-response function estimatesfrom Zanobetti and Schwartz (2009) that have been
"shrunken" towardsthe appropriate regional means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A.
Zanobetti via email.
Slncidenceestimateswere calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et
al. (2008). Location-specific C-R function estimates were not available from this study.
9
10
11
12
13
14
February 2010
4-6
Draft - Do Not Quote or Cite
-------
1 Table 4-2 Estimated Percent of Total Annual Incidence of Selected Mortality and
2 Morbidity Endpoints Associated with Long- and Short-Term Exposure to
3 Ambient PM2.5 Concentrations that Just Meet the Current Standards, Based
4 on Adjusting 2007 PM2.5 Concentrations. 1,2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Incidence of Ischemic Heart
Disease Mortality Associated with Long-term
Exposure to PM2.53
Exposure Period:
1979-1983
1 3.2 %
(10.9% - 15.5%)
1 1 .7 %
(9.6% - 13.7%)
1 0.9 %
(8.9% - 12.9%)
9%
(7.3% - 10.6%)
9.1%
(7.4% - 10.7%)
6.7%
(5.5% -8%)
1 0.7 %
(8.8% - 12.6%)
6. 1 %
(4.9% - 7.2%)
9. 3%
(7.6%- 11%)
10.5%
(8.6% - 12.3%)
6.7%
(5.5% - 7.9%)
9.3%
(7.6%- 11%)
2.9%
(2. 4% -3. 4%)
1 1 .2 %
(9.2% - 13.2%)
3.7%
(3% - 4.4%)
Exposure Period:
1999-2000
16.7%
(13.7% - 19.5%)
14.7%
(12.1% - 17.3%)
13.8%
(1 1 .3% - 1 6.2 %)
11.4%
(9.3% - 13.4%)
11.5%
(9.4% - 13.5%)
8 .5%
(7%- 10.1%)
13.6%
(11.1%- 16%)
7 .7%
(6.3% -9.1%)
11.8%
(9.6% - 13.9%)
13.2%
(10.8% - 15.6%)
8.5%
(6.9% - 10.1%)
11.8%
(9.6% - 13.9%)
3.7%
(3% -4.4%)
14.2%
(11.6% - 16.7%)
4.7%
(3.8% -5.6%)
Percent of Incidence
of Cardiovascular
Mortality Associated
with Short-term
Exposure to PM2.54
0.8%
(-0.8% -2.4%)
1.6%
(-0.1% -3.2%)
0%
(-1.5% -1.5%)
0.8%
(-0.5% -2.2%)
1%
(-0.1% -2.2%)
0.7%
(-0.5% -2%)
0.9%
(-0.6% -2.4%)
-0.2%
(-0.7% -0.4%)
2.1%
(1.2% -3%)
2.1%
(0.5% - 3.6%)
1.3%
(-0.1% -2.7%)
1.1%
(-0.2% -2.3%)
0.8%
(-0.2% -1.7%)
1.9%
(0.4% - 3.3%)
0.7%
(-0.4% -1.8%)
Percent of Incidence
of Cardiovascular
Hospital Admissions
Associated with Short
term Exposu re to
PM2.55
0.4%
(-0.2%- 1%)
1 .3%
(1% -1.7%)
0.3%
(-0.2% -0.9%)
0.3%
(-0.2% -0.7%)
1.1%
(0.8% - 1.4%)
0.5%
(0%-0.9%)
0.3%
(-0.2% -0.8%)
0.5%
(0% -0.9%)
1 .2%
(0.8% - 1.5%)
1 .3%
(0.9% - 1.6%)
0.5%
(0% - 1 %)
1.1%
(0.8% - 1.4%)
0.4%
(0%-0.7%)
1 .3%
(0.9% - 1.6%)
0.5%
(-1.3% -2.3%)
1 The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35ug/m3.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
SEstimates Based on Krewskiet at (2009), Using Ambient PM2.5 from 1979-1983 and from 1999-2000 respectively
4Basedon location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been
"shrunken" towardsthe appropriate regional means. "Shrunken" coefficent estimates and their standard errors were sent to EPA by A
Zanobetti via email.
Slncidence estimateswere calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et
al. (2008). Location-specificC-R function estimates were not avaiablefrom this study.
February 2010
4-7
Draft - Do Not Quote or Cite
-------
1
2
Figure 4-1
Percent reduction in long-term exposure-related mortality risk (alternative standards and recent conditions relative to the current standards)
(Note: inset shows PM25 related incidence and percent of total incidence for IHD mortality under the current suite of standards)
03
"53
°?0
§ £
3 2
o *=
Q- c
x o
UJ £
E g
90%
70% -
50%
30%
10%
-10%
-30%
-50%
4
5
6
7
Atlanta, GA 277 (227-324); 16.7% (13.7% -19.5%)
Baltimore, MD 374 (307-440); 14.7% (12.1% -17.3%)
Birmingham, AL 165 (135-194); 13.8% (11.3% -16.2%)
Dallas, TX 247 (202-291); 11.4% (9.3%-13.4%)
Detroit, Ml 478 (390-563); 11.5% (9.4%-13.5%)
Fresno, CA 98 (80-116); 8.5% (7%-10.1%)
Houston, TX 434 (355-511); 13.6% (11.1%-16%)
Los Angeles, CA 1094 (890-1296); 7.7% (6.3%-9.1%)
New York, NY 2222 (1814-2620); 11.8% (9.6%-13.9%)
Philadelphia, PA 330 (270-389); 13.2% (10.8% -15.6%)
Phoenix, AZ 402 (327-476); 8.5% (6.9%-10.1%)
Pittsburgh, PA 324 (264-382); 11.8% (9.6%-13.9%)
Salt Lake City, UT 19 (16-23); 3.7% (3%-4.4%)
St. Louis, MO 563 (461-662); 14.2% (11.6% -16.7%)
Tacoma, WA 49 (40-58); 4.7% (3.8%-5.6%)
SalfLake City,
/ /Tacoma, \
2007 air
quality
15/35*
14/35
13/35
12/35
13/30
12/25
Recent Air Quality, Current Standard and Alternative Standards
*Based on Krewski et al. (2009), exposure period from 1999 - 2000. The legend contains, for each urban area, the incidence estimate (and 95% CI) and the
estimate of percent of total incidence (and 95% CI) under the current standards.
**The current standards consist of an annual standard of 15 ug/m3 and a daily standard of 35 ug/m3. Combinations of an annual standard (n) and a daily standard
(m) are denoted n/m in this figure.
February 2010
4-8
Draft - Do Not Quote or Cite
-------
1
2
Figure 4-2
Percent reduction in long-term exposure-related mortality risk (recent conditions relative to the current standards) (Note: inset shows
PM2 5 related incidence and percent of total incidence for IHD mortality under the current suite of standards)
3
4
5
6
7
31
7 4-1
o> c
rn *w
II
I E
3 S
o *-
^§
HI ~
^1
I*
O oi
-1 p
o%
-50%
-100%
-150%
-200%
-250%
Phoenix, AZ
Dallas, TX
Houston, T
Atlanta, GA
Philadelphia, PA
New York, N
Jirmingham, AL
Detroit, Ml
Los Angeles, CA
Salt Lake City, UT
^- Atlanta, GA 277 (227-324); 16.7% (13.7% -19.5%)
- Baltimore, MD 374 (307-440); 14.7% (12.1% -17.3%)
* Birmingham, AL 165 (135-194); 13.8% (11.3% -16.2%)
* Dallas, TX 247 (202-291); 11.4% (9.3%-13.4%)
- Detroit, Ml 478 (390-563); 11.5% (9.4%-13.5%)
Fresno, CA 98 (80-116); 8.5% (7%-10.1%)
-- Houston, TX 434 (355-511); 13.6% (11.1%-16%)
Los Angeles, CA 1094 (890-1296); 7.7% (6.3%-9.1%)
NewYork, NY 2222 (1814-2620); 11.8% (9.6%-13.9%)
Philadelphia, PA 330 (270-389); 13.2% (10.8% -15.6%)
-*- Pittsburgh, PA 324 (264-382); 11.8% (9.6%-13.9%)
-m- Salt Lake City, UT 19 (16-23); 3.7% (3%-4.4%)
St. Louis, MO 563 (461 -662); 14.2% (11.6% -16.7%)
Tacoma, WA 49 (40-58); 4.7% (3.8%-5.6%)
--»- Phoenix, AZ 402 (327-476); 8.5% (6.9%-10.1%)
2007 air quality
Recent Air Quality and Current Standard
15/35*
*Based on Krewski et al. (2009), exposure period from 1999 - 2000. The legend contains, for each urban area, the incidence estimate (and 95% CI) and the
estimate of percent of total incidence (and 95% CI) under the current standards.
**The current standards consist of an annual standard of 15 ug/m3 and a daily standard of 35 ug/m3. Combinations of an annual standard (n) and a daily standard
(m) are denoted n/m in this figure.
February 2010
4-9
Draft - Do Not Quote or Cite
-------
1 Figure 4-3 Percent reduction in long-term exposure-related mortality risk (alternative standards relative to the current standards) (Note: inset shows
2 PM2 5 related incidence and percent of total incidence for IHD mortality under the current suite of standards)
60% -
2. B
40% -
X
UJ
o oi 20%
-1 P
0%
-*- Atlanta, GA 277 (227-324); 16.7% (13.7% -19.5%)
Baltimore, MD 374 (307-440); 14.7% (12.1% -17.3%)
-+- Birmingham, AL 165 (135-194); 13.8% (11.3% -16.2%)
-m- Dallas, TX 247 (202-291); 11.4% (9.3%-13.4%)
-m- Detroit, Ml 478 (390-563); 11.5% (9.4%-13.5%)
-- Fresno, CA 98 (80-116); 8.5% (7%-10.1%)
-m- Houston, TX 434 (355-511); 13.6% (11.1%-16%)
Los Angeles, CA 1094 (890-1296); 7.7% (6.3%-9.1%)
NewYork, NY 2222 (1814-2620); 11.8% (9.6%-13.9%)
-- Philadelphia, PA 330 (270-389); 13.2% (10.8% -15.6%)
Phoenix, AZ 402 (327-476); 8.5% (6.9%-10.1%)
Pittsburgh, PA 324 (264-382); 11.8% (9.6%-13.9%)
-m- Salt Lake City, UT 19 (16-23); 3.7% (3%-4.4%)
St. Louis, MO 563 (461-662); 14.2% (11.6% -16.7%)
--Tacoma, WA 49 (40-58); 4.7% (3.8%-5.6%)
15/35**
14/35
13/30
12/25
13/35 12/35
Current and Alternative Standards
, for each urban area, the incidence estimate (and 95% CI) and the
*Based on Krewski et al. (2009), exposure period from 1999 - 2000. The legend contains, lor eacii uroan area, uie incidence esumaie (aiiu y^7o v^i; anu me
estimate of percent of total incidence (and 95% CI) under the current standards.
**The current standards consist of an annual standard of 15 ug/m3 and a daily standard of 35 ug/m3. Combinations of an annual standard (n) and a daily standard
(m) are denoted n/m in this figure.
***The percent reductions for Salt Lake City and Tacoma at the 12/25 standard are 100% and 93%, respectively.
February 2010
4-10
Draft - Do Not Quote or Cite
-------
1
2
Figure 4-4
Percent reduction in short-term exposure-related mortality and morbidity risk (alternative standards and recent conditions relative to the
current standards) (Note: inset shows PM2 5 related incidence and percent of total incidence for CV under the current suite of standards)
4
5
6
7
30%
20%
10%
0%
-10%
-20%
-30%
-40%
-50%
New York
f
Tacoma '
WA
Birmingham,
AL
Pittsburgh
PA
Fresno, Cf
Salt Lake fcity/ UT
Atlanta, GA 32 (-33 - 95); 0.8% (-0.8% - 2.4%)
Baltimore, MD 62 (-4-126); 1.6% (-0.1%-3.2%)
Birmingham, AL -1 (-42-40); 0% (-1.5% - 1.5%)
Dallas, TX 29 (-19-76); 0.8% (-0.5% - 2.2%)
Detroit, Ml 60 (-8 -127); 1 % (-0.1 % - 2.2%)
Fresno, CA 12 (-9-33); 0.7% (-0.5% - 2%)
Houston, TX 46 (-31 -122); 0.9% (-0.6%-2.4%)
Los Angeles, CA -30 (-132-72); -0.2% (-0.7% - 0.4%)
New York, NY 473 (276-668); 2.1% (1.2% - 3%)
Philadelphia, PA 84 (22-145); 2.1% (0.5%-3.6%)
Phoenix, AZ 84 (-4-170); 1.3% (-0.1%-2.7%)
Pittsburgh, PA 43 (-9-93); 1.1% (-0.2%-2.3%)
Salt Lake City, UT 9 (-2 - 20); 0.8% (-0.2% - 1.7%)
St. Louis, MO 106 (24-187); 1.9% (0.4% - 3.3%)
Tacoma, WA 11 (-6 - 27); 0.7% (-0.4% - 1.8%)
2007 air
quality
15/35**
14/35
13/35
12/35
13/30
12/25
Recent Air Quality, Current Standard and Alternative Standards
*Based on Zanobetti and Schwartz (2009). The legend contains, for each urban area, the incidence estimate (and 95% CI) and the estimate of percent of total
incidence (and 95% CI) under the current standards.
**The current standards consist of an annual standard of 15 ug/m3 and a daily standard of 35 ug/m3. Combinations of an annual standard (n) and a daily standard
(m) are denoted n/m in this figure.
*** The percent reductions from 2007 air quality to the current standard for Salt Lake City and Fresno are -58% and -81%, respectively.
February 2010
4-11
Draft - Do Not Quote or Cite
-------
1
2
Figure 4-5
Percent reduction in short-term exposure-related mortality and morbidity risk (recent conditions relative to the current standards) (Note:
inset shows PM25 related incidence and percent of total incidence for CV under the current suite of standards)
-10%
1 -20%
TJ
§ - -30%
+5 ^
J2 £
o " -40%
D g
O £
ul .1 -50%
£ «
IT ^ -60%
o "£
co o
« -70%
-80%
-90%
Houston, TX
Philadelphia,?/
Atlanta, GA
New York.
Fresno, CA
Pittsburgh, B,
Salt Lake City, UT/
X
Atlanta, GA 32 (-33-95); 0.8% (-0.8%-2.4%)
Baltimore, MD 62 (-4-126); 1.6% (-0.1%-3.2%)
* Birmingham, AL -1 (-42-40); 0% (-1.5%-1.5%)
Fresno, CA 12 (-9-33); 0.7% (-0.5%-2%)
---<--- Houston, TX 46 (-31 -122); 0.9% (-0.6%-2.4%)
Los Angeles, CA -30 (-132-72); -0.2% (-0.7%-0.4%)
- New York, NY 473 (276-668); 2.1% (1.2%-3%)
Philadelphia, PA 84 (22-145); 2.1% (0.5%-3.6%)
Phoenix, AZ 84 (-4-170); 1.3% (-0.1%-2.7%)
^- Pittsburgh, PA 43 (-9-93); 1.1% (-0.2%-2.3%)
-m- Salt Lake City, UT 9 (-2-20); 0.8% (-0.2%-1.7%)
-m- St. Louis, MO 106 (24-187); 1.9% (0.4%-3.3%)
-- Dallas, TX 29 (-19-76); 0.8% (-0.5%-2.2%)
-m- Detroit, Ml 60 (-8-127); 1% (-0.1%-2.2%)
-- Tacoma.WA 11 (-6-27); 0.7% (-0.4%-1.8%)
4
5
6
7
2007 air quality 15/35**
Recent Air Quality and Current Standard
*Based on Zanobetti and Schwartz (2009). The legend contains, for each urban area, the incidence estimate (and 95% CI) and the estimate of percent of total
incidence (and 95% CI) under the current standards.
**The current standards consist of an annual standard of 15 ug/m3 and a daily standard of 35 ug/m3. Combinations of an annual standard (n) and a daily standard
(m) are denoted n/m in this figure.
February 2010
4-12
Draft - Do Not Quote or Cite
-------
1 Figure 4-6 Percent reduction in short-term exposure-related mortality and morbidity risk (alternative standards relative to the current standards)
2 (Note: inset shows PM25 related incidence and percent of total incidence for CV under the current suite of standards)
35%
30%
IS
ra a;
2=20%
£ £
3 I
& =
x §
w £'\5%
£ =
5 §
h- «
t *
O **
5 110%
Atlanta, GA 32 (-33-95); 0.8% (-0.8%-2.4%)
Baltimore, MD 62 (-4-126); 1.6% (-0.1%-3.2%)
Birmingham, AL -1 (-42 - 40); 0% (-1.5% -1.5%)
Dallas, TX 29 (-19-76); 0.8% (-0.5%-2.2%)
Detroit, Ml 60 (-8-127); 1% (-0.1%-2.2%)
Fresno, CA 12 (-9-33); 0.7% (-0.5%-2%)
Houston, TX 46 (-31 -122); 0.9% (-0.6%-2.4%)
Los Angeles, CA -30 (-132-72); -0.2% (-0.7%-0.4%)
NewYork.NY 473 (276-668); 2.1% (1.2%-3%)
Philadelphia, PA 84 (22-145); 2.1% (0.5%-3.6%)
Phoenix, AZ 84 (-4-170); 1.3% (-0.1%-2.7%)
Pittsburgh, PA 43 (-9-93); 1.1% (-0.2%-2.3%)
Salt Lake City, UT 9 (-2-20); 0.8% (-0.2%-1.7%)
St. Louis, MO 106 (24-187); 1.9% (0.4%-3.3%)
Tacoma.WA 11 (-6-27); 0.7% (-0.4%-1.8%)
£
15/35*
14/35
13/35
12/35
13/30
12/25
3 Recent Air Quality, Current Standard and Alternative Standards
4 *Based on Zanobetti and Schwartz (2009). The legend contains, for each urban area, the incidence estimate (and 95% CI) and the estimate of percent of total
5 incidence (and 95% CI) under the current standards.
6 **The current standards consist of an annual standard of 15 ug/m3 and a daily standard of 35 ug/m3. Combinations of an annual standard (n) and a daily standard
7 (m) are denoted n/m in this figure.
February 2010
4-13
Draft - Do Not Quote or Cite
-------
1 As noted above, the risk assessment includes risk estimates for a range of short-term and
2 long-term exposure-related health effect endpoints. To focus the discussion of these risk
3 estimates, we have selected a subset of the health endpoints as examples to help illustrate
4 patterns in the risk estimates that might be of interest from a policy standpoint. Specifically, we
5 have focused on those endpoints that the ISA identifies as having the greatest support in the
6 literature (i.e., endpoints related to cardiovascular effects, including both mortality and
7 morbidity). The subset of health effect endpoints selected as illustrative examples for this
8 overview include: IHD-related mortality (for long-term exposure) and CV-related mortality and
9 HA (for short-term exposure). While the discussion does focus on these cardiovascular-related
10 endpoints, we do address other endpoints modeled in the risk assessment to a limited extent. The
11 full set of risk estimates generated is presented in the detailed tables in Appendix E.
12 For a subset of the urban case studies (e.g., Dallas and Phoenix), incremental reductions
13 across alternative standards are initially very low (or even zero) reflecting the fact that recent
14 ambient PM2.5 concentrations for these study areas are well below the current annual standard
15 levels. For these study areas, meaningful reductions in risk may not be seen until relatively
16 lower alternative standards are assessed (and results in the percent reduction from the current set
17 of standards tables and figures may be zero for several of the less stringent, alternative sets of
18 standards). The pattern of risk reductions across alternative standard levels for a given urban
19 study area is an important factor that is discussed in the integrative discussion in Chapter 6. To
20 set up that later discussion, in summarizing risk estimates below, we provide observations
21 regarding trends in risk estimates across alternative suites of standards (for a given urban study
22 area).
23 For a number of the urban study areas, confidence intervals (and in some instances, point
24 estimates) for short-term mortality and morbidity incidence and related risk metrics include
25 values that fall below zero. Population incidence estimates with negative lower-confidence
26 bounds (or point estimates) do not imply that additional exposure to PM2.5 has a beneficial effect,
27 but only that the estimated PM2.5 effect estimate in the C-R function was not statistically
28 significantly different from zero. In the case of short-term exposure mortality, where study area-
29 specific effects estimates were used (see section 3.4), several of the urban locations have non-
30 statistically significant effects estimates; these result in incidence estimates with non-positive
31 lower bounds and/or best estimates (e.g., Birmingham, Detroit, and Los Angeles for non-
32 accidental mortality). In the case of short-term morbidity (e.g., HAs), where regional effects
33 estimates were used, one of the regional coefficients (for the southeast) is not statistically
34 significant, producing incidence estimates including negative values in the confidence interval
35 for urban study areas falling within that region (e.g., Atlanta, Dallas, and Houston, for CV-
36 related HAs). Lack of statistical significance could mean that there is no relationship between
February 2010 4-14 Draft - Draft Do Not Quote or Cite
-------
1 PM2.5 and the health endpoint or it could mean that there was not sufficient statistical power to
2 detect a relationship that actually exists. In the case of PM2.5 and both short-term exposure
3 mortality and morbidity, recognizing that the ISA has concluded that there is either a causal or
4 likely causal relationship between short-term PM2.5 exposure and these health effects (see section
5 3.3.1), we believe it is reasonable to assume that instances where effects estimates are not-
6 statistically significant are likely to reflect insufficient sample size, rather than the absence of an
7 actual association. We note, however, that (as discussed in section 3.6.3) many factors can
8 potentially result in variations in the magnitude of effect estimates. In addition to sample size,
9 these include: source and compositional differences for PM2.5, exposure error associated with the
10 use of ambient monitors as a surrogate for actual exposure, and differences in population
11 susceptibility and vulnerability.
12 An important theme in discussing risk associated with both current and alternative
13 standard levels is the linkage between the nature and magnitude of risk reductions seen for a
14 particular study area (for a particular suite of 24-hour and annual standards) and the specific mix
15 of 24-hour and annual design values associated with that study area. Because design values
16 determine the degree to which the PM2.5 monitors in a study area are adjusted in simulating
17 attainment of both current and alternative standard levels, they play a central role in determining
18 the degree of risk reduction associated with a particular suite of standard levels. Given the
19 importance of design values in determining risk reduction under both current and alternative
20 standard levels, we have examined patterns in design values (specifically the relationship
21 between 24-hour and annual design values) across the 15 urban study areas, as a means for
22 enhancing our interpretation of patterns in risk reductions for the standard levels modeled. In
23 addition, we have contrasted the patterns of design values for the 15 urban study areas with
24 patterns of design values for the broader set of urban areas in the U.S.; this supporting efforts to
25 place risk estimates for the urban study areas in a broader national context. This exploration of
26 design values is discussed in section 4.5.
27 An additional factor to consider in examining patterns in risk estimates is the overall
28 spread in PM2.5 measurements across monitors at a particular urban study area, including
29 distributions of both 24-hour distributions and annual averages. This factor works in concert
30 with the patterns in design values mentioned earlier in determining the degree of risk reduction
31 associated with a particular suite of standard levels. In addition, the spread in monitor values for
32 a particular urban study area can also determine the degree to which alternative rollback methods
33 (proportional, hybrid and peak shaving) produce differences in risk estimates for a given study
34 area. Consequently, in concert with examining patterns in design values (see above) we have
35 also explored patterns in PM2.5 monitoring data for the 15 urban study areas in an effort to better
February 2010 4-15 Draft - Draft Do Not Quote or Cite
-------
1 understand how application of different rollback methods results in differing impacts on core risk
2 estimates. This topic is discussed in section 4.5.
3 The remainder of this section is organized as follows. Core modeling results for the
4 recent conditions air quality scenario are presented in section 4.1. Core modeling results for just
5 meeting the current NAAQS and just meeting alternative NAAQS are presented in section 4.2.
6 The results of the sensitivity analysis (including single-factor and multi-factor results) are
7 presented in section 4.3. The results of a representativeness analysis involving comparison of
8 counties associated with the 15 urban study area locations against the national distribution of
9 counties with regard to a set of PM-risk related attributes are presented in section 4.4. Section
10 4.5 discusses the consideration of design values in interpreting risk estimates generated for the
11 15 urban study areas and helping to place them in a broader national context (section 5.4.1), as
12 well as consideration for the patterns in ambient PM2.5 data within study areas as a factor
13 influencing patterns of risk estimates (section 4.5.2). Chapter 6 provides an integrative
14 discussion of the results of the core risk assessment for the 15 urban study areas informed by
15 consideration of: (a) the single- and multi-factors sensitivity analysis, (b) the qualitative analysis
16 of sources of variability and uncertainty, (c) the representativeness analysis (d) the national-scale
17 mortality analysis (presented in chapter 5), and (e) the role of design values (and patterns in
18 ambient PM2.5 monitoring data) in influencing overall patterns of risk estimates across alternative
19 suites of standards.
20 4.1 ASSESSMENT OF HEALTH RISK ASSOCIATED WITH RECENT CONDITIONS
21 (CORE ANALYSIS)
22 This section discusses core risk estimates generated for the recent conditions air quality
23 scenario, focusing on the 2007 simulation year. Specifically, it provides a set of key observations
24 regarding core risk estimates generated for the recent conditions air quality scenario. Note, that
25 while the focus of this section is on identifying key risk-related observations potentially relevant
26 to the current review of the PM NAAQS, additional review of the risk estimates provided in
27 Appendix E is likely to result in additional observations that might be relevant to the PM
28 NAQQS review (EPA staff will continue to review those results as they work on completing the
29 summary of the RA presented in the first draft PA).
30 In discussing results for the recent conditions air quality scenario, we have focused on
31 absolute risk (either above PRB or LML, depending on the health effect endpoint). This reflects
32 the fact that this air quality scenario represents recent conditions within the urban study areas and
33 therefore, does not lend itself to an incremental assessment. The section is organized by health
34 endpoint category, with results discussed in the following order: long-term exposure mortality,
February 2010 4-16 Draft - Draft Do Not Quote or Cite
-------
1 short-term exposure mortality and short-term exposure morbidity.45 In summarizing estimates
2 for each endpoint category, we fist focus on the central-tendency risk estimates (these are what is
3 discussed in each of the bullets focusing on a particular endpoint category). A discussion of the
4 broader risk range reflecting consideration for 95th% confidence interval risk estimates is
5 presented as a separate bullet towards the end of the discussion. Key observations include:
6 Long-term exposure-related mortality: Total incidence of PM2.5-related all-cause mortality
7 ranges from 50-60 (Salt Lake City) to 2,380-3,000 (New York) (Appendix E, Table E-21 and
8 E-30), with this range reflecting not only differences in baseline incidence across urban study
9 areas, but also the size of study populations which vary considerably across the study areas.
10 The percent of total incidence of IHD-related mortality attributable to PM2.5 ranges from 6.3-
11 8.0% (Tacoma) to 17.7-22.2% (Fresno) (Appendix E, Table E-24 and E-33). Total PM2.5-
12 attriutable incidence for all-cause mortality and cardiopulmonary mortality is larger than IHD
13 (for a given study area under recent conditions), while total PM2.5-attributable incidence for
14 lung-cancer mortality is lower than for IHD. However, the percent of total incidence
15 attributable to PM2.5 exposure is larger for IHD-related mortality than for any of the other
16 mortality categories modeled (Appendix E, Tables E-24 and E-33).
17 Short-term exposure-related mortality: Total incidence of PM2.5-related mortality for
18 short-term exposure (for all categories modeled) is substantially smaller than estimates for
19 long-term exposure-related mortality. Estimates for CV mortality for short-term exposure
20 ranges from 14 (Salt Lake City) to 570 (New York) (Appendix E, Table E-84). The percent
21 of total non-accidental mortality attributable to PM2 5 ranges from 0.9% (Tacoma) to 2.5%
22 (New York). (Appendix E, Table E-87). Percent of total incidence attributable to PM2.5
23 exposure is generally lower for total non-accidental mortality (compared with CV), ranging
24 from 0.2% (Los Angeles) to 1.8% (Baltimore) (Appendix E, Table E-78). Estimates for
25 respiratory mortality are usually higher than for CV mortality, ranging from 0.9% (Dallas) to
26 2.8% (Fresno and New York) (Appendix E, Table E-96). Of the 15 urban study areas
27 modeled for CV mortality, 12 locations had negative lower bound estimates of incidence
28 (and two of these head negative point estimates), reflecting use of non-statistically significant
29 effects estimates (see section 4.0 for additional discussion). The number of study areas
30 modeled with non-statistically significant effects estimates was lower for the other two short-
31 term exposure-related mortality endpoints.
32 Short-term exposure-related morbidity (hospital admissions for respiratory and
33 cardiovascular illness): Total incidence of PM2.5-related cardiovascular HA range from 15
34 (Salt Lake City) to 910 (New York City) and are significantly larger than estimates of
35 respiratory HA attributable to PM2 5 exposure (Appendix E, Tables E102 and E-l 11).
36 Similarly, the percent of total cardiovascular HA attributable to PM2.5 is larger than estimates
37 for respiratory HA and ranges from 0.28% (Dallas) to 1.6% (Pittsburgh) (Appendix E, Table
38 E-105). In this case, the pattern of risk across urban study areas reflects both differences in
45 Note, that as discussed earlier, for long-term exposure-related mortality, two risk estimates are provided for each
urban study area, reflecting application of the two C-R functions used in modeling each mortality endpoint in the
core analysis - i.e., C-R function derived using 1979-1983 PM25 monitoring data and the C-R function derived using
1999-2000 data, with the latter function having the larger effect estimates and therefore, producing higher risk
estimates.
February 2010 4-17 Draft - Draft Do Not Quote or Cite
-------
1 underlying baseline incidence for these endpoints as well as the use of regionally-
2 differentiated effect estimates obtained from Bell et al., 2008 (see Appendix C, Table C-l).
3 Of the 15 urban study areas modeled for cardiovascular-related HAs, five locations had
4 negative lower bound estimates of incidence, reflecting use of non-statistically significant
5 effects estimates (see section 4.0 for additional discussion).
6 Patterns of recent conditions risk across the three simulation years: A comparison of
7 IHD mortality incidence estimates (based on the C-R function derived using 1979-1982
8 monitoring data) across the three years (see Appendix E, Tables E-22 through E-24) shows
9 that, while 2007 does produce incidence estimates that fall between those estimated for 2005
10 and 2006 for some urban areas (e.g., Tacoma, St. Louis, LA), results for 2007 can be the
11 highest of the three years (e.g., Fresno) or the lowest (e.g., Baltimore) for some locations.
12 Generally, results for the same urban study area across the three years are fairly similar
13 (results for Birmingham vary by less than 7% across the years), although they can vary by as
14 much as 30% or more in some locations (see results for Tacoma in 2005 and 2006). All of
15 this temporal variation results from year-to-year variation in the annual average PM2 5 levels
16 for the study areas (see Appendix A). This is because other candidate input parameters,
17 which could also involve temporal variability (e.g., demographics and baseline incidence
18 rates) were not modeled with year-specific values, but rather using one representative year
19 (see section 3.4.1.3 and 3.5 for demographics and baseline health effects incidence rates,
20 respectively). In terms of short-term exposure-related morbidity and mortality endpoints, the
21 pattern is similar to that described above for long-term mortality, with risk estimates for 2007
22 generally falling between those generated for 2005 and 2006 (in terms of magnitude),
23 although the magnitude of variations across the three simulation years for a given health
24 endpoint/case study combination was notably lower for the short-term exposure-related
25 endpoints than for the long-term endpoints. For example, with CV mortality, one of the
26 urban study area with the greatest variation across the three years (New York) had a 15%
27 difference in PM2.5 -related risk across the three years (see Appendix E, Tables E-82 through
28 E-84). This compares with a spread of 30% for some of the urban study areas modeled for
29 long-term exposure-related IFID mortality - see above. As with the long-term mortality risk
30 metrics, all of this temporal variation results from year-to-year variation in the daily PM2.5
31 levels for the study areas (see Appendix A), given that other candidate input parameters,
32 which could have temporal variability (e.g., demographics and baseline incidence rates) were
33 not modeled with year-specific values, but rather using one representative year.
34 Consideration for the 95th percentile confidence interval risk estimates in assessing
35 uncertainty related to the statistic fit of effect estimates: As noted above, all of the risk
36 metrics generated for this analysis include 95th percentiles, reflecting uncertainty in the
37 statistical fit of the underlying effect estimates in the C-R functions. These results suggest
38 that this source of uncertainty can be notable. In the case of recent conditions risk estimates,
39 for long-term mortality, while the central tendency risk estimate for all-cause (long term
40 exposure-related) mortality incidence in New York range from 2,380-3,000, the 95th
41 percentile confidence interval for this estimates is 1,960 to 3,500 (Appendix E, Table E-21
42 and E-30). In this case, this source of uncertainty results in estimates that are -18% lower to
43 -17% higher than the central tendency estimate range. Using the criteria we applied in
44 assessing the results of the sensitivity analysis, these would translate as having a "small"
45 impact on the core risk estimate (see Section 4.3.1). The impact of statistical fit uncertainty
February 2010 4-18 Draft - Draft Do Not Quote or Cite
-------
1 on the IHD-related long-term exposure-related mortality results (see Appendix E, Tables E-
2 24 and E-33) are similar in magnitude to those seen for all-cause mortality and also results in
3 a classification of this uncertainty having a "small" impact on core risk estimates. For short-
4 term exposure-related mortality and morbidity, the impact of statistical fit (as reflected in the
5 95th percentile CI risk estimate ranges) is greater than for long-term mortality. For example
6 with CV-related mortality, the central tendency estimate for New York is 570 cases, while
7 the 95th percentile CI is 332 to 902 (i.e., -40% lower and -40% higher than the core central-
8 tendency estimates). This translates into a "moderate" impact by this source of uncertainty
9 on core risk estimates using the classification scheme developed for the sensitivity analysis.
10 This suggests that uncertainty related to the statistical fig of effect estimates used in risk
11 characterization has twice as greater an impact on short-term mortality as long-term mortality
12 risk estimates.
13 4.2 ASSESSMENT OF HEALTH RISK ASSOCIATED WITH JUST MEETING THE
14 CURRENT AND ALTERNATIVE SUITES OF STANDARDS (CORE ANALYSIS)
15 This section discusses core risk estimates generated for just meeting the current suite of
16 standards and alternative suites of standards, focusing on the 2007 simulation year (although
17 general trends in observations across the three simulated years are discussed to a limited extent).
18 In discussing risk estimates for the current and alternative suites of standards, we include
19 discussion of risk metric which characterize both incremental reductions in risk (across standard
20 levels) as well as absolute risk for a particular standard level. In presenting these two categories
21 of risk metric, we recognize that there is greater uncertainty in estimates of absolute risk relative
22 to estimates of incremental risk. This reflects the fact that we have greater confidence in the
23 ability of the risk models to differentiate risk between sets of standards, since this requires the
24 models to estimate risk for ambient air PM2 5 levels likely near or within the range of ambient air
25 quality data used in the underlying epidemiology studies. By contrast, estimates of absolute risk
26 (for a given air quality scenario) require the models to perform at the lower boundary of ambient
27 air PM2.5 levels reflected in the studies (i.e., down to the LML reflected in the long-term
28 exposure mortality epidemiology studies or down to PRB levels in the short-term exposure
29 morbidity and mortality studies). There is greater overall uncertainty in risk estimates generated
30 based on the contribution to risk of exposures at these lower ambient air PM2.5 levels. While
31 there is greater uncertainty associated with estimates of absolute risk, these estimates are of
32 potential use in informing consideration of the magnitude of risk (and therefore public health
33 impact) for a particular standard level. The overall level of confidence associated with different
34 risk metrics (and implications for informing their use in the context of the PM NAAQS review)
35 is discussed in Chapter 6.
36 This section discusses risk estimates generated for the current standard level first,
37 followed by discussion of risk estimates associated the set of alternative standard levels assessed.
38 Each of these discussions is further organized by health endpoint category, with results discussed
February 2010 4-19 Draft - Draft Do Not Quote or Cite
-------
1 in the following order: long-term exposure mortality, short-term exposure mortality and short-
2 term exposure morbidity. Observations presented in the previous section regarding the statistical
3 significance of effects estimates used in generating risk estimates and their implications for
4 interpretation of those risk estimates also hold for estimates presented in this section.
5 Consequently, observations regarding risk results with confidence intervals including negative
6 estimates are not presented here and the reader is referred back to the earlier discussion in section
7 4.1.
8 We note that the lower magnitude of risk reductions (in terms of percent change in PM2.5-
9 attributable risk) generally seen for short-term exposure-related endpoints relative to long-term
10 exposure-related endpoints primarily reflects the fact that PM2.5-attributable risk is modeled
11 down to PRB for short-term, but only down to LML for long-term. This means that an
12 incremental change (reduction) in long-term risk will be a larger fraction of overall risk
13 compared with short-term risk and hence, the magnitude of risk reductions for long-term
14 exposure-related risk is notably larger compared with short-term risk
15 An important factor to consider in interpreting the risk estimates for both the current set
16 of standards and sets of alternative standards is whether the annual or 24-hour standard for a
17 given pairing of standards is controlling for a particular area.46 This factor can have a significant
18 impact on the pattern of risk reductions predicted for a given location under the simulation of just
19 meeting a specific set of standards. In addition, the approach used to simulate ambient PM2 5
20 levels under current and alternative standard levels (i.e., use of proportional, hybrid, or peak
21 shaving) can significantly impact the magnitude risk reduction seen across standard levels
22 (particularly the degree to which a particular standard produces notable reductions in long-term
23 exposure-related mortality).47 The potential for different rollback strategies (reflecting
24 potentially different combinations of local and/or regional controls) to impact patterns of risk
25 reduction is not discussed here, but rather reserved for discussion as part of the sensitivity
26 analysis (section 4.3) and the integrative chapter (chapter 6).
27 An overview of which urban study areas are predicted to have risk reductions under the
28 current and alternative suites of standards included in the risk assessment is presented below
46 For a given pairing of standard levels (e.g., 13/35), the controlling standard can be identified by comparing these
levels to the design values for a given study area (see section 4.5.1). The controlling standard is the standard (annual
or 24 hr) that requires the greatest percent reduction in the matching design value to meet that standard.
47 Approaches such as hybrid rollback or peak-shaving which simulate more localized control strategies have the
potential to reduce PM2 5 levels at monitors exceeding the daily standard, while leaving other monitors (which may
have elevated annual-average PM2 5 levels) relatively or totally unadjusted. This can result in the 24-hour standard
not providing coverage for the annual standard, even when the 24-hour standard is controlling (i.e., additional
reduction focused on monitors with high annual design values may be required to attain the annual) - see discussion
in Section 4.3 and Chapter 6.
February 2010 4-20 Draft - Draft Do Not Quote or Cite
-------
1 (Appendix E contains tables presenting the full set of detailed core risk estimates generated for
2 the current and alternative suites of standards).
3 4.2.1 Core Risk Estimates for Just Meeting the Current Suite of Standards
4 This section summarizes risk estimates generated for the 15 urban study areas based on
5 simulating just meeting the current suite of standards (including the magnitude of risk reductions
6 relative to recent conditions, where applicable).
7 Long-term exposure-related mortality: Total incidence of PM2.5-related IHD mortality
8 ranges from 15-20 (Salt Lake City) to 1,760-2,220 (New York) (Table 4-1). The percent of
9 total incidence of IHD mortality attributable to PM2.5 ranges from 3.7-4.7% (Tacoma) to
10 13.2-16.7% (Atlanta) (Table 4-2). These levels of IHD mortality risk attributable to PM2.5
11 exposure reflect reductions in risk relative to recent condition ranging from 8.7% (Houston)
12 to 68.6% (Salt Lake City). Two of the urban study areas (Dallas and Phoenix) do not exhibit
13 reductions in risk in simulating just meeting the current suite of standards since these two
14 locations meet the current suite of standards based on recent air quality data. As referenced
15 above for the recent conditions scenario, total PM2.5-attriutable incidence for all-cause
16 mortality and cardiopulmonary mortality is larger than IHD (for a given study area under
17 recent conditions), while total PM2.5-attributable incidence for lung-cancer mortality is lower
18 than for IHD. However, the percent of total incidence attributable to PM2.5 exposure is larger
19 for IHD-related mortality than for any of the other mortality categories modeled (Appendix
20 E, Tables E-24 and E-33).
21 Short-term exposure-related mortality: As with the recent conditions analysis, total
22 incidence of PM2.s-related mortality for short-term exposure is substantially smaller than
23 estimates for long-term exposure-related mortality. Estimates for CV mortality for short-
24 term exposure ranges from 9 (Salt Lake City) to 470 (New York) (Table 4-1). The percent of
25 CV mortality attributable to PM2.5 ranges from 0.7% (Fresno) to 2.1% (Philadelphia and New
26 York). (Table 4-2). The level of risk reduction (comparing risk under the current standard
27 with risk under recent conditions) is generally lower for short-term exposure-related CV
28 mortality compared with long-term exposure-related all-cause mortality and ranges from
29 5.5% (Baltimore) to 36.9% (Los Angeles). As mentioned for long-term exposure-related
30 risk, both Phoenix and Dallas did not exhibit any risk reduction since these two locations
31 meet the current suite of standards based on recent air quality data. Percent of total incidence
32 attributable to PM2.5 exposure is generally lower for total non-accidental mortality (compared
33 with CV), ranging from 0.1% (Los Angeles) to 1.7% (Baltimore) (Appendix E, Table E-78).
34 Estimates for respiratory mortality are usually higher than for CV, ranging from 0.9%
35 (Dallas) to 2.6% (Baltimore) (Appendix E, Table E-96). As noted above, of the 15 urban
36 study areas modeled for CV mortality, 12 locations had negative lower bound estimates of
37 incidence (and two of these had negative point estimates), reflecting use of non-statistically
38 significant effects estimates (see section 4.0 for additional discussion).
39 Short-term exposure-related morbidity (hospital admissions for respiratory and
40 cardiovascular illness): Total incidence of PM2.5-related cardiovascular HA range from 9
41 (Salt Lake City) to 750 (New York City) and are significantly larger than estimates of
42 respiratory HA attributable to PM2 5 exposure (Appendix E, Tables E102 and E-l 11).
43 Similarly, the percent of total cardiovascular HA attributable to PM2 5 is larger than estimates
February 2010 4-21 Draft - Draft Do Not Quote or Cite
-------
1 for respiratory HA and ranges from 0.28% (Dallas) to 1.33% (Baltimore). As noted above,
2 the pattern of risk across urban study areas reflects both differences in underlying baseline
3 incidence for these endpoints as well as the use of regionally-differentiated effect estimates
4 obtained from Bell et al., 2008 (see Appendix C, Table C-l). The level of risk reduction
5 (comparing risk under the current standard with risk under recent conditions) for both
6 respiratory and cardiovascular hospital admissions ranges from 5.5% (Baltimore) to 44.8%
7 (Fresno), again with Phoenix and Dallas not exhibiting any risk reduction since these two
8 locations meet the current suite of standards based on recent air quality data. As noted above,
9 of the 15 urban study areas modeled for cardiovascular-related HAs, five locations had
10 negative lower bound estimates of incidence, reflecting use of non-statistically significant
11 effects estimates (see section 4.0 for additional discussion).
12 Patterns of recent conditions risk across the three simulation years: Observations made
13 earlier regarding patterns of risk across the three simulation years for the recent conditions
14 simulations generally hold for the current standard level analysis. In other words, (a) 2007
15 generally represents risks in between the other two years in terms of magnitude, (b) there are
16 exceptions where 2007 had the highest risks and lowest risk (depending on study area and
17 endpoint), and (c) generally, long-term exposure-related mortality endpoints showed greater
18 cross year variation then the short-term exposure-related endpoints (with the magnitude of
19 this variation similar to what is reported above for the recent conditions simulation).
20 Consideration for the 95th percentile confidence interval risk estimates in assessing
21 uncertainty related to the statistic fit of effect estimates: Uncertainty related to the
22 statistical fit of effect estimates has the same magnitude of effect in modeling risk under the
23 current standard as it did under recent conditions (i.e., an impact of about +1-18% on the core
24 risk estimates, translating into a "small" impact based on classification used in the sensitivity
25 analysis) (see (Appendix E, Table E-21 and E-30 for risk estimates used to reach this
26 conclusion). The impact of this source of uncertainty on short-term exposure-related CV
27 morality was similar (although slightly larger) compared with what was seen with risk
28 estimates generated for the recent conditions air quality scenarios (i.e., 48% lower to 42%
29 higher than the core risk estimate - see estimates in Appendix E, Table E-84). This results in
30 a classification of "moderate" for this source of uncertainty and its impact on short-term
31 exposure-related mortality, based on the classification scheme developed for the sensitivity
32 analysis.
33 4.2.2 Core Risk Estimates for Just Meeting Alternative Suites of Standards
34 This section summarizes risk estimates generated for the 15 urban study areas when
35 ambient PM2.5 levels under the alternative standard levels are simulated. As noted in section 4.2,
36 this discussion focuses on the magnitude of incremental risk reductions for individual standard
37 levels relative to the current standard, given that overall confidence in incremental risk metrics is
38 considered higher than estimates of absolute risk for a given standard level. Note, however, that
39 we do provide limited discussion of absolute risk levels attributable to PM2.5 exposure for
40 alternative standard levels, with the provision that these be interpreted in the context of their
41 greater levels of uncertainty. In discussing risk estimates for the alternative standard levels, we
42 focus first on patterns of risk reduction across alternative annual levels (i.e., 14/35, 13/35 and
February 2010 4-22 Draft - Draft Do Not Quote or Cite
-------
1 12/35) and then discuss patterns across a combination of alternative 24 hour and annual
2 standards (i.e., 13/30 and 12/25).
3 As noted in Section 4.1, although reductions in absolute incidence will differ for health
4 effect endpoints associated with a particular averaging period across alternative suites of
5 standards for a given urban study area, the patterns of reduction in terms of percent change in
6 PMg_j-attributable risk are very similar for a given urban study area across health endpoints. This
7 reflects the fact that the C-R functions used in the core analysis are close to linear across the
8 range of ambient PM2.5 levels considered in this analysis, and consequently the main factor
9 producing percent reductions in risk across alternative standards is the reduction in the air quality
10 metric for a given study area (i.e., reductions in annual average PM2.5 concentrations or
11 reductions in the distribution of 24-hour estimates for a year). Consequently, in discussing
12 incremental risk reduction in terms of percent change relative to the current suite of standards,
13 we speak more generally in terms of the category of annual-average risk or 2-4hour average
14 risk, with the assumption that these observations hold for individual health effects endpoints
15 assessed for each averaging period. These observations regarding patterns of percent risk
16 reduction for the two averaging periods are reflected in Figures 4-1 through 4-6 which are
17 referenced in the discussion below.
18 Alternative annual standard levels (14/35. 13/35. and 12/35) 48
19 Percent reductions in long-term exposure-related mortality: Reductions in all long-term
20 exposure-related mortality categories were more limited under the 14/35 alternative standard,
21 with only 5 of the 15 urban study areas demonstrating notable reductions ranging from 9%
22 (Baltimore) to 12% (Houston and Birmingham) (see Figure 4-3 and Appendix E, Table E-9).
23 Reducing the annual standard level to 13 |ig/m3 (i.e., the 13/35 alternative suite of standards)
24 produced a notable increase in the number of locations (9 of the 15) with risk reductions
25 relative to the current standard ranging from 5% (New York) to 24% (Houston and
26 Birmingham). The lowest annual standard evaluated (12 |ig/m3 as reflected in the 12/35
27 alternative suite of standards) resulted in additional study areas (now 12 of the 15 study
28 areas) experiencing risk reductions with percentage risk reductions now ranging from 11%
29 (Phoenix) to 26% (Houston and Baltimore). Note, that even in the 12/35 case, three of the
30 urban study areas (Tacoma, Fresno and Salt Lake City) did not experience any decreases in
31 risk, although risk reductions were seen for these three study areas when alternative 24-hour
32 standards were considered - see below. The specific pattern of risk reduction (including
33 importantly, the magnitude of risk reduction as well as residual risk associated with a
48 The three alternative annual standards considered in the risk assessment (12, 13 and 14 ug/m3) were each paired
with the current 24-hour standard of 35 ug/m3 for purposes of generating risk estimates. A separate set of
alternative suites of standards (i.e., 13/30 and 12/25) were also considered - see next section below. In discussing
risk estimates associated with the alternative annual standards, each alternative annual standard level was paired
with the current 24-hour standard of 35 ug/m3 in determining which standard level was controlling and,
consequently, whether the alternative annual standard would produce any notable reductions in risk.
February 2010 4-23 Draft - Draft Do Not Quote or Cite
-------
1 particular standard level) reflects whether daily or annual standard levels were controlling -
2 see discussion below regarding patterns of risk reduction.
3 Percent reduction in short-term exposure-related mortality and morbidity: The pattern
4 of reductions in the percent of risk attributable to PM2.5 for mortality and morbidity
5 associated with short-term exposure is similar to that described above for long-term mortality
6 (see Figures 4-4 through 4-6). Specifically, the same five urban study areas (Atlanta,
7 Baltimore, Birmingham, Houston and St. Louis) had notable risk reductions under the full set
8 of alternative annual standards, with the degree of risk reduction for PM2.s-related
9 cardiovascular mortality for the lowest alternative annual standard level (12/35) compared to
10 the current standard level, ranging from 20% (St. Louis) to 23% (Birmingham) (see Figure 4-
11 4 and 4-6 and Appendix E, Table E-90). A number of the other study areas did not exhibit
12 notable risk reductions until the lowest alternative annual standard was considered (i.e.,
13 Detroit, Los Angeles, New York, Philadelphia, Pittsburgh), with the degree of reduction in
14 risk for the lowest alternative suite of standards (12/35) compared with the current standards
15 ranging from 5% (Phoenix) to 16% (Detroit) (see Figure 4-4 and 4-6 and Appendix E, Table
16 E-90). As with long-term exposure-related mortality, a number of additional study areas
17 (Fresno, Salt Lake City, Tacoma) did not exhibit any notable risk reduction under the set of
18 alternative annual standards considered and only experienced risk reductions when the 24-
19 hour standard level was reduced. Because the same air quality metric (annual distributions of
20 24-hour PM2.5 concentrations) is used in generating short-term exposure-related mortality
21 and morbidity endpoints, patterns of risk reduction are similar for both sets of endpoints (see
22 Figures 4-4 through 4-6. Specifically, the same groups of urban study areas experience the
23 same magnitude of risk reductions (in terms of percent changes in PM25-related risk relative
24 to the current standard level) across the alternative standard levels for short-term exposure-
25 related morbidity (HAs). The specific pattern of risk reduction reflects whether daily or
26 annual standard levels are controlling - see discussion below regarding patterns of risk
27 reduction.
28 Pattern of risk reduction linked to design values: The patterns of risk reduction across the
29 15 urban study areas for the set of alternative annual standard levels considered here depends
30 on whether the alternative annual (12, 13 or 14 |ig/m3) or the current 24-hour standard of 35
31 Hg/m3 is controlling. The approach used to simulate just meeting alternative 24-hour
32 standards (i.e., proportional, hybrid, or peak shaving) can have an impact on the magnitude
33 of risk reduction, although it does not influence whether the annual or 24-hour design value
34 was controlling for a given alternative suite of standards (see sensitivity analysis discussion
35 in 4.3 and the integrative discussion in Chapter 6). The pattern in risk reduction seen across
36 the 15 urban study areas (given the set of alternative annual standards considered) can be
37 divided into three categories: (a) all of the alternative annual standard levels are controlling,
38 resulting in notable risk reductions for all of the annual standard levels considered
39 (Birmingham, Atlanta, Houston), (b) alternative annual standards only control at lower levels
40 (i.e., 13/35 and/or 12/35) and consequently notable risk reductions are only seen at the lower
41 or lowest annual standard level(s) considered (Dallas, Los Angeles, New York, Philadelphia,
42 Phoenix, Pittsburgh), and (c) none of the alternative annual standard levels is controlling and
43 therefore there is no estimated risk reduction for the alternative annual standard levels
44 considered (Salt Lake City, Tacoma, Fresno).
February 2010 4-24 Draft - Draft Do Not Quote or Cite
-------
1 Absolute levels of PM2.s-attributable risk under alternative annual standards: As
2 discussed above, we have greater confidence in estimating incremental reductions in risk
3 between the current and alternative suites of standards, then the estimation of absolute
4 incidence under a given suite of standards. Nonetheless, we provide a summary of that risk
5 metric here for long-term and short-term exposure-related mortality and short-term exposure-
6 related morbidity endpoints:
7 o Long-term exposure-related mortality: The four study areas displaying the greatest
8 degree of reduction across the alternative annual standards (Atlanta, Baltimore,
9 Birmingham and Houston) have PM2.5-related IHD mortality estimates (under the
10 lowest alternative annual standard of 12/35) ranging from 85-110 (Birmingham) to
11 220-280 (Houston) (see Appendix E, Table E-21 and E-30). The two urban study
12 areas with the greatest degree of PM2.5-related risk in absolute terms (Los Angeles
13 and New York) do not exhibit significant reductions in risk until the lowest annual
14 standard level of 12/35 is considered, with PM2.5-related IHD mortality estimated at
15 750-950 and 1,420-1,800, respectively under that alternative standard (see Appendix
16 E, Table E-21 and E-30).
17 o Short-term exposure-related mortality: The four study areas displaying the greatest
18 degree of reduction across the alternative annual standards (Atlanta, Baltimore,
19 Birmingham and Houston), have PM2.5-related CV mortality estimates (under the
20 lowest alternative standard of 12/35) ranging from 25 (Atlanta) to 50 (Baltimore) (see
21 Appendix E, Table E-84). We note that Birmingham has an incidence estimate of-1,
22 reflecting application of a non-statistically significant effect estimate in modeling this
23 endpoint (see section 4.1). The urban study area with the greatest degree of PM2 5-
24 related risk in absolute terms (New York) does not exhibit significant reductions in
25 risk until the lowest annual standard level of 12/35 is considered with PM2.5-related
26 CV mortality estimated at 420 under that alternative standard level (see Appendix E,
27 Table E-84).
28 o Short-term exposure-related morbidity: The four study areas displaying the greatest
29 degree of reduction across the alternative annual standard levels (Atlanta, Baltimore,
30 Birmingham and Houston), have PM2.5-related cardiovascular HA (under the lowest
31 alternative standard of 12/35) ranging from 12 (Birmingham) to 170 (Baltimore) (see
32 Appendix E, Table E-102). The two urban study areas with the greatest degree of
33 PM2 5-related risk in absolute terms (Los Angeles and New York) do not exhibit
34 significant reductions in risk until the lowest annual standard level of 12/35 is
35 considered with PM25-related all-cause mortality estimated at 240 and 670,
36 respectively under that alternative standard level (see Appendix E, Table E-102).
37 Patterns of recent conditions risk across the three simulation years: Observations made
38 above regarding patterns of risk across the three simulation years for the recent conditions
39 and current standards simulations generally hold for the alternative standards analysis. In
40 other words, (a) 2007 generally represents risks between the other two years in terms of
41 magnitude, (b) there are exceptions where 2007 had the highest risks and lowest risk
42 (depending on study area and endpoint), and (c) generally, long-term exposure-related
43 mortality endpoints showed greater cross-year variation then the short-term exposure-related
44 endpoints in terms of both absolute PM2 5 risk for a particular alternative suite of standards,
45 as well as incremental risk reductions relative to the current suite of standards.
February 2010 4-25 Draft - Draft Do Not Quote or Cite
-------
1 Consideration of the 95th percentile confidence interval risk estimates in assessing
2 uncertainty related to the statistic fit of effect estimates: Continuing the pattern seen with
3 the current standard level, uncertainty related to the statistical fit of effect estimates has the
4 same magnitude of effect in modeling risk under alternative standards involving reduction of
5 the annual level as it did under recent conditions (i.e., an impact of about +/-18% on the core
6 risk estimates, translating into a "small" impact based on classification used in the sensitivity
7 analysis) (see Appendix E, Table E-21 and E-30 for risk estimates used to reach this
8 conclusion). Similarly, the pattern of impact this source of uncertainty on short-term
9 exposure-related CV morality continues to be similar compared with what was seen for risk
10 estimates generated for the recent conditions air quality scenarios (i.e., 42% lower to
11 42%higher than the core risk estimate - see estimates in Appendix E, Table E-84). This
12 continues to result in a classification of "moderate" for this source of uncertainty based on
13 the classification scheme developed for the sensitivity analysis.
14 Combinations of alternative 24-hour and annual standard levels (13/30, 12/25)
15 Percent reductions in long-term exposure-related mortality: The combination of suites
16 of alternative 2-hour and annual standards produced notable reductions in long-term
17 exposure-related mortality for 14 of the 15 urban study areas, with the lower combination
18 (12/25) producing a notable reduction in risk relative to the first combination of 13/30. The
19 only study area that did not exhibit a reduction in risk under the first combination (13/30)
20 was Dallas, reflecting the fact that its 24-hour and annual design values are below 30 |ig/m3
21 and 13 |ig/m3, respectively (and consequently, the 13/30 did not produce a reduction in
22 ambient air PM2.5, or a resulting reduction in risk). Reductions in long-term exposure-related
23 mortality (across all endpoints) under the 13/30 combination ranged from 14% (Phoenix) to
24 55% (Salt Lake City), while reductions for the 12/25 combination ranged from 12% (Dallas)
25 to -100% (Salt Lake City) (see Figure 4-1 and 4-3 and Appendix E, Table E-27). The
26 reduction for Salt Lake City reflects a very high 24-hour design value which, when reduced
27 to meet the 24-hour standard of 25 |ig/m3 produced a very large reduction in the annual
28 design value (given application of the proportional adjustment to simulate rollback), such
29 that the value was very close to 5.8 |ig/m3 (the LML below which long-term exposure-related
30 mortality is not estimated). The specific pattern of risk reduction reflects whether the 24-hour
31 or annual standard was controlling - see discussion below regarding patterns of risk
32 reduction.
33 Percent reduction in short-term exposure-related mortality and morbidity: The pattern
34 of reductions in the percent of risk attributable to PM2.5 for mortality and morbidity
35 associated with short-term exposure is similar to that described above for long-term mortality
36 in terms of the ordering of sites, however the magnitude of risk reduction (in terms of percent
37 change in PM25-related risk) is lower for short-term exposure-related health endpoints
38 compared with long-term exposure-related mortality (see Figures 4-4 through 4-6).
39 Specifically, 14 of the 15 urban study areas (Dallas being the exception), had notable risk
40 reductions under both the 13/30 and 12/35 alternative suites of standards (Dallas only was
41 estimated to have reductions in risk under the lower 12/25 combination - see Figure 4-4 and
42 4-6 and Appendix E, Table E-108). Reductions in short-term exposure-related mortality and
43 morbidity (across all endpoints) under the 13/30 combination ranged from 6% (Phoenix) to
44 15% (Salt Lake City), while reductions for the 12/25 combination ranged from 7% (Dallas)
45 to 30% (Birmingham).
February 2010 4-26 Draft - Draft Do Not Quote or Cite
-------
1 Pattern of risk reduction linked to design values: As with the set of alternative annual
2 standards discussed in the previous section, the pattern of risk reduction seen for the two
3 combinations of alternative 24-hour and annual standards described here depends on which
4 standard is controlling. In addition, the magnitude of the reduction in risk reflects (a) the
5 magnitude of the difference between the controlling design value and the standard level
6 (which determines the degree of reduction in ambient air PM2.5 levels) and (b) the method
7 used to simulate ambient PM2.5 levels under alternative suites of standards (i.e., proportional,
8 hybrid or peak shaving). For this set of alternative suites of standards, 10 of the 15 study
9 areas had the alternative 24-hour standard controlling under the 13/30 case and that number
10 was increased to 12 out of the 15 study areas with the 12/25 case (Table 3-5). As expected,
11 those study areas with the greatest reduction in risk (in terms of percent reduction compared
12 with the current suite of standards) under the 12/25 case had a controlling 24-hour standard
13 (e-g-, Tacoma, Salt Lake City, Los Angeles and Fresno - see Figure 4-4 and 4-6 and
14 Appendix E, Table E-90).
15 Absolute levels of PM2.s-attributable risk under alternative suites of annual and 24-
16 hour standards: As with the alternative annual standards, below we provide a brief
17 overview of the magnitude of PM2.5-attributable risk (i.e., absolute risk) associated with the
18 two alternative suites of annual and 24-hour standards:
19 o Long-term exposure-related mortality: The four study areas displaying the greatest
20 degree of reduction across these two alternative suites of standards (Tacoma, St.
21 Louis, Los Angeles and Fresno), have PM2.s-related IFID mortality estimates (under
22 the 12/25 case) ranging from 3-4 (Tacoma) to 290-360 (Los Angeles) (see Appendix
23 E, Table E-21 and E-30). The other urban study area with the greatest degree of
24 PM2 5-related risk in absolute terms besides New York (New York) has PM2 5-related
25 all-cause mortality estimated at 820-1,040 under the 12/25 case.
26 o Short-term exposure-related mortality: eleven of the 15 study areas had percent
27 reductions in risk for the 12/25 case (relative to the current standards) of
28 approximately 29% (the other study areas had lower percent reductions). Of the
29 locations with -29% reductions in risk, PM2.5-attributable CV mortality for the 12/25
30 case ranged from 6 (Salt Lake City) to 340 (New York) (see Appendix E, Table E-
31 84). New York City also represents the study area with the greatest residual risk for
32 short-term exposure-related mortality under the 12/25 case.
33 o Short-term exposure-related morbidity Of the 11 urban study areas with -29%
34 reduction in risk (for the 12/25 case relative to the current standards), the incidence of
35 PM2.5-attributable cardiovascular HA emissions ranges from 7 (Salt Lake City) to 530
36 (New York) (see Appendix E, Table E-102). New York City also represents the
37 study area with the greatest residual risk for short-term exposure-related morbidity
38 under the 12/25 case.
39 Consideration for the 95th percentile confidence interval risk estimates in assessing
40 uncertainty related to the statistic fit of effect estimates: As with the alternative standards
41 considering lower annual levels, risk estimates generated for the two standards considering
42 lower annual and 24-hour levels also suggest that uncertainty related to the statistical fit of
43 effect estimates will have a greater impact on short-term exposure-related mortality (+/-
44 -40%) compared with long-term exposure-related mortality (+/- -18%) (see Appendix E,
February 2010 4-27 Draft - Draft Do Not Quote or Cite
-------
1 Tables E-84 and E-21 plus Table E-30, respectively). Again, this results in a classification of
2 this source of uncertainty as having a "lower" impact for long-term exposure-related
3 mortality and a "moderate" impact on short-term exposure-related mortality.
4 4.3 SENSITIVITY ANALYSIS RESULTS
5 As noted in section 3.6.4 and section 4.1, the sensitivity analysis was conducted in order
6 to gain insights into which of the identified sources of uncertainty and variability in the risk
7 assessment model may have significant impacts on risk estimates. A second goal of the
8 sensitivity analysis was to generate an additional set of reasonable risk estimates to supplement
9 the core set of risk estimates to inform staffs characterization of uncertainty and variability
10 associated with those core estimates.
11 The first goal can be achieved by considering the magnitude of the impact of individual
12 modeling elements based on results from the sensitivity analysis and identifying those elements
13 which have the greatest impact on the core risk estimates. Use of the sensitivity analysis results
14 in this context is addressed in section 4.3.1. Use of the results of the sensitivity analysis as an
15 additional set of reasonable risk estimates to augment the core risk estimates in considering the
16 impact of uncertainty and variability in the core risk model is discussed in section 4.3.2.
17 In conducting the sensitivity analysis we modeled 2 of the 15 urban study areas
18 (Philadelphia and Los Angeles - representing east and west coast urban areas, respectively) for
19 most simulations. For some modeling elements (e.g., the hybrid and peak shaving alternative
20 rollback approaches) we included a larger number of urban study areas that were applicable to
21 the topic being assessed. In conducting the sensitivity analysis, we have also focused on long-
22 term exposure mortality and to a lesser extent on short-term exposure mortality and morbidity.
23 Although the sensitivity analysis simulations were completed for all three simulation
24 years (as reported in Appendix F), we have focused on results for 2007 in this presentation for
25 comparability with the core results discussed in sections 4.1 and 4.2.
26 4.3.1 Sensitivity Analysis Results to Identify Potentially Important Sources of Uncertainty
27 and Variability
28 The results of the sensitivity analysis are summarized in Table 4-3 (detailed results tables
29 are presented in Appendix F). In presenting the results of the sensitivity analysis, we have
30 compared the risk estimates for the particular simulation to the core set of risk estimates
31 generated for the same health effect endpoint/urban study area combination. Specifically, we
32 have calculated a percent difference between the sensitivity analysis result and the associated
33 core risk estimate to compare the results of the sensitivity analysis across the different modeling
34 elements that were considered. These percent difference results are emphasized in Table 4-1 and
35 in the discussion presented below.
February 2010 4-28 Draft - Draft Do Not Quote or Cite
-------
1 In discussing the results of the sensitivity analysis, we have developed four descriptive
2 categories, based on the general magnitude of the percent difference estimate generated for a
3 particular modeling element:
4 Modeling elements estimated to have percent differences of 20% or smaller (i.e., they
5 produced risk estimates that differed from the core risk estimates by no more than
6 20%) are classified as having a small contribution to uncertainty in the core risk
7 estimates.
8 Modeling elements estimated to have percent difference estimates in the range of 20 to
9 50% are classified as having a moderate contribution to uncertainty in the core risk
10 estimates.
11 Modeling elements estimated to have percent difference estimates in the range of 50 to
12 100% are classified as having a moderate-large contribution to uncertainty in the core
13 risk estimates.
14 Modeling elements estimated to have percent difference results >100% are classified as
15 having a large contribution to uncertainty in the core risk estimates.
16 The sensitivity analysis based on Moolgavkar's (2003) study in Los Angeles addressing
17 model specifications for both short-term mortality and morbidity (e.g., model selection, lag
18 structure and co-pollutant models) are discussed together as a group. This reflects the fact that
19 the Moolgavkar-based simulations were based on the same underlying dataset and focused on
20 Los Angeles. Furthermore, the discussion of the Moolgavkar-based sensitivity analysis results
21 presented below, as well as the summary of results presented in Table 4-1, focus on the
22 difference in the spread of risk results across the Moolgavkar-based model specifications (for a
23 particular endpoint), rather than the percent difference results based on comparison against the
24 core result that are emphasized with the other sensitivity analyses.49
25 The sensitivity analysis examining the impact of alternative rollback approaches for
26 simulating ambient PM2.5 concentrations in urban study areas under both the current and
27 alternative suites of standards also deserves additional discussion before presenting the results.
28 For the first draft RA, we considered the impact of using a hybrid rollback approach in addition
29 to the proportional rollback approach which has been traditionally used in PM NAAQS risk
49 Comparison of the Moolgavkar-based risk estimates with the core risk estimates consistently produce percent
difference estimates that range to levels well above +100%, resulting in a general conclusion, based on this metric,
that all of the factors considered in the Moolgavkar-based sensitivity analysis are large contributors to uncertainty in
the core risk estimates. However, there is significant uncertainty in assuming that the behavior of the Moolgavkar-
based risk models (reflecting consideration for alternate design elements) would be representative of how models
derived from either of the key short-term studies considered in this risk assessment (Zanobetti and Schwartz., 2009
and Bell et al., 2008) would respond to variations in design. Therefore, while sensitivity analysis results based on
comparing Moolgavkar-based risk estimates against the core risk estimates are included in the detailed sensitivity
analysis results tables presented in Appendix F (see Tables F-31 through F-33), we do not discuss these results here
due to the degree of uncertainty associated with them.
February 2010 4-29 Draft - Draft Do Not Quote or Cite
-------
1 assessments. For this second draft, as discussed in sections 2.6, 3.2.3 and 3.5.4, we have
2 included consideration of a peak shaving rollback approach in addition to the hybrid as non-
3 proportional methods to contrast with proportional rollback.50
4 As discussed in Section 3.2.3, for the second draft risk assessment, we have calculated
5 composite monitor estimates based on proportional rollback and hybrid and/or peak shaving,
6 where appropriate. The composite monitor values are surrogates for long-term exposure-related
7 mortality.51 Therefore, by comparing composite monitor values generated for the same study
8 area/suite of standards (using different rollback methods), we can obtain insights into the
9 potential impact of the rollback method used on long-term exposure-related mortality (see
10 Section 3.5.4 for additional discussion of how the composite monitor values generated using the
11 different rollback methods are used in the sensitivity analysis). These sensitivity analysis results
12 based on consideration for composite monitor values generated using the different rollback
13 methods (which are presented in detail in Appendix F, Tables F-49 and F-50) form the basis for
14 summary information related to rollback presented in Table 4-3. Due to the complexity of the
15 sensitivity analysis conducted examining the issue of rollback, the discussion of results from that
16 particular analysis presented in section 4.3.1.1 is more detailed than for the other factors
17 considered as part of the sensitivity analysis.
18 In discussing the results of the sensitivity analysis, results of the single-factor simulations
19 are presented first (section 4.3.1.1), followed by the results of the multi-factor simulations
20 (section 4.3.1.2). Within these categories, results are further organized by health effect endpoint
21 with results for long-term exposure mortality discussed first and then short-term exposure
22 mortality, followed by short-term exposure morbidity. An overall conclusion regarding which of
23 the factors included in the sensitivity analysis represent potentially significant sources of
24 uncertainty and variability impacting the core risk estimates is presented at the end of each sub-
25 section.
50 The peak shaving approach involves proportional reduction in 24-hour PM2 5 levels only at those urban study
areas where the 24-hour standard is controlling (and only at those specific monitors with design values exceeding
that 24-hour standard level) - see Section 3.2.3 for additional detail.
51 The composite monitor is essentially the mean of the annual averages across the PM25 monitors in a study area. It
is this air quality metric that is used in calculating long-term exposure-related mortality. Given that the same C-R
function is used across all study areas, differences in long-term mortality across study areas (and/or across standard
levels) reflect to a great extent underlying differences in the composite monitor values. Therefore, comparison of
composite monitors (in terms of percent difference for example) can provide insights into potential percent
differences in long-term mortality related to PM2 5 exposure across study areas and/or standard levels (see Section
3.5.4).
February 2010 4-30 Draft - Draft Do Not Quote or Cite
-------
Table 4-3 Overview of Sensitivity Analysis Results
Sensitivity Analysis1
Health Endpoint and Risk
Assessment Location
Summary of Results
(percent difference in risk estimate relative
to the core estimate)
Appendix F
Tables with
Detailed
Results (for
2007)
Single-Factor Sensitivity Analyses (long-term exposure mortality):
Impact of using different model choices:
fixed effects log-linear (the core) vs. random effects
log-linear C-R function
Impact of using different model choices:
fixed effects log-linear (the core) vs. random effects
log-log C-R function
Impact of using different model choices:
Single vs. multi-pollutant models
Impact of estimating risks down to PRB rather than
down to LML (the core)
Impact of using alternative C-R function from
another long-term exposure mortality study
Impact of using alternative hybrid rollback approach
reflecting more localized patterns of ambient PM2 5
reduction (evaluated across current and alternative
standard levels) - based on the composite monitor
analysis described in Section 3.5.4 considering both
hybrid and peak shaving approaches as alternatives
to proportional rollback
All-cause, CPD, IHD
Los Angeles and
Philadelphia
All-cause, CPD, IHD
Los Angeles and
Philadelphia
All-cause
Los Angeles and
Philadelphia
All cause
All 15 urban study areas
All-cause, CPD, lung
cancer
Los Angeles, Philadelphia
Surrogate for long-term
mortality (composite
monitor-based analysis)
All study areas except
Dallas had either hybrid
and/or peak shaving
applied as an alternative
Random effects log-linear C-R model:
all-cause: +23%
IHD: +12%
Random effects log-log C-R model:
All-cause: +123 to +159%
CPD: +50 to +74%
IHD: +80 to +111%
Lung Cancer: +67 to +94%
Model with CO: +45%
Model with NO2: +73%
Model with O3: +45%
Model with SO2: -74%
All-cause: +47 to +273%
All-cause: +119 to +121%
CPD: +29 to +30%
Lung cancer: +29 to +30%
Trend in incremental risk reduction
(alternative standard level compared to
current standard): rollback method did not
appear to have a significant impact on this
metric (those urban study areas with
different trends in reduction did not
demonstrate a consistent pattern related to
Table F-3
Table F-3
F-43
Table F-6
Table F-9
Tables F-49
andF-50
February 2010
4-31
Draft - Do Not Quote or Cite
-------
Sensitivity Analysis1
Health Endpoint and Risk
Assessment Location
rollback method to the
proportional
Summary of Results
(percent difference in risk estimate relative
to the core estimate)
the type of hybrid method used)
Absolute risk for a given standard level:
use of alternative rollback methods did
appear to impact estimation of PM2 5 risk
remaining for a given standard level: <1%
to >+50%
Has implications for degree to which 24-
hour standard levels produce reductions
in annual-average PM2 5 levels (and
consequently on long-term and short-term
exposure-related risk). Results suggest
that use of peek shaving rollback method
can result in smaller degree of reduction
in annual-average values compared with
proportional rollback, (see discussion in
text- section 4.3. 1.1)
Appendix F
Tables with
Detailed
Results (for
2007)
Single-Factor Sensitivity Analyses (short term exposure mortality):
Impact of using season-specific C-R functions (vs.
an annual C-R function)
Impact of using alternative hybrid rollback approach
reflecting a combination of more localized and
regional patterns of ambient PM2 5 reduction (note,
this analysis is based exclusively on the hybrid
rollback - the composite monitor analysis described
Non-accidental mortality,
CV, respiratory
All 15 urban study areas
Non-accidental mortality
Baltimore, Birmingham,
Detroit, Los Angeles,
New York and St. Louis
Non-accidental: -1 16 to +179%
CV: -82 to +500%
Respiratory: -48 to +162%
(Note, overall incidence estimates, particularly
for the locations with higher percent change
estimates, is very low, raising concerns over
the stability of these sensitivity analysis
results)
Results for all seven urban study areas
(across the current and alternative
standard levels) do not exceed +17%,
with most <+10%.
Table F-15
Table F-18
Table F-21
Table F-36
February 2010
4-32
Draft - Do Not Quote or Cite
-------
Sensitivity Analysis1
above pertains only to long-term mortality -related
risk)
Health Endpoint and Risk
Assessment Location
Summary of Results
(percent difference in risk estimate relative
to the core estimate)
Appendix F
Tables with
Detailed
Results (for
2007)
Single-Factor Sensitivity Analyses (short-term morbidity: hospital admissions (HA) and ED visits):
Impact of using season-specific C-R functions (vs.
an annual C-R function)
Impact of using an annual C-R function (applied to
the whole year) vs. a seasonal function for April
through August (applied only to that period) (using a
single pollutant model)
Impact of considering models with different lags
HA (unscheduled), CV
and respiratory
All 15 urban study areas
Asthma ED visits
New York
HA (CV and respiratory
LA and New York
HA (CV): -105 to +9%
HA (respiratory): -54 to +74%
(Note, overall incidence estimates, particularly
for the locations with higher percent change
estimates, is very low, raising concerns over
the stability of these sensitivity analysis
results)
NA (although incidence estimates were
generated for this simulation, "percent
difference from the core" were not generated
since the alternate simulation focused on a
subset of the year).
NA (although incidence estimates were
generated for this simulation, "percent
difference from the core" were not generated
since the lag-differentiated C-R functions used
are not regionally -differentiated, and
therefore, do not allow a focused
consideration of the lag factor alone in
impacting risk estimates)
Table F-24
Table F-27
Table F-30
Table F-48
Single-Factor Sensitivity Analysis (short-term exposure mortality and morbidity in LA based on Moolgavkar, 2003 study model options) (Note, results
presented here reflect spread in risk estimates across Moolgavkar-based model specifications and not percent difference from core risk estimates,
unless so stated - see text)
Impact of model selection (e.g., log-linear GAM
with 30 df; log-linear GAM with 100 df; and log-
linear GLM with 100 df)
Mortality (non-accidental,
CV); HA (CV)
Los Angeles
Non-accidental mortality: +80%
CV mortality: +49
CVHA:+36%
Table F-33
February 2010
4-33
Draft - Do Not Quote or Cite
-------
Sensitivity Analysis1
Impact of lag structure (0-day, 1-day, 2-day, 3 -day,
4-day, 5-day)
Impact of single- vs. multi-pollutant models (PM2 5
with CO)
Health Endpoint and Risk
Assessment Location
Mortality (non-accidental)
Los Angeles
Mortality (CV); HA (CV)
Los Angeles
Summary of Results
(percent difference in risk estimate relative
to the core estimate)
Non-accidental mortality: +55%
CV mortality : + 1 06%
CVHA:+140%
Appendix F
Tables with
Detailed
Results (for
2007)
Table F-33
Table F-33
Multi-Factor Sensitivity Analyses (long-term mortality):
Impact of using a fixed effects log-linear vs. a
random effects log-log model, estimating incidence
down to the lowest measured level (LML) in the
study vs. down to PRB, and using a proportional vs.
hybrid rollback to estimate incidence associated with
long-term exposure to PM2 5 concentrations that just
meet the current standards (note consideration of
rollback in the multi-factor analysis did not
incorporate the hybrid-based rollback approach)
All-cause, IHD long-term
mortality
Los Angeles and
Philadelphia
All-cause : +27 to + 1 ,089%
IHD: +256to +673%
F-39
Multi-Factor Sensitivity Analyses (short term mortality):
Impact of using season-specific vs. all-year C-R
functions and proportional vs. hybrid rollbacks to
estimate incidence associated with short-term
exposure to PM2 5 concentrations that just meet the
current standards
Non-accidental
Baltimore, Birmingham,
Detroit, Los Angeles,
New York and St. Louis
Non-accidental (four seasons + hybrid): -
116 to +179%
F-42
1
2
3
1 Unless otherwise noted, sensitivity analysis results are based on the scenario reflecting just meeting the current suite of PM2 5 standards.
2 This metric is the percent spread in risk estimates across the Moolgavkar-based model specifications (not the percent difference estimates - see text discussion
above).
February 2010
4-34
Draft - Do Not Quote or Cite
-------
1 4.3.1.1 Single-factor Sensitivity Analysis
2 This section presents the results of the single-factor sensitivity analysis, which involved
3 consideration of alternate model inputs on the core risk estimates, when those alternate inputs are
4 considered one at a time (consideration of the combined effect of several model inputs being
5 varied is covered by the multi-factor sensitivity analysis discussed in section 4.3.1.2). The
6 results of the single-factor sensitivity analysis are characterized qualitatively using the four-
7 category approach described above (i.e., low, moderate, moderate-large and large, with each of
8 these representing a defined range of percent difference from the core risk estimates).
9 Long-term exposure mortality
10 This section summarizes the results of the sensitivity analysis focused on long-term
11 exposure-related mortality endpoints (see Table 4-1 for the specific modeling elements
12 considered in the sensitivity analysis).
13 Impact of using different model choices for C-R function -fixed effects log-linear (the core
14 approach) vs. random effects log-linear or random effects log-log models: This simulation
15 considered two alternative C-R model forms obtained from Krewski et al., 2009 for modeling
16 all-cause, CPD, IHD and lung cancer mortality, including (a) random effects log-linear
17 model and (b) a random effects log-log model (note, the core effect estimate was derived
18 using a fixed effects log-linear model obtained from Krewswki et al., 2009). The simulation
19 also considered the use of multi-pollutant models that control for CO, NC>2, 63 or 862. The
20 results of the simulation suggest that the use of a random effects log-linear model, rather than
21 the core fixed effects model, has a relatively small effect on risk estimates, increasing them
22 by 12 to 23% across the mortality categories and urban study areas modeled (Appendix F,
23 Table F-3). However, use of a random effects log-log model has a larger impact on risk
24 estimates, increasing them by 50 to 159% (Appendix F, Table F-3). The greater impact of
25 the log-log model results from this function having an incrementally steeper slope at lower
26 PM levels, which quickly increases incidence estimates compared with the core log-linear
27 model (whose slope has a much more gradual incremental increase in slope at lower PM
28 levels). The use of multi-pollutant models that control for co-pollutants was shown to have
29 moderate-large impact on risk estimates, with control for CO, NC>2, or Os resulting in
30 increased PM2.5-attributable risk estimates, while control for SO2 resulted in a moderate-large
31 decrease in estimated PM2.5 risk.52
32 Impact of estimating risks down to PRB rather than down to LML: This simulation
33 compared long-term exposure mortality incidence associated with modeling risk down to
34 PRB (which varies by region - see section 3.2.1) with the core approach of modeling down
35 to LML (5.8 |ig/m3 for long-term mortality - see section 3.1). This simulation involved all
36 15 urban study areas, given that PRB is stratified by region and therefore, results of the
52 Sensitivity analysis results generated using the copollutant model involving PM2 5 and SO2 have been de-
emphasized since it is likely that control for SO2 may be capturing a portion of PM2 5 -attributable risk related to the
secondary formation of sulfate, which is a component of the PM25 mixture (i.e., the two pollutants are often highly
correlated).
February 2010 4-35 Draft - Do Not Quote or Cite
-------
1 simulation could differ significantly across the 15 urban study areas, or at least across the six
2 PM regions represented by those study areas. The results of this simulation suggest that
3 modeling risk down to PRB could have a moderate to large impact on long-term exposure
4 mortality incidence, with estimates ranging from 47 to 273% higher than the core estimates
5 (for matching urban locations) (Appendix F, Table F-6). Note, however, that risk metrics
6 based on considering the incremental reduction in risk (incidence) between two alternative
7 suites of standards would not be impacted by this source of uncertainty, since it only affects
8 estimates of absolute risk.
9 Impact of C-R function from alternative long-term exposure mortality study: This simulation
10 considered use of alternative C-R functions (and effect estimates) based on the reanalysis of
11 the Six Cities study (Krewski et al., 2000). The results suggest that use of the alternative C-R
12 function could have a moderate to moderate-large effect on CPD mortality (+45 to +74%), a
13 large effect on all-cause mortality (+123 to +159%), a moderate-large to large effect on IHD
14 mortality (+80 to +111%) and a moderate-large effect on lung cancer mortality (+67 to
15 +94%) (Appendix F, Table F-9). The results of this simulation suggest that (at least with
16 regard to application of C-R functions obtained from the Six Cities study) the potential
17 impact of functions from alternative studies on long-term exposure mortality depends on the
18 mortality category being considered. In this analysis, use of the alternative C-R functions
19 was shown to have a significant impact on all of the long-term mortality categories
20 considered.
21 Impact of using alternative rollback approaches (hybrid and peak shaving) to simulate just
22 meeting the current and alternative suites of standards. This sensitivity analysis assessed the
23 impact of estimating risk for the current and alternative sets of standards using two
24 alternatives to the proportional rollback strategy: (a) the hybrid rollback approach that
25 reflects an initial localized pattern of ambient PM2.5 reduction (resulting in non-proportional
26 rollbacks of monitored PM2 5 concentrations) with a second phase of more regional
27 reductions in ambient PM2.5 levels (based on proportional adjustments) and (b) peak shaving
28 which represents a primarily local pattern of reductions in ambient PM2 5 (see Section 3.5.4
29 for additional discussion of how these alternative rollback methods were integrated into the
30 sensitivity analysis). We note that the core analysis utilized proportional rollback exclusively
31 in simulating conditions for the current and alternative sets of standards, with this approach
32 representing a regional pattern of ambient PM2 5 reduction. A number of observations can be
33 drawn from this sensitivity analysis including:
34 o Impact on estimates ofPM2.s-related risk remaining after simulation of just
35 meeting a given suite of standards: The sensitivity analysis results suggest that
36 the use of alternative rollback methods can have a notable impact on estimates of
37 the PM2.5-attributable risk remaining after simulation of a given suite of standards
38 (see Appendix F, Table F-50 and discussion in section 3.5.4). Generally, use of
39 the hybrid approach had a small to moderate impact on absolute PM2 5-
40 attributable risk estimates, compared with the core approach of using proportional
41 rollback. By contrast, use of the peak shaving approach had a moderate to
42 moderate-large impact on absolute PM2 5-attributable risk estimates. For example,
43 Los Angeles had composite monitor values for the current suite of standards and
44 several of the alternative suites of standards that were 40 to 60% greater when the
45 peak shaving rollback method was used, compared with the proportional rollback
February 2010 4-36 Draft - Do Not Quote or Cite
-------
1 method (see Appendix F, Table F-50). By contrast, composite monitor values
2 generated using hybrid rollback for Los Angeles, were between 13 and 38%
3 higher than the proportional rollback methods.
4 o Impact on degree of reduction across alternative suites of standards: When the
5 same rollback methods is used to simulate both the current and any alternative
6 suite of standards, the pattern of risk reduction across alternative standards is
7 generally similar regardless of the rollback approaches used (see Table F-49, in
8 Appendix F). However, if one looks at meeting the current suite of standards with
9 application of the peak-shaving approach, followed by application of proportional
10 rollback to simulate alternative suites of standards, we can see notable differences
11 in the pattern of risk reduction. This is particularly true for areas with peaky PM2.5
12 distributions (i.e., areas with relatively high 24-hour design values and lower
13 annual average design values). For example, with Los Angeles, which represents
14 a study area with a relatively peaky PM2.5 distribution, application of proportional
15 rollback in simulating both the current suite of standards and the alternative
16 annual standard of 12 |ig/m3 results in a 13% reduction in long-term exposure-
17 related mortality (see Figure 4-3 and Table E-27 in Appendix E). By contrast,
18 application of peak shaving in simulating the current suite of standard levels
19 followed by proportional reduction in simulating the same alternative annual
20 standard results in an estimated 48% reduction in long-term exposure-related
21 mortality.53
22 Based on the simulations discussed above covering potential sources of uncertainty and
23 variability impacting long-term mortality, we conclude that the following factors contribute
24 potentially large sources of uncertainty to the core risk estimates: (a) use of alternative form of
25 the C-R function, specifically use of a random-effects log-log model form obtained from the
26 updated ACS study (Krewski et al., 2009) (b) use of an alternative C-R function with effects
27 estimates obtained from the reanalysis of the Six Cities study (Krewski et al. 2000), and (c)
28 estimation of risk down to PRB.54 Other factors considered in the sensitivity analysis had
29 smaller impacts on core risk estimates.
53 The difference in risk reductions based on application of different rollback methods in simulating the current suite
of standards reflects the fact that peak shaving rollback, when applied to a location where the 24hr standard level is
controlling , such as Los Angeles, will produce a smaller degree of reduction in the composite monitor annual-
average PM2 5 level. By contrast, application of proportional rollback will produce a larger degree of rollback in the
composite monitor annual-average (i.e., a level equal to that needed to get the 24hr design value to meet the 24hr
standard). We also note that the risk reductions cited here reflecting application of peak-shaving in simulating the
current suite of standards are based on comparison of composite monitor annual-averages presented in Table F-49 in
Appendix F. In generating this surrogate for reduction in long-term exposure-related mortality between the two
standard levels, we compared composite monitor annual-averages with consideration for the fact that long-term
exposure-related mortality is only calculated down to LML.
54 Use of peak-shaving as an alternative method for simulating ambient PM25 concentrations for alternative
standards had a moderate-large impact on risk estimates.
February 2010 4-37 Draft - Do Not Quote or Cite
-------
1 Short-term exposure mortality
2 This section summarizes the results of the sensitivity analysis focused on short-term
3 exposure-related mortality endpoints (see Table 5-1 for the specific modeling elements
4 considered in the sensitivity analysis).
5 Impact of using season-specific C-R functions (vs. an annual C-R function): This
6 simulation considered the impact on short-term exposure mortality risk of using seasonally-
7 differentiated effects estimates rather than the core approach of using a single C-R function
8 for the whole year (note, that the seasonal models were based on the same study as the model
9 used in the core analysis - Zanobetti and Schwartz, 2009). The results of the simulation
10 suggest that this source of uncertainty can have a wide range of effects across urban study
11 areas (including not only variation in the magnitude of effect, but also in the direction).
12 Percent changes compared with the core risk estimate were large, ranging from -116% (Los
13 Angeles) to +179% (Birmingham) (these results are for non-accidental mortality - see
14 Appendix F, Table F-15). We note that these two locations also have relatively low overall
15 incidence estimates, which does raise concerns over the degree of stability in the sensitivity
16 analysis estimates. Furthermore, for 9 of the 15 urban study areas (for non-accidental
17 morality), percent changes from the core were small, with absolute values of 12% or less
18 (Appendix F, Tables F-15). The results for CV and respiratory mortality also demonstrate
19 considerable variation across locations, but are generally smaller than results cited above for
20 non-accidental, with one exception. Birmingham is estimated to have short-term CV
21 mortality that is +500% higher using seasonal effects estimates compared with the core
22 results (We note, however, that this endpoint category also has very small incidence, again
23 raising concerns over the stability of the sensitivity analysis results) (see Table F-18). The
24 results for respiratory-related mortality also demonstrate considerable variability with results
25 that could suggest a moderate to large impact (i.e., -48 to +162% - see Appendix F, Table F-
26 21). We note, however, that small incidence estimates again raise concerns regarding the
27 stability of these percent difference results.
28 Impact of using alternative hybrid rollback approach: This simulation evaluates the
29 potential impact of using the hybrid (non-proportional) approach for simulating just meeting
30 current and alternative sets of standards, as an alternative to the proportional approach used
31 in the core analysis.55 The results of this simulation (as contrasted with the impact of using
32 the hybrid approach on long-term exposure mortality) suggest that use of the hybrid rollback
33 approach has relatively little effect on short-term mortality risk (e.g., percentage differences
34 relative to the core risk estimates were in the low single digits for most locations, with one
35 location having a difference of+17% - see Appendix F, Table F-36).
Note, that the peak shaving rollback method was only assessed in the context of the composite monitor values
used in generating long-term exposure-related mortality estimates. Consequently, consideration of the peak shaving
rollback method is only assessed in terms of its impact on long-term risk and not short-term exposure-related
mortality. Note, however, that the impact of using peak shaving versus proportional rollback on short-term
exposure-related risk is expected to be smaller than the impact on long-term exposure-related risk, since the latter is
linked to composite annual averages which are expected to experience the greatest impact from application of
alternative rollback methods.
February 2010 4-38 Draft - Do Not Quote or Cite
-------
1 The sensitivity analysis results discussed above, result in a number of overall
2 observations regarding sources of uncertainty potentially impacting short-term exposure morality
3 endpoints. The results of using the seasonally-differentiated effect estimates in modeling short-
4 term exposure mortality appear to generally have a relatively small impact (e.g., <15%) in most
5 study areas. For some study areas, the impact does appear to be much larger, with results
6 including both substantial negative and positive percent differences from the core estimates.
7 However, in all of these cases, the total incidence estimates involved are very small, raising
8 concerns over the stability of the risk estimates generated as part of this particular sensitivity
9 analysis (in many of these instances, the estimates include negative lower bounds, reflecting the
10 use of non-statistically significant effects estimates). For these reasons, the results of this
11 sensitivity analysis, while initially appearing to be notable in terms of magnitude in some study
12 areas, need to be interpreted with care. At this point, we are uncertain as to how important this
13 source of uncertainty is in the context of short-term exposure mortality estimation. Regarding
14 the use of the alternative hybrid (non-proportional) approach for simulating conditions under
15 alternative standard levels, the results suggest that this factor has a modest impact on short-term
16 exposure mortality (significantly less impact than with the use of the hybrid approach in
17 estimating long-term exposure mortality). With the exception of factors examined using the
18 Moolgavkar et al., (2003) study in Los Angeles (see section 4.3.1.4), it would appear that the
19 factors examined here do not have a large impact on risk estimates generated for short-term
20 exposure mortality. However, we note that the overall scope of the sensitivity analysis completed
21 for short-term exposure-related mortality and morbidity is far more limited than that completed
22 for long-term exposure-related mortality.
23 Short-term exposure morbidity
24 This section summarizes the results of the sensitivity analysis focused on short-term
25 exposure-related morbidity endpoints (see Table 5-1 for the specific modeling elements
26 considered in the sensitivity analysis). The results of individual sensitivity analysis simulations
27 are presented below, with overall observations presented at the end of the section.
28 Impact of using season-specific C-R functions (vs. an annual C-R function): This
29 simulation considered the impact on short-term exposure morbidity (HAs) of using
30 seasonally-differentiated effects estimates rather than the core approach of using a single C-R
31 function for the whole year (we note that the seasonal models were obtained from the same
32 study as the model used in the core analysis - Bell et al, 2008). The results of the simulation
33 suggest that, as with short-term exposure mortality this source of uncertainty can have a wide
34 range of impacts on the risk estimates across urban study areas (including not only variation
35 in the magnitude of risk, but also in the direction) depending on the specific health endpoint
36 examined. We note, however, that the magnitude of impact appears to be less for short-term
37 morbidity than for short-term mortality. Percent changes for most of the 15 urban study
February 2010 4-39 Draft - Do Not Quote or Cite
-------
1 areas were small for CV HAs (generally less than a 20% difference in either direction,
2 although there was a large impact for Tacoma (-105%)) (see Appendix F, Table F-24). This
3 source of uncertainty has a moderate to moderate-large impact for respiratory-related HAs
4 with most locations having greater than a 54% to 74% absolute effect (see Appendix F, Table
5 F-27).
6 Impact of using a seasonal function for April through August (applied only to that
7 period) in modeling asthma-related ED visits in New York, relative to the core
8 approach of using a single annual effect estimate (and applying that to the whole year):
9 This sensitivity analysis involved the approach of using a season-specific estimate to model
10 incidence for the period April through August (obtained from Ito et al., 2007). Because this
11 sensitivity analysis estimate covers a period shorter than a year, we have not directly
12 compared it with the annual estimate generated for this endpoint in the core risk assessment
13 (i.e., we have not generated percent difference estimates as is done with other sensitivity
14 analysis simulations). However, the results of this sensitivity analysis do suggest that the use
15 of seasonally-differentiated estimates in modeling this endpoint can impact risk.
16 Impact of considering models with different lags: To examine the impact of lag on
17 modeling of short-term exposure-related morbidity, we used a range of effects estimates
18 obtained from Bell et al., 2008 based on application of different lags, including 0-, 1- and 2-
19 day lags, (for both respiratory and cardiovascular-related morbidity). Because lag-
20 differentiated effects estimates were only available as national-averages and were not
21 regionally-differentiated, we could not directly compare the results using different lag models
22 to the results generated for the core analysis (i.e., the sensitivity analysis results would have
23 mixed both the lag effect and the effect of regional differentiation, thereby preventing clear
24 assessment of the importance of either factor considered in isolation). However,
25 consideration of the magnitude of the risk estimates generated using different lag models, for
26 the same endpoint at the same urban study are, suggests that choice of lag does effect
27 estimates of short-term exposure-related morbidity (see Appendix F, Table F-48).
28 Given the results of the set of simulations completed for short-term exposure morbidity,
29 both of which focused on the use of seasonally-differentiated effects estimates, it would appear
30 that this factor does not have a substantial impact on risk estimates. The analysis considering
31 different lag models does suggest that this factor could have a notable impact on risk estimates
32 and should be carefully considered when specifying C-R functions to use in the risk assessment.
33 Additional factors potentially impacting short-term exposure morbidity are addressed below in
34 relation to the sensitivity analysis based on alternative models from Moolgavkar et al. (2003). As
35 noted earlier, the scope of the sensitivity analysis completed for short-term exposure-related
36 morbidity is limited.
37 Short-term exposure-related mortality and morbidity (Moolgavkar et al., 2003 study-based
38 analysis)
39 As noted earlier in the introduction to section 4.3, the results of sensitivity analysis based
40 on Moolgavkar et al., (2003) include percent difference estimates based on considering the range
February 2010 4-40 Draft - Do Not Quote or Cite
-------
1 of risk estimates generated using alternative model specifications from this study for a given
2 health endpoint and it is these results that are discussed below.
3 Impact of model selection (e.g., log-linear GAM with 30df, log-linear GAM with lOOdf,
4 and log-linear GLM with lOOdf) on estimating short-term exposure mortality and
5 morbidity: Application of models obtained from Moolgavkar et al., (2003) with various
6 formulations related to model selection (degrees of freedom, GLM vs. GAM) to the Los
7 Angeles urban case study location results in a range of short-term exposure mortality
8 estimates (for non-accidental and CV) that differ by 80% and 49%, respectively (see
9 Appendix F, Table F-33). In the case of short-term exposure morbidity (specifically, CV-
10 related HAs), incidence estimates differ by 36% (see Appendix F, Table F-33). These results
11 suggest that these elements of model specification represent a moderate source of uncertainty
12 in estimating short-term mortality and morbidity.
13 Impact of lag structure (0-day through 5-day) on estimating short-term exposure
14 mortality: Consideration of the range of risk estimates for non-accidental mortality
15 generated using different lag structures (and associated effect estimates) provided in
16 Moolgavkar et al., (2003), suggest that this factor could have a moderate impact on risk (in
17 the range of 55% when comparing the lowest and highest positive incidence estimates
18 generated), (see Appendix F, Table F-33).
19 Impact of considering multi-pollutant models on estimating short-term exposure
20 mortality and morbidity: The results of the Moolgavkar-based simulations (when
21 considering the spread in risk estimates specifically across these simulations) suggest that the
22 multi-pollutant versus single-pollutant model issue (i.e., including CO in addition to PM2.5),
23 could have a large impact on the estimation of short-term exposure mortality (106% for all-
24 cause) and morbidity (140% for CV-related HAs).
25 Overall observations regarding key sources of uncertainty impacting short-term exposure
26 mortality and morbidity risk estimates (based on the Moolgavkar et al., 2003 study) include the
27 following. The spread in risk estimates generated across the Moolgavkar-based model
28 specifications suggests that factor related to specifying the C-R model may have a moderate to
29 large impact. More specifically, variation in the lag structure has a moderate impact on risk and
30 use of single versus multi-pollutant models could have a potentially large impact on risk. Note,
31 however, that as discussed earlier, the relevance of these sensitivity analysis results to the
32 interpretation of core risk estimates is not clear and may be relatively low (see Section 4.3.1).
33 4.3.1.2 Multi-factor Sensitivity Analysis Results
34 The results of the multi-factor sensitivity analyses are intended to support both goals of
35 the sensitivity analysis: (a) identify which factors (now in combination), appear to have a
36 significant impact on estimation of the core estimates and (b) to derive a set of reasonable
37 alternative risk estimates for use in considering uncertainty and variability associated with the
38 core risk estimates. Regarding the latter application, because these multi-factor simulations
39 combine multiple factors reflecting uncertainty and variability together in generating alternative
February 2010 4-41 Draft - Do Not Quote or Cite
-------
1 risk estimates, they are likely to produce the highest sensitivity analysis results. Therefore, it is
2 particularly important to consider the reasonableness of the results of these multi-factor
3 simulations, to insure that only credible estimates are included in the set of reasonable alternative
4 risk estimates. Consequently, we emphasize consideration for the reasonableness of these multi-
5 factor simulations in the discussion presented below.
6 Long-term exposure mortality
7 This section summarizes the results of the sensitivity analysis focused on long-term
8 exposure-related mortality endpoints (see Table 4-1 for the specific modeling elements
9 considered in the sensitivity analysis).
10 Impact of using log-linear vs. log-log C-R model with fixed or random effects,
11 estimating incidence down to the LML vs. PRB, and using proportional vs. hybrid
12 rollback to estimate long-term exposure mortality: This multi-factor sensitivity
13 analysis focused on a number of model design choices related to modeling long-term
14 exposure mortality (all-cause and IHD). Modeling elements reflected in the
15 simulations included: (a) model form (log-linear vs log-log and random vs fixed
16 effects), (b) modeling risk down to PRB (vs LML), and (c) use of an alternative hybrid
17 rollback approach (vs proportional rollback) to simulate just meeting the current and
18 alternative sets of standards. Various permutations of these design elements choices
19 (relative to the elements selected for the core analysis) were considered. Percent
20 difference estimates (for all-cause mortality) ranged from 27% (for a model estimating
21 risk down to PRB and use of the hybrid rollback approach) to 1,089% (for a model
22 with random effects log-log model, risk estimated down to PRB, and use of the hybrid
23 rollback approach).
24 We believe that application of a log-log model with random effects is a reasonable
25 alternative to the core model (fixed-effects log-linear model), based on our review of the
26 discussion in Krewski et al. (2009). Similarly, the use of a hybrid rollback approach involving
27 non-proportional adjustment where there is the potential for greater use of local control strategies
28 to address local-sources is a reasonable alternative to solely using a proportional rollback
29 approach in all study areas. Therefore, we believe that the combinations of modeling elements
30 including these alternative choices are reasonable. However, there is more concern in predicting
31 risk down to PRB. This is not because there is evidence for a threshold, but rather because we
32 do not have data to support characterization of the nature of the C-R function in the vicinity of
33 PRB. Specifically, there is increasing uncertainty in predicting the nature of the C-R function as
34 you move below the LML. So, while we believe it is reasonable conceptually to estimate risk
35 down to PRB, the quantitative process of doing this requires use of a function with very high
36 uncertainty. Therefore, we concluded that those alternative risk estimates generated using risk
37 estimated down to PRB should not be used in creating the reasonable alternative set of risk
38 estimates in considering uncertainty associated with the core risk estimates.
February 2010 4-42 Draft - Do Not Quote or Cite
-------
1 A key limitation of the multi-factor sensitivity analysis is that the approach used did not
2 allow us to consider the peak-shaving rollback method in concert with the other modeling
3 elements described above. This means that the combined impact of peak shaving (which has a
4 greater impact than the hybrid rollback method) with other model specifications is not
5 characterized. However, as part of the integrative discussion in Chapter 6, we will consider the
6 results of the single-factor sensitivity analysis examining rollback (with its consideration for
7 peak shaving) along with the multi-factor sensitivity analysis results described here.
8 Short-term exposure mortality
9 This section summarizes the results of the sensitivity analysis focused on short-term
10 exposure-related mortality endpoints (see Table 4-1 for the specific modeling elements
11 considered in the sensitivity analysis).
12 Impact of using season-specific vs. annual effect estimates and proportional vs.
13 hybrid rollback approaches in modeling short -term exposure mortality: This
14 multi-factor sensitivity analysis focused on a number of model design choices related
15 to modeling short-term mortality (non-accidental). Modeling elements included in this
16 sensitivity analysis were use of seasonal vs. annual effects estimates and use of hybrid
17 vs proportional rollback to simulate just meeting current and alternative standard
18 levels. Percent difference estimates (for non-accidental mortality) across the 7 urban
19 study areas included in the simulation ranged from -109% (LA) to +119%
20 (Birmingham) (see Appendix F, Table F-42). However, we note that the total
21 incidence estimates associated with these higher-impact locations were relatively low,
22 again raising the concern for the stability in relative differences with the core estimates.
23 Because of the more limited scope of the multi-factor sensitivity analysis completed for
24 short-term exposure-related mortality, we have concluded that these results should not be used as
25 an additional set of reasonable risk estimates to inform consideration of uncertainty associated
26 with this category of risk estimates.
27 4.3.2 Additional Set of Reasonable Risk Estimates to Inform Consideration of
28 Uncertainty in Core Risk Estimates
29 This section discusses the use of the output of the sensitivity analysis completed as part
30 of this risk assessment as an additional set of reasonable risk estimates to inform consideration of
31 uncertainty associated with the core risk estimates. Specifically, in the case of long-term
32 exposure-related mortality endpoints, staff has concluded that the results of the sensitivity
33 analysis represent a reasonable set of alternate risk estimates that fall within an overall set of
34 plausible risk estimates surrounding the core estimates.56
56 As noted in section 4.3.2 and in the integrative discussion in Section 6.4, while staff believes that the sensitivity
analysis does provide insights into the potential impact of certain sources of uncertainty on short-term exposure-
related mortality and morbidity risk, the sensitivity analysis conducted for short-term exposure-related endpoints
February 2010 4-43 Draft - Do Not Quote or Cite
-------
1 While not representing a formal uncertainty distribution, the output of the sensitivity
2 analysis, when combined with the core risk estimates, represent a set of plausible risk estimates,
3 which reflect consideration for uncertainty in various elements of the risk assessment model.
4 Therefore, while the discussion of risk estimates in the context of assessing the degree of risk
5 reduction associated with suites of alternative standards (see Chapter 6) does focus on the core
6 risk estimates since these are judged to have the greatest overall confidence, the output of the
7 sensitivity analysis can be used to provide additional perspective on the potential range of
8 uncertainty around the core estimates. Note however, that we do not know the confidence
9 interval captured by this uncertainty set, or the specific percentiles of the risk distribution are
10 represented by points within that set.
11 As noted earlier, the quantitative single- and multi-factors sensitivity analyses generated
12 an additional set of risk estimates for a subset of the urban study areas, air quality scenarios and
13 health endpoints included in the core risk analysis (i.e., Los Angeles and Philadelphia assessed
14 for the current standard level). However, the part of the sensitivity analysis focusing on
15 alternative methods for simulating ambient PM2.5 levels (i.e., rollback), did consider a larger
16 number of study areas and air quality scenarios. In presenting the alternative sets of reasonable
17 risk estimates, we focus on Los Angeles and Philadelphia for many of the modeling elements,
18 although we expand the discussion in the context of discussing results related to conducting
19 rollback..
20 In using the additional set of reasonable risk results to augment the core risk estimates,
21 we begin by presenting both the core and alternative sets of estimates for Los Angeles and
22 Philadelphia in Table 4-4. Then, in Figures 4-7 and 4-8, we present graphical display of the full
23 uncertainty set comprising the core plus additional reasonable risk estimates for Los Angeles and
24 Philadelphia, differentiated by mortality category (Figure 4-7 present results for IHD and Figure
25 4-8 presents results for all cause mortality). This section concludes with a set of observations
26 resulting from consideration of information depicted in Table 4-4 and Figures 4-7 and 4-8 in the
27 context of interpreting uncertainty in the core risk estimates.57
was not as comprehensive as that conducted for long-term exposure-related endpoints. Therefore, we do not believe
that the results of the sensitivity analysis can be used as an additional set of reasonable risk estimates to supplement
the core set in the case of short-term exposure-related endpoints.
57 As noted earlier in 3.4.1, we have excluded several of the sensitivity analysis results in defining the set of
alternative reasonable risk estimates. Specifically, we consider estimates based on modeling risk down to PRB to be
less reasonable than the other scenarios included in the sensitivity analysis, since there is substantial uncertainty
associated with the C-R function shape below the LML. In addition, as discussed in Section 4.3.1.1 risk estimates
generated using the copollutant model involving PM2 5 and SO2 have been de-emphasized since it is likely that
control for SO2 may be capturing a portion of PM2 5-attributable risk related to the secondary formation of sulfate.
February 2010 4-44 Draft - Do Not Quote or Cite
-------
1
2
Table 4-4 Derivation of a set of reasonable alternative risk estimates to
supplement the core risk estimates (Los Angeles and Philadelphia,
current standards, for long-term IHD mortality).
Core risk estimate
Percent of total incidence
for IHD and all cause
mortality (current suite of
standards):
Los Angeles:
IHD: 6.1 to 7.7%
All cause: 1.6 to 2.0%
Philadelphia:
IHD: 10.5 to 13.2%
All cause: 2.8 to 3.6%
(note, two core estimates
are presented for each
combination of urban
study area and mortality
endpoint category
reflecting use of C-R
functions derived using
different periods of
ambient data from
Krewskietal., 2009 -see
section 3.3.3)
Sensitivity analysis
Description of
simulation
Results
(percent difference:
sensitivity analysis versus
core estimate)4
Adjusted set of risk estimate
to supplement core risk
estimates1
Single-element sensitivity analysis results
(A) Impact of using
different model
choices: random
effects log-linear
model
(B) Impact of using
different model
choices: random
effects log-log model
(C) Impact of using
different model
choices (single vs.
multi-pollutant - NO2
Vs 03/CO)3
(D) Impact of C-R
function from
alternative long-term
exposure study
(Krewski et al, 2000)
(E) Impact of using
alternative roll-back
approach (hybrid and
peak shaving) to
simulate just meeting
alternative standards
Los Angeles and Philadelphia:
IHD: +12%;
All cause: +23%
Los Angeles:
IHD: +1 1 1%; All cause: +159
Philadelphia:
IHD: +80%; All cause: +123%
Los Angeles and Philadelphia:
All cause: +45 to +74%
(O3/CO and NO2 , respectively)
and -74% for SO2
Los Angeles:
All cause: +121%
Philadelphia:
All cause: +11 9%
Los Angeles:
Both all cause & IHD: +21 to
+40% (hybrid and peak
shaving, respectively)
Philadelphia:
Both all cause & IHD: +8%
(peak shaving only)
Los Angeles and
IHD: 8.6%, All cause: 2.5%
Philadelphia:
IHD: 14.8%, All cause: 4.4%
Los Angeles and
IHD: 16.2%, All cause: 5.2%
Philadelphia:
IHD: 23.8%, All cause: 8.0%
Los Angeles and
All cause: 2.9% and 3.5%
(forO3/COandNO2,
respectively), 0.52% (SO2)
Philadelphia:
All cause: 5.2% and 6.3% (for
O3/COandNO2,
respectively), 0.94% (SO2)
Los Angeles:
All cause: 4.4%
Philadelphia:
All cause: 7.9%
Los Angeles and
Hybrid: IHD: 9.3%, All cause:
2.4%
Peak shaving: IHD: 10.8%,
All cause: 2.8%
Philadelphia:
IHD: 14.3%
All cause: 3.9%
Multi-element sensitivity analysis results
(F) Random effects
log-log & hybrid
non-proportional
rollback
Los Angeles:
IHD: +149%
All cause: +211
Philadelphia:
NA2
Los Angeles:
IHD: 19.2%
All cause: 6.2%
Philadelphia:
NA2
4
5
6
1 Percent of total incidence that is PM2 5- related (note, the set of estimates for each entry reflect adjustment to the
two core estimates generated for IHD and all-cause mortality)
2 hybrid not run for Philadelphia, so multi-element sensitivity analysis not completed
Note, that the risk estimates for SO2 are presented as open circles in Figures 4-6 and 4-7, to signify that they have
lower confidence and are de-emphasized relative to the other alternative risk estimates presented.
February 2010
4-45
Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
3 the two pollutant model for PM2 5 and CO and PM2 5 and O3 had the same sensitivity result, so both models are
referenced here with the same impact on mortality estimates.
4 Sensitivity analysis based on comparison of alternative model formulations to the core risk estimates based on the
C-R function derived using 1999-2001 ambient monitoring data (see section 3.5.4).
Figure 4-7 Comparison of core risk estimates with reasonable alternative set of
risk estimates for Los Angeles and Philadelphia (IHD mortality).
geles
Blphia
* *
* *«
5 10 15 20 25
Percent total incidence attributable to PM25 (IHD mortality)
KEY:
if - core risk estimate
0 - alternative reasonable risk estimate (letter
matches simulation in Table 4-2)
Figure 4-8 Comparison of core risk estimates with reasonable alternative set of
risk estimates for Los Angeles and Philadelphia (all cause mortality).
10
Los Angeles o
Philadelphia
0.5 1 2 3 4 5 6 78
Percent total incidence attributable to PM25 (all-cause mortality)
KEY:
- core risk estimate
- alternative reasonable risk estimate (letter
matches simulation in Table 4-2)
- alternative reasonable risk estimate for copollutant modelin including PM25and
SO2 (de-emphasized relative to other alternative risk estimates - see text)
February 2010
4-46
Draft - Do Not Quote or Cite
-------
1
2 Review of the set of risk estimates presented in Table 4-4 and displayed in Figures 4-7
3 and 4-8 results in a number of observations regarding uncertainty associated with the core risk
4 estimates:
5 Consideration for uncertainty and variability in the core risk estimates results in a
6 notable spread in risk estimates: Given the factors considered in generating the
7 alternative set of reasonable risk estimates, there appears to be a factor of 3 to 4 spread in
8 risk estimates if we consider the lowest (core) estimates generated and the highest
9 alternative risk estimates generated. This observation holds for both urban study areas
10 considered, as well as for the two mortality endpoint categories. As noted earlier in this
11 section, we have de-emphasized risk estimates generated using the copollutant model
12 involving PM2.5 and SC>2 due to concerns with collinearity between the two pollutants and
13 the potential that SC>2 represents risk attributable to secondarily formed PM2 5.
14 Uncertainty set of risk estimates generated to supplement the core risk estimates are
15 skewed towards higher risk: It appears that, given the factors considered in generating
16 the alternative set of reasonable risk estimates, consideration of uncertainty could result
17 in higher (more elevated) risk estimates, compared with the core risk estimates. In other
18 words, most if not all of the alternative model specifications we considered resulted in
19 risks that are higher than our core estimates.
20 Sensitivity analysis is limited in its scope (potentially important sources of uncertainty
21 not considered): As noted earlier, the sensitivity analysis did not consider a number of
22 potentially important sources of uncertainty, some of which were addressed as part of the
23 qualitative analysis of uncertainty (see Table 3-13). For example, information is not
24 available to consider compositional differences in PM2.5 and the potential for
25 differentiation of effects estimates. Further, not considering more refined patterns of
26 intra-urban exposure to PM2.5 in deriving effects estimates could result in under-
27 estimation of risk.
28 It is important to reiterate that this set of alternative realizations presented in Table 4-4
29 and depicted in Figures 4-6 and 4-7, does not represent an uncertainty distribution. Therefore,
30 we can not assign percentiles to the individual data points presented and (importantly), we do not
31 draw any conclusions based on any clustering of the alternative risk estimates seen in Figures 4-6
32 and 4-7. Further, we do not know whether any of the higher-end estimates generated actually
33 represent true bounding risk estimates given overall uncertainty associated with the core risk
34 estimates. Despite these key caveats, having a set of risk estimates reflecting the impact of
35 modeling element uncertainties does provide information that helps to inform our
36 characterization of uncertainty related to the core risk estimates.
February 2010 4-47 Draft - Do Not Quote or Cite
-------
1 4.4 EVALUATING THE REPRESENTATIVENESS OF THE URBAN STUDY AREAS
2 IN THE NATIONAL CONTEXT
3 The goal in selecting the 15 urban study areas included in this risk assessment was two-
4 fold: (a) to choose urban locations with relatively elevated ambient PM levels (in order to
5 evaluate risk for locations likely to experience some degree of risk reduction under alternative
6 standards) and (b) to include a range of urban areas reflecting heterogeneity in other PM risk-
7 related attributes across the country. To further support interpretation of risk estimates generated
8 in this analysis, we are assessing the degree to which urban study areas represent the range of
9 key PM2.5 risk-related attributes that spatially vary across the nation. We have partially
10 addressed this issue by selecting urban study areas that provide coverage for different PM
11 regions of the country (see section 3.3.2). In addition, we are considering how well the selected
12 urban areas represent the overall U.S. for a set of spatially-distributed PM2.s risk related variables
13 (e.g., PM2.5 composition, weather, demographics including SES, baseline health incidence rates).
14 This analysis will help to inform how well the urban study areas reflect national-level variability
15 in these key PM risk-related variables. Based on generally available data (e.g. from the 2000
16 Census, Centers for Disease Control (CDC), or other sources), distributions for risk-related
17 variables across U.S. counties and for the specific counties represented in the urban study areas
18 are generated. The specific values of these variables for the selected urban study areas are then
19 plotted on these distributions, and an evaluation is conducted of how representative the selected
20 study areas are with respect to these individual variables, relative to the national distributions.
21 Estimates of risk (either relative or absolute, e.g. number of cases) within our risk
22 assessment framework are based on four elements: population, baseline incidence rates, air
23 quality, and the coefficient relating air quality and the health outcome (i.e., the PM2.5 effect
24 estimates). Each of these elements can contribute to heterogeneity in risk across urban locations,
25 and each is variable across locations. In addition, there may be additional identifiable factors
26 that contribute to the variability of the four elements across locations. In this assessment, we
27 examine the representativeness of the selected urban area locations for the four main elements,
28 and also provide additional assessment of factors that have been identified as influential in
29 determining the magnitude of the C-R function across locations.
30 The specific choice of variables which may affect the PM2.5 effect estimates for which we
31 will examine urban study area representativeness is informed by an assessment of the
32 epidemiology literature. We particularly focused on meta-analyses and multi-city studies which
33 identified variables that influence heterogeneity in PM2.5 effect estimates, and exposure studies
34 which explored determinants of differences in personal exposures to ambient PM2.5. While
35 personal exposure is not incorporated directly into PM epidemiology studies, differences in the
36 PM2.s effect estimates between cities clearly is impacted by differing levels of exposure and
February 2010 4-48 Draft - Do Not Quote or Cite
-------
1 differences in exposure are clearly related to a number of exposure determinants. Broadly
2 speaking, determinants of PM2.5 effect estimates can be grouped into three areas: demographics,
3 baseline health conditions, and climate and air quality. Based on a review of these studies, we
4 identified the following variables within each group as potentially determining the PM2.5 effect
5 estimates:
6 Demographics: education (see Zeka et al, 2006; Ostro et al, 2006), age and gender (see
7 Zeka et al, 2006), population density (see Zeka et al, 2005), unemployment rates (see Bell
8 and Dominici, 2008), race (see Bell and Dominici, 2008), public transportation use (see
9 Bell and Dominici, 2008),
10 Baseline health conditions: disease prevalence (diabetes - Bateson and Schwartz, 2004;
11 Ostro et al, 2006; Zeka et al, 2006; pneumonia - Zeka et al, 2006; stroke - Zeka et al,
12 2006; heart and lung disease - Bateson and Schwartz, 2004; acute myocardial infarction
13 - Bateson and Schwartz, 2004).
14 Climate and air quality: PM2 5 levels (average, 98th percentiles, and numbers of days over
15 the level of the 24-hour standard, e.g. 35 jig/m3), co-pollutant levels, PM composition
16 (see Bell et al, 2009; Dominici et al, 2007; Samet, 2008; Tolbert, 2007), temperatures
17 (temp) (days above 90 degrees, variance of summer temp, mean summer temp, 98th
18 percentile temp, mean winter temp see Roberts, 2004; Medina-Ramon et al, 2006; Zeka
19 et al., 2005), air conditioning prevalence (see Zanobetti and Schwartz, 2009; Franklin et
20 al, 2007; Medina-Ramon et al, 2006), ventilation (see Sarnat et al, 2006), percent of
21 primary PM from traffic (see Zeka et al., 2005),
22 Based on these identified potential risk determinants, we identified possible datasets that
23 could be used to generate nationally representative distributions for each parameter. We were
24 not able to identify readily available national datasets for all variables. In these cases, if we were
25 able to identify a broad enough dataset covering a large enough portion of the U.S., we used that
26 dataset to generate the parameter distribution. In addition, we were not able to find exact
27 matches for all of the variables identified through our review of the literature. In cases where an
28 exact match was not available, we identified proxy variables to serve as surrogates. For each
29 parameter, we report the source of the dataset, its degree of coverage, and whether it is a direct
30 measure of the parameter or a proxy measure. The target variables and sources for the data are
31 provided in Table 4-2. Summary statistics for the most relevant variables are provided in Table
32 D-3.
February 2010 4-49 Draft - Do Not Quote or Cite
-------
1 Table 4-5 Data Sources for PM NAAQS Risk Assessment Risk Distribution
2 Analysis.
Potential Risk
Determinant
Metric
Year Source
Degree of
National
Coverage
Demographics
Age
Age
Age
Education
Unemployment
Income
Race
Population
Population density
Urbanicity
Median Age
Percent over 65
Percent under 15
Population with less
than HS diploma
Percent unemployed
Per Capita Personal
Income
Percent nonwhite
Total population
Population/square mile
ERS Classification
Code
County Characteristics, 2000-2007
2005 Inter-university Consortium for
Political and Social Research
County Characteristics, 2000-2007
2005 Inter-university Consortium for
Political and Social Research
County Characteristics, 2000-2007
2005 Inter-university Consortium for
Political and Social Research
USDA/ERS,
2000 http://www.ers.usda.gov/Data/Edu
cation/
County Characteristics, 2000-2007
2005 Inter-university Consortium for
Political and Social Research
County Characteristics, 2000-2007
2005 Inter-university Consortium for
Political and Social Research
County Characteristics, 2000-2007
2006 Inter-university Consortium for
Political and Social Research
Cumulative Estimates of Resident
Population Change for the United
States, States, Counties, Puerto
2008 Rico, and Puerto Rico Municipios:
April 1, 2000 to July 1, 2008,
Source: Population Division, U.S.
Census Bureau
Cumulative Estimates of Resident
Population Change for the United
States, States, Counties, Puerto
2008 Rico, and Puerto Rico Municipios:
April 1, 2000 to July 1, 2008,
Source: Population Division, U.S.
Census Bureau
County Characteristics, 2000-2007
2003 Inter-university Consortium for
Political and Social Research
All counties
All counties
All counties
All counties
All counties
All counties
All counties
All counties
All counties
All counties
Climate and Air Quality
PM2 5 Levels
PM2 5 Levels
PM2 5 Levels
PM2 5 Levels -
Monitored Ann Mean
PM2 5 Levels -
Monitored 98th %ile
Average MCAPS
2007 AQS
2007 AQS
MCAPS website 204 counties
617 Monitored
counties
617 Monitored
counties
204 MCAPS
counties
February 2010
4-50
Draft - Do Not Quote or Cite
-------
Potential Risk
Determinant
PM2 5 Levels
Copollutant Levels
Roadway
emissions/Exposure
Temperature
Temperature
Relative Humidity
Ventilation
Metric
% days exceeding 35
ug/m3
Ozone
% of primary emissions
from traffic
Annual Average
Mean July Temp 1941-
1970
Mean July RH 1941-
1970
Air conditioning
prevalence
Year Source
MCAPS website 204 counties
AQS
1999 NEI
MCAPS website 204 counties
County Characteristics, 2000-2007
Inter-university Consortium for
Political and Social Research
County Characteristics, 2000-2007
Inter-university Consortium for
Political and Social Research
American Housing Survey, with
2005 additional processing as in Reid et
al (2009)
Degree of
National
Coverage
204 MCAPS
counties
725 Monitored
counties
All counties
204 MCAPS
counties
All counties
All counties
83 urban areas
Baseline Health Conditions
Baseline Mortality
Baseline Mortality
Baseline Mortality
Baseline Mortality
Baseline Morbidity
Baseline Morbidity
Baseline Morbidity
Baseline Morbidity
Baseline Morbidity
Baseline Morbidity
Obesity
Level of exercise
Level of exercise
Respiratory Risk
Factors
Smoking
All Cause
Non Accidental
Cardiovascular
Respiratory
AMI prevalence
Diabetes Prevalence
Pneumonia Prevalence
Stroke Prevalence
CHD Prevalence
COPD Prevalence
BMI
vigorous activity 20
minutes
moderate activity 30
minutes or vigorous
activity 20 minutes
Current Asthma
Ever Smoked
CDC Wonder 1999-2005
CDC Wonder 1999-2006
CDC Wonder 1999-2007
CDC Wonder 1999-2008
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
2007 BRFSS MSA estimates
All counties
All counties
All counties
All counties
184 BRFSS
MSA
m BRFSS
J-J-EVL OO
MSA
184 BRFSS
MSA
184 BRFSS
MSA
184 BRFSS
MSA
184 BRFSS
MSA
184 BRFSS
MSA
184 BRFSS
MSA
184 BRFSS
MSA
C-R Estimates
Mortality Risk
Mortality Risk
Mortality Risk
All Cause
Respiratory
Cardiovascular
Zanobetti and Schwartz (2009) 212
cities
20Q9 Zanobetti and Schwartz (2009) 212
cities
20Q9 Zanobetti and Schwartz (2009) 212
cities
212 cities
212 cities
212 cities
February 2010
4-51
Draft - Do Not Quote or Cite
-------
Table 4-6 Summary Statistics for Selected PM Risk Attributes.
Risk Attributes
Demographics
Population
Population Density (Pop/sq mile)
Median Age (years)
% Age 65 Plus
Unemployment rate (%)
% with Less than High School Diploma
Income ($2005)
Air conditioning prevalence (%)
% Non-white
Health Conditions
Prevalence of CHD (%)
Prevalence of Obesity (%)
Prevalence of Stroke (%)
Prevalence of Smoking (ever) (%)
Prevalence of Exercise (20 minutes) (%)
All Cause Mortality (per 100,000 population)
Non-accidental Mortality (per 100,000 population)
Cardiovascular Mortality (per 100,000 population)
Respiratory Mortality (per 100,000 population)
Air Quality and Climate
AQ - PM25 Annual Mean (ug/m3)
AQ - PM25 98th %ile 24-hour Average (ng/m3)
AQ - O3 4th High Maximum 8-hour Average (ppm)
% Mobile Source PM Emissions
Average
Urban study U.S.
areas counties
1,410,331 97,020
7,212 258
35.5 38.6
11.3 14.9
5.4 5.4
21.8 22.6
35691 27367
85.8 83.3
29.5 13.0
3.9 4.3
26.4 26.0
2.7 2.7
18.4 19.6
28.4 28.0
833.7 1022.3
774.1 950.6
317.5 392.1
70.8 97.3
15.1 11.7
38.7 30.7
0.087 0.077
34.0 44.4
Standard Deviation
Urban study U.S.
areas counties
1,870,237 312,348
14,960 1,757
2.6 4.4
2.6 4.1
1.5 1.8
7.7 8.8
12605 6604
13.3 21.5
18.2 16.2
0.9 1.3
3.0 4.1
0.8 1.0
3.1 4.0
3.6 4.8
241.1 258.6
227.3 249.6
100.6 121.0
23.0 32.3
2.2 3.1
11.6 9.3
0.009 0.010
11.2 21.9
Maximum
Urban study U.S.
areas counties
9,862,049 9,862,049
71,758 71,758
41.5 55.3
17.2 34.7
9.0 20.9
37.7 65.3
93377 93377
99.4 100.0
68.3 95.3
5.2 8.7
32.7 35.7
4.1 6.5
23.1 34.4
33.9 44.1
1342.9 2064.2
1242.0 1958.4
535.7 970.4
130.3 351.0
19.6 22.5
79.2 81.1
0.105 0.126
56.6 97.6
Minimum
Urban study U.S.
areas counties
57,441 42
87 0
30.2 20.1
5.8 2.3
2.7 1.9
11.2 3.0
23492 5148
58.6 9.9
2.7 0.0
1.8 1.8
22.2 14.0
1.1 0.7
14.2 6.5
20.5 15.4
402.5 176.8
361.6 117.7
122.4 37.5
34.8 13.3
9.7 3.4
26.8 9.1
0.064 0.033
13.7 0.3
Sample Size
Urban study U.S.
areas (number
(number of of
counties) counties)
31 3143
31 3143
31 3141
31 3141
31 3133
31 3141
31 3086
10 70
31 3141
14 184
14 182
14 184
14 184
14 183
31 3142
31 3142
31 3142
31 3136
29 617
29 617
27 725
31 3141
February 2010
4-52
Draft - Do Not Quote or Cite
-------
Risk Attributes
July Temperature Long Term Average (°F)
July Relative Humidity Long Term Average (°F)
C-R Estimates
All Cause Mortality PM2.5 Risk Estimate
Respiratory Mortality PM2.5 Risk Estimate
Cardiovascular Mortality PM2.5 Risk Estimate
Average
Urban study U.S.
areas counties
78.1 75.9
58.2 56.2
0.000971 0.000974
0.001606 0.001670
0.001013 0.000842
Standard Deviation
Urban study U.S.
areas counties
4.5 5.4
14.0 14.6
0.000340 0.000216
0.000419 0.000305
0.000586 0.000324
Maximum
Urban study U.S.
areas counties
91.2 93.7
70.0 80.0
0.001349 0.001508
0.002157 0.002221
0.001958 0.001958
Minimum
Urban study U.S.
areas counties
64.8 55.5
19.0 14.0
0.000159 -0.000099
0.000931 -0.000346
-0.000180 -0.000180
Sample Size
Urban study U.S.
areas (number
(number of of
counties) counties)
31 3104
31 3104
15 112
15 112
15 112
February 2010
4-53
Draft - Do Not Quote or Cite
-------
1 Formal comparisons of parameter distributions for the set of urban study areas and the
2 national parameter distributions are conducted using standard statistical tests, e.g. the
3 Kolmogorov-Smirnov non-parametric test for equality of distributions. In addition, visual
4 comparisons are made using cumulative distribution functions, and boxplots.
5 The formal Kolmogorov-Smirnov test results are provided in Table 4-4. The K-S tests
6 the hypotheses that two distributions are not significantly different. A high p-value indicates a
7 failure to reject the null hypotheses that the case-study and national distributions are the same.
8 We used a rejection criterion of p<0.05, which is a standard rejection criteria. It should be noted
9 that the K-S test provides a good overall measure of fit, but will not provide a test of how well
10 specific percentiles of the distributions are matched. As such, the K-S test results will not be
11 sufficient to determine whether the urban study areas adequately capture the tails of the
12 distributions of specific risk related variables. Additional visual analyses are used to assess
13 representativeness for the tails of the distributions. Overall, the K-S test results show that for
14 many of the important risk variables such as population, air quality, age, and baseline mortality
15 rates, the urban study areas are not representative of the distributions of these variables for the
16 U.S. as a whole. However, for some important potential risk determinants, such as prevalence of
17 underlying hear and lung diseases, the case study areas are representative of the national
18 distributions. However, for these specific variables, the national distribution is represented
19 primarily by large urban areas, so it is more accurate in these cases to suggest that the urban
20 study areas are representative of the overall distribution across urban areas.
21 Figures 4-14 through 4-17 show for the four critical risk function elements (population,
22 air quality, baseline incidence, and the PM2 5 effect estimate) the cumulative distribution
23 functions plotted for the nation, as well as for the urban study areas. These four figures focus on
24 critical variables representing each type of risk determinant, e.g. we focus on all-cause mortality
25 rates, but we also have conducted analyses for cardiovascular and respiratory mortality
26 separately. The complete set of analyses is provided in Appendix D. The vertical black lines in
27 each graph show the values of the variables for the individual urban study areas. These figures
28 show that the selected urban study areas represent the upper percentiles of the distributions of
29 population and air quality, while not representing lower population locations with lower 24-hour
30 PM2 5 levels. This is consistent with the objectives of our case study selection process, e.g. we
31 are characterizing risk in areas that are likely to be experiencing excess risk due to PM levels
32 above alternative standards. The urban case study locations represent the full distribution of
33 PM2.5 risk coefficients, but do not capture the upper end of the distribution of baseline all-cause
34 mortality. The interpretation of this is that the case study risk estimates may not capture the
35 additional risk that may exist in locations that have the highest baseline mortality rates.
February 2010 4-54 Draft - Do Not Quote or Cite
-------
1 Figures 4-18 through 4-21 shows for several selected potential risk attributes the CDF
2 plotted for the nation as well as for the urban study areas. These potential risk attributes do not
3 directly enter the risk equations, but have been identified in the literature as potentially affecting
4 the magnitude of the PM2.5 C-R functions reported in the epidemiological literature. The
5 selected urban study areas do not capture the higher end percentiles of several risk
6 characteristics, including populations over 65, income, and baseline cardiovascular disease
7 prevalence. Comparison graphs for other risk attributes are provided in Appendix D.
8 Summarizing the analyses of the other risk attributes, we conclude that the urban study areas
9 provide adequate coverage across population, population density, annual and 24-hour PM2.5
10 levels, ozone co-pollutant levels, temperature and relative humidity, unemployment rates,
11 percent non-white population, asthma prevalence, obesity prevalence, stroke prevalence, exercise
12 prevalence, and less than high school education. We also conclude that while the urban study
13 areas cover a wide portion of the distributions, they do not provide coverage for the upper end of
14 the distributions of age (all case study locations are below the 85th %ile), % of population 65 and
15 older (below 85th %ile), percent of primary PM emissions from mobile sources (below 8oth
16 %ile), prevalence of angina/coronary heart disease (below 85th %ile), prevalence of diabetes
17 (below 85th %ile), prevalence of heart attack (below 80th %ile), prevalence of smoking (below
18 85th %ile), all-cause mortality rates (below 90th %ile), cardiovascular mortality rates (below 90th
19 %ile) and respiratory mortality rates (below 90th %ile). In addition, all of the case study
20 locations were above the 25th percentile of the distribution of personal income.
21 Based on the above analyses, we can draw several inferences regarding the
22 representativeness of the urban case studies. First, the case studies represent urban areas that are
23 among the most populated and most densely population in the U.S. Second, they represent areas
24 with relatively higher levels of annual mean and 24-hour 98th percentile PM2.5. Third, they
25 capture well the range of effect estimates represented in the Zanobetti and Schwartz (2009)
26 study. These three factors would suggest that the urban study areas should capture well overall
27 risk for the nation, with a potential for better characterization of the high end of the risk
28 distribution. However, there are several other factors that suggest that the urban study areas may
29 not be representing areas that may have a high risk per microgram of PM2.5. The analysis
30 suggests that the urban study areas are not capturing areas with the highest baseline mortality
31 risks, nor those with the oldest populations. These areas may have higher risks per microgram of
32 PM2.5, and thus the high end of the risk distribution may not be captured, although the impact on
33 characterization of overall PM risk may not be as large, for the following reasons.
34 It should be noted that several of the factors with underrepresented tails, including age
35 and baseline mortality (R=0.81) are spatially correlated, so that certain counties which have high
36 proportions of older adults also have high baseline mortality and high prevalence of underlying
February 2010 4-55 Draft - Do Not Quote or Cite
-------
1 chronic health conditions. Because of this, omission of certain urban areas with higher
2 percentages of older populations, for example, cities in Florida, may lead to underrepresentation
3 of high risk populations. However, with the exception of areas in Florida, most locations with
4 high percentages of older populations have low overall populations, less than 50,000 people in a
5 county. And even in Florida, the counties with the highest PM2.5 levels do not have a high
6 percent of older populations. This suggests that while the risk per exposed person per microgram
7 of PM2 5 may be higher in these locations, the overall risk to the population is likely to be within
8 the range of risks represented by the urban case study locations.
9 Table 4-7 Results of Kolomogrov-Smirnoff Tests for Equality Between National
10 and Urban Study Area Distributions for Selected National Risk
11 Characteristic Variables
12 (null hypothesis is no difference between the distributions)
Risk Attributes
Demographics
Population
Population Density (Pop/sq mile)
Median Age
% Age 65 Plus
Unemployment rate
% with Less than High School Diploma
Income
Air Conditioning Prevalence (%)
% Non-white
Health Conditions
Prevalence of CHD
Prevalence of Obesity
Prevalence of Stroke
Prevalence of Smoking (ever)
Prevalence of Exercise (20 minutes)
All Cause Mortality
Non-accidental Mortality
Cardiovascular Mortality
Respiratory Mortality
Air Quality and Climate
AQ - PM25 Annual Mean
AQ - PM25 98th %ile 24-hour Average
AQ - PM25 % of days above 35 ug/m3
AQ - O3 4th High Maximum 8-hour
Average
% Mobile Source PM Emissions
Reject HO?
p-value
Y
Y
Y
Y
N
N
Y
N
Y
0.0001
0.0001
0.0001
0.0001
0.5850
0.8535
0.0001
0.9592
0.0001
N
N
N
N
N
Y
Y
Y
Y
0.7705
0.9180
0.7064
0.5748
0.7649
0.0001
0.0002
0.0060
0.0001
Y
Y
Y
Y
Y
0.0001
0.0001
0.0248
0.0003
0.0133
February 2010
4-56
Draft - Do Not Quote or Cite
-------
1
2
3
4
Risk Attributes
July Temperature Long Term Average
July Relative Humidity Long Term
Average
C-R Estimates
All Cause Mortality PM25Risk
Respiratory Mortality PM2 5 Risk
Cardiovascular Mortality PM2 5 Risk
Reject HO?
Y
N
p-value
0.0003
0.0614
N
N
N
0.1585
0.2864
0.1161
Figure 4-9 Comparison of distributions for key elements of the risk equation:
total population.
Comparison of Urban Case Study Area Population with U.S. Distribution of
Population (all U.S. Counties)
100% -i
90%
80% -
70%
Urban case study areas are
all above the 65th Percentile
of county populations
100
1000
10000 100000
Population
1000000
10000000
All Counties CDF ^Case Study Counties CDF Case Study Counties
February 2010
4-57
Draft - Do Not Quote or Cite
-------
1
2
Figure 4-10 Comparison of distributions for key elements of the risk equation:
»th
98 percentile 24-hour average PM2.s
Comparison of Urban Case Study Area 98th %ile PM2.5 with U.S. Distribution of
98th %ile PM2.5
(617 U.S. Counties with PM2.5 Monitors)
100%
90%
«, 80%
.0)
1 70%
O 60%
§ 40%
5 30% -
10 20 30 40 50 60
98th Percentile Daily PM2.5
70
80
90
All Counties CDF Case Study Counties CDF Case Study Counties
February 2010
4-58
Draft - Do Not Quote or Cite
-------
1
2
Figure 4-11 Comparison of distributions for key elements of the risk equation: all
use mortality rate.
Comparison of Urban Case Study All Cause Mortality Rate to U.S. Distribution of All
Cause Mortality Rate
(3143 U.S. Counties)
IUU /O
90%
80%
tn 70%
0) /U/0
^
1 60%
o
° 50%
CO
2 40%
0
£ 30%
20%
10%
.
^^^^_^^^^_i
^
1 1
l-l
1 1 I
\t
1
/
/
\ , 1 1
X
1 1 1 1
Ul
l-l
*
.-"
^^^^^
Urban case study areas are
all below the 90th percentile
of county all cause mortality
100
300
500
700 900 1100 1300 1500
All Cause Mortality per 100,000 Population
1700 1900 2100
All Counties CDF ^Case Study Counties CDF Case Study Counties
February 2010
4-59
Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
Figure 4-12 Comparison of distributions for key elements of the risk equation:
Mortality risk effect estimate from Zanobetti and Schwartz (2008).
Comparison of Urban Case Study PM All-cause Mortality Risk ((3) to
U.S. Distribution of PM All-cause Mortality Risk
(212 U.S. Urban Areas)
IUU70
90% -
% 80% -
CD
< 70% -
ra 60% -
§ 50% -
% 40% -
!g 30% -
^ 20% -
10% -
n% -
^^^^ i
x '
i i 1
S
X^
-"
1 r , 1
y^
^^^
i , i
i , i
/
i i
H
1 1
^^^
f
l-i 1
1 1
/
H
T
i , ,
0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012 0.0014 0.0016
Bayesian Shrunken PM All Cause Mortality Risk Coefficient ((3)
All Z&S Urban Areas CDF ^Case Study Urban Area CDF Case Study Urban Areas
February 2010
4-60
Draft - Do Not Quote or Cite
-------
1
2
3
Figure 4-13 Comparison of distributions for selected variables expected to
influence the relative risk from PMi.s: long term average July
temperature.
Comparison of Urban Case Study Area Long Term Average July Temperature to
U.S. Distribution of Long Term Average July Temperature
(3141 U.S. Counties)
4
5
in
0)
"c
3
O
O
CO
=)
5
100% -
90% -
80% -
70% -
60% -
50% -
40% -
30% -
20% -
10% -
0% -
50
60 70 80 90
July 30 Year Average Temperature, 1941-1970
100
All Counties CDF ^Case Study Counties CDF Case Study Counties
February 2010
4-61
Draft - Do Not Quote or Cite
-------
1
2
3
Figure 4-14 Comparison of distributions for selected variables expected to
influence the relative risk from PMi.s: percent of population 65 and
older.
Comparison of Urban Case Study Area % 65 and Older to U.S. Distribution of % 65
and Older
(3141 U.S. Counties)
1 UU /O -
90% -
80% -
w 70% -
1 60% -
o
°. 50% -
2 40% -
£ 30%
20% -
1 0% -
no/.
^
^/
/*
^
f
Urban case study areas are
all below the 75th percentile
of county % of population 65
and older
10 15 20 25
% of Population 65 and Older, 2005
30
All Counties CDF ^Case Study Counties CDF Case Study Counties
35
February 2010
4-62
Draft - Do Not Quote or Cite
-------
1
2
3
Figure 4-15 Comparison of distributions for selected variables expected to
influence the relative risk from PMi.s: per capita annual personal
income.
Comparison of Urban Case Study Area Per Capita Personal Income to U.S.
Distribution of Per Capita Personal Income
(3141 U.S. Counties)
4
5
90%
80%
£ 70%
1 60%
£ 50%
2 40%
0
£ 30%
10%
1 1 1
l-l 1
1 II
1 1
l-l 1
II
1 1
^
l-l
X
i i
Urban case study areas are
all above the 25th percentile
of county per capita income
i, , 1 1
$10,000 $20,000 $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 $90,000
Per Capita Personal Income, 2005
All Counties CDF ^Case Study Counties CDF Case Study Counties
February 2010
4-63
Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Figure 4-16 Comparison of distributions for selected variables expected to
influence the relative risk from PMi.s: per capita annual personal
income.
Comparison of Urban Case Study Area Angina/CHD Prevalence to U.S. Distribution
of Angina/CHD Prevalence
(183 U.S. MSA)
CO
CO
U-
o:
m
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
L
Urban case study areas are
all below the 85th percentile
of MSA angina or CHD
prevalence
34567
% Prevalence of Angina or CHD, BRFSS, 2007
All BRFSS MSA CDF ^Case Study County MSA CDF Case Study County MSA
4.5 CONSIDERATION OF DESIGN VALUES AND PATTERNS OF PM2.5
MONITORING DATA IN INTREPRETING CORE RISK ESTIMATES
The degree of risk reduction associated with the current and alternative suites of
standards at a particular urban study area depends to a great extent on the degree of reduction in
PM2.5 concentrations simulated for that location. This in turn depends on the interplay between
the 24-hour and annual design values and the monitoring data used to characterize ambient PM2.5
concentrations, since these factors determine the composite annual average and composite 24-
hour PM2.5 profiles used in modeling long-term and short-term exposure related risk for that
study area. Because of the role that design values and underlying patterns in PM2.5 monitoring
data play in determining the degree of risk reductions, these factors can be used in helping to
interpret risk estimates generated for the 15 urban study areas under the various standard levels
considered in this risk assessment. Further, it is possible to consider, more broadly, patterns of
design values across urban areas in the U. S. and contrast these with patterns seen for the 15
February 2010
4-64
Draft - Do Not Quote or Cite
-------
1 urban study areas to help to place risk estimates for the 15 urban study areas in a broader national
2 context.
3 This section discusses consideration of patterns of design values (section 4.5.1) and
4 underlying ambient monitoring PM2.5 data (section 4.5.2) for the 15 urban study areas in the
5 context of helping to interpret risk estimates. Each of these discussions begins by describing the
6 methods used in each analysis and concludes with a set of key observations.
7 4.5.1 Design Values
8 The set of design values for an urban study area determines whether the 24-hour or
9 annual standard will be controlling as well as the degree of reduction in ambient PM2.5
10 concentrations associated with a particular suite of standards. Therefore, by plotting the
11 relationship between 24-hour and annual design values for each of the 15 urban study areas, we
12 can obtain a quick visual perspective on (a) which study areas will experience reductions in risk
13 for a particular suite of standards, (b) whether the 24-hour or annual standard will control, and
14 (c) the general magnitude of risk reduction. The last observations result from comparing the
15 controlling standard level with the matching design value, which will determine the fractional
16 reduction in PM2.5 levels at monitors exceeding the standard level (for peak shaving rollback), or
17 more broadly across all monitors (for proportional rollback).
18 Figures 4-17 through 4-19 present scatter plots of 24-hour and annual design values for a
19 combination of the 15 urban study areas (red stars) and the broader set of larger urban areas in
20 the U.S. (green circles). In addition to depicting the set of design values for these urban areas,
21 each figure also includes a set of superimposed lines representing the current suite of standards
22 (Figure 4-17) and three of the alternative suites of standards considered in the risk assessment
23 (12/35 - Figure 4-18, and 12/25 - Figure 4-19). In each figure, the horizontal line represents the
24 24-hour standard level, while the vertical line represents the annual standard level. The line that
25 intercepts the origin (i.e., the "35/15 line" in Figure 4-17) represents the point of demarcation
26 between those study areas where the 24-hour standard controls (to the left of the intercept line)
27 and those study area where the annual standard level controls (to the right of the intercept line).
28 By superimposing these lines related to the current standard level on the scatter plot, we have
29 created five zones within each figure including:
30 Zone A: 24-hour design values exceeding the 24-hour standard level, but annual design
31 values below the annual standard level (i.e., 24-hour standard is controlling). Urban study
32 areas in this zone are predicted to experience risk reduction with the degree of reduction
33 reflecting the degree to which the 24-hour design value exceeds the 24-hour standard level.
34 For example, in Figure 4-17 (depicting the current suite of standards), Tacoma and Salt Lake
35 City fall in this zone, along with 20-30 additional urban areas in the U.S.
February 2010 4-65 Draft - Do Not Quote or Cite
-------
1 ZoneB: 24-hour design values and annual design values exceed 24-hour and annual
2 standard levels, respectively, and the 24-hour standard is controlling. We have further
3 transected this zone into Bl and B2, with the former representing those urban areas with
4 notably high 24-hour design values (Fresno, Los Angeles in Figure 4-17) and B2 those with
5 lower, although still controlling, 24-hour design values (Baltimore, New York, Detroit,
6 Philadelphia, St. Louis in Figure 4-17). Those urban areas in Bl have exceptionally peaky
7 PM2.s distributions relative to urban areas in B2 (i.e., relatively high 24-hour design values
8 and lower annual average design values).
9 ZoneC: 24-hour design values and annual design values exceed 24-hour and annual
10 standard levels, respectively, and the annual standard is controlling. Atlanta, Birmingham
11 and Houston fall into this zone and represent a relatively small number of urban areas in the
12 U.S..
13 ZoneD: annual design values exceed the annual standard level, but 24-hour design values
14 are below the 24-hour standard level (i.e., annual standard is controlling). Houston is the only
15 urban study area falling into this zone for the current standard level, along with a small
16 number of additional urban areas in the U.S..
17 ZoneE: both the 24-hour and annual design values are below their respective standard levels
18 (i.e., this is the only zone where urban areas would not be expected to experience risk
19 reductions under the suite of standards being considered). The majority of urban areas in the
20 U.S. depicted in these scatter plots fall into Zone E in Figure 4-17.
21
22 The five zones presented above are useful in interpreting the risk results generated for the
23 current suite of standards (for the 15 urban study areas). Specifically, as noted above, they allow
24 us to (a) quickly identify which of the 15 urban study areas experience risk reductions under the
25 current standard level, (b) determine whether those reductions are due primarily to a controlling
26 24-hour or annual standard and (c) to see how well our set of urban study areas provide coverage
27 for the broader set of urban areas in the U.S..
28 In addition to presenting Figures 4-17 through 4-19 as a means for supporting the
29 interpretation of risk estimates generated for the 15 urban study areas (based on consideration for
30 patterns in design values), we have also included Table 4-8 for this purpose. Table 4-8 presents
31 the annual and 24-hour design values for each urban study area and also identifies which
32 standard is controlling for a given suite of standards. For example, we see that in Atlanta (which
33 has design values of 16.2 |ig/m3 and 35 |ig/m3, annual and 24-hour, respectively), the annual
34 standard controls for the current suite of standards (15/35) as well as the first 4 alternative suites
35 of standards considered (14/35, 13/35, 12/35 and 13/30). However, the 24-hour standard controls
36 for the final suite of standards considered (12/25). This matches with information presented in
37 Figures 4-17 through 4-19 (e.g., Figure 4-17 shows that the Atlanta is just inside of zone C,
38 suggesting that it meets the 24-hour standard, but not the annual standard.
39
February 2010 4-66 Draft - Do Not Quote or Cite
-------
2
3
4
5
6
7
Figure 4-17 Design values in 15 urban study areas and broader set of U.S. urban
areas relative to the current suite of standards (15/35)
70 i
65 -
60 -
55
Annual Design Value (ug/m3)
8 10 12 14 16 18 20 22 24 26 28 30
Key:
if- urban study area included in risk assessment
-MSA's within the U.S.
February 2010
4-67
Draft - Do Not Quote or Cite
-------
2
3
Figure 4-18 Design values in 15 urban study areas and broader set of U.S. urban
areas relative to the 12/35 alternative suite of standards
4
5
6
7
St. Louis
-------
2
3
4
5
6
7
Figure 4-19 Design values in 15 urban study areas and broader set of U.S. urban
areas relative to the 12/25 alternative suite of standards)
70
65
Annual Design Value (ug/m-3)
8 10 12 14 16 18 20 22 24 26 28 30
Key:
if- urban study area included in risk assessment
-MSA's within the U.S.
February 2010
4-69
Draft - Do Not Quote or Cite
-------
2
3
Table 4-8 Identification of controlling standard (24-hour or annual) for
alternative suites of standard levels
4
5
6
7
8
9
10
11
12
13
14
1 5
16
17
18
19
20
21
22
Urban study area
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA 5
Salt Lake City,
UT
St. Louis, MO
Tacoma, WA
Design Value
Annual
16.2
15.6
18.7
12.8
17.2
17.4
15.8
19.6
15.9
15.0
12.6
19.8
11.6
16.5
10.2
24-Hr
35
37
44
26
43
63
31
55
42
38
32
60
55
39
43
Combination of annual and 24-hour design values*
Current
standard levels
15/35
A
24hr
A
-
24hr
24hr
A
24hr
24hr
24hr
-
24hr
24hr
24hr
24hr
Alternative annual standard
levels
14/35
A
A
A
-
A
24hr
A
24hr
24hr
24hr
-
24hr
24hr
A
24hr
13/35
A
A
A
-
A
24hr
A
24hr
A
A
-
24hr
24hr
A
24hr
12/35
A
A
A
A
A
24hr
A
A
A
A
A
24hr
24hr
A
24hr
Combinations of
alternative 24-hour
and annual standard
levels
13/30
A
24hr
A
A
24hr
24hr
A
24hr
24hr
24hr
A
24hr
24hr
24hr
24hr
12/25
24hr
24hr
24hr
A
24hr
24hr
A
24hr
24hr
24hr
24hr
24hr
24hr
24hr
24hr
* "24hr" denotes that the 24-hour standard is controlling. "A" denotes that the annual standard is
controlling
Based on consideration of the zones defined in Figures 4-17 through 4-19, we can make
the following observations regarding potential patterns of risk reduction across urban study areas
in the U.S., given the current and alternative suites of standards considered. Further, we can
characterize the degree to which the 15 urban study areas provide coverage for these groupings
of U.S. urban study areas:
For the current suite of standards (see Figure 4-17), Based on 2005-2007 air quality data,
most urban areas in the country meet the current standards based on 2005-2007 air quality
data (zone E). A smaller but still notable number meet the current annual standard but do not
meet the current 23hr standard (Zone A). A similar number of areas do not meet either
current standard (zones B and C). Only a few areas do not meet the current annual standard,
but do meet the current 24hr standard (zone D). Of the 15 urban study areas included in the
risk assessment most fall into zones that do not meet either standard (zones B and C)
although some study areas are in each of the other zones.
Alternative suites of standards involving reduction of the annual standard levels (see Figure
4-18) Based on 2005-2007 air quality data, as shown in Figure 4-18, reduction in the annual
standard level down to 12 |ig/m3 results in a significant increase in the number of areas that
February 2010
4-70
Draft - Do Not Quote or Cite
-------
1 do not meet the annual standard (zones C and D). And of those areas, roughly similar
2 numbers of urban areas do meet the 24hr standard as do not meet the 24hr standard
3 (comparing numbers of urban areas in B and C to the number in zone D).
4 Alternative suite of standards involving reductions in both annual and 24-hour levels (see
5 Figure F-19): Based on 2005-2007 air quality data, a large fraction of urban areas are
6 predicted not to meet the 24hr standard (zones A, B and C). Furthermore, the majority of
7 these have the 24hr controlling (zone A and B). We also note that there are virtually no urban
8 areas that exceed the annual standard while not meeting the 24hr standard (zone C). Of the
9 15 urban study areas, most do not meet either the 24hr or annual standards, while the 24hr is
10 controlling in most (zone B).
11
12 4.5.2 Patterns in PM2.s Monitoring Data
13 As noted earlier, patterns in PM2 5 monitoring data for each of the 15 urban study areas
14 can be used (together with consideration of design values as described in section 4.5.1) to
15 support interpretation of risk estimates generated for current and alternative standard levels. This
16 is particularly true when considering the impact of using different rollback methods in
17 supporting risk characterization for current and alternative standard levels, as discussed below.
18 To facilitate consideration of patterns in PM2.5 monitoring data across the 15 urban study
19 areas, we have developed Figures 4-20 and 4-21. Each of these figures presents 24-hour and
20 annual design values (blue and green dots, respectively) for each PM2.5 monitor within each
21 study area. The figures also flag the highest design values for each study area (red and brown
22 stars for the annual and 24-hour standard levels, respectively).58 Each figure has been scaled to
23 represent a particular suite of standards, with Figure 4-20 scaled to represent the current suite of
24 standards (15/35) and Figure F-21 scaled to represent the 12/25 alternative suite of standards.59
25 In addition, the figures allow identification of whether a study area had the highest design value
26 (for the 24-hour and annual averaging periods) occurring at the same or at different monitors.
27 This factor can influence the degree to which simulation of a controlling 24hr standard level,
28 given application of peak shaving, results in reduction in annual-average PM2.5 levels for that
29 study area. If an area has both 24hr and annual design values occurring at the same monitor, then
30 application of peak shaving to reduce the controlling 24hr standard will also bring down the
58 Note, that it is the highest viable study-area level design values (represented as stars in the diagram) that were
used as the basis for determining the degree of rollback needed to simulate a particular standard level in the risk
assessment.
59 For example, in Figure 4-20, the left y-axis, which represents the annual standard level extends from the 15/35
line up to a maximum of 30, with this representing a factor of two spread in the annual design value (i.e., from the
current 15 up to 30). Similarly, the right hand y-axis represents the 24-hour standard level with the 15/35 line
extending from 35 to a maximum of 70 (again a factor of 2 above the current standard of 35). This allows 24-hour
and annual standard levels for a given study area to be compared directly in terms of how far they are above (or
below) the 15/35 line in order to determine which standard is controlling (i.e., the standard which is higher on the
plot).
February 2010 4-71 Draft - Do Not Quote or Cite
-------
1 annual design value (i.e., the annual-average PM2.5 level for that study area is likely to be
2 reduced to a greater extent). By contrast, if 24hr and annual design values are located at different
3 monitors, then peak shaving focused on reduction of the 24hr design value monitor will
4 potentially not impact the annual design value (i.e., there will be a smaller impact on the annual-
5 average PM2.5 level for that study area).60
6 To gain a better understanding of the information provided in Figures 4-20 and 4-21, we
7 will provide a walkthrough for one of the urban study areas, highlighting key attributes related to
8 24-hour and annual design values. With Los Angeles (in Figure 4-20) we see that the study area
9 has a relatively wide spread in 24-hour and annual design values across the monitors (i.e., it has a
10 relatively peaky PM2.5 distribution), with 24-hour values ranging from -15 to -55 and annual
11 design values ranging from -7 to -19 (exact values are presented in Appendix A). In addition,
12 we see that the 24-hour standard is clearly controlling, given how much farther the highest viable
13 24-hour design value is from the 15/35 line compared with the highest annual design value. In
14 addition, we can compare these trends in 24-hour and annual design values for Los Angeles to
15 those for the other urban study area and see that generally, Los Angeles (a) has some of the
16 widest spreads in both 24-hour and annual design values (i.e., it has one of the more peaky
17 PM2.5 distributions across monitors) and (b) has one of the highest 24-hour design value of the 15
18 urban study areas (i.e., it will require more rollback in simulating just meeting the current suite
19 of standards compared with most of the other study areas). The attributes described above match
20 well with urban areas falling into zone B1 in Figure 4-17 (i.e., the zone where urban areas do not
21 meet both the current 24-hour and annual standards, and where the 24-hour standard is
22 controlling).
60 When a star in either Figure 4-20 or 4-21 (signifying the highest design value for that study area) is placed over a
point estimate, then the highest design value (for both 24-hour and annual levels) occurs at different monitors. This
is the case, for example, with Phoenix, while Los Angeles represents a location where the highest 24-hour and
annual design values occur at the same monitor.
February 2010 4-72 Draft - Do Not Quote or Cite
-------
1
2
Figure 4-20 Annual and 24-hour design values (for individual monitors and at the study-area level) for the 15 urban
study areas (with the presentation of values scaled to reflect current standard of 15/35)
30:
28:
26:
24:
- 22 :
CO
-^ 20 :
-^
CD
D :
CO 16 :
C 14~
O)
CD 12
"co
< 6-
4:
2:
0 :
i/>
Q in
T
c
c
5 Q
3 ,_
: n
s|/ 4
'
1
1
CM
*
L
1
*!
i!
^
..
! t
li
A
it
^1
1
r
i
7
15/35 standard level line
*
A
A *
A
A A
Q
A
4
. j
^
A A
A
t
t
P
A
>
tf
A
A
t
*
* i
. *
ti
f
*
^
a
^
1
I
I
A
1
A
A
*!
l!
A
.
*
Stars (*) = Area DVsite Dots () = Other valid site DVs Triangles (A) = Invalid site DVs
A B B D D
t a i a e
1 1 r 1 t
a t m 1 r
n i i a o
t m n s i
a o g t
r R
e a
m
F H L N
r o o e
e u s »
s s
n t A Y
o o n o
n g r
e k
1
e
s
-70
-65
-60
- 55
^ ^
-50 "
-5
-45 ^
-40 ^
D
CO
-35 >
C
-30 .??
W
CD
-25 Q
b=
-20 o
1
CNJ
-10
-5
- n
P P P P S S T
h h i i a t a
i 0 t t 1 . C
1 e t t t c
a n s - L m
d i b L L o E
e x u i a u
1 r b k i
p gees
h R r
i t
a y
February 2010
4-73
Draft - Do Not Quote or Cite
-------
1
2
Figure 4-21 Annual and 24-hour design values (for individual monitors and at the study-area level) for the 15 urban
study areas (with the presentation of values scaled to reflect current standard of 12/25)
36:
34:
32:
30:
28:
(""x OC
CO ^D "
E :
-0) 24:
5 22:
CD
_D 20:
CD
^
O) 16:
CD 14:
CD
g 10:
^
< 8 :
6:
4:
2:
0 :
V)
> 10
D
o
3
C
<
4.
I
>
Q
=
o
CM
4>
[I
i-
.
*
i"
A A
ft
I!
*
j-
/
/
*
o
*
^
12/25 standard level line
*
A
V "
A
V
1
a
*!
i
\-
A
1 |
*
t
1
A
^
!'
*.
t
? *
0 Q
*
A A
^
^
.
A
I
^
S
s
!
'
*
!
*
*
Stars (*) = Area DV site Dots () = Other valid site DVs Triangles (A) = Invalid site DVs
:75
:70
:65
:60
~ 55 "f s-
CO
; E
r50 -|
-45 "-'
CD
; ^ CD
:35 c
: O)
r30 '$
: D
:20 °
H5 CN
:10
:5
- n
ABBDDFHLNPPPPSST
taiaerooehhiiata
llrlteuswiottl.c
atmlrss lettt o
niiaontAYans- Lm
t mnsi oonodi bLLoa
aog t ngrexuiau
r n e k I r b k i
e a I p gees
m e h n r
s i t
a y
February 2010
4-74
Draft - Do Not Quote or Cite
-------
1 The sensitivity analysis examining uncertainty related to conducting rollback
2 demonstrated that for some of the study areas (e.g., Los Angeles and Salt Lake City) use of the
3 peak shaving rollback method reflecting application of more localized controls resulted in
4 composite monitor values that differed notably from values generated when the proportional
5 rollback approach was used.61 In contrast, many of the other urban study areas displayed little
6 difference in composite monitor values based on application of proportional or peak shaving
7 rollback methods.
8 Design value information provided in Figures 4-20 and 4-21 provide explanations for
9 these sensitivity analysis results. For Los Angeles (which had composite monitor values 40%
10 higher when using the peak shaving rollback method compared with the proportional approach -
11 see Section 4.3.1.1), the 24-hour standard is controlling. This can be seen by noting that the
12 maximum 24-hour design value is significantly further away from the 15/35 line in Figure 4-20
13 compared with the maximum annual design value. In addition, these two maximum design
14 values do not occur at the same monitor.62 This means that when the proportional rollback
15 method is used, a relatively large fractional reduction is uniformly applied to all monitors,
16 resulting in a new (adjusted) composite monitor value that has been reduced to a relatively large
17 extent. However, if peak shaving rollback is used, then only those monitors with 24-hour design
18 values exceeding the current 24-hour standard level are adjusted and only by the fraction
19 required to get each 24-hour design value down to the current 24-hour standard level.63 This
20 means that in an overall sense, there is less adjustment to PM2.5 levels, such that with peak
21 shaving we will see higher composite monitor annual averages than with proportional rollback.
22 In the case of Salt Lake City (which also has significantly higher composite monitor
23 annual averages with peak shaving than with proportional rollback), while the highest 24-hour
24 and annual design values occur at the same monitor, which means that even with peak shaving,
25 the monitor with the highest annual averages will be adjusted downward substantially, because
26 the annual design values for monitors are closer to each other, the impact of peak shaving on the
27 composite annual average is smaller. Specifically, while some of the monitors with 24-hour
28 design values above the current 24-hour standard level will have their annual averages adjusted
61 Recall that differences in composite monitor estimates represent surrogates for differences in long-term exposure-
related mortality - long-see section 4.3.1.1.
62 In figures 4-20 and 4-21, when the max viable 24-hour and annual design values occur at the same monitor, this is
signified by showing the red stars for the max viable standard level superimposed over a green dot.
63 With the peak shaving approach, many of the monitors will not have their PM2 5 levels adjusted because their 24-
hour levels do not exceed the current standard. Furthermore, because Los Angeles has its max 24-hour and annual
standard levels occurring at different monitors, the max adjustment applied (that associated with the highest 24-hour
monitor) will not be applied to the monitor having the highest annual design value, resulting in a lower overall
impact to the composite annual average, compared with proportional rollback.
February 2010 4-75 Draft - Do Not Quote or Cite
-------
1 down, there is a fraction of the monitors (with 24-hour design values below the current standard)
2 that will not be adjusted under peak shaving.
3 These two examples illustrate different conditions under which the type of rollback
4 applied can have a significant impact on the degree of public health protection assessed for a
5 particular standard level. By contrast, conditions at some of the other urban study areas result in
6 little difference in risk from application of different rollback methods (i.e., simulation of more
7 regional versus local control strategies). Specifically, if an urban location has 24-hour and annual
8 design values at each monitor that display little variation, we expect to see less impact on risk
9 from varying the type of rollback method used. Examples that fall into this latter category
10 include Atlanta, Dallas, and St. Louis (see Figure 4-20).
February 2010 4-76 Draft - Do Not Quote or Cite
-------
1 5 NATIONAL-SCALE ASSESSMENT OF LONG-TERM MORTALITY
2 RELATED TO PM2 5 EXPOSURE
3 5.1 OVERVIEW
4 In this section we present the estimated nationwide premature mortality resulting from
5 recent exposures to ambient PM2.5. The goal of this assessment is twofold: (1) estimate the
6 incidence of premature mortality within the U.S. related to long-term PM2.5 exposure; and (2)
7 identify where the subset of counties assessed in the urban case study areas analysis fall along
8 the distribution of national county-level risk.64 To perform this assessment we use 2005 PM2 5
9 fused air quality estimates from the Community Model for Air Quality (CMAQ) (Byun and
10 Schere, 2006) in conjunction with the environmental Benefits Mapping and Analysis Program
11 (BenMAP, Abt Associates Inc, 2008) to estimate long-term PM2.5-related premature mortality
12 nationwide.
13 To address the first goal of the assessment, we estimate excess PM2.5-related long-term
14 mortality by applying two estimates of all-cause mortality risk found in the Krewski et al. (2009)
15 PM2.s mortality extended analysis of the American Cancer Society (ACS) cohort, and an estimate
16 of all-cause mortality risk found in the Laden et al. (2006) PM2 5 mortality extended analysis of
17 the Six-Cities cohort. We estimate that total PM2.5-related premature mortality ranges from
18 63,000 (39,00087,000) (95th percentile confidence interval) and 88,000 (49,000130,000),
19 respectively; in each case we estimated deaths per year down to the lowest measured levels
20 (LMLs) in each epidemiological study.
21 In addressing the second goal of this assessment, we observe that the subset of 31
22 counties for the 15 urban study areas considered in the urban case study fall toward the upper
23 end of the national distribution. Specifically, all of the 31 counties were above the median of the
24 national risk distribution and 23 of the 31 fell within the upper 5th percentile of the national
25 distribution. Therefore, according to this analysis, we appear to be capturing high-end
26 percentiles of the national risk distribution with the set of urban case study areas we are
27 evaluating in the PM2 5 NAAQS risk assessment.
28 We had considered expanding the national-scale mortality to include additional health
29 endpoints (related to short-term PM2 5 exposure) or additional air quality scenarios that simulate
30 just meeting the current and alternative suites of standards. However, as noted in section 2.3, we
64 We do not directly compare the estimated county-level risks generated in the urban case study
assessment and the county-level risks generated in the national-scale analysis. Rather, we identify where the 31
counties modeled for urban case study fell along the national risk distribution. This assessment revealed whether the
baseline PM2 5 mortality risks in the 3 Icounties modeled in the urban case study areas represented more typical or
higher-end risk relative to the national risk distribution.
February 2010 5-1 Draft - Do Not Quote or Cite
-------
1 continue to conclude that any expansion of this assessment, is beyond the scope of what is
2 needed or can reasonably be done within the time and resources available for this review. Here
3 we provide additional discussion of the rationale for our decision not to expand the scope of the
4 nati onal - seal e analy si s.
5 The goal of the national-scale analysis is two-fold: to provide perspective on the
6 magnitude of PM2.5 health impacts on a nati onal-scale and to help to place the risk estimates
7 generated for the urban study areas in a national context. The analysis as currently implemented
8 achieves the first goal by providing estimates of long-term exposure-related all-cause mortality
9 under recent conditions. While simulation of risk for the current and alternative standard levels
10 would provide additional perspective on the magnitude of national-scale risk, that assessment
11 would be resource-intensive and subject to considerably uncertainty if it were conducted using
12 air quality simulation methods similar to those used in the urban study area analysis (i.e.,
13 application of a combination of rollback methods that reflects both local and regional patterns in
14 ambient PM2.5 reductions implemented at the monitor-level). A particular area of uncertainty
15 (and technical complexity) related to air quality simulation would be addressing the interplay
16 between regional-scale reductions in ambient PM2.5 in adjacent urbanized areas. In the urban
17 study area analysis, each location is treated independently with regard to simulating ambient
18 PM2.5 under alternative suites of standards. However, if we were to expand the national analysis
19 to include alternative standards, then simulation of rollbacks in ambient PM2 5 levels would
20 necessarily have to address this contiguity issue between adjacent urban areas and even between
21 suburban areas and adjacent urbanized areas in the context of simulating monitor rollback.
22 In addition, because long-term exposure-related mortality dominates PM2 5 in terms of
23 total incidence, providing coverage for this endpoint category ensures that the majority of PM2.5-
24 related mortality incidence is reflected in the analysis, without including short-term exposure-
25 related mortality.
26 The national-scale mortality analysis, as currently implemented, also achieves its second
27 goal: to help place risk estimates for the urban study areas in a national context. Because the
28 national-scale analysis focuses on the long-term exposure-related mortality, which is the primary
29 driver for PM2.s-related health impacts, the analysis allows us to assess how the urban study
30 areas "fall" across a national distribution of risk for this key health endpoint category (see
31 discussion below). This then allows us to characterize the degree to which the set of urban study
32 areas provides coverage for areas of the country likely to experience relatively elevated levels of
33 PM2.5-related health impacts.
February 2010 5-2 Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
11
13
14
15
16
17
18
19
20
21
22
23
24
25
5.2 METHODS
This assessment combines information regarding estimated PM2.s air quality levels,
population projections, baseline mortality rates, and mortality risk coefficients to estimate PM2.5-
related premature mortality. Figure 5-1 below provides a conceptual diagram, detailing each of
the key steps involved in performing this BenMAP-based health impact assessment. Appendix
G contains additional information regarding the data inputs to this analysis.
Figure 5-1 Conceptual diagram of data inputs and outputs for national long-term
mortality risk assessment
Census Population
Data
Air Quality
Monitoring
Mortality Functions
Population
Estimates
Population
Exposure
PM25-ReIated /
Premature Mortality
Population
Projections
Air Quality
Modeling
Incidence and
Prevalence Rates
5.2.1 Population Estimates
The starting point for estimating the size and demographics of the potentially exposed
population is the 2000 census-block level population, which BenMAP aggregates up to the same
grid resolution as the air quality model. Using county-level growth factors based on economic
projections (Woods and Poole Inc., 2001), BenMAP projects this 2000 population to the analysis
year of 2005; we selected this population year because it matches both the year in which the
emissions inventory was developed for the air quality modeling and the year to which the
baseline mortality rates were projected (see below).
5.2.2 Population Exposure
Having first estimated the size and geographic distribution of the potentially exposed
population, BenMAP then matches these population projections with estimates of the ambient
levels of PM2.5. In contrast to the urban study areas analysis, the national-scale analysis
February 2010
5-3
Draft - Do Not Quote or Cite
-------
1 employed a data fusion approach, which joined 2005 monitored PM2.5 concentrations with 2005
2 CMAQ-modeled air quality levels using the Voronoi Neighbor Averaging (VNA) technique
3 (Abt, 2003). CMAQ was run at a horizontal grid resolution of 12km for the east and 36km in the
4 west using 2005 estimated emission levels and meteorology. More information on this model
5 run can be found in Appendix G of this document. Figure 5-2 shows the geographic distribution
6 of baseline annual mean PM2.5 concentrations across the continental U.S. The maximum
7 predicted value within the U.S. is 31 |ig/m3, the mean PM2.5 value is 8.7 |ig/m3, median is 8.8
8 ng/m3 and the 95th percentile value is about 14 |ig/m3.
9
11 Figure 5-2 2005 fused surface baseline PM2.s concentrations
2005 Fused Surface Baseline Concentrations (ug/m3)
^B 1.03 to 4.2
^H 4.3 to 6.5
6.6 to 9.34
9.35 to 12,30
^H 1131 to 20.57
20.58 to 59.42
12
13
14
15
16
This assessment applies PM2 5 mortality risk coefficients drawn from long-term cohort
studies which estimate changes in risk based on annual mean changes in PM2.5 concentration.
For this reason, EPA used the CMAQ model to estimate annual mean concentrations at each grid
cell. These grid-level annual average concentrations were then input to BenMAP.
February 2010
5-4
Draft - Do Not Quote or Cite
-------
1 5.2.3 Premature Mortality Estimates
2 In this assessment of PM2.5-related premature mortality we considered risk estimates
3 drawn from studies based on two prospective cohorts. The first study is the recently published
4 Krewski et al. (2009) extended reanalysis of the ACS cohort. To remain consistent with the
5 urban study areas analysis, we applied the two log-linear all-cause mortality risk coefficients
6 based on the 1979-1983 and the 1999-2000 time periods that control for 44 individual and 7
7 ecologic covariates. We also applied a log-linear all-cause mortality risk coefficient drawn from
8 the extended analysis of the Six Cities cohort as reported by Laden et al. (2006). When
9 estimating premature mortality using these functions we considered air quality levels down to the
10 lowest measured levels (LML) in each study; for the Krewski et al. (2009) study this is 5.8 |ig/m3
11 and for the Laden et al. (2006) study this is 10 |ig/m3. In general, we place a higher degree of
12 confidence in health impacts estimated at air quality levels at or above the LML because the
13 portion of the concentration-response curve below this point is extrapolated beyond the observed
14 data. We also estimated health impacts down to Policy Relevant Background (PRB) levels
15 (EPA, 2008). The final ISA presents estimates of annual mean PRB for each of 7 Health Effects
16 Institute PM regions; this value ranges from 0.62 |ig/m3 in the southwest to 1.72 |ig/m3 in the
17 southeast.
18 BenMAP contains baseline age-, cause- and county-specific mortality rates drawn from
19 the CDC-WONDER. Current baseline mortality estimates are an average of a three year period
20 from 1996-1998. EPA is in the process of updating these rates with 2006-2008 data; a sensitivity
21 analysis suggests that the results reported here are largely insensitive to the use of more current
22 mortality rates.
23 5.3 RESULTS
24 Table 5-1 and figures 5-3 through 5-4 below summarize the results of the national-scale
25 analysis. Table 5-1 summarizes the total PM2.s-related premature mortality associated with
26 modeled 2005 PM2.5 levels.
27 Estimated PM2.5 -Related Premature Mortality Associated with Incremental Air Quality
28 Differences Between 2005 Ambient Mean PM2 5 Levels and LML from the Epidemiology
29 Studies or PRB (90th percentile confidence interval)
30
31
February 2010 5-5 Draft - Do Not Quote or Cite
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Table 5-1 Estimated PM2.s-related premature mortality associated with
incremental air quality differences between 2005 ambient mean
PM2.s levels and lowest measured level from the epidemiology studies
or policy relevant background (90th percentile confidence interval)
Air Quality
Level
10 ug/m3 (LML
for Laden et al.,
2006)
5.8 ug/m3 (LML
for Krewski et
al., 2009)
Policy-Relevant
Background
Estimates Based on Krewski et al. (2009)
'79- '83 estimate
(90th percentile
confidence interval)
26,000
(16,00036,000)
63,000
(39,00087,000)
110,000
(68,000150,000)
'99-' 00 estimate
(90th percentile
confidence interval)
33,000
(22,000^44,000)
80,000
(54,000110,000)
140,000
(94,000180,000)
Estimates Based on
Laden et al. (2006)
(90th percentile
confidence interval)
88,000
(49,000130,000)
210,000
(120,000300,000)
360,000
(200,000500,000)
Bold indicates that the minimum air quality level used to calculate this estimate corresponds to the
lowest measured level identified in the epidemiological study
In this table, the bold figures indicate the estimate that corresponds with the LML
identified in the epidemiological study. The bold estimates in the column Krewski et al. (2009)
were calculated using the same risk coefficients as the urban case study analysis. We place a
greater emphasis on those results calculated using the LML reported in the epidemiological
studies.65 Figure 3 illustrates the percentage of baseline mortality attributable to PM2.5 exposure
in each of the grid cells according to the 2005 PM2.5 air quality levels, using the Krewski et al.
(2009) estimate based on 1999-2000 air quality levels.
Note, that as stated in Section 4.3.2, modeling of risk down to PRB is subject to considerable uncertainty. While
there is no evidence for a threshold (which conceptually supports estimation of risk below LML), we do not have
information characterizing the nature of the C-R function for long-term mortality below the LML and consequently
estimates of mortality based on incremental exposure below LML (and down to PRB) is subject to greater
uncertainty.
February 2010
5-6
Draft - Do Not Quote or Cite
-------
1 Figure 5-3
2
Percentage of premature mortality attributable to PM2.s exposure at various
2005 annual average PM2.s levels*
n=12
n=13,14S
10 to 15 u g/m 3 15 to 2 0 u g/ m 3
2005 Average Ambient Baseline PM25 Levels
Attributable mortality calculated usin^Krevvski et ai. (2009) risk estimate based on '99-'00 follovv-up period.
n= number of 12km or 36km grid cells at each baseline air qualitylevel
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
This figure illustrates the number of deaths attributable to PM2.5 according to the baseline
level of ambient average PM2.5 levels down to 5.8 |ig/m3 (the LML for the Krewski et al. (2009)
analysis). Each of four box plots characterizes the range of premature mortality attributable to
PM2.5 according to the baseline level of annual mean PM2.5 levels in that model grid cell. Note
that while the lower whisker of the box plots for the baseline air quality values of 5.8 |ig/m3 to 10
|ig/m3 appear to extend to zero, the minimum value is greater than zero. The number above each
box plot indicates the number of grid cells summarized by that plot.
Figure 5-4 displays the cumulative distribution of total mortality attributable to PM2.5
exposure at the county level developed as part of the national-scale analysis. The location of the
31 counties included in the urban case study analysis is then superimposed on top of the
cumulative distribution.
February 2010
5-7
Draft - Do Not Quote or Cite
-------
1
2
3
5
Figure 5-4 Cumulative distribution of county-level percentage of total
mortality attributable to PM2.s for the U.S. with markers
identifying where along that distribution the urban case study area
analysis fall*
M arker klentif ie $ -.v he re, and
how many, counties considered
by the Risk and Exposure
Assessment fall along the
distribution of national baseline
mortality risk.
Percentage of mortality attributable to PM^ ? exposure
'AtmbuUbUmartallly
-------
1 6 INTEGRATIVE DISCUSSION OF URBAN CASE STUDY ANALYSIS
2 OF PM2 s-RELATED RISKS
3 This chapter provides an integrative discussion of the risk-related analyses presented
4 throughout this second draft RA, including the PM2.s-related risk estimates generated for the set
5 of urban study areas and the related uncertainty and sensitivity analyses, the representativeness
6 analyses, and the national-scale long-term exposure PM2.5 mortality assessment. The goal of this
7 integrative discussion is to inform our understanding of important policy-relevant risk-based
8 questions, including: (a) what is the magnitude of risk likely to remain if the urban study areas
9 were just meeting the current suite of PM2.5 standards, and what level of confidence do we have
10 in those estimates?; (b) what is the degree and nature of risk reduction likely to be associated
11 with just meeting the alternative suites of annual and 24-hour PM2.5 standards considered in this
12 risk assessment, and what roles do the annual and 24-hour standards play in bringing about such
13 reductions?; and c) what is the distribution of risks associated with recent PM2 5 air quality in
14 areas across the U.S., and how representative are the risk results for the urban study areas from a
15 national perspective?
16 In addressing the risk-based questions listed above, we have placed primary focus on risk
17 associated with long-term exposure to PM2.5. This choice reflects the fact that long-term
18 exposure to PM2 5 has been shown in this and previous quantitative risk assessments to produce
19 substantially larger mortality risk (in terms of overall incidence and percent of total mortality)
20 compared with short-term PM2 5 exposure. Because of the emphasis placed on long-term PM2 5
21 exposure-related mortality risk, the risk assessment has been designed to generate robust
22 estimates for this risk category, including comprehensive analysis of uncertainty. For the
23 assessment of mortality and morbidity risks related to short-term PM2 5 exposure, the assessment
24 of uncertainty and its impact on risk estimates has been more limited.
25 In characterizing risks associated with both long- and short-term exposure to PM2.5
26 throughout this document, we have included those health endpoints for which sufficient
27 information was available to generate quantitative risk estimates with a reasonable degree of
28 confidence. It is important to emphasize that beyond the health endpoints evaluated
29 quantitatively in this risk assessment, there is an array of additional health endpoints potentially
30 associated with PM2.5 that will be discussed as part of the evidence-based considerations
31 presented in the policy assessment now being prepared..
32 The following discussion begins with a summary of analytical approaches used in this
33 quantitative risk assessment, emphasizing the degree of confidence we have in the data, models,
34 and assumptions we have used in developing our core risk estimates and in the results of our
35 sensitivity analyses (section 6.1). We then summarize our core risk results for the urban study
February 2010 6-1 Draft - Do Not Quote or Cite
-------
1 areas, and the confidence we have in those results in light of our uncertainty and variability
2 analyses, and provide insights into how those results inform the policy-relevant considerations
3 described above (section 6.2). Next we place these results into a national perspective (section
4 6.3). In so doing, we provide insights into how well the set of urban study areas represent the
5 broader set of urban areas in the U.S. likely to experience increased risk from PM2.5. We also
6 integrate the results from the urban study areas with the national-scale mortality assessment to
7 provide insights into the degree to which the PM2.5-related risks estimated in the urban study
8 areas are likely to be characteristic of risks in the broader U.S. population. Finally, in section
9 6.4, we highlight key points that address the policy-relevant questions that began this chapter.
10 6.1 KEY ANALYTICAL ELEMENTS IN THIS RISK ASSESSMENT
11 This quantitative risk assessment has been designed to generate estimates of risk for a set
12 of urban study areas likely to represent those urban areas in the U.S. experiencing higher PM2.5-
13 related risk due to elevated PM2.5 levels and/or other attributes related to PM2.5 risk (e.g.,
14 meteorology, baseline health effects incidence rates, differences in PM2.5 emissions sources and
15 composition).66 In addition, the risk assessment is designed to produce robust risk estimates that
16 reflect consideration of the latest research into PM2.s-related exposure and risk. To achieve these
17 goals, a deliberative process has been used in specifying each of the analytical elements
18 comprising the risk model, including selection of urban study areas as well as specification of
19 other inputs such as C-R functions. This deliberative process involved rigorous review of
20 available literature addressing both PM2.5 exposure and risk combined with the application of a
21 formal set of criteria to guide development of each of the key analytical elements in the risk
22 assessment. In addition, the risk assessment design reflects consideration of CASAC and public
23 comments on the initial risk assessment plan, as well as the first draft risk assessment. The
24 application of this deliberative process increases overall confidence in the risk estimates by
25 insuring that the estimates are based on the best available science and data characterizing PM2.5
26 exposure and risk, and that they reflect consideration of input from experts on PM exposure and
27 risk through CASAC and public reviews.
28 The approach used in specifying several of the key analytical elements used in the risk
29 assessment is highlighted below for purposes of illustrating the systematic approach used in
30 developing the model.
66 As discussed in section 3.3.2, the seven PM regions were designed to capture regional differences in factors
potentially related to PM risk. By providing coverage for these regions with the set of urban study areas selected, we
have provided some degree of coverage for regional differences in attributes potentially related to PM risk. In
addition, the representativeness analysis discussed in section 4.4 also allowed us to assess the degree to which the
set of urban study areas captured key patterns in PM risk-related attributes across urban areas in the U.S..
February 2010 6-2 Draft - Do Not Quote or Cite
-------
1 Selection of the 15 urban study areas included consideration of (a) whether a city of county
2 had been included in multi-city epidemiology studies used in specifying C-R functions used
3 in the core risk estimates, (b) providing coverage for urban areas with relatively high annual
4 and 24-hour design values, and (c) providing coverage for the seven PM regions which
5 reflect differences in key PM risk-related attributes (e.g., meteorology, demographic
6 attributes, PM sources and composition). See section 3.3.2 for additional detail on selection
7 of study areas.
8 Simulation of ambient PM^ levels under current and alternative standard levels included
9 application of the proportional rollback approach used in previous risk assessments, which
10 generally represents regional patterns of reductions in ambient PM2.5 concentrations.
11 Recognizing that simulating regional patterns in ambient PM2.5 reductions alone does not
12 capture the potential variability in future patterns of reductions that may occur, we also
13 considered alternative rollback approaches, including hybrid and peak shaving approaches.
14 Both of these approaches simulate more localized patterns of ambient PM2 5 reductions
15 combined with additional regional patterns of reductions in ambient PM2.5. Including these
16 three rollback approaches allowed us to assess the degree to which differences in the spatial
17 pattern of ambient PM2.5 reductions resulting from simulations of just meeting current and
18 alternative suites of PM25 standards can impact risk profiles.
19 Selection of health endpoints reflected consideration of the degree of support in the literature
20 for a causal relationship between PM2.5 exposure and the health effect of interest as assessed
21 in the ISA, together with consideration of the health significance of the endpoint. In addition,
22 we considered whether sufficient information existed in the literature to develop C-R
23 functions and whether we could obtain the baseline incidence data necessary to generate risk
24 estimates with a reasonable degree of confidence (see section 3.3.1).
25 The selection of epidemiological studies and specification of C-R functions for use in
26 modeling risk for these endpoints involved a rigorous review of existing literature based on
27 application of criteria we identified for specifying robust C-R functions. These criteria took
28 into account both study design as well as the potential scope of the C-R functions that could
29 be drawn from the studies (e.g., geographic coverage, demographic groups covered and
30 health endpoints involved). We outlined our rationale for the set of epidemiology studies we
31 selected and the choices made in specifying C-R functions, and we discussed our rationale
32 for not including other potential studies and/or forms of C-R functions in the risk assessment.
33 The systematic approach described above resulted in a core risk model which included
34 those model inputs that in our judgment have the greatest degree of support in the literature.
35 These core risk estimates are emphasized in addressing the policy-related questions outlined
36 above. To provide a more comprehensive assessment of risk for the urban study areas, we have
37 included an assessment of uncertainty and variability and their impact on the core risk estimates
38 as part of this analysis. This assessment of uncertainty and variability includes both qualitative
39 and quantitative elements, the latter taking the form of single- and multi-factor sensitivity
February 2010 6-3 Draft - Do Not Quote or Cite
-------
1 analysis.67 The goal of these assessments was to evaluate the robustness of the core risk
2 estimates given identified sources of uncertainty and variability. Inclusion of the qualitative
3 analysis of uncertainty, in additional to the sensitivity analyses, helped insure that a more
4 complete list of potentially important sources of uncertainty was considered in the risk
5 assessment and not only those sources for which it is possible to conduct a sensitivity analysis.
6 The assessment of uncertainty and variability completed for this analysis is more
7 comprehensive than had been done for previous risk assessments. This reflects, in part, the
8 development of methods by EPA staff to address potentially important sources of variability and
9 uncertainty. For example, to more fully explore potential variability in the patterns of reductions
10 in ambient PM2.5 that may occur upon just meeting the current and alternative suites of standards,
11 we incorporated as part of the sensitivity analysis two additional rollback approaches (hybrid and
12 peak shaving) in addition to the proportional rollback used in the core analyses. In addition,
13 recently published literature has allowed us to more rigorously examine the impact of uncertainty
14 related to specifying C-R functions for long-term exposure-related mortality (i.e., the Krewski et
15 al., 2009 study which provided extensive analysis of alternative model specifications for
16 mortality which could be readily incorporated into our sensitivity analysis).68
17 In addition to enhanced sensitivity analyses, we also included a number of national-scale
18 assessments that had not been done in past risk assessments (i.e., the representativeness analysis
19 and national-scale assessment of long-term mortality). These national-scale assessments allowed
20 us to more fully consider the degree to which the selected urban study areas are representative of
21 the broader set of urban areas within the U.S., thereby allowing us to place risk estimates for the
22 urban study areas in the broader national context.
23 6.2 INTERPREATION OF URBAN STUDY AREA RESULTS
24 This section describes the core risk estimates generated for the 15 urban study areas,
25 focusing on the policy-relevant questions outlined above. An important factor to consider in
26 interpreting these results is that the magnitude of both long- and short-term exposure-related risk
27 depends primarily on annual-average PM2.5 concentrations. Furthermore, reductions in both
28 categories of risk, as we consider simulating just meeting alternative suites of standards, also
29 depend on changes in annual-average PM2.5 concentrations.
67 As discussed in section 4.1, available information did not support a full probabilistic analysis of uncertainty and
variability in the risk model and consequently, a combination of single- and multi-factor sensitivity analyses was
used to assess the potential impact of these factors on core risk estimates.
68 Given increased emphasis placed in this analysis on long-term exposure-related mortality, the uncertainty analyses
completed for this health endpoint category are somewhat more comprehensive than those conducted for short-term
exposure-related mortality and morbidity, which to some extent reflects limitations in study data available for
addressing uncertainty in the later category.
February 2010 6-4 Draft - Do Not Quote or Cite
-------
1 The role of annual-average ambient PM2.5 concentrations in driving long-term exposure-
2 related risk is intuitive given that this risk category is modeled using the annual-average air
3 quality metric.69 The fact that short-term exposure-related risk is also driven by changes in long-
4 term average PM2.5 concentrations is less intuitive, since changes in average daily PM2.5
5 concentrations are used to estimate changes in risk for this category.70 Analyses in previous PM
6 NAAQS risk assessments have shown that short-term exposure-related risks are not primarily
7 driven by the small number of days with PM2.5 concentrations in the upper tail of the air quality
8 distribution, but rather by the large number of days with PM2 5 concentrations at and around the
9 mean of the distribution. Consequently, consideration of changes in annual-average PM2.5
10 concentrations will explain to a large extent changes in short-term exposure-related risk.
11 Therefore, in interpreting patterns of long-term exposure-related risk, and the similar patterns we
12 observe in short-term exposure-related risk, we focus primarily on how simulating just meeting
13 specific suites of PM2 5 standards impacts the annual-average PM2 5 concentration for the study
14 areas.
15 In the case of simulating just meeting the current and alternative annual standards, this is
16 straight forward, since the simulation produces a direct change in the annual-average PM2 5
17 concentration. However, simulating just meeting the current and alternative 24-hour standards
18 has a less direct effect on annual average PM2 5 concentrations across study areas, which depends
19 on a number of factors, including: (a) the type of rollback used to simulate just meeting the
20 current or alternative standards, (b) the combination of 24-hour and annual design values in each
21 study area (Table 4-8), and (c) the pattern of PM2 5 monitoring data across each study area. If
22 proportional rollback is used, the annual-average PM2 5 concentrations will be reduced by the
23 same percentage as was needed to lower the 24-hour design value to the level of the controlling
24 24-hour standard. However, our sensitivity analysis examining alternative rollback methods
25 showed that application of a peak shaving rollback approach (reflecting more localized patterns
26 of PM2 5 reductions) can, under certain circumstances, produce notably smaller changes to annual
27 average concentrations, which in turn, translate into smaller changes in both long-term and short-
28 term exposure-related risks. Specifically, for those urban study areas where a peak shaving
69 As noted in section 3.2.1, estimates of long-term exposure-related mortality are actually based on an average
annual PM25 level across monitors in a study area (i.e., the composite monitor annual-average). Therefore, in
considering changes in long-term exposure-related mortality, it is most appropriate to compare composite monitor
estimates generated for a study area under each suite of standards. The maximum monitor annual-average for a
study area (i.e., the annual design value) determines the percent reduction in PM2 5 levels required to attain a
particular standard. Both types of air quality estimates are provided in Tables F-49 and F-50 in Appendix F and
both are referenced in this discussion of core risk estimates, as appropriate.
70 Estimates of short-term exposure-related mortality and morbidity are based on composite monitor daily PM2 5.
concentrations. However, similar to the case with long-term exposure-related mortality, it is the maximum monitor
98th percentile 24-hour concentration (the 24-hour design value) that will determine the degree of reduction required
to meet a given 24-hour standard.
February 2010 6-5 Draft - Do Not Quote or Cite
-------
1 rollback approach was applied to a PM2.5 distribution that was more "peaky" in nature (i.e.,
2 relatively high 24-hour design values and lower annual average design values), the resulting
3 change in annual-average PM2 5 concentrations was notably smaller than when proportional
4 rollback was used.71 We note also that an additional factor introducing variation in risk across
5 urban study areas is the relationship between the annual-average PM2.5 concentrations at the
6 maximum monitor and the composite monitor, which varies across study areas. For this reason,
7 two study areas that are simulated to just meet the same annual standard (and consequently will
8 have the same adjusted maximum monitor annual-average PM2.5 concentration) can have notably
9 different composite monitor values.
10 In discussing the core risk estimates below, we focus on cardiovascular-related endpoints
11 given the greater overall degree of confidence assigned to this category in the ISA relative to
12 other health effect categories (e.g., respiratory-related effects). This means that for long-term
13 exposure-related risk, we focus our discussion on IHD-related mortality; the related categories
14 for short-term exposure-related risk include CV-related mortality and morbidity (the latter in the
15 form of HA related to CV symptoms).
16 Finally, we note that the set of urban study areas selected for this assessment reflect the
17 profile of urban areas in the U.S. with regard to the mix of annual and 24-hour design values. As
18 illustrated in Figure 4-18, only a few urban areas have controlling annual standard levels
19 exceeding the current standard level (i.e., fall into zones C or D in Figure 4-18). Therefore, there
20 are not a large number of areas that will experience risk reductions due to simulation of the
21 current annual standard alone. By contrast, there are a lot more urban areas in the U.S. in which
22 the 24-hour standard is controlling and the 24-hour design value exceeds the level of the current
23 standard (i.e., fall into zones A and B in Figure 4-18). Therefore, more of the urban study areas
24 available for analysis are likely to see risk reductions under the current suite of standards driven
25 by simulation of the 24-hour standard. Recognition of the profile of urban areas in the U.S. with
26 regard to the interplay between the 24-hour and annual design values is important in fully
27 understanding the core risk estimates summarized below and how those risk estimates can be
28 interpreted in the national context.
29 The discussion below is organized as follows. First, we present observations regarding
30 core risk estimates generated for the current suite of standards. We then present observations
71 The results of the sensitivity analysis examining the hybrid rollback approach, which represents a combination of
an initial localized pattern of ambient PM2 5 reduction, followed by a more regional pattern of reduction, showed this
approach not to vary substantially from the proportional approach in terms of its impact on annual-average PM2 5
concentrations and consequently risk (i.e., the peak shaving rollback method was found to result in more substantial
differences in annual-average PM2 5 concentrations and consequently risk, relative to the proportional) (see section
4.3.1.1). Therefore, in discussing the results of the sensitivity analysis examining rollback, we focus here on
contrasting results for the proportional approach with those for peak shaving.
February 2010 6-6 Draft - Do Not Quote or Cite
-------
1 related to simulation of alternative annual standards at levels of 14, 13, and 12 |ig/m3 in
2 conjunction with the current 24-hour standard (35 |ig/m3). Finally, we discuss simulation of
3 alternative suites of standards involving combinations of alternative annual and 24-hour levels
4 (i.e., an annual standard of 13 |ig/m3 paired with a 24-hour standard of 30 |ig/m3 (denoted as the
5 13/30 suite of standards); an annual standard of 12 |ig/m3 paired with a 24-hour standard of 25
6 |ig/m3 (denoted as the 12/25 suite of standards).
7 6.2.1 Simulation of Just Meeting the Current Suite of PM2.s Standards
8 In characterizing PM2.5-related risks likely to remain upon just meeting the current PM2.5
9 annual and 24-hour standards in the 15 areas included in this assessment, we focus on the 13
10 areas that would not meet the current standards based on recent (2005-2007) air quality. These
11 13 areas have annual and/or 24-hour design values that are above the levels of the current
12 standards (Table 4-8).72 Based on the core risk estimates for these areas presented above in
13 section 4.2.1, we make the following general observation regarding the magnitude of risk
14 remaining upon simulation (using proportional rollback) of just meeting the current suite of
15 standards:
16 Long-term exposure-related mortality: Total incidence of long-term exposure-related IHD
17 mortality attributable to PM2.5 ranges from 15-20 deaths per year (Salt Lake City) to 1,760-
18 2,220 deaths per year (New York) (Table 4-1). This translates into a percent of total
19 mortality incidence attributable to PM2 5 ranging from 3.7-4.7% (Tacoma) to 13.2-16.7%
20 (Atlanta) (Table 4-2).
21 Variability in incidence estimates is obviously driven in large part by differences in the
22 population in each study area, as well as by other factors such as differences in baseline
23 incidence rates and in exposure patterns. Substantially less variability would be expected in
24 estimates of the percent of total mortality attributable to PM2 5 when each area is simulated to
25 just meet the current suite if standards, since this risk metric should normalize for population and
26 baseline incidence rates. Nonetheless, we see appreciable variability across study areas for this
27 risk metric as well.
28 In considering the source of this variability, we recognize that, as noted above, the
29 magnitude of long-term PM2 5 exposure-related mortality estimated to remain upon just meeting
30 the current suite of standards depends directly on the annual-average PM2 5 concentrations that
31 result from the simulated changes in air quality patterns. In the case of the three urban study
32 areas out of the 13 experiencing risk reductions in which the annual standard is controlling
33 (Atlanta, Birmingham, and Houston), simulation of the current suite of standards results in
34 virtually the same annual-average PM2.5 concentration (-15 |ig/m3) and, consequently, estimates
72 Of the 15 study areas, only Dallas and Phoenix have both annual and 24-hour design values below the levels of
the current standards based on 2005-2007 air quality.
February 2010 6-7 Draft - Do Not Quote or Cite
-------
1 of the percent of IHD-related mortality attributable to PM2.5 for these study areas is similar
2 (Table 4-2).73
3 However, the remaining 10 study areas in which the 24-hour standard is controlling
4 display substantially greater variability in this risk metric when the proportional rollback
5 approach is applied for the core analysis. This results because the simulation of just meeting the
6 current 24-hour standard produces varying impacts on annual-average PM2.5 concentrations. For
7 example, the urban study area with the highest estimated risk remaining upon just meeting the
8 current suite of standards (Baltimore, with 11.7 to 14.7% of total mortality incidence attributable
9 to PM2.5 - Table 4-2) has annual and 24-hour design values very close to the current suite of
10 standard levels (Table 4-8). Therefore, simulating just meeting the current 24-hour standard does
11 not much change the annual-average PM2 5 concentration, which is fairly close to 15 |ig/m3, and
12 therefore, long-term exposure-related IHD mortality (as a percent of total incidence) is reduced
13 only by a very small amount below that estimated for recent air quality. In contrast, Salt Lake
14 City, which has one of the lowest estimates of the percent of total mortality incidence attributable
15 to PM2.5 upon just meeting the current suite of standards (2.9 to 3.7% of total incidence - Table
16 4-2), has a relatively low annual design value (11.6 |ig/m3) and a relatively high 24-hour design
17 value (55 |ig/m3) (Table 4-8). Therefore, simulating just meeting the current 24-hour standard
18 results in a substantial change in the annual average, using proportional rollback, since the same
19 fractional reduction required to get the 24-our design value to meet the current standard (i.e., a
20 35% reduction) is applied to the annual design value of 11.6 |ig/m3, resulting in an annual
21 average of 7.7 |ig/m3. These two examples illustrate the varying impact that the 24-hour
22 standard, if controlling, can have on annual-average PM2 5 concentrations and consequently on
23 the magnitude of long-term (and short-term) PM2.5 exposure-related mortality associated with
24 just meeting the current suite of standards.74
25 As discussed above, the sensitivity analysis examining alternative rollback approaches
26 showed that in instances where PM2.5 distributions are relatively peaky, application of peak
27 shaving (reflecting more localized patterns of ambient PM2 5 reductions) can result in a
28 controlling 24-hour standard having a substantially smaller impact on annual-average PM2.5
29 concentrations. Sensitivity analysis results for the examples referenced above (Baltimore and
30 Salt Lake City) illustrate this issue related to application of alternative rollback methods. In the
31 case of Baltimore, which has a less peaky PM2.5 distribution (in that its 24-hour and annual
73 Although, as noted earlier, composite monitor annual-averages will display differences across urban study areas,
even in those cases where the maximum monitor annual-average has been adjusted to meet the same annual standard
(see Table F-49 in Appendix F).
74 As noted above, variation in the relationship between the maximum monitor annual-average and the composite
monitor annual-average across study areas adds an additional degree of variability to the estimated long-term
exposure-related mortality seen across the 10 study areas.
February 2010 6-8 Draft - Do Not Quote or Cite
-------
1 design values are both fairly close to the current suite of standard levels), application of peak
2 shaving in simulating just meeting the current suite of standards resulted in an annual average
3 only slightly higher than that simulated using proportional rollback (i.e., 15.2 |ig/m3 compared
4 with 14.8 |ig/m3 - Table F-49). This means that long-term exposure-related IHD mortality for
5 Baltimore would be relatively similar if either proportional or peak shaving rollback approaches
6 were applied. In contrast, application of peak shaving in Salt Lake City resulted in annual-
7 average concentrations substantially higher than those simulated by proportional rollback (i.e.,
8 10.8 |ig/m3 compared with 7.7 |ig/m3, respectively - Table F-49). Therefore, for this study area,
9 use of peak shaving rollback would result in estimates of IHD mortality risk that are larger than
10 with proportional rollback (i.e., >50% higher than with proportional rollback - Table F-49).
11 These examples further illustrate that variability in the pattern of estimated reductions in ambient
12 PM2.5 concentrations based on simulation of just meeting the current suite of standards can result
13 in quite different percentage reductions in long-term PM2.5 exposure-related mortality.
14 Additional sensitivity analyses considering sources of uncertainty impacting the core risk
15 estimates focused on specification of the C-R function for long-term PM2.5 exposure-related
16 mortality. This analysis suggested that most of the alternative model specifications supported by
17 available literature would produce risk estimates that were higher (by up to a factor of 2 to 3)
18 than the core risk estimates. These findings would apply both to estimates of PM2.5-attributable
19 IHD mortality incidence, as well as to estimates of the percent of total IHD mortality incidence
20 attributable to PM2.5 exposure.
21 Taken together, the sensitivity analyses completed for this risk assessment, including
22 those considering variability in rollback methods as well as uncertainty in the form of C-R
23 functions, suggest that the set of alternative risk model specifications that we identified generally
24 produced risk estimates that are higher than the core risk estimates. Furthermore, our decision to
25 model risk down to the LML (rather than to lower PRB levels) for long-term PM2.5 exposure-
26 related mortality, despite the lack of evidence for a threshold, results in lower estimates of risk
27 that would have resulted from modeling risk down to PRB. These considerations increase our
28 overall confidence that we did not over-state risks with the core risk estimates.
29 In considering the results of the quantitative sensitivity analyses summarized above, we
30 note that the qualitative analysis of uncertainty did identify areas of ongoing research which
31 could impact risk estimates, including: (a) more refined characterization of intra-urban variability
32 in ambient PM2.5 concentrations and the resulting impact on risk characterization and (b)
33 consideration of specific components within the mix of PM2.5, including regional differences in
34 composition, and potential implications for risk characterization. These considerations introduce
35 further uncertainty into the overall risk assessment, although we do not believe that these
February 2010 6-9 Draft - Do Not Quote or Cite
-------
1 additional sources of uncertainty are likely to alter the fundamental observations resulting from
2 the core risk assessment of the current suite of standards.
3 6.2.2 Simulation of Just Meeting Alternative Annual Standards
4 In characterizing PM2.s-related risks associated with simulation of the alternative annual
5 standards, we estimate both the magnitude of risk reductions (relative to risk remaining upon just
6 meeting the current suite of standards) as well as the magnitude of risk remaining upon just
7 meeting the alternative standards. In discussing these risks, we focus on the set of urban study
8 areas experiencing risk reductions under each alternative annual standard.
9 Based on the risk estimates for these areas presented in section 4.2.2 and in Appendix E,
10 we make the following general observations regarding the magnitude of risk remaining upon
11 simulation (using proportional rollback) of just meeting the alternative annual standards (in
12 combination with the current 24-hour standard):
13 Patterns of risk reduction across alternative annual standard levels: There is a consistent
14 pattern of increasing risk reduction with decreasing alternative annual standard levels, both in
15 terms of the number of study areas experiencing risk reductions and the magnitude of those
16 reductions. Specifically, 5 of the 15 urban study areas experience risk reductions under the
17 alternative annual standard level of 14 |ig/m3, with percent reductions in PM2.5-attributable
18 long-term exposure-related mortality ranging from 9% (Baltimore) to 12% (Houston) (Figure
19 4-3 and Table E-27 in Appendix E). For an annual standard level of 12 |ig/m3, 12 of the 15
20 urban study areas experience risk reductions, with percent reductions ranging from 11%
21 (Phoenix) to 35% (Houston and Birmingham) (Figure 4-3 and Table E-27 in Appendix E).
22 Estimates of long-term PM2.s exposure-related mortality remaining upon just meeting
23 alternative annual standards: For an annual standard level of 14 |ig/m3, the percent of total
24 incidence of IHD mortality attributable to PM2.5 in the 5 urban study areas experiencing risk
25 reductions ranges from 9-11.3% (Detroit) to 11.8-14.9% (Atlanta) (Tables E-24 and E-33 in
26 Appendix E). For an annual standard of 12 |ig/m3, estimated risk remaining in the 12 urban
27 study areas experiencing risk reductions ranges from 6-7.6% (Phoenix) to 9-11.4% (Atlanta)
28 in terms of PM2.5-attributable long-term exposure-related mortality (Tables E-24 and E-33 in
29 Appendix E).
30 While there is a consistent pattern of risk reduction across the alternative annual standards
31 with lower standard levels resulting in more urban study areas experiencing increasingly larger
32 risk reductions, there is considerable variability in the magnitude of these reductions across study
33 areas for a given alternative annual standard level (e.g., as noted above, for the alternative annual
34 standard level of 12 |ig/m3, risk reduction ranges from 11% for Phoenix to 35% for Houston).
35 This variability in risk reflects differing degrees of reduction in annual-average concentrations
36 across the study areas. These differences in annual-averages result in part because the study
37 areas begin with varying annual-average PM2.s concentrations after simulating just meeting the
38 current suite of standards (see section 6.2.1). Therefore, even if study areas have similar
February 2010 6-10 Draft - Do Not Quote or Cite
-------
1 "ending" annual average PM2.5 concentrations after simulation of just meeting the a given
2 alternative annual standard, because the starting point in the calculation (the annual-average
3 PM2 5 concentrations upon just meeting the current suite of standards) can be variable, the overall
4 reduction in annual-average PM2.5 concentrations across the standards can also be variable. This
5 translates into variation in reductions in long-term exposure-related risk upon just meeting
6 alternative annual standard levels across the study areas.75
7 The sensitivity analysis involving application of peak shaving rollback reveals that the
8 pattern of reductions in ambient PM2 5 concentrations upon just meeting the current suite of
9 standards can impact the magnitude of additional risk reductions estimated for just meeting
10 alternative (lower) annual standard levels. Specifically, for those study areas with more peaky
11 PM2 5 distributions, application of peak shaving rollback will result in higher annual-average
12 PM2 5 levels remaining upon just meeting the current suite of standards. If proportional rollback
13 is then used to simulate just meeting alternative annual standard levels, a greater degree of
14 reduction in annual-average PM2.5 concentrations will result, since the "starting point" for the
15 calculation (annual-average PM2.5 levels upon just meeting the current suite of standards) will be
16 higher.
17 For example, with Los Angeles, which represents a study area with a relatively peaky PM2.5
18 distribution, application of proportional rollback in simulating both the current suite of standards
19 and the alternative annual standard of 12 |ig/m3 results in a 13% reduction in long-term
20 exposure-related mortality (see Figure 4-3 and Table E-27 in Appendix E - this calculations
21 represents the approach used in the core risk assessment model, since proportional rollback was
22 used in simulating both suites of standards). In contrast, application of peak shaving in
23 simulating the current suite of standards followed by proportional reduction in simulating the
24 alternative annual standard of 12 |ig/m3 results in an estimated 48% reduction in long-term
25 exposure-related mortality.76 This example illustrates that application of peak shaving in
26 simulating just meeting the current suite of standards for urban areas such as Los Angeles which
27 have relatively peaky PM2 5 distributions can substantially increase the magnitude of risk
28 reduction simulated for an alternative (lower) annual standard level.
75 We note that additional variation in the risk estimates, in terms of both risk reduction across standard levels and
residual risk for each of the alternative annual standard levels, results from differences across study areas in the
relationship between the maximum monitor annual-averages values used in estimating percent reductions under an
alternative standard and the composite monitor annual-average values used in estimating long-term exposure-related
risk.
76 These risk reductions reflecting application of peak-shaving in simulating the current suite of standards are based
on comparison of composite monitor annual-averages presented in Table F-49 in Appendix F. In generating this
surrogate for reduction in long-term exposure-related mortality between the two standard levels, we compared
composite monitor annual-averages taking into account that long-term exposure-related mortality is only calculated
down to the LML.
February 2010 6-11 Draft - Do Not Quote or Cite
-------
1 Observations made above in the context of the current suite of standards regarding
2 uncertainty and its impact on risk estimates apply in this context as well. Specifically, given the
3 results of the sensitivity analysis examining the form of the C-R functions for long-term
4 exposure-related mortality, combined with only modeling risk down to the LML, we have
5 increased confidence that we have not overstated either the magnitude of risk reductions across
6 alternative standard levels, or the magnitude of risk remaining for a given standard level.
7 6.2.3 Simulation of Just Meeting Alternative Suites of Annual and 24-hour Standards
8 The two suites of standards involving alternative annual and alternative 24-hour
9 standards can be used to consider the impact on risk of reducing the 24-hour standard.
10 Specifically, by comparing risks estimated for the 13/30 and 13/35 suites of standards, we can
11 consider a reduction of 5 |ig/m3 in the 24-hour standard. Similarly if we compare the 12/25 and
12 12/35 suites of standards we can consider a 10 |ig/m3 reduction. In both cases, the reduction in
13 the 24-hour standard level is associated with a fixed annual standard level (i.e., 13 and 12 |ig/m3,
14 respectively). These two comparisons of suites of alternative standards form the basis for the
15 discussion presented below. As with the alternative annual standard levels, we address both the
16 magnitude of risk reductions as well as the magnitude of risk remaining upon just meeting the
17 alternative suites of standards. In discussing these risks, we also continue to focus on the set of
18 urban study areas experiencing risk reductions under each alternative suite of standards.
19 Based on the risk estimates for these areas presented in section 4.2.2 and in Appendix E,
20 we make the following general observations regarding the magnitude of risk remaining upon
21 simulation (using proportional rollback) of these alternative suites of standards:
22 Patterns of reduction in long-term exposure-related mortality across alternative standards:
23 Comparing risks associated with just meeting the 13/35 and 13/30 suites of alternative
24 standards, we see considerable variation in the magnitude of risk reduction across urban
25 study areas. For example, St Louis, under with the 13/35 suite of alternative standards has
26 IHD mortality risk attributable to PM2.5 reduced by 22% relative to risk under the current
27 suite of standards. Very little additional risk reduction (24%) is estimated under the 13/30
28 alternative suite of standards. In contrast, with Salt Lake City, we estimate that the 13/35
29 suite of alternative standards will produce no risk reduction relative to the current suite of
30 standards, while the 13/30 suite would produce a 55% reduction in IHD mortality risk
31 relative to risk under the current standard level (see Figure 4-3 and Table E-27 in Appendix
32 E). The additional risk reduction provided by an alternative 24-hour standard is even more
33 pronounced in comparing the 12/25 and 12/35 alternative suites of standards. In this case we
34 see that for nine of the study areas (Detroit, Fresno, Los Angeles, New York, Philadelphia,
35 Phoenix, Pittsburgh, Salt Lake City and Tacoma) the 12/25 suite of alternative standards
36 produced estimated reductions in risk (relative to risk associated with just meeting the current
37 suite of standards) that are twice as large as for the 12/35 suite of alternative standards (see
38 Figure 4-3 and Table E-27 in Appendix E).
39 Estimates of long-term exposure-related mortality remaining upon just meeting the
February 2010 6-12 Draft - Do Not Quote or Cite
-------
1 alternative 24-hour standards: There is appreciable variation in the estimated magnitude of
2 risk remaining upon simulation of the 13/30 suite of alternative standards. For example, the
3 percent of total IHD mortality incidence attributable to PM2.5 (again, for urban study areas
4 experiencing risk reductions) ranges from 2-2.5% (for Tacoma) to 8.9-11.3% (for Baltimore)
5 (see Tables E-24 and E-33, in Appendix E). There continues to be variation in the levels of
6 residual risk under the 12/25 alternative suite of standards with estimates ranging from 0.3-
7 4.7% (for Tacoma) to 8.8-11.1% (for Atlanta) (see Tables E-24 and E-33, in Appendix E).
8 The observations presented above again highlight variability both in the magnitude of
9 risk reduction as well as in the residual risk estimated from the simulation of just meeting
10 alternative 24-hour standards. This reflects the fact that, as noted above, alternative 24-hour
11 standards can produce different degrees of reduction in the annual-average PM2.5 concentrations,
12 depending on the relationship between 24-hour and annual design values at a particular location.
13 For example, the fact that Salt Lake City is predicted to have a 55% reduction in long-term
14 exposure-related mortality risk with the 13/30 suite of alternative standards (compared with risk
15 under the current suite of standards), reflects the peaky nature of its PM2.5 distribution.
16 Specifically, simulating just meeting the 24-hour standard using proportional rollback will
17 produce a substantial reduction in the annual-average PM2.5 concentrations (i.e., from a recent
18 conditions annual-average of 11.6 |ig/m3, to 7.7 |ig/m3 under the current suite of standards, to 6.7
19 |ig/m3 with the 13/30 suite of alternative standards-see Table F-49 in Appendix F). In
20 contrast, with St Louis, which does not experience as substantial a risk reduction under the 13/30
21 suite of alternative standards, there is a far less peaky PM2.5 distribution (i.e., the annual and 24-
22 hour design values are relatively closer to each other - see Table F-49 in Appendix F).
23 Therefore, simulation of the alternative 24-hour standard level of 30 |ig/m3 does not have as
24 substantial an effect on annual-average concentrations (i.e., from a recent conditions annual-
25 average of 16.5 |ig/m3, to 14.9 |ig/m3 under the current suite of standards, to 12.8 |ig/m3 under
26 the 13/30 suite of alternative standards).
27 It is possible to stratify the set of urban study areas based on patterns of risk reduction
28 estimated under the alternative 24-hour standards. In this discussion, we focus on risk estimates
29 generated for the 12/25 suite of alternative standards, focusing on how risks under this scenario
30 compare with risks under the current suite of standards.77 The stratification of the study areas
31 based on the magnitude of risk reduction highlights factors responsible for these differences
32 across study areas. For example, when the 24-hour standard is controlling (in simulating the
33 12/25 suite of alternative standards) and the PM2.5 distribution is relatively peaky, there is a
34 greater potential for the annual-average PM2.5 concentrations to be reduced more in simulating
35 just meeting the alternative 24-hour standard (in some instances, well below 12 |ig/m3) resulting
77 Further, in considering risk reduction, we are comparing risk under the alternative suites of standards to risk under
the current suite of standards based solely on application of proportional rollback.
February 2010 6-13 Draft - Do Not Quote or Cite
-------
1 in larger estimated risk reductions. In fact, we see that the urban study areas having the largest
2 risk reductions have annual-average PM2.5 concentrations simulated under the 12/25 suite of
3 standards (using proportional rollback) well below 12 |ig/m3, with some locations ranging down
4 to -6 |ig/m3.
5 We identified four strata in considering patterns of risk reduction across the 15 urban study
6 areas under the 12/25 suite of alternative standards (all of the percent reductions presented are in
7 terms of long-term exposure-related IHD mortality).
8 -100% reduction in risk: Those study areas where the 24-hour standard was controlling
9 and where the resulting annual-average PM2.5 concentrations (under the 12/25 suite of
10 standards) were ~ 6 |ig/m3. Because annual-average concentrations for these study areas
11 are at or below the LML for long-term exposure-related mortality (5.8 |ig/m3), little to no
12 risk is predicted under the alternative suite of standards, resulting in a near 100%
13 reduction in risk relative to the current suite of standards. These study areas have the
14 most peaky PM2.5 distributions of the 15 urban study areas (i.e., relatively high 24-hour
15 design values and lower annual average design values) and include study areas Tacoma
16 and Salt Lake City.78
17 -70% reduction in risk: Those study areas where the 24-hour standard is controlling and
18 where the resulting annual-average PM2.5 levels (under the 12/25 suite of standards) were
19 -7-9 |ig/m3. These study areas also have relatively peaky PM2.5 distributions and include
20 Los Angeles and Fresno.79
21 -50-60% reduction in risk: Those study areas where the 24-hour standard is controlling
22 and where the resulting annual-average PM2.5 levels (under the 12/25 suite of standards)
23 were -9-11 |ig/m3. These study areas have less peaky PM2.5 distributions (24-hour
24 standard still controls, but there is not as great a disparity with the annual design values)
25 and include the majority of the study areas (Detroit, NYC, Philadelphia, Pitts, St Louis,
26 Baltimore, Birmingham, and Phoenix).80
27 -35-45% reduction in risk: This category includes some study areas where the 24-hour
28 standard controls and some where the annual standard controls. Annual average PM2.5
29 concentrations under the 12/25 suite of standards are generally in the 12 |ig/m3 range.
30 These study areas have relatively less peaky PM2.5 distributions and include Atlanta and
78 These study areas fall in zone A in Figure 4-20, which represents the largest grouping of urban areas in the U.S.
predicted to be exceeding this alternative suite of standards (12/25). However, we note that Tacoma and Salt Lake
City have some of the most peaky PM25 distributions of the urban areas in this zone and therefore are likely to
experience greater risk reductions than most of the urban areas in zone A.
79 Los Angeles and Fresno fall in zone B and specifically, subarea B1, in Figure 4-20 (subarea B1 represents those
study areas that exceed the 12/25 suite of alternative standards and that also have a greater degree of peakiness in
their PM2 5 distributions relative to other urban areas in zone B - see section 4.5.1). Consequently, these study areas
are likely to experience greater risk reductions relative to other urban areas in zone B.
80 These eight study areas fall in zone B in Figure 4-20 and specifically, subarea B2, which includes a relatively
large fraction of those urban areas in the U.S. predicted to exceed the 12/25 suite of alternative standards. Urban
areas in subarea B2 have less peaky PM25 distributions compared to areas in subarea Bl.
February 2010 6-14 Draft - Do Not Quote or Cite
-------
1 Houston.81'82
2 Observations made earlier regarding the impact of variability in simulating changes in
3 PM2.5 distributions using different rollback approaches, and its impact on the degree of risk
4 reduction, also hold here. Specifically, in those instances where PM2.5 distributions are more
5 peaky, application of peak shaving rollback would result in smaller reductions in annual-average
6 PM2 5 concentrations and consequently, smaller reductions in estimates of long-term exposure-
7 related mortality. For example, with Salt Lake City, which has a peaky PM2.5 distribution, under
8 the 12/25 suite of standards application of proportional rollback results in an annual average
9 PM2.s concentration of 5.7 |ig/m3, while application of peak shaving results in an estimate of 8.9
10 |ig/m3. In contrast, simulation of the 12/25 suite of standards for Baltimore, which has a less
11 peaky PM2 5 distribution, results in an annual average PM2 5 concentration of 10.7 |ig/m3 for
12 proportional rollback compared to 10.8 |ig/m3 with peak shaving (see Table F-49 in Appendix F).
13 A key observation made above in relation to the current suite of standards, that is even
14 more relevant in considering the results discussed here, is that simulated annual-average PM2 5
15 concentrations upon just meeting alternative suites of standards for many of the urban study
16 areas are considerably lower than 12 |ig/m3. For example, with the current suite of standards,
17 Fresno and Salt Lake City are simulated to have annual average PM2.5 concentrations of 9.9 and
18 7.7 |ig/m3, respectively, which are in turn reflected in the risk estimates generated (see Table F-
19 49, in Appendix F). Annual average concentrations in these study areas are even lower under the
20 alternative suites of standards with lower 24-hour standard levels. For example, under the 13/30
21 suite of standards, simulated annual average concentrations range down to 6.7 |ig/m3 (Salt Lake
22 City), with a number of urban study areas having annual-average concentrations simulated in the
23 range of 7 to 11 |ig/m3 (Fresno, Los Angeles, and Tacoma). Under the 12/25 suite of standards,
24 simulated annual-average concentrations are even lower, ranging down to 5.7 |ig/m3 (Salt Lake
25 City). These very low annual-average PM2 5 concentrations reflect lower annual design values to
26 begin with as well as relatively peaky PM2 5 distributions, which means that simulation of the 24-
27 hour standard (when controlling) will produce appreciable impacts on the annual average
28 concentration.
29 The results discussed above show that simulating just meeting alternative 24-hour
30 standard levels in the range of 25 to 30 |ig/m3 can produce substantial reductions in estimated
81 Atlanta and Houston fall into zones B and C in Figure 4-20, and specifically portions of those zones including
urban areas with less peaky PM2 5 distributions.
82 We note that Dallas has a substantially smaller estimate of risk reduction (-13%) compared with the other 14
urban study areas. The relatively low risk reduction for this location reflects the fact that Dallas has annual and 24-
hour design values (12.8 and 26 ug/m3, respectively) that are well below the current suite of standards and only just
exceed the 12/25 suite of standards. Therefore, the estimated risk reduction under this suite of standards is expected
to be very low. Dallas just barely falls into Zone C in Figure 4-19.
February 2010 6-15 Draft - Do Not Quote or Cite
-------
1 risk, beyond that produced by simulations of just meeting lower annual standard level down to
2 12 |ig/m3 (combined with a 24-hour standard of 35 |ig/m3). This results from the simulations
3 producing substantially lower annual-average PM2 5 concentrations, which drive reductions in
4 both long-term and short-term exposure-related risk. The results also show that there can be
5 considerable variability across study areas in the degree to which alternative 24-hour standard
6 levels produce reductions in annual average PM2.5 concentrations and, consequently, reductions
7 in risk. This variability is seen to depend largely on the peakiness of the PM2.5 distribution in an
8 area and on the rollback approach used to simulate just meeting the current and alternative suites
9 of standards. These results suggest that while lowering the 24-hour standard can be used to
10 reduce annual-average PM2 5 concentrations, and thus to reduce estimated risk, the results are
11 likely to be highly variable across urban areas. This analysis also suggests that more consistent
12 annual-average PM2.5 concentrations, and thus more consistent reductions in estimated risk,
13 would result from simulating just meeting alternative annual standards at levels below 12 |ig/m3
14 which was the lowest annual standard level considered in this assessment. In general,
15 considering suites of standards in which the annual standard is the controlling standard would be
16 expected to provide more consistent reductions in annual-average PM2 5 concentrations, thereby,
17 providing more uniform public health protection across urban areas.
18 Observations made earlier regarding overall confidence in the estimates of long-term
19 exposure-related mortality also hold for these estimates (i.e., the sensitivity analysis results
20 combined with the fact that we modeled risk down to LML result in our concluding that it is
21 unlikely we have overstated either the degree of risk reduction or the degree of residual risk).
22 6.3 NATIONAL PERSPECTIVE ON PM2.5-RELATED RISKS
23 This section places the core risk estimates in the broader national-context by considering
24 the degree to which the 15 urban study areas are representative of larger urban areas within the
25 U.S., particularly areas likely to experience elevated risk related to PM exposure. As such, it
26 draws on information presented in several sections of the risk assessment including: (a) the
27 representativeness analysis discussed in section 4.4, (b) consideration of patterns of design
28 values for the 15 urban study areas as contrasted with the broader set of larger urban areas within
29 the US (section 4.5.1), and (c) the national-scale mortality analysis discussed in Chapter 5.
30 The representativeness analysis presented in section 4.4, compared attributes of the 15 urban
31 study eras (assessed at the county-level) against national distributions for the same attributes.
32 The analysis suggests that the 15 urban study areas represent areas in the U.S. that are among
33 the most densely populated, have relatively higher levels of annual and 24-hour 98th
34 percentile PM2.5 concentrations, and capture well the range of effect estimates represented by
35 the Zanobetti and Schwartz (2009) study. Together, these factors suggest that the urban
36 study areas should capture well the overall distribution of risk for the nation, with the
February 2010 6-16 Draft - Do Not Quote or Cite
-------
1 potential for better characterization of the high end of that distribution.83
2 Consideration of the mix of design values across the 15 urban study areas as contrasted with
3 design values for the broader set of urban study areas in the U.S. suggests that the 15 urban
4 study areas do a good job of capturing the key groupings of urban areas in the U.S. likely to
5 experience elevated risk due to PM (i.e., we have coverage for each of the zones containing
6 urban study areas likely to experience risk reductions under the suites of alternative standard
7 levels considered - see section 4.5.1). Furthermore, this analysis suggested that we have also
8 included study areas likely to experience relatively greater degrees of PM2.s-related risk,
9 considering the pattern of design values across urban areas in the U.S..
10 Consideration of where the 15 urban study areas fell along the distribution of U.S. counties
11 included in the national-scale mortality analysis further suggests that we have captured
12 counties likely to experience elevated PM25-related risk. As part of the national-scale
13 mortality analysis (see Chapter 5), we created a cumulative distribution of the percentage of
14 mortality attributable to PM2.s based on the county-level estimates for the U.S.84 We then
15 identified where along this cumulative distribution the 31 counties comprising our 15 urban
16 study areas fell. This analysis suggests that our urban study areas capture the upper end of
17 the tail with regard to PM2.5-attributable risk, with 23 of these counties falling within the
18 upper 5th percentile of the distribution. These findings support the assertion based on the
19 other analyses described above that the urban study areas are likely to capture risk at urban
20 areas experiencing relatively elevated levels of PM2 5_attributable mortality.
21 Our overall assessment of the representativeness of the 15 urban study areas in the
22 national context, based on the three analyses summarized above, is that our study areas do a good
23 job of representing urban areas in the U.S. experiencing elevated levels of risk related to ambient
24 PM2.5 exposure. The results of the national-scale mortality analysis also suggest that, while our
25 15 urban study areas do provide coverage for urban areas in the U.S. experiencing elevated
26 levels of PM2.5-related risk, there are many additional areas (counties) not modeled in the risk
27 assessment that experience elevated PM2.5-related risk. In other words, it should not be
28 construed that significant PM2.s-related risk is limited only to the urban study areas included in
29 the risk assessment.
83 This analysis also showed that the urban study areas do not capture areas with the highest baseline morality risks
or the oldest populations (both of which can result in higher PM2 5-related mortality estimates). However, some of
the areas with the highest values for these attributes have relatively lower PM2 5 levels (e.g., urban areas in Florida)
and consequently failure to include these areas in the set of urban study areas is unlikely to bias the risk estimates in
terms of excluding high PM2 5-risk locations.
84 Note that by using this risk metric, we avoid influence by difference in overall population size (as would be the
case with raw incidence) and focus on a unitized estimate of PM2 5-related mortality which reflects differences in (a)
baseline mortality incidence, and (b) the annual PM2 5 levels average for each county.
February 2010 6-17 Draft - Do Not Quote or Cite
-------
1 6.4 KEY OBSERVATIONS
2 Key observations from this quantitative risk assessment for PM2.s, with emphasis on the
3 observations made above in this chapter, are outlined below. These observations are organized
4 around the three policy-relevant questions outlined at the beginning of this chapter.
5
6 (1) What is the magnitude of risk likely to remain if the urban study areas were just meeting the
7 current suite of PM2.5 standards (an annual standard of 15 |ig/m3 and a 24-hour standard of 35
8 Hg/m3), and what level of confidence do we have in those estimates?
9 Upon simulation of just meeting the current suite of standards, the core analysis estimates
10 that the urban study areas would have IHD-related mortality attributable to long-term PM2.5
11 exposure ranging from <100 to approximately 2,000 cases per year, with this variability
12 reflecting to a great extent differences in the size of study area populations. These estimates
13 represent from 4 to 17% of all IHD-related mortality in a given year for the urban study
14 areas, which is a measure of risk that takes into account differences in population size and
15 baseline mortality rates. Estimates were also developed for other long-term exposure-related
16 mortality endpoints, including all-cause, cardiopulmonary-related, and lung cancer mortality.
17 Generally comparable estimates of CV-related mortality attributable to short-term PM2.5
18 exposure are substantially lower than for long-term exposure-related IHD mortality. The
19 core analysis estimates that the urban study areas would have CV-related mortality
20 attributable to short-term PM2 5 exposure ranging from approximately 10 to 470 cases per
21 year. Estimates were also developed for other short-term exposure-related endpoints,
22 including non-accidental and respiratory-related mortality, CV- and respiratory-related
23 hospital admissions, and asthma-related emergency department visits.
24 A broader array of health effects has also been associated with PM2.5 exposures, including in
25 particular effects on children, such as reproductive and developmental effects. While
26 information was too limited to consider these effects in this quantitative risk assessment, such
27 effects are appropriately considered based on the related evidence in the broader
28 characterization of risks to be discussed in a separate Policy Assessment document.
29 Given the quantitative and qualitative assessments of uncertainty and variability that we have
30 completed as part of our quantitative risk assessment, we believe that it is unlikely that we
31 have over-stated the degree of risk remaining upon simulation of just meeting the current
32 suite of standards. While this conclusion applies to all quantitative estimates of risk, it applies
33 most strongly for long-term PM2.5 exposure-related mortality for which more extensive
34 uncertainty and variability assessment has been done.
35 Estimated risks remaining upon just meeting the current suite of standards vary substantially
36 across study areas, even when considering risks normalized for differences in population size
37 and baseline incidence rates. This variability in estimated risks is a consequence of the
38 substantial variability in the annual-average PM2.5 concentrations across study areas that
39 result from simulating just meeting the current standards. This is important because annual-
40 average concentrations are highly correlated with both long-term and short-term exposure-
41 related risk. This variability in annual-average PM2.5 concentrations occurs especially in
February 2010 6-18 Draft - Do Not Quote or Cite
-------
1 those study areas in which the 24-hour standard is the "controlling" standard.85 In such
2 areas, the variability across study areas in estimated risks is largest when regional patterns of
3 reductions in PM2.5 concentrations are simulated (using proportional rollback, as was done in
4 the core analyses), with less variability when more localized patterns of PM2.5 reductions are
5 simulated (using peak shaving rollback, as was done in a sensitivity analysis). When
6 simulations are done using peak shaving rollback, estimated risks remaining upon just
7 meeting the current suite of standards can be appreciably larger than those estimated in the
8 core analysis.
9 In simulating just meeting the current suite of standards, the resulting annual-average PM2.5
10 concentrations range from about 15 |ig/m3 (for those study areas in which the annual
11 standard was controlling) down to as low as about 8 |ig/m3 (for those study areas in which
12 the 24-hour standard was controlling or the annual average was well below 15 |ig/m3 based
13 on recent air quality). Thus, estimates of risk remaining upon just meeting the current
14 standards are, in many cases, reflective of annual average PM2.5 concentrations that are well
15 below the level of the current annual standard.
16 The 15 urban study areas included in this risk assessment are generally characteristic of
17 urban areas across the U.S. that do not meet the current suite of standards. Of those urban
18 areas in the U.S. that do not meet the current suite of standards (based on 2005-2007 air
19 quality data), the 24-hour standard is controlling in most such areas - a pattern that is
20 reflected in the urban study areas included in this assessment. Two areas are included in this
21 assessment that meet the current suite of standards (reflective of the majority of urban areas
22 in the U.S.), although these two areas fail to meet some of the alternative suites of standards
23 considered in this assessment.
24
25 (2) What is the degree and nature of risk reduction likely to be associated with just meeting the
26 alternative suites of annual and 24-hour PM2.5 standards considered in this risk assessment, and
27 what roles do the annual and 24-hour standards play in bringing about such reductions?
28 Upon simulation of just meeting the alternative annual standard levels considered (14, 13,
29 and 12 |ig/m3) in conjunction with the current 24-hour standard (denoted as 14/35, 13/35 and
30 12/35 suites of standards), the core analysis estimates reductions in long-term exposure-
31 related mortality for 12 of the 15 urban study areas, with the degree of risk reduction
32 increasing incrementally across the alternative standard levels (both in terms of the number
33 of study areas experiencing risk reduction and the magnitude of those reductions). For the
34 alternative annual standard level of 12 |ig/m3 (in conjunction with the current 24-hour
35 standard), the core analysis estimates that these study areas have reductions in risk (relative
36 to risk remaining upon just meeting the current suite of standards) ranging from about 11 to
37 35%. For some of those areas in which the 24-hour standard is controlling, larger risk
38 reductions would have been estimated in this case (12/35 suite of standards) if peak shaving
39 rollback had been used to simulate just meeting the current suite of standards. This result
40 would be expected since the magnitude of risk remaining upon just meeting the current suite
41 of standards would have been higher than that estimated based on the proportional rollback
85 The controlling standard is the standard (either 24-hour or annual) that requires the largest percent reduction in the
related design value to just meet that standard.
February 2010 6-19 Draft - Do Not Quote or Cite
-------
1 used in the core analysis. Therefore, while we are going down to the same level of risk
2 (under the 12/35 suite of standards), we are starting with a higher level of risk from the
3 current standard.
4 Upon just meeting the alternative suites of standards that included lower levels of both the
5 annual and 24-hour standards (denoted as 13/30 and 12/25 suites of standards), the core
6 analysis estimates that the lower 24-hour standard levels produce additional risk reductions
7 beyond the reductions estimated for the lower annual standard levels alone. In the case of the
8 12/25 suite of standards, estimated risk reductions compared with reductions for the annual
9 standard alone (12 |ig/m3), were roughly twice as large in many of the study areas, although
10 in a few areas risk reductions were much higher (ranging up to -100%) and in a few other
11 areas, there was little to no risk reduction. These results show that lower 24-hour standards
12 can have an appreciable and highly variable impact on long-term exposure-related mortality,
13 particularly when just meeting the lower standards is simulated using a more regional pattern
14 of PM2.5 reductions (i.e., the proportional rollback used in the core analysis). However, the
15 magnitude of risk reductions estimated for the lower 24-hour standards was reduced when
16 simulations using a more localized pattern of PM2.5 reductions (i.e., the peak shaving rollback
17 used in a sensitivity analysis).
18 The results of simulating alternative suites of standards including lower levels of both annual
19 and24hr standards suggest that while lowering the 24-hour standard can be used to reduce
20 annual-average PM2 5 concentrations, and thus to reduce estimated risk, the results are likely
21 to be highly variable across urban areas. More consistent annual-average PM2.5
22 concentrations across study areas, and thus more consistent reductions in estimated risk,
23 would result from simulating just meeting a specific alternative annual standard level. In
24 general, considering suites of standards in which the annual standard is the controlling
25 standard would be expected to provide more consistent reductions in annual-average PM2.5
26 concentrations, thereby, providing more uniform public health protection across urban areas.
27 In simulating just meeting the alternative suites of standards, especially those with lower 24-
28 hour standard levels, the resulting annual-average PM2.5 concentrations are substantially
29 lower than the lowest annual standard level considered in the analysis (12 |ig/m3). For
30 example, under the 12/25 suite of standards, estimated annual-average PM2.5 concentrations
31 ranged down to approximately 6 |ig/m3, with eight urban study areas having annual average
32 PM2.5 levels in the 8-11 |ig/m3 range.
33 Addressing overall confidence in risk estimates generated for just meeting the alternative
34 suites of standards, as with the current suite of standards, we conclude based on our
35 quantitative and qualitative analysis of uncertainty and variability that we have likely not
36 over-stated risk reductions or levels of residual risk estimated for just meeting these
37 alternative suites of standards.
38
39 (3) What is the distribution of risks associated with recent PM2.5 air quality in areas across the
40 U.S., and how representative are the risks estimated for the urban study areas from a national
41 perspective?
42 Based on recent air quality from 2005 to 2007, we estimate that within the continental U.S.,
43 total PM2.5-related premature mortality ranges from 63,000 and 88,000 per year. Further, we
February 2010 6-20 Draft - Do Not Quote or Cite
-------
1 estimate that the percent of total mortality attributable to PM2.5 long-term exposure ranges
2 from approximately 3 to 9% in about half of the counties in the U.S., with a, range from
3 approximately 0 to 3% in the other half of counties.
4 Efforts to place the 15 urban study areas and the core risk estimates generated for those areas
5 into a broader national context suggest that these study areas likely capture well the full set of
6 urban areas in the U.S. likely to experience relatively higher PM2.5-related risk.
7 It is important to recognize that there are many additional areas besides those included in the
8 risk assessment that experience elevated PM2.5-related risk of similar magnitude to the risks
9 estimated for the urban study areas included in this assessment.
10
February 2010 6-21 Draft - Do Not Quote or Cite
-------
1 7 REFERENCES
2 Abt Associates Inc. (2003), "Environmental Benefits Mapping and Analysis Program (BenMAP), User's Manual.
3 Available on the Internet at .
4
5 Abt Associates Inc. (2005). Paniculate Matter Health Risk Assessment for Selected Urban Areas. Prepared for
6 Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle
7 Park, NC. June 2005. Available electronically on the internet at:
8 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr td.html.
9 Abt Associates Inc. (2008). Environmental Benefits Mapping and Analysis Program (Version 3.0). Bethesda, MD.
10 Prepared for Environmental Protection Agency, Office of Air Quality Planning and Standards, Air Benefits
11 and Costs Group. Research Triangle Park, NC.
12 American Farm Bureau Federation v. EPA, 559 F. 3d 512 (D.C. Cir. 2009).
13 Bateson, T.F.; Schwartz, J. (2004). Who is sensitive to the effects of paniculate air pollution on mortality? A case-
14 crossover analysis of effect modifiers. Epidemiology. 15(2):143-9.
15 Bell, M.L.; Ebisu, K.; Peng, R.D.; Walker, J.; Samet, JM; Zeger, S.L.; Dominici, F. (2008). Seasonal and regional
16 short-term effects of fine particles on hospital admissions in 202 US counties, 1999-2005. Am J
17 Epidemiol. 168(11):1301-1310.
18 Bell, M.L.; Dominici, F. (2008). Effect modification by community characteristics on the short-term effects of ozone
19 exposure and mortality in 98 US communities. Am J Epidemiol. 167(8):986-97.
20 Bell (2009). Personal communication via e-mail between Dr. Michelle Bell and Beth Hasset-Sipple, June 26th, 2009
21 (e-mail with attachments placed in docket #EPA-HQ-OAR-2007-0492).
22 Bell M.L.; Ebisu K.; Peng R.D.; Samet J.M.; Dominici, F. (2009). Hospital admissions and chemical composition of
23 fine particle air pollution. Am J Respir Crit Care Med. 179(12): 1115-20.
24 Berkey, C.S.; Hoaglin, D.C.; Antczak-Bouckoms, A.; Mosteller, F.; Colditz, G.A. (1998). Meta-analysis of multiple
25 outcomes by regression with random effects. Stat Med. 17(22):2537-2550.
26 Byun, D., and K.L. Schere, (2006). Review of the Governing Equations, Computational Algorithms, and Other
27 Components of the Models-3 Community Multiscale Air Quality (CMAQ) Modeling System. Applied
28 Mechanics Reviews, 59, 51-77.
29 Dominici, F.; Peng, R.D.; Ebisu, K.; Zeger, S.L.; Samet, J.M.; Bell, M.L. (2007). Does the effect of PM10 on
3 0 mortality depend on PM nickel and vanadium content? A reanaly sis of the NMMAPS data. Environ Health
31 Perspect. 115(12):1701-3.
32 Franklin, M.; Zeka, A.; Schwartz, J. (2007). Association between PM25 and all-cause and specific-cause motrality in
33 27 US communities. J Expo Sci Environ Epidemiol. 17:279-287.
34 Henderson, R. (2005). EPA's Review of the National Ambient Air Quality Standards for Paniculate Matter (Second
3 5 Draft PM Staff Paper, January 2005): A review by the Paniculate Matter Review Panel of the EPA Clean
36 Air Scientific Advisory Committee. June 6, 2005. Available electronically on the internet at:
3 7 http://www.epa.gov/sab/pdf/casac-05-007.pdf.
38 Henderson, R. (2006). Clean Air Scientific Advisory Committee Recommendations Concerning the Proposed
39 National Ambient Air Quality Standards for Paniculate Matter. March 21, 2006. Available electronically
40 on the internet at: http://www.epa.gov/sab/pdf/casac-ltr-06-002.pdf
February 2010 7-1 Draft - Do Not Quote or Cite
-------
1 Henderson, R; Poirot, R.L.; Cowling, E.; Speizer, F.; Crapo, J.D.; Zielinska, B.; Miller, F.J. (2006). Clean Air
2 Scientific Advisory Committee Recommendations Concerning the Final National Ambient Air Quality
3 Standards for Paniculate Matter. September 29, 2006. Available electronically on the internet at:
4 http://www.epa.gov/sab/pdf/casac-ltr-06-003.pdf.
5 Hopke, P. (2002). Letter from Dr. Phil Hopke, Chair, Clean Air Scientific Advisory Committee (CASAC)
6 Paniculate Matter Review Panel, to Honorable Christine Todd Whitman, Administrator, U.S. EPA. Final
7 advisory review report by the CASAC Paniculate Matter Review Panel on the proposed paniculate matter
8 risk assessment. EPA-SAB-CASAC-ADV-02- 002. May 23. Available electronically on the internet
9 at:http://www.epa.gov/sab/pdf/casacadv02002.pdf.
10 Hopke, P. (2004). Letter from Dr. Phil Hopke, Clean Air Scientific Advisory Committee
11 (CASAC) Paniculate Matter (PM) Review Panel, to Honorable Michael O. Leavitt. CASAC PM Review Panel's
12 Ongoing Peer Review of the Agency' s Fourth External Review Draft of Air Quality Criteria for P articulate
13 Matter (June 2003); and Peer Review of the Review of the National Ambient Air Quality Standards for
14 P articulate Matter: Policy Assessment of Scientific and Technical Information (OA QPS Staff Paper - First
15 Draft) (August 2003) and a Related Draft Technical Report, Particulate Matter Health Risk Assessment for
16 Selected Urban Areas (Draft Report) (August 2003). EPA-SAB-CASAC-04-004. February 18. Available
17 electronically on the internet at: http://www.epa.gov/sab/pdf.
18 Ito, K.; Thurston, G.; Silverman, R.A. (2007). Characterization of PM2 5 gaseous pollutants and meteorological
19 interactions in the context of time-series health effects models. J Expo Sci Environ Epidemiol. 17: 45-60.
20 Jerrett. M.; Burnett, R.T.; Ma, R.; Pope, C.A. Ill; Krewski, D.; Newbold, K.B.; Thurston, G.; Shi, Y.; Finkelstein,
21 N.; Calle, E.; Thun, M.J. (2005). Spatial analysis of air pollution and mortality in Los Angeles.
22 Epidemiology. 6(6): 1-10.
23 Kottek, M.; Grieser, J.; Beck, C.; Rudolf, B.; Rubel, F. (2006). World map of the Koppen-Geiger climate
24 classification updated. Meteorol Zeit. 15(3): 259-263.
25 Kottek, M. (2008). World Map of the Koppen-Geiger climate classification updated. Available: http://koeppen-
26 geiger.vu-wien.ac.at/ [accessed 12/5/2008].
27 Krewski (2009). Letter from Dr. Daniel Krewski to HEFs Dr. Kate Adams (dated July July 7, 2009) regarding
28 "EPA queries regarding HEI Report 140". Dr. Adams then forwarded the letter on July 10, 2009 to EPA's
29 Beth Hassett-Sipple. (letter placed in docket #EPA-HQ-OAR-2007-0492).
30 Krewski, D.; Burnett, R.T.; Goldberg, M.S.; Hoover, K.; Siemiatycki, J.; Jerrett, M.; Abrahamowicz, M.; White,
31 W.H. (2000). Reanalysis of the Harvard Six Cities study and the American Cancer Society study of
32 paniculate air matter pollution and mortality. HEI Research Report, Health Effects Institute, Boston, MA.
33 Krewski, D.; Jerrett, M.; Burnett, R.T.; Ma, R.; Hughes, E.; Shi, Y.; Turner, M.C.; Pope, C.A. Ill; Thurston, G.;
34 Calle, E.E.; Thun, M.J. (2009). Extended follow-up and spatial analysis of the American Cancer Society
3 5 study linking paniculate air pollution and mortality. HEI Research Report, 140, Health Effects Institute,
36 Boston, MA.
37 Laden, F., J. Schwartz, F.E. Speizer, and D.W. Dockery. (2006). Reduction in Fine Particulate Air Pollution and
3 8 Mortality. American Journal of Respiratory and Critical Care Medicine 173:667-672
39 Le Terte, A.; Schwartz, J.; Touloumi, G. (2005). Empirical bayes and adjusted estimates approach to estimating the
40 relation of mortality to exposure of PM10. Risk Anal. 25(3):711-718.
February 2010 7-2 Draft - Do Not Quote or Cite
-------
1 Medina-Ramon, M; Zanobetti, A.; Schwartz, J. (2006). The effect of ozone and PM10 on hospital admissions for
2 pneumonia and chronic obstructive pulmonary disease: a national multicity study. Am J Epidemiol.
3 163(6):579-88.
4 Moolgavkar, S. (2000). Air pollution and hospital admissions for chronic obstructive pulmonary disease in three
5 metropolitan areas in the United States. Inhal Toxicol. 12(Supplement 4): 75-90.
6 Moolgavkar, S. (2003). Air pollution and daily deaths and hospital admissions in Los Angeles and Cook counties.
7 pp. 183-198. In: Health Effects Institute Special Report, Revised Analyses of Time-Series Studies of Air
8 Pollution and Health. Health Effects Institute, Cambridge, MA.
9 Ostro, B.; Broadwin, R.; Green, S.; Feng, W.Y.; Lipsett, M. (2006). Fine paniculate air pollution and mortality in
10 nine California counties: results from CALFINE. Environ Health Perspect. 114(l):29-33.
11 Pope, C.A. Ill; Burnett, R.T.; Thun, M.J. ; Calle, E.E. ; Krewski, D. ; Ito, K. ; Thurston G.D. (2002). Lung cancer,
12 cardiopulmonary mortality, and long-term exposure to fine paniculate air pollution. JAMA. 287(9): 1132-
13 1141.
14 Roberts, S. (2004). Interactions between paniculate air pollution and temperature in air pollution mortality time
15 series studies. Environ Res. 96(3):328-37.
16 Samet, J.M. (2008). Air pollution risk estimates: determinants of heterogeneity. J Toxicol Environ Health A. 71(9-
17 10):578-82.
18 Samet, 2009. Letter from Dr. Jonathan M. Samet, Chair Clean Air Scientific Advisory Committee, to Lisa P.
19 Jackson, Administrator U.S. Environmental Protection Agency regarding "Consultation on EPA's
20 Paniculate Matter National Ambient Air Quality Standards: Scope and Methods Plan for Health Risk and
21 Exposure Assessment." May, 21, 2009. (letter placed in docket #EPA-HQ-OAR-2007-0492).
22 Sarnat, S.E.; Coull, B.A.; Schwartz, J.; Gold, D.R.; Suh, H.H. (2006). Factors affecting the association between
23 ambient concentrations and personal exposures to particles and gases. Environ Health Perspect.
24 114(5):649-54.
25 Schwartz, J. (2000). The distributed lag between air pollution and daily deaths. Epidemiology 11(3):320-326.
26 Tolbert, P.E. (2007). Invited commentary: heterogeneity of paniculate matter health risks. Am J Epidemiol.
27 166(8):889-91; discussion 892-3.
28 Tolbert, P.E.; Klein, M.; Peel, J.L.; Sarnat, S.E.; Sarnat, J.A. (2007). Multipollutant Modeling Issues in a Study of
29 Ambient Air Quality and Emergency Department Visits in Atlanta. J Expo Sci Environ Epidemiol. 17:829-
30 S35.
31 U.S. Environmental Protection Agency. (1996). Air Quality Criteria for Paniculate Matter. National Center for
32 Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency,
33 Research Triangle Park, NC 27711; report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October
34 2004. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr cd.html
35 U.S. Environmental Protection Agency. (2001). Risk assessment guidance for Superfund. Vol. Ill, Part A. Process
36 for conducting probabilistic risk assessment (RAGS 3 A). Washington, DC, United States Environmental
37 Protection Agency, EPA 540-R-02-002; OSWER 9285.7-45; PB2002 963302;
38 http://www.epa.gov/oswer/riskassessment/rags3adt/index.htm.
39 US Environmental Protection Agency. (2004a). Air Quality Criteria for Paniculate Matter. National Center for
40 Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency,
February 2010 7-3 Draft - Do Not Quote or Cite
-------
1 Research Triangle Park, NC 27711; report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October
2 2004. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr cd.html
3 U.S. Environmental Protection Agency. (2004b) EPA's risk assessment process for air toxics: History and overview.
4 In: Air toxics risk assessment reference library. Vol. 1. Technical resource manual. Washington, DC,
5 United States Environmental Protection Agency, pp. 3-1 - 3-30, EPA-453-K-04-001A;
6 http://www.epa.gov/ttn/fera/data/risk/vol I/chapter 03.pdf.
7 U.S. Environmental Protection Agency. (2005). Review of the National Ambient Air Quality Standards for
8 Paniculate Matter: Policy Assessment of Scientific and Technical Information - OAQPS Staff Paper,
9 Office of Air Quality Planning and Standards, Research Triangle Park, NC, EPA-452/R-05-005. Available:
10 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr sp.html
11 U.S. Environmental Protection Agency. (2007). Draft Integrated Review Plan for the National Ambient Air Quality
12 Standards for Particulate Matter. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA
13 452/P-08-006.
14 U.S. Environmental Protection Agency. (2008a). Integrated Review Plan for the National Ambient Air Quality
15 Standards for Paniculate Matter. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA
16 452/R-08/004.
17 U.S. Environmental Protection Agency. (2008b). Integrated Science Assessment for Paniculate Matter: First
18 External Review Draft. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA/600/R-
19 08/139 and 139A. Available: http://www.epa.gOv/ttn/naaqs/standards/pm/sjm_2007_isa.html.
20 U.S. Environmental Protection Agency. (2009a). Integrated Science Assessment for Paniculate Matter: Second
21 External Review Draft. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA 600/R-
22 08/139B. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm 2007 isa.html.
23 U.S. Environmental Protection Agency. (2009b). Paniculate Matter National Ambient Air Quality Standards: Scope
24 and Methods Plan for Health Risk and Exposure Assessment. U.S. Environmental Protection Agency,
25 Research Triangle Park, NC, EPA-452/P-09-002.
26 U.S. Environmental Protection Agency. (2009c). Paniculate Matter Urban-Focused Visibility Assessment - External
27 Review Draft. Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency,
28 Research Triangle Park, NC. EPA-452/P-09/005. September 2009. Available:
29 http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_index.html.
30 U.S. Environmental Protection Agency. (2009d). Integrated Science Assessment for Paniculate Matter: Final. U.S.
31 Environmental Protection Agency, Research Triangle Park, NC, EPA/600/R-08/139F. Available
32 http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=216546.
3 3 Woods & Poole Economics Inc. (2001). Population by Single Year of Age CD. CD-ROM. Woods & Poole
34 Economics, Inc. Washington, D.C.
3 5 World Health Organization. (2008). Part 1: Guidance Document on Characterizing and Communicating Uncertainty
36 in Exposure Assessment, Harmonization Project Document No. 6. Published under joint sponsorship of the
3 7 World Health Organization, the International Labour Organization and the United Nations Environment
38 Programme. WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland
39 (tel.: +41 22 791 2476).
40 Zanobetti (2009). Shrunken estimates for PM2 5 Personal Communication, June 1, 2009. (e-mail with attachments
41 placed in docket #EPA-HQ-OAR-2007-0492).
February 2010 7-4 Draft - Do Not Quote or Cite
-------
1 Zanobetti, A.; Schwartz, J. (2009). The effect of fine and coarse paniculate air pollution on mortality: A National
2 Analysis. Environ Health Perspect. 117(6): 898-903.
3 Zeka, A.; Zanobetti, A.; Schwartz, J. (2005). Short term effects of paniculate matter on cause specific mortality:
4 effects of lags and modification by city characteristics. Occup Environ Med. 62(10):718-25.
5 Zeka, A.; Zanobetti, A.; Schwartz, J. (2006). Individual-level modifiers of the effects of paniculate matter on daily
6 mortality. American Journal of Epidemiology. 163(9): 849-859.
February 2010 7-5 Draft - Do Not Quote or Cite
-------
APPENDIX A: AIR QUALITY ASSESSMENT
A-l
-------
1 Appendix A. Air Quality Assessment
2
3 This Appendix describes the PM data for the 15 urban study areas evaluated in the risk
4 assessment, including summaries of PM2.5 monitoring data associated with each study area as
5 well as the composite monitor estimates generated for each study area based on that monitoring
6 data (see section 3.2 for additional detail regarding selection of monitors and derivation of
7 composite monitor values).
9
10
11
12
13
14
15
16
17
18
A-2
-------
Table A-1. Air Quality Data for Atlanta
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
130630091 (3)
1 30670003 (1'2'3)
1 30670004 (1'2'3)
1 30890002 (1'2'3)
130892001 (1'2'3)
131210032 (1'2'3)
131210039 (1'2'3)
131210048 (1'2'3)
131 350002 (1'3)
1 32230003 (3)
Composite monitor for Atlanta - 1
Composite monitor for Atlanta - 2
Composite monitor for Atlanta - 3
27
27
30
82
80
84
27
0
13
28
90
90
90
30
30
28
84
75
89
30
0
14
29
91
91
91
25
29
26
81
67
76
23
0
12
26
92
92
92
30
29
27
88
85
80
29
0
14
26
92
92
92
112
115
111
335
307
329
109
0
53
109
365
365
365
12.63
13.75
12.98
12.72
12.84
13.64
15.03
14.35
11.41
13.62
13.49
13.26
16.83
17.39
17.17
15.72
15.10
16.00
18.35
14.62
15.52
16.34
16.62
16.30
21.22
18.57
18.03
18.81
20.44
19.43
17.97
20.39
18.62
19.09
18.87
19.27
15.92
15.62
13.98
14.56
14.83
14.38
16.56
15.16
12.99
15.01
14.99
14.89
16.65
16.33
15.54
15.45
15.80
15.86
16.98
16.13
14.63
16.01
15.99
15.93
36.09
34.94
30.28
32.82
36.72
33.40
30.29
31.66
34.52
31.03
31.52
31.06
2006
130630091 (3)
1 30670003 (1'2'3)
1 30670004 (1'2'3)
1 30890002 (1'2'3)
130892001 (1'2'3)
131210032 (1'2'3)
131210039 (1'2'3)
131210048 (1'2'3)
131 350002 (1'3)
1 32230003 (3)
Composite monitor for Atlanta - 1
Composite monitor for Atlanta - 2
Composite monitor for Atlanta - 3
29
28
28
85
86
88
29
0
12
29
90
90
90
29
29
29
86
84
86
28
0
14
27
91
91
91
31
31
27
81
77
84
26
2
13
31
92
92
92
30
30
28
81
81
90
0
30
15
29
92
92
92
119
118
112
333
328
348
83
32
54
116
365
365
365
12.94
12.22
12.09
12.25
11.94
12.46
15.12
15.21
10.91
13.04
12.68
12.79
17.91
17.88
17.75
16.09
15.75
15.99
19.15
18.98
15.20
17.37
17.10
17.19
21.32
21.52
21.04
19.86
18.31
19.28
20.88
15.25
20.31
18.90
20.17
20.15
20.16
14.49
14.20
12.39
13.43
12.18
13.74
15.00
12.93
10.77
13.41
13.49
13.24
16.67
16.46
15.82
15.41
14.54
15.37
16.86
13.95
16.00
15.86
15.84
30.84
32.66
33.34
31.65
28.89
31.44
30.64
32.28
27.34
27.89
26.82
A-2
-------
Table A-1 cont'd. Air Quality Data for Atlanta
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2007
130630091 (3)
1 30670003 (1'2'3)
1 30670004 (1'2'3)
1 30890002 (1'2'3)
130892001 (1'2'3)
131210032 (1'2'3)
131210039 (1'2'3)
131210048 (1'2'3)
131 350002 (1'3)
1 32230003 (3)
Composite monitor for Atlanta - 1
Composite monitor for Atlanta - 2
Composite monitor for Atlanta - 3
29
29
26
85
69
87
0
28
27
29
90
90
90
30
30
27
83
79
88
0
28
27
30
91
91
91
30
29
30
90
76
91
0
31
29
29
92
92
92
29
29
30
85
75
85
0
28
29
30
92
92
92
118
117
113
343
299
351
0
115
112
118
365
365
365
13.87
13.49
12.50
12.78
12.48
12.99
13.45
13.05
12.21
12.96
12.95
12.98
16.51
17.03
17.47
15.54
17.11
17.95
18.97
14.03
17.12
16.87
17.35
16.86
18.83
19.49
18.77
19.38
20.04
19.64
18.24
17.97
18.95
19.08
19.26
19.03
13.02
13.41
11.39
12.15
12.38
13.08
12.83
11.68
10.64
12.42
12.54
12.29
15.56
15.85
15.03
14.96
15.50
15.91
15.87
14.18
14.73
15.33
15.52
15.29
36.04
35.51
33.54
34.22
37.42
35.10
37.52
30.19
33.82
31.82
31.35
30.59
Note 1: Different definitions of Atlanta include different monitors. The number(s) shown in the parenthesis next to the monitor indicates the location(s) in which it is
included. For example, monitor 130630091 is used in Atlanta- 3 only while 130670003 is used for all definitions of Atlanta.
Note 2: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-4
-------
Table A-2. Air Quality Data for Baltimore
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
240051007
240053001
245100006
245100007
245100008
245100035
245100040
245100049
Composite Monitor for Baltimore
30
75
28
27
24
79
79
26
90
28
80
31
27
30
75
81
30
91
27
85
27
30
30
78
90
25
92
27
92
28
30
29
70
76
27
92
112
332
114
114
113
302
326
108
365
14.78
16.09
15.76
16.09
18.85
17.58
18.47
17.72
16.91
11.86
12.60
12.47
12.50
14.16
13.59
14.68
13.19
13.13
20.66
18.27
20.18
20.05
20.99
20.24
19.40
20.62
20.05
12.34
13.44
11.67
13.00
14.80
14.12
13.42
12.77
13.19
14.91
15.10
15.02
15.41
17.20
16.38
16.49
16.07
15.82
33.76
35.77
33.17
35.27
39.16
37.49
39.45
36.43
32.98
2006
240051007
240053001
245100006
245100007
245100008
245100035
245100040
245100049
Composite Monitor for Baltimore
29
90
27
30
30
74
85
0
90
29
85
30
29
28
90
86
0
91
28
90
27
29
31
83
87
0
92
30
92
30
31
30
82
86
0
92
116
357
114
119
119
329
344
0
365
12.03
12.81
13.20
12.64
14.80
13.31
13.83
13.23
11.37
11.79
11.62
11.59
13.34
12.57
12.58
12.12
15.73
18.51
16.24
15.19
16.88
19.27
18.64
17.21
11.09
13.90
11.61
12.03
12.97
14.14
14.73
12.92
12.55
14.25
13.17
12.86
14.50
14.82
14.94
13.87
32.06
34.25
32.67
32.27
35.21
36.74
35.93
31.34
2007
240051007
240053001
245100006
245100007
245100008
245100035
245100040
245100049
Composite Monitor for Baltimore
29
74
30
29
30
79
82
0
90
29
87
29
30
30
85
85
0
91
31
83
31
30
31
74
89
0
92
30
89
27
28
27
76
76
0
92
119
333
117
117
118
314
332
0
365
12.09
12.53
12.10
12.07
13.53
12.11
13.42
12.55
13.54
12.95
12.83
13.20
14.68
14.03
13.66
13.55
15.53
16.93
16.28
15.84
16.90
17.23
16.32
16.43
12.04
13.70
11.16
12.44
14.79
13.23
13.35
12.96
13.30
14.03
13.09
13.39
14.97
14.15
14.19
13.87
31.46
34.01
31.55
33.31
35.25
33.77
34.39
28.41
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-5
-------
Table A-3. Air Quality Data for Birmingham
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
10730023*
10731005*
10731009*
10731010*
10732003*
10732006*
10735002*
10735003*
11170006
11270002
Composite Monitor for
Birmingham - 1
Composite Monitor for
Birmingham - 2
90
30
30
15
88
30
30
30
30
27
90
90
90
31
31
15
90
30
31
30
31
31
91
91
89
29
29
15
91
30
30
30
30
28
92
92
92
31
31
16
91
31
31
31
28
30
92
92
361
121
121
61
360
121
122
121
119
116
365
365
14.35
11.62
9.82
11.71
14.49
11.53
10.84
10.60
11.23
10.37
11.66
11.87
20.49
16.70
16.12
16.91
18.48
16.46
16.33
16.42
15.67
15.31
16.89
17.24
26.42
22.61
20.26
22.77
23.75
21.11
21.08
21.94
19.60
18.86
21.84
22.49
17.27
14.33
11.87
15.51
15.03
13.79
12.61
12.74
12.92
12.17
13.82
14.14
19.63
16.32
14.52
16.73
17.94
15.72
15.21
15.43
14.85
14.18
16.05
16.44
49.68
35.06
37.68
36.46
44.41
33.98
36.23
39.20
32.86
33.17
35.47
36.27
2006
10730023*
10731005*
10731009*
10731010*
10732003*
10732006*
10735002*
10735003*
11170006
11270002
Composite Monitor for
Birmingham - 1
Composite Monitor for
Birmingham - 2
89
30
30
15
89
30
30
29
30
29
90
90
91
30
29
15
90
30
30
30
30
30
91
91
92
31
30
15
90
31
31
30
31
30
92
92
92
31
30
16
92
31
31
30
31
29
92
92
364
122
119
61
361
122
122
119
122
118
365
365
13.61
10.51
8.81
11.57
14.41
10.76
9.87
10.37
9.95
9.85
10.97
11.24
20.57
18.84
17.16
18.63
20.48
18.08
17.15
17.42
16.37
17.49
18.22
18.54
22.35
19.59
17.78
18.71
21.62
20.02
19.61
18.84
18.38
17.38
19.43
19.82
17.02
13.38
10.02
12.37
15.67
12.33
10.60
11.31
11.65
11.83
12.62
12.84
18.39
15.58
13.44
15.32
18.05
15.30
14.31
14.48
14.09
14.14
15.31
15.61
39.55
33.14
31.69
32.28
40.18
31.69
33.16
33.22
29.79
34.53
30.49
30.91
A-6
-------
Table A-3 cont'd. Air Quality Data for Birmingham
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(ug/m3)
2007
10731010*
10732003*
10732006*
10735002*
10735003*
11170006
11270002
Composite Monitor for
Birmingham - 1
Composite Monitor for
Birmingham - 2
15
89
30
30
29
29
28
90
90
15
90
30
28
30
30
29
91
91
15
89
31
31
31
31
31
92
92
15
90
30
30
30
30
29
92
92
60
358
121
119
120
120
117
365
365
14.53
15.40
12.24
12.15
11.79
12.97
11.97
12.99
13.12
18.69
21.38
19.29
19.16
18.99
18.27
17.81
19.62
20.02
19.31
19.18
18.53
18.41
17.83
17.52
17.72
18.58
18.82
13.63
12.42
10.93
10.40
10.38
10.84
10.95
11.60
11.78
16.54
17.10
15.25
15.03
14.75
14.90
14.61
15.70
15.93
37.92
44.02
39.92
37.90
38.56
38.52
34.91
37.65
38.40
Note 1: The monitors marked with * are used for Birmingham - 2. All monitors shown in this
Note 2: The information on the composite monitors in this table is based on the composite
table are used for Birmingham -1.
monitors after missing values have been filled
in.
A-7
-------
Table A-4. Air Quality Data for Dallas
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(ug/m3)
2005
481130035
481130050
481130057
481130069
481130087
481133004
Composite Monitor for Dallas
30
15
27
78
27
88
90
31
30
21
88
31
89
91
20
27
22
90
30
61
92
0
31
0
91
30
0
92
81
103
70
347
118
238
365
11.78
11.95
12.00
11.07
9.87
10.86
11.26
15.16
15.01
16.07
13.80
13.32
13.58
14.49
13.90
15.64
14.41
14.03
13.45
12.82
14.04
12.47
11.11
10.18
11.25
13.77
12.50
11.70
12.76
28.55
27.44
24.55
26.93
2006
481130035
481130050
481130057
481130069
481130087
481133004
Composite Monitor for Dallas
0
28
0
84
30
0
90
0
30
0
90
30
0
91
0
31
0
92
30
0
92
0
31
0
90
28
0
92
0
120
0
356
118
0
365
10.99
9.97
9.22
10.06
12.53
12.15
11.66
12.11
12.98
11.73
10.89
11.87
10.68
9.26
8.45
9.46
11.79
10.78
10.05
10.88
22.16
21.99
19.55
19.22
2007
481130035
481130050
481130057
481130069
481130087
481133004
Composite Monitor for Dallas
0
29
0
88
28
0
90
0
28
0
91
21
0
91
0
30
0
91
29
0
92
0
0
0
79
30
0
92
0
87
0
349
108
0
365
11.54
10.13
9.96
10.54
11.76
10.91
11.16
11.27
15.42
13.78
12.70
13.97
10.14
9.30
9.72
11.24
10.78
11.38
23.24
20.03
21.87
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-8
-------
Table A-5. Air Quality Data for Detroit
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
261630001
261630015
261630016
261630019
261630025
261630033
261630036
261630038
261630039
Composite Monitor for Detroit
88
27
87
28
26
28
29
28
0
90
87
27
79
31
28
31
28
25
0
91
89
30
84
29
30
28
29
22
7
92
86
30
88
29
30
28
27
0
28
92
350
114
338
117
114
115
113
75
35
365
18.45
20.20
18.92
19.82
17.86
21.50
16.96
16.98
18.84
13.87
14.73
14.78
14.48
11.74
16.57
14.92
14.60
14.46
17.15
18.73
16.62
17.43
17.45
18.22
18.58
17.66
18.20
17.73
14.38
15.18
13.70
14.20
12.68
17.90
15.19
14.25
14.69
15.96
17.21
16.01
16.48
14.94
18.55
16.41
16.43
42.31
48.27
47.80
51.37
39.50
48.69
46.22
44.06
2006
261630001
261630015
261630016
261630019
261630025
261630033
261630036
261630038
261630039
Composite Monitor for Detroit
82
29
79
30
27
28
29
0
29
90
85
26
14
15
14
29
26
29
30
91
88
28
13
14
15
27
29
27
31
92
90
31
17
16
17
31
29
28
30
92
345
114
123
75
73
115
113
84
120
365
13.66
16.98
13.04
15.20
13.49
18.79
15.10
14.78
15.13
11.89
12.26
11.58
10.39
11.23
12.85
10.95
11.10
11.71
11.55
13.68
14.93
12.58
11.78
10.01
15.56
13.69
14.34
14.20
13.42
13.65
14.56
14.97
13.46
12.70
17.30
11.94
11.98
11.84
13.60
13.22
14.68
13.04
12.71
11.86
16.13
12.92
13.13
13.42
32.82
35.89
35.49
35.67
30.00
42.43
32.91
32.32
28.34
2007
261630001
261630015
261630016
261630019
261630025
261630033
261630036
261630038
261630039
Composite Monitor for Detroit
86
28
26
30
26
29
29
27
29
90
89
30
26
28
30
29
28
27
30
91
87
27
30
31
31
29
30
28
30
92
92
29
29
27
27
27
29
30
28
92
354
114
111
116
114
114
116
112
117
365
12.92
15.15
13.98
13.20
12.23
18.84
13.75
13.63
13.83
14.17
10.28
13.06
12.12
11.16
10.59
15.20
11.96
12.85
12.98
12.24
14.00
15.12
14.74
14.36
13.76
16.02
14.60
15.35
14.65
14.73
14.08
14.82
14.61
13.31
14.42
17.49
13.47
14.23
13.86
14.48
12.82
14.54
13.86
13.01
12.75
16.89
13.45
14.01
13.83
13.91
31.19
32.73
33.72
31.09
32.49
36.60
28.48
33.38
33.97
27.66
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-9
-------
Table A-6. Air Quality Data for Fresno
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
60190008
60195001
60195025
Composite Monitor for Fresno
85
30
30
90
78
15
15
91
89
15
13
92
91
22
31
92
343
82
89
365
19.53
17.11
20.24
18.96
7.19
7.55
8.29
7.68
11.42
10.78
11.24
11.14
28.65
29.95
27.92
28.84
16.70
16.35
16.92
16.65
67.64
64.56
71.90
63.26
2006
60190008
60195001
60195025
Composite Monitor for Fresno
89
30
30
90
87
15
15
91
87
14
12
92
85
29
31
92
348
88
88
365
21.82
18.38
20.13
20.11
9.10
9.47
9.81
9.46
12.39
12.99
13.66
13.01
23.85
24.96
26.87
25.22
16.79
16.45
17.62
16.95
50.06
53.69
57.60
47.46
2007
60190008
60195001
60195025
Composite Monitor for Fresno
87
29
29
90
90
13
14
91
88
14
15
92
91
27
30
92
356
83
88
365
27.61
23.70
24.91
25.41
8.32
7.16
8.73
8.07
10.70
9.91
9.65
10.09
28.71
24.91
24.10
25.90
18.84
16.42
16.85
17.37
66.95
61.01
57.53
57.42
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-10
-------
Table A-7. Air Quality Data for Houston
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
482010024
482010026
482010055
482010058
482011034
482011035
Composite Monitor for Houston
26
23
25
20
10
84
90
31
31
28
28
15
68
91
22
20
19
23
10
78
92
15
0
0
26
0
87
92
94
74
72
97
35
317
365
11.77
10.47
9.12
11.95
11.79
13.09
11.28
14.39
13.10
12.31
12.99
15.36
16.59
14.12
17.17
14.47
12.97
14.40
14.49
18.41
15.48
11.83
...
12.19
15.47
13.16
13.79
...
12.88
15.89
13.51
26.00
...
24.61
30.10
25.12
2006
482010024
482010026
482010055
482010058
482011034
482011035
Composite Monitor for Houston
15
0
0
26
0
85
90
13
0
0
29
0
87
91
13
0
0
29
0
88
92
13
0
0
29
0
88
92
54
0
0
113
0
348
365
10.92
9.74
13.98
11.55
11.66
12.34
18.15
14.05
15.97
9.04
17.38
14.13
12.58
9.82
14.48
12.29
12.78
10.24
16.00
13.01
23.80
21.93
32.01
23.67
2007
482010024
482010026
482010055
482010058
482011034
482011035
Composite Monitor for Houston
15
0
0
26
0
87
90
14
0
0
30
0
91
91
13
0
0
30
0
91
92
0
0
0
30
0
82
92
42
0
0
116
0
351
365
11.01
...
9.40
14.42
11.61
12.82
...
10.96
17.02
13.60
14.64
...
11.84
16.62
14.36
...
11.75
14.50
13.13
...
10.99
15.64
13.18
...
25.48
32.00
23.26
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-ll
-------
Table A-8. Air Quality Data for Los Angeles
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent Me
(ug/m3)
2005
60370002
60371002
60371103
60371201
60371301
60371602
60372005
60374002
60374004
60379033
Composite Monitor for Los
Angeles
65
29
90
25
29
29
30
87
90
28
90
78
25
84
29
26
9
26
82
84
30
91
87
30
87
28
28
9
26
88
87
27
92
62
22
89
22
31
29
31
67
83
18
92
292
106
350
104
114
76
113
324
344
103
365
11.37
17.01
15.26
12.27
16.68
16.90
12.98
13.39
12.64
8.18
13.67
13.97
13.75
13.78
11.97
13.28
11.63
12.95
11.54
10.83
8.27
12.26
20.71
18.55
19.62
15.01
18.15
17.13
17.15
16.21
15.63
9.96
16.78
21.78
21.95
22.48
16.18
21.75
22.31
17.28
22.56
19.59
9.00
19.49
16.96
17.82
17.79
13.86
17.46
16.99
15.09
15.93
14.67
8.85
15.55
51.56
50.47
52.91
35.69
47.18
52.65
42.71
40.11
37.44
15.96
38.75
2006
60370002
60371002
60371103
60371201
60371301
60371602
60372005
60374002
60374004
60379033
Composite Monitor for Los
Angeles
66
25
89
20
28
29
29
73
89
15
90
73
24
82
27
28
28
27
81
86
15
91
84
30
85
28
27
31
28
73
79
14
92
55
25
74
17
24
28
29
63
66
14
92
278
104
330
92
107
116
113
290
320
58
365
12.62
15.33
14.49
11.19
17.62
16.82
12.85
15.19
14.35
6.13
13.66
16.17
18.34
14.69
14.21
14.76
13.92
14.64
12.27
11.99
7.27
13.83
16.95
15.87
16.34
12.95
15.11
17.19
13.46
13.53
14.21
8.36
14.40
15.87
16.66
16.80
13.00
19.26
18.57
12.51
15.57
17.22
8.00
15.35
15.40
16.55
15.58
12.84
16.69
16.63
13.37
14.14
14.44
7.44
14.31
36.83
43.21
38.55
30.42
43.98
42.34
31.95
33.89
34.17
12.86
29.93
A-12
-------
Table A-8 cont'd. Air Quality Data for Los Angeles
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent Me
(ug/m3)
2007
60370002
60371002
60371103
60371201
60371301
60371602
60372005
60374002
60374004
60379033
Composite Monitor for Los
Angeles
64
23
67
22
25
27
28
76
65
15
90
77
26
83
26
27
27
23
86
81
15
91
74
27
90
28
29
21
30
88
90
15
92
77
22
84
19
25
26
27
82
90
15
92
292
98
324
95
106
101
108
332
326
60
365
13.57
13.64
16.25
9.50
16.98
16.75
12.62
15.45
13.84
6.73
13.53
17.11
15.96
16.05
13.24
14.05
14.01
15.60
12.42
12.26
7.67
13.84
14.68
15.36
14.62
12.55
13.00
15.18
14.02
11.50
11.30
9.00
13.12
17.47
22.47
20.19
17.72
19.99
20.45
15.24
19.04
17.31
8.67
17.85
15.71
16.86
16.78
13.25
16.00
16.60
14.37
14.60
13.68
8.02
14.59
48.71
45.32
49.41
28.90
45.22
49.40
43.62
39.96
33.25
19.28
35.51
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-13
-------
Table A-9. Air Quality Data for New York
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m )
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
360050080
360050083
360050110
360470122
360610056*
360610062*
360610079*
360610128*
360610134*
360810124
360850055
360850067
Composite Monitor for New
York City -1
Composite Monitor for New
York City -2
28
30
90
28
30
27
30
25
0
89
28
24
90
90
31
31
91
30
31
31
31
31
0
79
25
28
91
91
29
30
91
28
30
30
30
30
0
62
28
28
92
92
27
31
91
27
31
31
31
31
0
74
27
30
92
92
115
122
363
113
122
119
122
117
0
304
108
110
365
365
18.59
13.77
14.93
16.04
18.44
17.14
14.60
17.74
13.02
14.92
12.60
15.62
16.98
14.78
12.21
12.17
13.74
15.51
13.84
13.12
14.11
10.44
12.49
10.75
13.02
14.15
18.42
16.90
15.38
17.31
19.16
18.34
17.03
18.37
15.21
17.81
16.17
17.28
18.22
15.68
12.71
12.30
14.13
15.17
13.54
12.56
15.21
10.84
12.91
10.41
13.22
14.12
16.87
13.90
13.69
15.31
17.07
15.71
14.33
16.36
12.38
14.53
12.48
14.78
15.87
37.50
36.05
36.58
35.94
39.93
38.96
36.18
37.66
34.28
33.37
33.00
31.19
32.81
2006
360050080
360050083
360050110
360470122
360610056*
360610062*
360610079*
360610128*
360610134*
360810124
360850055
360850067
Composite Monitorfor New
York City -1
Composite Monitorfor New
York City -2
29
30
86
28
30
30
30
26
0
69
25
30
90
90
30
30
91
30
30
28
30
30
0
86
27
26
91
91
27
29
84
29
27
28
31
29
0
84
29
31
92
92
29
29
86
25
30
27
29
29
0
76
29
29
92
92
115
118
347
112
117
113
120
114
0
315
110
116
365
365
16.57
13.44
13.10
15.00
16.61
14.33
14.12
15.79
11.17
12.27
10.01
13.86
15.21
13.17
11.06
11.15
12.49
14.03
13.00
12.08
13.07
10.67
12.07
10.49
12.12
13.04
13.95
13.34
14.49
14.75
14.41
13.82
13.32
14.39
13.68
14.06
12.60
13.89
13.99
11.88
10.33
11.40
9.00
12.59
9.86
10.59
12.64
10.91
10.56
8.54
10.75
11.42
13.89
12.04
12.53
12.81
14.41
12.75
12.53
13.97
11.61
12.24
10.41
12.65
13.42
38.89
34.80
36.51
37.06
40.60
35.73
36.92
37.84
33.10
35.89
31.85
30.36
33.78
A-14
-------
Table A-9 cont'd. Air Quality Data for New York
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(ug/m3)
2007
360050080
360050083
360050110
360470122
360610056*
360610062*
360610079*
360610128*
360610134*
360810124
360850055
360850067
Composite Monitor for New
York City - 1
Composite Monitor for New
York City - 2
30
30
89
29
30
27
30
30
3
74
30
27
90
90
30
30
84
30
27
0
30
30
30
86
28
30
91
91
30
30
85
28
31
0
31
29
31
80
31
26
92
92
29
29
91
30
30
0
30
21
30
92
30
26
92
92
119
119
349
117
118
27
121
110
94
332
119
109
365
365
17.45
14.14
12.90
13.67
18.43
15.84
14.11
19.10
8.53
11.34
13.04
10.60
14.60
16.87
13.49
11.72
11.64
12.82
14.73
...
12.48
13.83
14.12
10.66
12.37
10.49
12.58
13.79
16.20
13.91
14.22
15.92
15.99
...
14.92
14.63
16.43
12.30
14.55
14.29
14.85
15.49
15.43
12.87
12.31
13.00
15.29
...
12.89
14.76
14.08
11.35
11.91
10.54
13.13
14.25
15.64
13.16
12.77
13.85
16.11
...
13.60
15.58
13.29
11.41
12.97
11.48
13.79
15.10
36.16
32.50
33.92
33.38
36.12
...
33.86
37.01
33.66
30.81
31.58
28.56
29.12
30.12
Note 1: The monitors marked with * are used for New York City - 2
Note 2: The information on the composite monitors in this table is
:. All monitors in the table are used for New York City -1.
based on the composite monitors after missing values have been filled
in.
A-15
-------
Table A-10. Air Quality Data for Philadelphia
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(ug/m3)
2005
421010003
421010004
421010020
421010024
421010047
421010057
421010136
Composite Monitor for Philadelphia
0
55
19
37
19
0
86
90
0
61
0
54
28
0
89
91
0
78
0
67
26
0
29
92
62
74
0
71
12
0
33
92
62
268
19
229
85
0
237
365
13.23
15.51
12.68
16.99
13.57
14.40
13.06
10.76
12.04
11.40
11.81
17.26
16.26
18.91
19.06
17.87
14.35
13.28
12.02
12.31
12.91
12.97
14.21
12.93
15.06
14.23
14.26
35.83
34.57
37.70
31.13
32.12
2006
421010003
421010004
421010020
421010024
421010047
421010057
421010136
Composite Monitorfor Philadelphia
85
81
0
34
40
0
47
90
26
70
0
70
67
0
50
91
0
53
0
71
45
0
79
92
0
84
0
80
47
0
73
92
111
288
0
255
199
0
249
365
12.21
12.74
11.52
14.44
11.97
12.58
8.74
11.85
10.56
14.57
12.06
11.55
17.23
16.17
18.04
16.29
16.93
12.41
11.34
15.04
12.25
12.76
13.56
12.40
15.52
13.14
13.46
38.08
34.60
35.91
36.36
33.46
2007
421010003
421010004
421010020
421010024
421010047
421010057
421010136
Composite Monitorfor Philadelphia
0
87
0
87
71
0
75
90
0
71
0
58
59
0
65
91
0
86
0
86
90
18
72
92
0
90
0
90
92
90
82
92
0
334
0
321
312
108
294
365
...
13.61
12.05
14.49
12.60
13.19
...
13.19
12.76
13.05
13.38
13.09
...
15.15
14.88
16.33
10.96
14.36
14.33
...
12.96
11.73
13.43
13.13
12.99
12.85
...
13.73
12.85
14.32
13.33
13.37
...
34.61
33.42
35.07
31.53
32.44
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-16
-------
Table A-11. Air Quality Data for Phoenix
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Ann ual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
40130019
40131003
40134003
40137020
40139997
Composite Monitor for Phoenix
32
0
29
0
29
90
32
22
31
30
31
91
30
30
27
29
30
92
31
29
31
31
31
92
125
81
118
90
121
365
11.04
10.94
9.04
10.34
10.78
8.77
13.04
8.08
8.69
9.87
11.11
8.26
10.40
7.72
7.58
9.01
18.37
9.72
16.98
9.46
13.56
13.62
12.83
12.84
9.72
10.71
39.88
34.73
27.48
26.03
2006
40130019
40131003
40134003
40137020
40139997
Composite Monitor for Phoenix
30
26
28
29
29
90
30
28
28
30
29
91
31
31
31
31
30
92
31
31
29
30
30
92
122
116
116
120
118
365
14.17
8.87
13.53
8.09
10.74
11.08
13.58
9.52
10.34
7.98
8.66
10.01
8.07
8.92
9.31
7.14
7.46
8.18
17.82
11.33
17.58
9.12
14.04
13.98
13.41
9.66
12.69
8.08
10.22
10.81
28.51
20.07
28.38
15.35
24.29
26.84
2007
40130019
40131003
40134003
40137020
40139997
Composite Monitorfor Phoenix
32
29
30
30
30
90
30
28
29
30
29
91
31
30
30
31
32
92
30
30
29
20
30
92
123
117
118
111
121
365
10.26
7.66
10.54
5.85
8.85
8.63
8.85
10.45
11.76
7.81
8.12
9.40
8.63
9.50
11.32
7.35
8.21
9.00
15.42
11.27
15.45
8.21
12.75
12.62
10.79
9.72
12.27
7.31
9.48
9.91
26.63
18.20
27.33
13.44
22.02
18.70
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-17
-------
Tabl e A-12. Ai r Qual ity Data for Pi tts bu rg h
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
420030008
420030021
420030064
420030067
420030093
420030095
420030116
420030133
420031008
420031301
420033007
420039002
Composite Monitor for Pittsburgh
89
28
88
26
13
14
23
14
30
29
15
13
90
90
27
90
28
11
13
29
13
29
29
13
13
91
92
30
92
29
12
14
28
13
30
29
14
14
92
89
27
86
27
13
15
26
9
29
26
15
15
92
360
112
356
110
49
56
106
49
118
113
57
55
365
13.80
12.91
16.28
12.32
10.66
12.79
13.82
13.54
12.79
14.39
14.13
12.95
13.37
15.29
14.99
22.26
13.95
13.83
14.49
16.42
12.62
15.60
16.86
14.25
14.01
15.38
20.72
22.00
25.94
20.35
23.66
21.55
21.68
20.51
21.90
23.90
24.36
21.32
22.32
13.40
11.49
21.10
10.26
9.63
9.83
12.66
9.51
13.52
13.37
12.71
11.25
12.58
15.80
15.35
21.40
14.22
14.44
14.67
16.15
14.04
15.95
17.13
16.36
14.88
15.91
42.23
35.01
69.46
33.87
41.68
36.09
38.72
27.32
40.11
38.22
30.68
37.93
41.92
2006
420030008
420030021
420030064
420030067
420030093
420030095
420030116
420030133
420031008
420031301
420033007
420039002
Composite Monitor for Pittsburgh
85
0
85
23
14
13
0
0
27
26
15
0
90
89
0
90
26
6
13
0
0
23
28
15
0
91
91
0
87
28
13
13
0
0
28
29
14
0
92
92
0
89
21
13
14
0
0
25
29
15
0
92
357
0
351
98
46
53
0
0
103
112
59
0
365
11.60
14.86
9.61
10.37
10.02
11.87
12.56
12.93
11.49
13.28
17.89
9.52
9.85
10.97
14.30
14.55
13.51
13.05
20.19
22.78
16.39
16.38
18.22
18.32
19.89
19.16
18.69
12.54
--
20.97
9.06
9.41
10.31
--
--
11.63
13.11
12.36
_.
11.95
14.40
19.13
11.14
11.50
12.38
14.03
15.03
14.49
13.79
37.44
55.70
28.04
29.46
36.70
37.54
37.73
34.73
33.16
A-18
-------
Tabl e A-12. Ai r Qual ity Data for Pi tts bu rg h
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2007
420030008
420030021
420030064
420030067
420030093
420030095
420030116
420030133
420031008
420031301
420033007
420039002
Composite Monitor for Pittsburgh
85
0
88
19
15
14
0
0
27
28
14
0
90
86
0
90
25
12
13
0
0
27
27
14
0
91
86
0
91
28
14
15
0
0
30
31
14
0
92
89
0
90
26
14
14
0
0
27
26
13
0
92
346
0
359
98
55
56
0
0
111
112
55
0
365
11.80
14.16
10.28
9.67
10.96
12.79
14.02
12.36
11.87
14.72
18.64
13.40
10.50
9.89
14.55
15.18
13.03
13.51
20.30
25.16
19.46
19.35
20.79
19.68
21.90
21.19
20.74
12.74
-..
17.57
10.73
12.57
12.90
-..
-..
13.23
15.16
13.85
13.36
14.89
18.88
13.47
13.02
13.64
15.06
16.56
15.11
14.87
39.35
54.67
40.80
32.56
32.40
39.60
43.57
34.74
36.08
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-19
-------
Table A-13. Air Quality Data for Salt Lake City
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Q uarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(ug/m3)
2005
490350003
490350012
490351001
490353006
490353007
490353008
490353010
Composite Monitor for Salt Lake City
30
82
29
88
28
30
0
90
29
89
30
90
27
31
0
91
30
85
28
90
29
24
0
92
31
85
30
85
28
31
0
92
120
341
117
353
112
116
0
365
14.16
16.73
11.85
13.95
13.64
9.90
13.37
6.58
9.59
5.47
6.27
7.40
6.03
6.89
8.98
12.68
8.61
9.56
10.57
7.76
9.69
14.49
17.24
11.35
14.17
16.36
7.45
13.51
11.06
14.06
9.32
10.99
11.99
7.79
10.87
41.66
43.36
36.25
43.23
39.37
26.61
36.45
2006
490350003
490350012
490351001
490353006
490353007
490353008
490353010
Composite Monitor for Salt Lake City
28
76
27
88
30
29
0
90
28
87
28
90
30
26
0
91
29
82
29
90
31
30
0
92
30
90
27
88
29
30
0
92
115
335
111
356
120
115
0
365
10.76
11.80
7.95
10.59
10.11
6.14
9.56
6.98
11.22
5.65
7.21
7.18
6.85
7.51
9.41
14.19
8.65
8.54
11.56
9.26
10.27
13.58
14.91
9.29
12.37
13.61
7.09
11.81
10.18
13.03
7.88
9.68
10.61
7.33
9.79
38.67
37.93
27.72
37.54
35.69
21.97
29.80
2007
490350003
490350012
490351001
490353006
490353007
490353008
490353010
Composite Monitor for Salt Lake City
30
80
24
89
29
23
0
90
30
86
30
85
29
28
80
91
29
0
31
78
29
28
83
92
28
0
26
89
31
30
92
92
117
166
111
341
118
109
255
365
18.12
20.84
11.42
18.17
17.72
10.03
16.05
6.97
11.45
6.44
6.11
7.17
6.06
7.68
7.41
10.99
10.08
9.42
11.53
9.66
11.62
10.55
13.89
9.71
12.05
13.42
7.09
13.00
11.53
12.49
9.41
11.44
12.46
8.21
11.39
55.65
29.84
54.28
50.13
23.02
49.06
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-20
-------
Table A-14. Air Quality Data for St. Louis
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
171190023*
171190024*
171191007*
171192009*
171193007*
171630010*
171634001*
290990012
291890004*
291892003*
295100007*
295100085*
295100086*
295100087*
295100093*
Composite Monitor for St Louis - 1
Composite Monitor for St Louis - 2
28
0
26
12
29
13
30
90
29
57
88
90
84
90
0
90
90
28
0
31
12
31
15
30
87
29
30
88
86
26
87
0
91
91
29
0
29
13
27
14
29
90
28
29
83
78
30
82
0
92
92
29
0
30
12
29
15
28
91
31
31
81
88
29
81
0
92
92
114
0
116
49
116
57
117
358
117
147
340
342
169
340
0
365
365
18.01
18.40
14.94
16.42
17.31
17.86
15.22
16.01
16.73
16.99
16.78
15.11
17.02
...
16.68
16.80
19.10
16.49
16.35
15.20
16.81
14.17
14.69
12.64
14.15
14.67
14.46
14.34
14.80
...
15.22
15.27
21.49
21.47
20.82
19.99
19.97
17.20
19.26
17.80
18.44
18.92
19.67
18.43
18.74
...
19.40
19.41
16.95
16.27
11.98
12.49
14.47
14.69
12.42
11.87
12.65
12.87
13.33
13.14
12.94
...
13.54
13.64
18.89
18.16
16.02
16.02
17.14
15.98
15.40
14.58
15.49
15.86
16.06
15.26
15.88
...
16.21
16.28
41.17
43.68
39.63
41.08
39.59
37.61
39.86
37.57
40.00
38.44
39.81
39.57
40.80
...
37.87
37.78
2006
171190023*
171190024*
171191007*
171192009*
171193007*
171630010*
171634001*
290990012
291890004*
291892003*
295100007*
295100085*
295100086*
295100087*
295100093*
Composite Monitor for St Louis - 1
Composite Monitor for St Louis - 2
30
0
27
15
28
12
28
82
30
29
78
86
30
85
0
90
90
26
0
24
15
30
14
28
81
29
29
88
77
30
90
0
91
91
31
0
24
14
31
15
31
91
0
28
91
84
31
86
0
92
92
29
0
27
16
31
14
29
89
0
26
90
92
29
91
0
92
92
116
0
102
60
120
55
116
343
59
112
347
339
120
352
0
365
365
15.21
14.95
12.59
13.08
14.18
13.43
11.62
10.56
11.36
12.27
13.04
11.94
12.92
...
12.86
12.96
17.34
16.12
13.35
12.00
13.75
12.87
11.79
10.49
10.69
11.82
12.46
11.55
12.32
...
12.81
12.90
19.40
20.18
13.49
16.47
15.72
15.20
15.46
...
13.87
15.89
15.26
15.48
16.17
...
16.05
16.10
12.11
14.05
12.92
10.87
14.48
12.00
11.49
...
11.00
12.51
12.68
10.90
13.18
...
12.35
12.43
16.02
16.32
13.08
13.11
14.53
13.38
12.59
...
11.73
13.12
13.36
12.47
13.65
...
13.52
13.60
32.81
36.24
27.28
27.54
29.18
27.92
30.20
...
27.61
29.39
28.52
30.46
29.60
...
25.08
24.78
A-21
-------
Table A-14 cont'd. Air Quality Data for St. Louis
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(ug/m3)
2007
171190023*
171190024*
171191007*
171192009*
171193007*
171630010*
171634001*
290990012
291890004*
291892003*
295100007*
295100085*
295100086*
295100087*
295100093*
Composite Monitor f or St Louis - 1
Composite Monitor f or St Louis - 2
0
0
29
15
29
13
26
82
0
89
88
90
27
90
0
90
90
0
0
27
12
28
13
30
81
0
90
91
88
30
86
0
91
91
0
6
29
14
26
14
31
90
0
91
91
89
0
92
24
92
92
0
29
26
13
30
14
29
86
0
90
92
90
0
86
29
92
92
0
35
111
54
113
54
116
339
0
360
362
357
57
354
53
365
365
...
14.28
14.31
12.42
14.94
13.35
11.94
11.63
12.56
12.59
11.79
13.24
13.00
13.11
...
15.31
16.02
14.84
17.65
13.95
14.44
12.96
14.50
13.79
14.50
14.43
14.76
14.79
15.07
17.61
15.66
17.39
15.94
14.83
16.23
15.25
16.13
16.09
16.61
17.26
16.27
16.28
...
14.94
13.23
13.51
12.32
13.79
10.90
12.13
12.49
12.97
13.30
13.10
13.82
13.04
13.13
...
15.11
14.88
14.24
15.58
13.26
13.68
13.09
14.04
13.94
14.34
14.27
14.33
...
35.86
34.98
34.45
33.08
32.27
31.92
30.28
31.61
32.06
33.72
31.51
31.52
Note 1: The monitors marked with * are used for St
Note 2: The information on the composite monitors
Louis - 2. All monitors shown in the table are used for St Louis -1.
in this table is based on the composite monitors after missing values have been filled in.
A-22
-------
Table A-15. Ai r Quality Data for Tacoma
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Ann ual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
530530029
Composite MonitorforTacoma
29
90
30
91
30
92
31
92
120
365
16.46
16.46
5.34
5.34
7.13
7.13
17.07
17.07
11.50
11.50
40.42
39.61
2006
530530029
Composite MonitorforTacoma
30
90
30
91
31
92
26
92
117
365
8.92
8.92
5.89
5.89
7.45
7.45
15.93
15.93
9.55
9.55
39.82
37.05
2007
530530029
Composite MonitorforTacoma
29
90
28
91
31
92
29
92
117
365
13.76
13.76
5.94
5.94
5.23
5.23
13.76
13.76
9.67
9.67
45.11
41.26
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
A-23
-------
APPENDIX B: HYBRID (NON-PROPORTIONAL) AND PEAK
SHAVING ROLLBACK APPROACHES
B-l
-------
1 Appendix B. Methodologies for Rolling Back PM2.s Concentrations Due to Local Source
2 Impacts (hybrid non-proportional and peak shaving approaches)
O
4 During the last review of the Particulate Matter National Ambient Air Quality Standards
5 (NAAQS), a technique was employed to simulate fine particulate concentrations under a series
6 of attainment scenarios to determine the risk associated with each. The "rolling back" of the
7 concentrations consisted of simply using a proportional rollback calculation where every
8 measured concentration value was multiplied by a constant to obtain a set of concentrations
9 which would meet alternative standard levels. This technique was reviewed by the Clean Air
10 Scientific Advisory Committee (CASAC) and was considered to be a satisfactory way to
11 simulate alternative PM2 5 distributions. The rolled back values, however, constituted only a
12 regional reduction in PM concentrations without accounting in any way for emission reductions
13 at local point sources.
14
15 The Hybrid Non-Proportional Approach
16
17 For the current review, an alternative rollback approach reflecting the combined effects
18 of both local and regional reduction strategies was considered (this alternative approach is
19 referred to as the hybrid non-proportional approach in the risk assessment). In addition to
20 utilizing a traditional proportional rollback to represent the regional PM reductions, a distance-
21 weighted rollback was conducted on a subset of the 15 study areas which contain source-oriented
22 monitors measuring concentrations higher than those observed at other sites within a particular
23 area.1
24 Unique sites with high design values exceeding the NAAQS were further investigated to
25 determine if they were in close proximity to a large source of PM2.5 (Figure B-l). The presence
26 of possible source-oriented sites in each area was visually determined using satellite photographs
27 provided by Google Earth. Areas where source-oriented adjustments were made include Detroit
28 MI, Pittsburgh PA, St. Louis MO-IL, Baltimore MD, New York NY, Los Angeles CA and
29 Birmingham AL.
30
31
32
33
1 In the risk assessment, as outlined in Section 3.1, the proportional rollback approach was used in
generating the core risk estimates, while the hybrid non-proportional approach described here, was considered as
part of the sensitivity analysis.
B-2
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Detroit, Ml (261630033)
' *'7^k\'VB^BI OR
Dearborn monitor
(marked by blue
circle) is located
adjacent to a large
rail yard and
Ford River
Rouge Plant
(encircled in red)
Figure B-l. Example of a monitor, in Dearborn MI, located near a large source of
emissions
For those sites that were within proximity to a large emitter, the site's measured
concentrations were reduced using a proportional rollback depending on the magnitude of the
reduction needed to either the highest 24-hour or annual design value of a non-source oriented
site within the area whose design values were close to those of the source oriented site (Figure B-
2).
B-2
-------
Drtnt. Ml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
u « is is
1 Concentration
values ofthe site
in the blue circle
get reduced
so that eiher
the 24-hour or
annual average
design value
matches the
design value
ofthe sie in
the red circle.
The reduction
amount is
dependent on
the difference
between the
two sites' 24-hour
and annual design
values and
concentrations are
adjusted by the
larger of the two
values.
Figure B-2. Plot of the 24-hour versus the annual average PM2.s design values for
individual sites in Detroit MI
The fractional reduction made to the site near the point source was then weighted by the
inverse distance in kilometers between the source-oriented site and all ofthe other individual
sites in the area to determine their fractional reductions in relation to the source-oriented site. If
more than one source-oriented site was reduced, a distance-weighted average fractional reduction
was calculated and implemented across the non-source-oriented sites. Sites within one kilometer
ofthe source oriented site received the same amount of reduction as the source oriented site. An
example ofthe effect of this reduction technique for Detroit is presented in Table B-l. For
Detroit, adjustments were based on the difference between the two sites' annual design values.
B-4
-------
1 Table B-l. Comparison of the original and adjusted design values for Detroit, MI
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Site ID
260490021
260990009
261150005
261250001
261470005
261610005
261610008
261630001
261630015
261630016
261630019
261630025
261630033
261630036
261630038
261630039
Original Annual
Design Value
(2005-2007)
11.6
12.5
13.8
13.6
13.2
13.2
13.7
14
15.5
14.3
14.1
13.2
17.2
14.3
14.3
14.4
Adjusted
Annual Design
Value (2005-
2007)
11.5
12.4
13.7
13.5
13.1
13.1
13.6
13.9
15.2
14.2
14
13.1
15.4
14.2
14.1
14.3
Original 24-hour
Design Value
(2005-2007)
29
35
38
40
41
39
39
36
40
41
40
34
43
36
40
37
Adjusted 24-hour
Design Value
(2005-2007)
29
35
38
40
40
39
39
36
39
41
40
34
39
36
39
37
Site in blue represents source-oriented site
Site in red represents reference site used for reduction
Reduction of the concentrations of the source-oriented site reduced either the 24-hour or
annual design value of the site to either the maximum non-source-oriented site's 24-hour or
annual design value. This did not necessarily mean that the adjusted values at the source-
oriented site met either the 24-hour or annual standard after the reduction. Since the adjusted
design values were calculated using the same data handling rules as contained within 40 CFR
Part 50 Appendix N, truncation or rounding of the adjusted concentrations could sometimes give
adjusted design values at the source-oriented site that were not exactly the same value as the
original design value at the reference site. However, they were usually within 1 ug/m3 for the
24-hour standard and a few tenths of a microgram per cubic meter for the annual standard.
The Peak Shaving Approach
The peak shaving approach was used to calculate annual averages for 2005, 2006, and
2007 at composite monitors for comparison with the composite monitor annual averages
B-5
-------
1 calculated using the proportional and hybrid rollback approaches. Because of time constraints,
2 we did not calculate health risks when alternative standards are just met using the peak shaving
3 approach. However, because the C-R functions used in the risk assessment are almost linear, a
4 comparison of annual averages at composite monitors using the three different methods for
5 simulating just attaining standards should give a good idea of the corresponding estimates of
6 health risks when alternative standards are just met (see Section 3.5.4 for additional detail on the
7 composite monitor-based comparison of the three rollback strategies completed as part of the
8 sensitivity analysis).
9 We applied the peak shaving method only in those cases in which the daily standard in a
10 location is controlling (i.e., the percent rollback necessary to meet the daily standard is greater
11 than the percent rollback necessary to meet the annual standard in that location). Like the
12 proportional and hybrid rollback methods, the peak shaving method for calculating annual
13 averages at composite monitors starts with monitor-specific quarterly averages that have been
14 calculated as described in Section 3.2.1.
15 In contrast to the proportional and hybrid rollback approaches, the peak shaving method
16 uses monitor-specific design values. For each monitor, we compared the monitor-specific daily
17 design value to the daily standard and calculated the percent rollback necessary to get each
18 monitor above the 24hr standard level into attainment (using a formula that is analogous to the
19 proportional rollback formula given in Section 3.2.3.1). We then rolled back each quarterly
20 average at the monitor by this percent rollback. We calculated the average quarterly average
21 across all monitors in the location, for each quarter. Finally, we calculated the annual average at
22 the composite monitor under the standard by averaging the four quarterly averages calculated on
23 the previous step. See Section 3.2.2 for more detail.
24 The results of the peak shaving analysis are presented, along with results based on the
25 hybrid and proportional rollback approaches, as part of the sensitivity analysis results (see
26 Appendix F, Tables F-49 and F-50).
B-6
-------
APPENDIX C: EPI STUDY SPECIFIC INFORMATION ON
PM25
C-l
-------
1 Appendix C. Epidemiology Study-Specific Information for PM2.s Risk Assessment
2
3 This Appendix provides detailed summary information for the epidemiological studies
4 used to obtain the concentration-response (C-R) functions used in the risk assessment. For
5 additional details on selection of epidemiological studies and specification of the C-R functions,
6 see section 3.3.3.
7
C-2
-------
Table C-l. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Ail-Year Functions
Study
Health Endpoint
ICD-9 or 10 Codes
Ages
Covered
Model
Other
Pollutants in
Model
Lag
Metric
Region
Covered
Coefficient
Lower
Bound
Upper
Bound
Health Effects Associated with Long-Term Exposure to PM25:
Krewski et al.
(2009) - exposure
period 1979-1983
Krewski et al.
(2009) - exposure
period 1999-2000
Krewski et al.
(2000) [reanalysis
of Six Cities Study]
Mortality, all-cause
Mortality,
cardiopulmonary
Mortality, ischemic
heart disease
Mortality, lung cancer
Mortality, all-cause
Mortality,
cardiopulmonary
Mortality, ischemic
heart disease
Mortality, lung cancer
Mortality, all-cause
Mortality, ischemic
heart disease
Mortality, all-cause
Mortality,
cardiopulmonary
Mortality, ischemic
heart disease
Mortality, lung cancer
Mortality, all-cause
Mortality,
cardiopulmonary
Mortality, lung cancer
All
401-440,460-519
410-414
162
All
401-440,460-519
410-414
162
All
410-414
All
401-440,460-519
410-414
162
All
400-440, 485-495
162
30+
30+
30+
30+
25+
log-linear
log-linear
log-linear (random
effects)
log-log
log-linear
none
none
none
none
none
n/a
n/a
n/a
n/a
n/a
annual mean
annual mean
annual mean
annual mean
annual mean
National
National
Six U.S. Cities
0.00431
0.00898
0.01689
0.00880
0.00554
0.01293
0.02167
0.01293
0.00686
0.02437
0.10966
0.17225
0.35942
0.19284
0.00414
0.00561
-0.01133
0.00276
0.00677
0.01363
0.00325
0.00354
0.01007
0.01748
0.00554
0.00315
0.01450
0.06758
0.11261
0.24629
0.09861
0.00414
0.00561
-0.01133
0.00583
0.01115
0.02005
0.01432
0.00760
0.01587
0.02585
0.02029
0.01053
0.03429
0.15306
0.23161
0.47210
0.28797
0.02071
0.02789
0.04525
C-3
-------
Table C-l cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: All-Year Functions
Study
Health Endpoint
ICD-9 or 10 Codes
Ages
Covered
Model
Other
Pollutants in
Model
Lag
Metric
Region
Covered
Coefficient
Lower
Bound
Upper
Bound
Health Effects Associated with Short-Term Exposure to PM2S:
Bell et al. (2008)
Ito et al. (2007)
Moolgavkar (2003)
[reanalysis of
Moolgavkar
(2000a)]
HA (unscheduled),
cardiovascular
HA (unscheduled),
respiratory
ER visits, asthma
Mortality,
cardiovascular
426^27, 428,
430-438; 41 0^1 4,
429; 440^49
490^92; 464-466,
480-487
493
390-429
65+
65+
all ages
all ages
log-linear
log-linear
log-linear
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
none
none
none
none
CO
none
CO
0-day
2-day
avg of 0-
and 1-day
Oday
1 day
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
Northeast
Northwest
Southeast
Southwest
Northeast
Northwest
Southeast
Southwest
New York
Los Angeles
0.00107
0.00074
0.00029
0.00053
0.00028
0.00019
0.00035
0.00094
0.00453
0.00099
0.00097
0.00097
0.00178
0.00188
0.00103
0.00080
0.00069
0.00091
0.00091
0.00079
-0.00176
-0.00019
0.00000
-0.00017
-0.00255
-0.00044
0.00022
0.00286
0.00010
0.00014
-0.00002
0.00075
0.00067
0.00015
-0.00003
-0.00032
-0.00013
-0.00035
0.00136
0.00324
0.00077
0.00104
0.00072
0.00294
0.00113
0.00166
0.00621
0.00188
0.00180
0.00196
0.00281
0.00309
0.00191
0.00163
0.00170
0.00195
0.00217
C-4
-------
Table C-l cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: All-Year Functions
Study
Moolgavkar (2003)
[reanalysis of
Moolgavkar
(2000a)]
Health Endpoint
Mortality, non-
accidental
ICD-9 or 10 Codes
<800
Ages
Covered
all ages
Model
log-linear, GAM
(stringent), 30 df
log-linear, GLM, 30
df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GLM, 30
df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
Other
Pollutants in
Model
none
none
CO
none
Lag
Oday
1 day
1 day
Oday
1 day
2 day
3 day
4 day
5 day
Metric
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
Region
Covered
Los Angeles
Coefficient
0.00054
0.00040
0.00032
0.00030
0.00059
0.00055
0.00010
-0.00001
-0.00053
-0.00033
-0.00033
0.00054
0.00059
0.00038
-0.00015
-0.00009
-0.00056
Lower
Bound
-0.00007
-0.00034
-0.00023
-0.00043
0.00000
-0.00017
-0.00046
-0.00099
-0.00131
-0.00105
-0.00117
-0.00007
0.00000
-0.00019
-0.00073
-0.00064
-0.00115
Upper
Bound
0.00115
0.00114
0.00087
0.00103
0.00118
0.00127
0.00066
0.00097
0.00025
0.00039
0.00051
0.00115
0.00118
0.00095
0.00043
0.00046
0.00003
C-5
-------
Table C-l cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: All-Year Functions
Study
Moolgavkar (2003)
[reanalysis of
Moolgavkar
(2000a)]
Moolgavkar (2003)
[reanalysis of
Moolgavkar
(2000b)]
Health Endpoint
Mortality, respiratory
(COPD+)
HA, cardiovascular
ICD-9 or 10 Codes
490-496
390-429
Ages
Covered
all ages
65+
Model
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
Other
Pollutants in
Model
none
none
none
CO
none
CO
Lag
Oday
1 day
Oday
Oday
1 day
1 day
Metric
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
Region
Covered
Los Angeles
Los Angeles
Coefficient
-0.00056
-0.00142
-0.00121
0.00038
0.00086
0.00020
0.00158
0.00116
0.00126
0.00039
0.00058
0.00139
0.00113
0.00120
0.00024
0.00027
Lower
Bound
-0.00300
-0.00380
-0.00407
-0.00210
-0.00158
-0.00282
0.00091
0.00050
0.00045
-0.00044
-0.00041
0.00069
0.00046
0.00038
-0.00065
-0.00075
Upper
Bound
0.00188
0.00096
0.00165
0.00286
0.00330
0.00322
0.00225
0.00182
0.00207
0.00122
0.00157
0.00209
0.00180
0.00202
0.00113
0.00129
C-6
-------
Table C-l cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: All-Year Functions
Study
Moolgavkar (2003)
[reanalysis of
Moolgavkar
(2000c)]
Tolbert et al.
(2007)
Health Endpoint
HA, respiratory
(COPD+)
ER visits,
cardiovascular
ER visits, respiratory
ICD-9 or 10 Codes
490-496
410-414,427,428,
433-437, 440,
443-445, 451-453
493, 786.07, 786.09;
491,492, and 496;
460-465, 460.0, and
477; 480-486; 466.1,
466.11, and 466.19
Ages
Covered
all ages
all ages
all ages
Model
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 30 df
log-linear, GAM
(stringent), 100df
log-linear, GLM,
100df
log-linear, GAM
(stringent), 100df
log -linear
log -linear
Other
Pollutants in
Model
none
none
none
NO2
none
none
Lag
Oday
1 day
2 day
Oday
1 day
2 day
3 day
avg of 0-,1-
day, and 2-
day
avg of 0-,1-
day, and 2-
day
Metric
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
24-hr avg.
Region
Covered
Los Angeles
Atlanta
Atlanta
Coefficient
0.00167
0.00138
0.00149
0.00119
0.00075
0.00077
0.00185
0.00114
0.00103
0.00042
-0.00004
0.00035
-0.00109
0.00046
0.00046
Lower
Bound
0.00068
0.00052
0.00041
0.00022
-0.00011
-0.00027
0.00082
0.00021
-0.00012
-0.00091
-0.00161
-0.00102
-0.00238
-0.00064
-0.00046
Upper
Bound
0.00266
0.00224
0.00257
0.00216
0.00161
0.00181
0.00288
0.00207
0.00218
0.00175
0.00153
0.00172
0.00020
0.00154
0.00136
C-7
-------
Table C-l cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: All-Year Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality,
cardiovascular
ICD-9 or 10 Codes
101-159
Ages
Covered
all ages
Model
log-linear
Other
Pollutants in
Model
none
Lag
avg of 0-
and 1-day
Metric
24-hr avg.
Region
Covered
Atlanta
Baltimore
Birmingham
Dallas
Detroit
Fresno
Houston
Los Angeles
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
St. Louis
Tacoma
Coefficient
0.00066
0.00128
-0.00002
0.00086
0.00097
0.00082
0.00084
-0.00018
0.00196
0.00179
0.00142
0.00102
0.00117
0.00158
0.00104
Lower
Bound
-0.00066
-0.00009
-0.00140
-0.00056
-0.00012
-0.00056
-0.00056
-0.00080
0.00114
0.00046
-0.00006
-0.00020
-0.00027
0.00035
-0.00055
Upper
Bound
0.00198
0.00265
0.00135
0.00228
0.00205
0.00219
0.00223
0.00044
0.00278
0.00313
0.00291
0.00225
0.00260
0.00282
0.00262
-------
Table C-l cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: All-Year Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, non-
accidental
ICD-9 or 10 Codes
AOO-R99
Ages
Covered
all ages
Model
log-linear
Other
Pollutants in
Model
none
Lag
avg of 0-
and 1-day
Metric
24-hr avg.
Region
Covered
Atlanta
Baltimore
Birmingham
Dallas
Detroit
Fresno
Houston
Los Angeles
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
St. Louis
Tacoma
Coefficient
0.00094
0.00135
0.00032
0.00112
0.00068
0.00096
0.00104
0.00016
0.00132
0.00126
0.00110
0.00104
0.00105
0.00105
0.00117
Lower
Bound
0.00018
0.00054
-0.00050
0.00027
-0.00012
0.00014
0.00021
-0.00023
0.00077
0.00046
0.00018
0.00030
0.00021
0.00030
0.00020
Upper
Bound
0.00170
0.00215
0.00115
0.00198
0.00147
0.00178
0.00188
0.00055
0.00186
0.00206
0.00202
0.00177
0.00188
0.00180
0.00214
C-9
-------
Table C-l cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: All-Year Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, respiratory
ICD-9 or 10 Codes
JOO-J99
Ages
Covered
all ages
Model
log-linear
Other
Pollutants in
Model
none
Lag
avg of 0-
and 1-day
Metric
24-hr avg.
Region
Covered
Atlanta
Baltimore
Birmingham
Dallas
Detroit
Fresno
Houston
Los Angeles
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
St. Louis
Tacoma
Coefficient
0.00121
0.00211
0.00096
0.00093
0.00169
0.00175
0.00211
0.00112
0.00216
0.00157
0.00194
0.00149
0.00194
0.00132
0.00179
Lower
Bound
-0.00048
0.00039
-0.00076
-0.00084
0.00008
0.00006
0.00033
0.0001 1
0.00075
-0.00015
0.00015
-0.00014
0.00024
-0.00034
-0.00005
Upper
Bound
0.00290
0.00384
0.00268
0.00270
0.00330
0.00344
0.00388
0.00213
0.00356
0.00329
0.00374
0.00313
0.00364
0.00298
0.00363
C-10
-------
Table C-2. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Season-Specific Functions
Study
Bell et al. (2008)
Ito et al. (2007)
Health Endpoint
HA (unscheduled),
cardiovascular
HA (unscheduled),
respiratory
ER visits, asthma
ICD-9or10
Codes
426-427, 428,
430-438;
410-414,429;
440-449
490-492;
464-466,
480-487
493
Ages
Covered
65+
65+
all ages
Other
Pollutants
in Model
none
none
O3
NO2
CO
SO2
Lag
0-day
2-day
avg of 0-
and 1-day
Region
Covered
Northeast
Northwest
Southeast
Southwest
Northeast
Northwest
Southeast
Southwest
New York
New York
New York
New York
New York
Season
Covered
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
April-August
April-August
April-August
April-August
April-August
Coefficient
0.00199
0.00095
0.00055
0.00102
0.00085
-0.00007
-0.00156
-0.00067
0.00105
0.00075
-0.00067
0.00017
0.00076
0.00176
-0.00121
0.00030
0.00079
0.00004
0.00077
0.00012
-0.00006
0.00226
0.00074
-0.00074
0.00040
0.00075
-0.00052
0.00014
0.00119
0.00104
0.00238
0.00097
0.00759
0.00602
0.00334
0.00647
0.00469
Lower Bound
0.00138
0.00032
0.00008
0.00048
-0.00420
-0.01324
-0.01651
-0.00721
-0.00007
-0.00026
-0.00161
-0.00072
-0.00025
-0.00087
-0.00502
-0.00098
-0.00021
-0.00088
-0.00001
-0.00082
-0.00674
-0.01539
-0.02074
-0.01062
-0.00146
-0.00082
-0.00209
-0.00130
-0.00010
-0.00220
-0.00264
-0.00137
0.00486
0.00322
0.00029
0.00356
0.00163
Upper Bound
0.00260
0.00157
0.00101
0.00157
0.00589
0.01309
0.01337
0.00587
0.00219
0.00176
0.00026
0.00106
0.00177
0.00441
0.00262
0.00158
0.00178
0.00097
0.00155
0.00106
0.00663
0.01991
0.02220
0.00915
0.00224
0.00231
0.00105
0.00158
0.00249
0.00430
0.00741
0.00330
0.01032
0.00883
0.00640
0.00939
0.00775
C-ll
-------
Table C-2 cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Season-Specific Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, short-term
cardiovascular
ICD-9or10
Codes
101-159
Ages
Covered
all ages
Other
Pollutants
in Model
none
Lag
avg of 0-
and 1-day
Region
Covered
Atlanta
Atlanta
Atlanta
Atlanta
Baltimore
Baltimore
Baltimore
Baltimore
Birmingham
Birmingham
Birmingham
Birmingham
Dallas
Dallas
Dallas
Dallas
Detroit
Detroit
Detroit
Detroit
Fresno
Fresno
Fresno
Fresno
Houston
Houston
Houston
Houston
Los Angeles
Los Angeles
Los Angeles
Los Angeles
Season
Covered
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Coefficient
0.00135
0.00076
0.00062
-0.00018
0.00104
0.00085
0.00067
0.00296
0.00080
0.00016
-0.00004
-0.00189
0.00120
0.00125
0.00115
-0.00022
-0.00006
0.00166
0.00136
0.00226
-0.00033
0.00050
0.00019
0.00071
0.00070
0.00013
0.00183
0.00046
-0.00014
0.00007
-0.00106
0.00000
Lower Bound
-0.00193
-0.00273
-0.00222
-0.00293
-0.00196
-0.00269
-0.00251
-0.00017
-0.00283
-0.00333
-0.00301
-0.00485
-0.00214
-0.00222
-0.00223
-0.00349
-0.00203
-0.00045
-0.00099
-0.00001
-0.00201
-0.00138
-0.00173
-0.00105
-0.00285
-0.00347
-0.00142
-0.00246
-0.00109
-0.00113
-0.00253
-0.00099
Upper Bound
0.00462
0.00425
0.00347
0.00257
0.00405
0.00438
0.00384
0.00609
0.00443
0.00365
0.00293
0.00106
0.00454
0.00472
0.00453
0.00306
0.00191
0.00378
0.00371
0.00452
0.00135
0.00238
0.00211
0.00248
0.00425
0.00373
0.00509
0.00337
0.00080
0.00127
0.00042
0.00099
C-12
-------
Table C-2 cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Season-Specific Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, short-term
cardiovascular
ICD-9or10
Codes
101-159
Ages
Covered
all ages
Other
Pollutants
in Model
none
Lag
avg of 0-
and 1-day
Region
Covered
New York
New York
New York
New York
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Phoenix
Phoenix
Phoenix
Phoenix
Pittsburgh
Pittsburgh
Pittsburgh
Pittsburgh
Salt Lake City
Salt Lake City
Salt Lake City
Salt Lake City
St. Louis
St. Louis
St. Louis
St. Louis
Tacoma
Tacoma
Tacoma
Tacoma
Season
Covered
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Coefficient
0.00204
0.00231
0.00202
0.00205
0.00214
0.00153
0.00178
0.00300
... 1
...
...
0.00150
0.00284
0.00085
0.00047
...
...
...
-0.00013
0.00278
0.00188
0.00253
0.00006
0.00020
0.00025
0.00053
Lower Bound
0.00048
0.00050
0.00038
0.00047
-0.00042
-0.00135
-0.00082
0.00044
...
...
...
-0.00102
0.00026
-0.00148
-0.00185
...
...
...
-0.00297
-0.00013
-0.00084
-0.00022
-0.00182
-0.00173
-0.00168
-0.00136
Upper Bound
0.00360
0.00412
0.00366
0.00363
0.00470
0.00441
0.00438
0.00555
...
...
...
0.00401
0.00543
0.00318
0.00279
...
...
...
0.00270
0.00568
0.00459
0.00527
0.00193
0.00212
0.00219
0.00242
C-13
-------
Table C-2 cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Season-Specific Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, short-term non-
accidental
ICD-9or10
Codes
AOO-R99
Ages
Covered
all ages
Other
Pollutants
in Model
none
Lag
avg of 0-
and 1-day
Region
Covered
Atlanta
Atlanta
Atlanta
Atlanta
Baltimore
Baltimore
Baltimore
Baltimore
Birmingham
Birmingham
Birmingham
Birmingham
Dallas
Dallas
Dallas
Dallas
Detroit
Detroit
Detroit
Detroit
Fresno
Fresno
Fresno
Fresno
Houston
Houston
Houston
Houston
Los Angeles
Los Angeles
Los Angeles
Los Angeles
Season
Covered
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Coefficient
0.00133
0.00123
0.00078
0.00069
0.00126
0.00119
0.00100
0.00129
0.00097
0.00105
0.00049
0.00035
0.00099
0.00090
0.00106
0.00132
-0.00009
0.00174
0.00090
0.00072
0.00002
0.00225
0.00054
0.00088
0.00106
0.00129
0.00092
0.00092
0.00012
0.00059
-0.00084
-0.00002
Lower Bound
0.00020
0.00007
-0.00027
-0.00035
0.00016
0.00002
-0.0001 1
0.00017
-0.00022
-0.00012
-0.00061
-0.00074
-0.00017
-0.00027
-0.00008
0.00018
-0.00125
0.00043
-0.00053
-0.00066
-0.00159
-0.00021
-0.00217
-0.00090
-0.0001 1
0.00010
-0.00023
-0.00015
-0.00059
-0.00031
-0.00208
-0.00067
Upper Bound
0.00246
0.00238
0.00184
0.00172
0.00236
0.00236
0.00212
0.00240
0.00216
0.00222
0.00160
0.00144
0.00215
0.00208
0.00221
0.00247
0.00107
0.00304
0.00233
0.00210
0.00163
0.00471
0.00325
0.00266
0.00223
0.00248
0.00207
0.00199
0.00083
0.00149
0.00039
0.00064
C-14
-------
Table C-2 cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Season-Specific Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, short-term non-
accidental
ICD-9or10
Codes
AOO-R99
Ages
Covered
all ages
Other
Pollutants
in Model
none
Lag
avg of 0-
and 1-day
Region
Covered
New York
New York
New York
New York
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Phoenix
Phoenix
Phoenix
Phoenix
Pittsburgh
Pittsburgh
Pittsburgh
Pittsburgh
Salt Lake City
Salt Lake City
Salt Lake City
Salt Lake City
St. Louis
St. Louis
St. Louis
St. Louis
Tacoma
Tacoma
Tacoma
Tacoma
Season
Covered
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Coefficient
0.00168
0.00123
0.00074
0.00181
0.00195
0.00078
0.00064
0.00200
...
...
...
0.00135
0.00193
0.00090
0.00062
0.00113
0.00152
0.00106
0.00131
0.00054
0.00136
0.00097
0.00129
0.00006
0.00154
0.00088
0.00145
Lower Bound
0.00061
0.00001
-0.00029
0.00078
0.00041
-0.00090
-0.00089
0.00050
...
...
...
-0.00013
0.00034
-0.00047
-0.00073
-0.00013
-0.00047
-0.00095
-0.00051
-0.00055
0.00025
-0.00009
0.00022
-0.00236
-0.00123
-0.00203
-0.00099
Upper Bound
0.00275
0.00245
0.00177
0.00285
0.00350
0.00247
0.00217
0.00350
...
...
...
0.00283
0.00352
0.00227
0.00197
0.00240
0.00352
0.00308
0.00314
0.00164
0.00247
0.00203
0.00236
0.00249
0.00431
0.00378
0.00389
C-15
-------
Table C-2 cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Season-Specific Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, short-term
respiratory
ICD-9or10
Codes
JOO-J99
Ages
Covered
all ages
Other
Pollutants
in Model
none
Lag
avg of 0-
and 1-day
Region
Covered
Atlanta
Atlanta
Atlanta
Atlanta
Baltimore
Baltimore
Baltimore
Baltimore
Birmingham
Birmingham
Birmingham
Birmingham
Dallas
Dallas
Dallas
Dallas
Detroit
Detroit
Detroit
Detroit
Fresno
Fresno
Fresno
Fresno
Houston
Houston
Houston
Houston
Los Angeles
Los Angeles
Los Angeles
Los Angeles
Season
Covered
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Coefficient
0.00093
0.00035
0.00077
0.00096
0.00107
0.00144
0.00116
0.00103
0.00043
0.00079
-0.00018
0.00145
0.00040
0.00106
0.00060
0.00038
0.00104
0.00226
0.00253
0.00247
-0.00022
0.00496
0.00263
0.00099
0.00138
0.00129
0.00100
0.00092
0.00165
0.00237
-0.00134
-0.00003
Lower Bound
-0.00144
-0.00205
-0.00155
-0.00134
-0.00127
-0.00097
-0.00120
-0.00134
-0.00197
-0.00160
-0.00252
-0.00087
-0.00198
-0.00135
-0.00180
-0.00202
-0.00128
-0.00015
0.00009
0.00001
-0.00423
-0.00093
-0.00375
-0.00383
-0.00102
-0.00114
-0.00140
-0.00143
-0.00016
-0.00018
-0.00500
-0.00190
Upper Bound
0.00329
0.00275
0.00310
0.00325
0.00340
0.00384
0.00353
0.00340
0.00282
0.00318
0.00217
0.00377
0.00278
0.00347
0.00300
0.00278
0.00335
0.00467
0.00498
0.00492
0.00380
0.01085
0.00900
0.00580
0.00377
0.00372
0.00341
0.00327
0.00345
0.00493
0.00233
0.00183
C-16
-------
Table C-2 cont'd. Information about the Concentration-Response Functions Used in the PM2 5 Risk Assessment: Season-Specific Functions
Study
Zanobetti and
Schwartz (2009)
Health Endpoint
Mortality, short-term
respiratory
ICD-9or10
Codes
JOO-J99
Ages
Covered
all ages
Other
Pollutants
in Model
none
Lag
avg of 0-
and 1-day
Region
Covered
New York
New York
New York
New York
Philadelphia
Philadelphia
Philadelphia
Philadelphia
Phoenix
Phoenix
Phoenix
Phoenix
Pittsburgh
Pittsburgh
Pittsburgh
Pittsburgh
Salt Lake City
Salt Lake City
Salt Lake City
Salt Lake City
St. Louis
St. Louis
St. Louis
St. Louis
Tacoma
Tacoma
Tacoma
Tacoma
Season
Covered
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Winter
Spring
Summer
Fall
Coefficient
0.00334
0.00172
0.00157
0.00235
0.00217
0.00219
0.00182
0.00186
0.00251
0.00538
0.00577
0.00887
0.00134
0.00223
0.00188
0.00231
0.00301
0.00438
-0.00353
-0.00138
0.00019
0.00123
0.00060
0.00127
0.00011
0.00287
0.00190
0.00138
Lower Bound
0.00122
-0.00058
-0.00066
0.00013
-0.00030
-0.00033
-0.00068
-0.00062
-0.00253
-0.00140
-0.00083
0.00285
-0.00110
-0.00024
-0.00052
-0.00009
-0.00088
-0.00459
-0.01304
-0.00915
-0.00212
-0.00112
-0.00171
-0.00106
-0.00563
-0.00349
-0.00467
-0.00458
Upper Bound
0.00547
0.00403
0.00381
0.00457
0.00463
0.00471
0.00432
0.00435
0.00755
0.01215
0.01238
0.01489
0.00377
0.00470
0.00428
0.00472
0.00690
0.01336
0.00598
0.00639
0.00250
0.00357
0.00292
0.00360
0.00585
0.00924
0.00848
0.00733
1 indicates that results were not available.
C-17
-------
APPENDIX D: SUPPLEMENT TO THE REPRESENTATIVENESS
ANALYSIS OF THE 15 URBAN STUDY AREAS
D-l
-------
1 Appendix D. Supplement to the Representativeness Analysis of the 15 Urban Study Areas
2 (additional graphical comparisons of distributions for key contributors to PMi.s risk )
O
4 Following the analysis discussed in Section 4.4, this appendix provides graphical
5 comparisons of the empirical distributions of components of the risk function, and additional
6 variables that have been identified as potentially influencing the risk associated with PM
7 exposures.
8 In each graph, the orange line represents the empirical cumulative distribution function
9 (CDF) for the complete set of data available for the variable. In some cases, this may encompass
10 all counties in the U.S., while in others it may be based on a subset of the U.S., usually for large
11 urban areas. The green line in each graph represents the empirical cumulative distribution
12 function for the variable based only on the data available for the set of urban case study
13 locations. The black squares at the bottom of each graph represents the specific value of the
14 variable for one of the case study locations, with the line showing where that value intersects the
15 two empirical CDFs.
D-2
-------
D.I Elements of the Risk Equation
Figure D-l. Comparison of Distributions for Key Elements of the Risk Equation:
Total Population
Comparison of Urban Case Study Area Population with U.S. Distribution of
Population (all U.S. Counties)
100%
90% -
i/)
.0)
'^
c
o
o
Urban case study areas are
all above the 65th Percentile
100
1000
10000 100000
Population
1000000
10000000
All Counties CDF Case Study Counties CDF Case Study Counties
D-3
-------
Figure D-2. Comparison of Distributions for Key Elements of the Risk Equation:
Percent of Population Under 15 Years of Age
Comparison of Urban Case Study Area % Under 1 5 to U.S. Distribution of % Under
15
100% -i
90% -
80% -
£ 70% -
1 60% -
0
O
2 40% -
o
£ 30% -
20% -
10% -
0% -
1
^^^^^^M
0
(3141 U.S. Counties)
^
-------
Figure D-3. Comparison of Distributions for Key Elements of the Risk Equation:
Percent of Population 65 Years of Age and Older
Comparison of Urban Case Study Area % 65 and Older to U.S. Distribution of % 65
and Older
(3141 U.S. Counties)
*=
c
o
CO
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Urban case study areas are
all below the 75th percentile
of county % of population 65
and older
10 15 20 25
% of Population 65 and Older, 2005
30
35
All Counties CDF ^ Case Study Counties CDF Case Study Counties
D-5
-------
Figure D-4. Comparison of Distributions for Key Elements of the Risk Equation:
Percent of Population 85 Years of Age and Older
Comparison of Urban Case Study Area % 85 and Older to U.S. Distribution of % 85
and Older
100% ^
90% -
80% -
8 70%-
1 60% -
ri 50%
2 40% -
0
£ 30% -
20% -
10% -
0% H
1
(3141 U.S. Counties)
i
ill 11
Mi
2
H
1 1
Urban case study areas are
all below the 85th percentile
of county % of population 85
and older
34567
% of Population 85 and Older, 2005
All Counties CDF Case Study Counties CDF Case Study Counties
8
D-6
-------
Figure D-5. Comparison of Distributions for Key Elements of the Risk Equation: Annual
Mean PM2.5
1 00% -
90% -
80% -
i/)
0)
1 70% -
0 60% -
| 50%-
§ 40%
g
£ 30% -
^ 20% -
10% -
0% -
C
Comparison of Urban Case Study Area Annual Mean PM2.5 with U.S.
Distribution of Annual Mean PM2.5 (617 U.S. Counties with PM2.5 Monitors)
)
^-
5
>
j*
h 1
10
f
^
I 1
1 1
1 1
U i
15
H
H
I 1
h- 1
^
20
Annual Mean PM2.5
All Counties CDF Case Study Counties CDF Case Study Counties
25
D-7
-------
Figure D-6. Comparison of Distributions for Key Elements of the Risk Equation: 98
%ile Daily Average PM2.s
th
Comparison of Urban Case Study Area 98th %ile PM2.5 with U.S. Distribution of
100% -
90% -
«, 80% -
0)
1 70% -
0 60% -
1 50%-
§ 40%
g 30% -
^ 20% -
10% -
0% -
(
)
98th %ile PM2.5
(617 U.S. Counties with PM2.5 Monitors)
J
^
10 20
^4"
30
Li
^^
h I
40
-
X
X
L "1
m
50 60 70 80
98th Percentile Daily PM2.5
All Counties CDF Case Study Counties CDF Case Study Counties
90
D-8
-------
Figure D-7. Comparison of Distributions for Key Elements of the Risk Equation: % of
Days with PM2.5> 35 ug/m3
Comparison of Urban Case Study Area % of Days with PM2.5>35 ug/m3 to U.S.
Distribution of % of Days with PM2.5>35 ug/m3
100% -
90% -
80% -
in
j= 70% -
c
o 60% -
£ 50%
^ 40% -
o 30% -
20% -
10% -
0% -
(
(204 U.S. Counties in MCAPS Study)
i
)
i
H
H
H
H
.,r
/^
^X
^
5 10 15 20
% of Days Above 35 ug/m3
All Counties CDF Case Study Counties CDF Case Study Counties
25
D-9
-------
Figure D-8. Comparison of Distributions for Key Elements of the Risk Equation: All
Cause Mortality Rate
(
100% -,
90%
80%
w 70%
1 60%
0
O
2 40%
o
£ 30%
20%
10%
0%
1C
Comparison of Urban Case Study All Cause Mortal ty Rate to U.S. Distribution of All
Cause Mortality Rate
(31 43 U.S. Counties)
)0
300
^
50
H
0
H
I «
70
m
D
r-
I !
900
.,
m i
11
H
00
H
**
^-"
Urban case study areas are
all below the 90th percentile
of county all cause mortality
1300 1500 1700 1900
All Cause Mortality per 100,000 Population
All Counties CDF Case Study Counties CDF Case Study Counties
2100
D-10
-------
Figure D-9. Comparison of Distributions for Key Elements of the Risk Equation: Non-
Accidental Mortality Rate
Comparison of Urban Case Study Non-accidental Mortality Rate to U.S. Distribution of
Non-accidental Mortality Rate
(3143 U.S. Counties)
IUU /O -
90% -
80% -
w 70% -
~
1 60% -
o
o
50% -
CO
2 40% -
0
£ 30% -
20% -
10% -
no/, ^
-^
.---
I I
l_l
i i
r
__i
Ul
i i
i , 1
LJ
Ul
/
1 1 r-l
/
Ul
U 1
X"
1 , 1
Urban case study areas are
all below the 90th percentile
of county non-accidental
mortality
200 400 600 800 1000 1200 1400 1600
Non-accidental Mortality per 100,000 Population
1800
2000
All Counties CDF - Case Study Counties CDF Case Study Counties
D-ll
-------
Figure D-10. Comparison of Distributions for Key Elements of the Risk Equation:
Cardiovascular Mortality Rate
Comparison of Urban Case Study Cardiovascular Mortality Rate to U.S. Distribution of
Cardiovascular Mortality Rate
(3143 U.S. Counties)
90% -.
80%
£ 70%
'-1 i
1 60%
o
o
50%
CO
2 40%
0
£ 30%
20%
10%
0%
^^^H
50
H
A
W
H
H
!
\m,m
,'
tmt
^-9
H
U
H
y
l-rl
X"*
I 1
^^^
Urban case study areas are
all below the 90th percentile
of county cardiovascular
mortality
|
150 250 350 450 550 650 750 850 950
Cardiovascular Mortality per 100,000 Population
All Counties CDF ^ Case Study Counties CDF Case Study Counties
D-12
-------
Figure D-ll. Comparison of Distributions for Key Elements of the Risk Equation:
Respiratory Mortality Rate
100% -,
90%
80%
8 70%
1 60%
^ 50%
2 40%
0
£ 30%
20%
10%
0%
1
Comparison of Urban Case Study Respiratory Mortality Rate to U.S. Distribution of
Respiratory Mortality Rate
(31 43 U.S. Counties)
^^^M
0
-X
ii
i
lM"H
60
I ll
I
H
Urban case study areas are
all below the 90th percentile
of county respiratory
mortality
110 160 210
Respiratory Mortality per 100,000 Population
All Counties CDF Case Study Counties CDF Case Study Counties
260
D-13
-------
Figure D-12. Comparison of Distributions for Key Elements of the Risk Equation: All
Cause Mortality Risk Effect Estimate from Zanobetti and Schwartz (2008)
CO
CD
c
CO
.Q
CO
N
"o
c
100% -
90% -
80% -
70% -
60% -
50% -
40% -
30% -
20% -
10% -
0% -
C
Comparison of Urban Case Study PM All-cause Mortality Risk ((3) to
U.S. Distribution of PM All-cause Mortality Risk
(2 12 U.S. Urban Areas)
x '
^
X^
/
J
f
^^^
s
S
) 0.0002 0.0004 0.0006 0.0008 0.001 0.0012 0.0014 0.0016
Bayesian Shrunken PM All Cause Mortality Risk Coefficient ((3)
All Z&S Urban Areas CDF Case Study Urban Area CDF Case Study Urban Areas
D-14
-------
Figure D-13. Comparison of Distributions for Key Elements of the Risk Equation:
Cardiovascular Mortality Risk Effect Estimate from Zanobetti and Schwartz
(2008)
Comparison of Urban Case Study PM Cardiovascular Mortality Risk (p) to U.S.
Distribution of PM Cardiovascular Mortality Risk
(212 U.S. Urban Areas)
0.0002 0.0004 0.0006 0.0008 0.001 0.0012 0.0014 0.0016 0.0018
Bayesian Shrunken PM Cardiovascular Mortality Risk Coefficient (p)
0.002
All Z&S Urban Areas CDF Case Study Urban Area CDF Case Study Urban Areas
D-15
-------
Figure D-14. Comparison of Distributions for Key Elements of the Risk Equation:
Respiratory Mortality Risk Effect Estimate from Zanobetti and Schwartz
(2008)
Comparison of Urban Case Study PM Respiratory Mortality Risk (p) to U.S.
Distribution of PM Respiratory Mortality Risk
100% -,
90%
en
ro
0)
c
(0
.0
CO
08
N
0
0s
80%
70%
60%
50%
40%
30%
20%
10%
0%
C
(21 2 U.S. Urban Areas)
^
) 0.0005
u
0
X
v"
r
s
X
^
/
1
/
^_J
-
'
.001 0.0015 0.002
Bayesian Shrunken PM Respiratory Mortality Risk Coefficient (p)
All Z&S Urban Areas CDF ^ Case Study Urban Area CDF Case Study Urban Areas
D-16
-------
D.2. Variables Expected to Influence the Relative Risk from PM2.s
D.2.1. Demographic Variables
Figure D-15. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PM2.s: Population Density
Comparison of Urban Case Study Area Population Density with U.S. Distribution of
Population Density (all U.S. Counties)
100%
90%
80%
70%
Urban case study areas are
all above the 65th Percentile
of county population density
10 100 1000 10000
Population Density (persons per sq mile)
100000
All Counties CDF ^Case Study Counties CDF ^^Case Study Counties
D-17
-------
Figure D-16. Comparison of Distributions for Selected Variables Expected to
Influence the Relative Risk from PMi.s: Unemployment Rate
100% -,
90%
80%
8 70%
1 60%
^ 50%
2 40%
0
£ 30%
20%
10%
0%
(
Comparison of Urban Case Study Area Unemployment Rate to U.S. Distribution of
Unemployment Rates
(3141 U.S. Counties)
)
^^
2
Jj
4
h*H
H
H
tl
6
H
^
i
ii
X
8 10 12 14 16 18
Unemployment Rate (%)
All Counties CDF Case Study Counties CDF Case Study Counties
20
D-18
-------
Figure D-17. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: % with Less than a High School Education
100% -i
90% -
80% -
£ 70% "
1 60% -
ri 50%
2 40% -
0
5S 30% -
20% -
10% -
0% -
Comparison of Urban Case Study % Less than High School Education to U.S.
Distribution of % Less than High School Education
(31 46 U.S. Counties)
^_^_
i
^
!
1
1-1
3
H
1 1
H
^
2
i i
3
i li
1 1
H
-^
.X*
S3 43 53
% of Population with Less than a High School Education, 2000
All Counties CDF Case Study Counties CDF Case Study Counties
63
D-19
-------
Figure D-18. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Per Capita Personal Income
Comparison of Urban Case Study Area Per Capita Personal Income to U.S.
100% -r^
90%
80%
$ 70%
"P
1 60%
O
° 50%
CO
2 40%
0
c£ 30%
20%
10% -]
n% 4
u /o ^^^
$10,000
Distribution of Per Capita Personal Income
(3141 U.S. Counties)
/
J
i i i
i i i
1 U
i i
LJ 1
LJ
1 1
1 1
X
1 1
Urban case study areas are
all above the 25th percentile
of county per capita income
L . . .
1
1
^^ ^ ^^ ~~ i I I I I
$20,000 $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 $90,000
Per Capita Personal Income, 2005
All Counties CDF Case Study Counties CDF Case Study Counties
D-20
-------
Figure D-19. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Air Conditioning Prevalence
w
CO
£
<
c
CD
.Q
D
M
o
^
(
100% -
90%
80%
70%
60%
50%
40%
30%
20%
10%
no/ _
Comparison of Urban Case Study Air Conditioning Prevalence to U.S. Distribution of
Air Conditioning Prevalence
(70 U.S. Urban Areas)
'
-~
~
i
^
/
1
r.
/
0 0.2 0.4 0.6 0.8 1
All
% Air Conditioning Prevalence
Urban Areas CDF ^ Case Study Urban Areas CDF Case Study Urban Areas
D-21
-------
Figure D-20. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: % Non-White Population
CO
£
<
c
CD
.Q
D
M
o
^
(
100% -
90%
80%
70%
60%
50%
40%
30%
20%
10%
no/ _
Comparison of Urban Case Study Air Conditioning Prevalence to U.S. Distribution of
Air Conditioning Prevalence
(70 U.S. Urban Areas)
'
-~
~
i
^
/
1
r.
/
0 0.2 0.4 0.6 0.8 1
All
% Air Conditioning Prevalence
Urban Areas CDF ^ Case Study Urban Areas CDF Case Study Urban Areas
D-22
-------
D.2.2. Health Conditions
Figure D-21. Comparison of Distributions for Selected Variables Expected to
Influence the Relative Risk from PMi.s: Angina/Coronary Heart
Disease Prevalence
Comparison of Urban Case Study Area Angina/CHD Prevalence to U.S. Distribution
of Angina/CHD Prevalence
(183 U.S. MSA)
90%
80%
< 70%
CO
5 60%
CO
CO
U- 50%
tr
DQ
£ 40%
0
^ 30%
20%
10%
0%
^^
/>
-------
Figure D-22. Comparison of Distributions for Selected Variables Expected to
Influence the Relative Risk from PMi.s: Asthma Prevalence
Comparison of Urban Case Study Area Current Asthma Prevalence to U.S.
Distribution of Current Asthma Prevalence
(183 U.S. MSA)
IUU /O
90%
80%
< 70%
CO
5 60%
CO
£ 50%
a:
2 40%
o
^ 30%
20%
10%
<:
f)0/ U i
'
i i
i
i_i 1 1
1 1 i_i
i i
x^
s
i i
^
i i
i
\J /O ^^^^ ^^^r^ ^ i -^^^^^ ^^ ^^ ^ | H |
3 5 7 9 11 13
% Current Prevalence of Asthma, BRFSS 2007
All BRFSS MSA CDF Case Study County MSA CDF Case Study County MSA
D-24
-------
Figure D-23. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Diabetes Prevalence
Comparison of Urban Case Study Area Diabetes Prevalence to U.S. Distribution of
Diabetes Prevalence
100% -i
90% -
80% -
< 70% -
CO
5 60% -
CO
L£ 50%
or
2 40% -
o
^ 30% -
20% -
10% -
0% -
(183 U.S. MSA)
^
>. 4
/
/
^
H 1
6
/
f
i i
8
L
/
1 1 i
i 1 i
10
^^
Urban case study areas are
all below the 85th percentile
of MSA diabetes prevalence
12 14 16
% Prevalence of Diabetes, BRFSS 2007
All BRFSS MSA CDF Case Study County MSA CDF Case Study County MSA
D-25
-------
Figure D-24. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Heart Attack Prevalence
Comparison of Urban Case Study Area Heart Attack Prevalence to U.S. Distribution of
Heart Attack Prevalence
(183 U.S. MSA)
1 nno/
90%
80%
< 70%
CO
5 60%
CO
CO
ul 50%
or
2 40%
o
Q
^ 30%
2
1
0%
0%
0%
,
/
'
1
/
f
^^
Urban case study areas are
all below the 80th percentile
of MSA heart attack
prevalence
23456789 10
% Prevalence of Heart Attacks, BRFSS 2007
All BRFSS MSA CDF Case Study County MSA CDF Case Study County MSA
11
D-26
-------
Figure D-25. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Obesity Prevalence
Comparison of Urban Case Study Area Obesity Prevalence to U.S. Distribution of
Obesity Prevalence
(183 U.S. MSA)
14
19 24 29
% Prevalence of Obesity, BRFSS 2007
34
All BRFSS MSA CDF - Case Study County MSA CDF Case Study County MSA
D-27
-------
Figure D-26. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Stroke Prevalence
Comparison of Urban Case Study Area Stroke Prevalence to U.S. Distribution of
Stroke Prevalence
100% -,
90%
80%
< 70%
CO
5 60%
CO
i2 50%
£ 40%
0
^ 30%
20%
10%
0%
(
(183 U.S. MSA)
/
/
/
/
/
X
X
^
) 1 2 3 4 5 e
% Prevalence of Stroke, BRFSS 2007
All BRFSS MSA CDF Case Study County MSA CDF Case Study County MSA
I
D-28
-------
Figure D-27. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Smoking Prevalence
Comparison of Urban Case Study Area Smoking Prevalence to U.S. Distribution of
Smoking Prevalence
(183 U.S. MSA)
1 UU /O ^
90%
80%
< 70%
CO
5 60%
CO
CO
ul 50% H
or
2 40%
0
^ 30%
20%
10%
^^^^^^0 I
ii ii
I
\ i
^
i i
S
^r
^ ^r
V
\ i
*
u
ui
i i i
'
\ i
/
i
^~^
'
Urban case study areas are
all below the 85th percentile
of MSA smoking prevalence
i
6 11 16 21 26
% Prevalence of Smoking (ever), BRFSS 2007
All BRFSS MSA CDF Case Study County MSA CDF Case Study County MSA
31
D-29
-------
Figure D-28. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Exercise Prevalence
Comparison of Urban Case Study Area Exercise Prevalence to U.S. Distribution of
Exercise Prevalence
100% -,
90%
80%
< 70%
CO
W 6°%
i2 50%
a:
2 40%
o
£ 30%
20%
(183 U.S. MSA)
0% J- H
15 20
^^
^
A
/
/
25
/
/
II ll
rl
30
^
i 1
^
-**
35 40
45
% Prevalence of Exercise (Adults with 20+ minutes of vigorous physical activity 3 or
more days per week), BRFSS 2007
All BRFSS MSA CDF Case Study County MSA CDF Case Study County MSA
D-30
-------
D.2.3. Air Quality and Climate Variables
Figure D-29. Comparison of Distributions for Selected Variables Expected to
Influence the Relative Risk from PM2.5: 4th Highest Daily Max 8-
hour Average
Comparison of Urban Case Study Area 4th Highest Daily 8-Hour Ozone with U.S.
1 00% -
90% -
«, 80% -
0
c 70% -
0 60% -
1 50% -
3
§ 40% ^
g
5 30%j
^ 20% \
10% -
0% -
0.
D2
Distribution of 4th Highest Daily 8-Hour Ozone
(725 U.S. Counties with Ozone Monitors)
^
0.04 0.06
J
^s
^
1
1 1
0
1
08
1 1
L
11
H
r
^f^^
0.1
4th Highest Daily 8-Hour Maximum
All Counties CDF ^ Case Study Counties CDF Case Study Counties
^^^^^M
0.12
D-31
-------
Figure D-30. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PM2.s: % Mobile Source Direct PM2.s Emissions
Comparison of Urban Case Study Mobile Source Carbon to U.S. Distribution of Mobile
Source Carbon
100% -,
90%
80%
8 70%
1 60%
ri 50%
2 40%
0
5S 30%
20%
(31 42 U.S. Counties)
10% -]
0% 4-
0
^
0.1
s
I 1
0
1 i
2
H
(
II-H
).3
h- 1
I-J4
0.4
ill
t
H
H
(
^
Urban case study areas are
all below the 80th percentile
of county mobile source
proportion of EC/OC
emissions
J.5 0.6 0.7 0.8 0.9
Mobile Source Proportion of Total Organic and Elemental Carbon Emissions
All Counties CDF Case Study Counties CDF Case Study Counties
1
D-32
-------
Figure D-31. Comparison of Distributions for Selected Variables Expected to
Influence the Relative Risk from PMi.s: July Temperature Long
Term Average
Comparison of Urban Case Study Area Long Term Average July Temperature to
U.S. Distribution of Long Term Average July Temperature
(3141 U.S. Counties)
100%
50
60 70 80 90
July 30 Year Average Temperature, 1941-1970
100
All Counties CDF ^Case Study Counties CDF Case Study Counties
D-33
-------
Figure D-32. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: July Relative Humidity Long Term Average
c
100% -,
90%
80%
8 70%
1 60%
^ 50%
2 40%
0
5S 30%
20%
10%
0%
(
Comparison of Urban Case Study Area Long Term Average July Relative Humidity to
U.S. Distribution of Long Term Average July Relative Humid ty
(3141 U.S. Counties)
^
)
-
10 20
^X
'^"^'
S
/
30 40 50
/
/
11 1
60
/
V
iU i
H
7
0 80
July 30 Year Average Relative Humidity, 1941-1970
All Counties CDF Case Study Counties CDF Case Study Counties
90
D-34
-------
APPENDIX E: RISK ESTIMATES (CORE ANALYSIS)
E-l
-------
1 Appendix E. Risk Estimates (core analysis)
2
3 This Appendix provides detailed risk estimates generated for the core analysis for the 15
4 urban study areas. The tables cover all of the air quality scenarios modeled, including recent
5 conditions, the current standard, and alternative standard levels. For additional detail on the types
6 of risk metrics (and figures summarizing key metrics) presented in this Appendix, see section
7 4.0.
E-l
-------
Table E-l. Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2.s Concentrations in a Recent
Year (2005) and PM25 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005 PM2 s
Concentrations: Estimates Based onKrewski et al. (2009), Using Ambient PM2 5from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dal las, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5
Concentrations that Just Meetthe Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2 :
Recent PM2.5
Concentrations
649
(421 - 873)
597
(387 - 803)
424
(275-571)
379
(245-511)
798
(518-1073)
254
(165-342)
609
(394 - 820)
2333
(1514-3141)
2000
(1297-2693)
521
(338 - 703)
483
(312-652)
593
(385 - 798)
102
(66 - 1 38)
826
(536 -1111)
123
(80 - 1 66)
15/353
575
(373 - 774)
548
(355 - 738)
297
(192-400)
379
(245-511)
580
(376 - 782)
89
(57 - 1 20)
557
(360-751)
1045
(676- 1413)
1477
(956 - 1 992)
455
(295-614)
483
(312 - 652)
387
(251 - 523)
29
(19-39)
700
(454 - 943)
80
(52 - 1 09)
14/35
513
(333 - 692)
502
(325 - 676)
262
(170-354)
379
(245 - 51 1 )
573
(371 - 772)
89
(57 - 1 20)
491
(318-663)
1045
(676-1413)
1477
(956 - 1 992)
455
(295-614)
483
(312-652)
387
(251 - 523)
29
(19-39)
634
(411 -855)
80
(52 - 1 09)
13/35
451
(292 - 608)
442
(286 - 596)
227
(147-307)
379
(245 - 51 1 )
502
(325 - 677)
89
(57 - 1 20)
426
(275 - 575)
1045
(676- 1413)
1410
(912- 1902)
406
(263 - 548)
483
(312-652)
363
(235 - 490)
29
(19-39)
557
(361 -752)
80
(52 - 1 09)
12/35
389
(252 - 525)
382
(247-516)
192
(124-260)
336
(218- 454)
431
(279 - 581 )
89
(57 - 1 20)
360
(233 - 486)
919
(593 - 1 242)
1205
(779 - 1 627)
348
(225 - 470)
433
(280 - 586)
318
(206 - 430)
29
(19-39)
480
(310-648)
80
(52 - 1 09)
13/30
451
(292 - 608)
426
(276 - 574)
227
(147-307)
379
(245 -511)
442
(286 - 597)
59
(38 - 79)
426
(275 - 575)
719
(464 - 972)
1100
(711 -1486)
345
(223 - 465)
420
(271 - 568)
289
(187-390)
10
(7-14)
543
(351 - 732)
53
(34 - 72)
12/25
379
(245-512)
303
(196-409)
160
(103-216)
336
(218-454)
303
(196 -410)
28
(18-38)
360
(233 - 486)
390
(252 - 528)
721
(465 - 975)
233
(151 -315)
263
(170-356)
190
(122-257)
0
(0-0)
383
(248-518)
26
(17-36)
1Based on follow-upthrough2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski etal., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
The current primary PM2 5 standards include an annual standard set at 1 5 ug/m and a daily standard set at 35 ug/m .
E-2
-------
Table E-2. Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2.5 Concentrations in a Recent
Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006 PM2 5
Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
668
(434 - 899)
483
(313-651)
399
(259 - 536)
284
(183-383)
576
(373 - 776)
265
(172-356)
589
(381 - 794)
2054
(1332-2767)
1548
(1002-2087)
471
(305 - 636)
512
(331 -691)
468
(303-631)
84
(54 - 1 1 3)
618
(401 - 834)
83
(54 - 1 1 2)
15/353
592
(384 - 797)
441
(285 - 594)
276
(179-373)
284
(183-383)
398
(257 - 537)
94
(61 -127)
537
(348 - 725)
863
(557-1166)
1096
(708-1481)
409
(265 - 552)
512
(331 -691)
290
(187-392)
16
(10-21)
514
(332 - 693)
47
(30 - 64)
14/35
528
(342-712)
400
(259 - 539)
243
(157-328)
284
(183-383)
392
(253 - 529)
94
(61 -127)
472
(306 - 638)
863
(557-1166)
1096
(708-1481)
409
(265 - 552)
512
(331 -691)
290
(187-392)
16
(10-21)
458
(296-619)
47
(30 - 64)
13/35
464
(301 - 626)
347
(225 - 469)
209
(135-283)
284
(183-383)
334
(216-451)
94
(61 -127)
407
(263 - 550)
863
(557-1166)
1038
(671 -1403)
363
(235 - 490)
512
(331 -691)
270
(174-364)
16
(10-21)
394
(255 - 532)
47
(30 - 64)
12/35
400
(259 - 540)
295
(191 -398)
176
(114-238)
247
(160-334)
276
(178-373)
94
(61 -127)
342
(221 - 462)
745
(481 -1008)
861
(556-1164)
308
(199-417)
460
(297 - 622)
231
(149-312)
16
(10-21)
329
(213-445)
47
(30 - 64)
13/30
464
(301 - 626)
333
(216-450)
209
(135-283)
284
(183-383)
285
(184-386)
63
(41 - 85)
407
(263 - 550)
561
(362 - 759)
771
(498 - 1 043)
305
(197-412)
446
(288 - 603)
205
(133-278)
0
(0-0)
382
(247-515)
25
(16-33)
12/25
390
(253 - 527)
225
(145-304)
144
(93-195)
247
(160-334)
172
(111 -233)
32
(20 - 43)
342
(221 - 462)
257
(166-348)
444
(286 - 601 )
200
(129-271)
281
(181 -380)
120
(78-163)
0
(0-0)
249
(160-336)
2
(1-3)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-3
-------
Table E-3. Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in a Recent
Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007 PM2 5
Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent Year and PM25
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
642
(41 6 - 864)
482
(31 3 - 650)
418
(271 - 563)
317
(205 - 428)
607
(393 - 81 8)
279
(181 -375)
615
(398 - 829)
2134
(1384-2874)
1812
(1174-2443)
466
(302 - 629)
433
(280 - 586)
527
(342 - 71 0)
120
(78-162)
679
(440 - 91 5)
87
(56 - 1 1 8)
15/353
567
(368 - 764)
440
(285 - 593)
291
(189-393)
317
(205 - 428)
424
(274 - 572)
101
(65-137)
561
(363 - 757)
911
(588-1232)
1316
(852-1777)
405
(262 - 546)
433
(280 - 586)
339
(219-457)
38
(24-51)
568
(368 - 766)
50
(32 - 68)
14/35
505
(327 - 680)
399
(258 - 538)
257
(166-347)
317
(205 - 428)
418
(270 - 564)
101
(65-137)
494
(320 - 667)
911
(588-1232)
1316
(852-1777)
405
(262 - 546)
433
(280 - 586)
339
(21 9 - 457)
38
(24-51)
509
(330 - 687)
50
(32 - 68)
13/35
442
(286 - 596)
347
(224 - 468)
222
(144-300)
317
(205 - 428)
358
(232 - 484)
101
(65-137)
427
(276 - 577)
911
(588-1232)
1253
(810-1692)
359
(232 - 484)
433
(280 - 586)
316
(205 - 427)
38
(24-51)
441
(285 - 596)
50
(32 - 68)
12/35
379
(245 - 51 2)
294
(190-398)
187
(121 -253)
278
(180-375)
299
(193-404)
101
(65-137)
359
(232 - 486)
791
(511 -1070)
1058
(684-1430)
304
(197-411)
385
(248 - 520)
274
(177-371)
38
(24-51)
373
(241 - 504)
50
(32 - 68)
13/30
442
(286 - 596)
333
(21 5 - 449)
222
(144-300)
317
(205 - 428)
308
(199-417)
69
(44 - 93)
427
(276 - 577)
601
(388 - 81 3)
959
(620 - 1 296)
301
(195-407)
371
(240 - 502)
247
(160-334)
17
(1 1 - 23)
428
(277 - 578)
27
(17-36)
12/25
370
(239 - 499)
225
(145-304)
155
(100-209)
278
(180-375)
192
(124-260)
36
(23 - 49)
359
(232 - 486)
289
(187-392)
600
(387 -811)
197
(127-266)
216
(139-292)
156
(100-211)
0
(0-0)
287
(186-389)
4
(2-5)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-4
-------
Table E-4. Estimated Percent of Total Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year
and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
4.3%
(2.8% - 5.8%)
4.2%
(2.7% - 5.7%)
4.3%
(2.8% - 5.8%)
3%
(1 .9% - 4%)
4.5%
(2.9% - 6%)
4.6%
(3% -6.1%)
3.3%
(2.1% -4. 4%)
4.1%
(2.7% - 5.5%)
3.8%
(2. 5% -5.1%)
3.6%
(2.3% - 4.8%)
2.1%
(1 .4% - 2.8%)
4.3%
(2.8% - 5.7%)
2.2%
(1 .4% - 2.9%)
4.4%
(2.8% - 5.9%)
2.4%
(1 .6% - 3.3%)
15/353
3.8%
(2. 5% -5.1%)
3.9%
(2.5% - 5.2%)
3%
(2% -4.1%)
3%
(1 .9% - 4%)
3.3%
(2.1% -4. 4%)
1 .6%
(1 % - 2.2%)
3%
(1 .9% - 4%)
1 .8%
(1 .2% - 2.5%)
2.8%
(1 .8% - 3.8%)
3.1%
(2% - 4.2%)
2.1%
(1 .4% - 2.8%)
2.8%
(1 .8% - 3.8%)
0.6%
(0.4% - 0.8%)
3.7%
(2.4% - 5%)
1 .6%
(1%-2.1%)
14/35
3.4%
(2.2% - 4.6%)
3.6%
(2.3% - 4.8%)
2.7%
(1 .7% - 3.6%)
3%
(1 .9% - 4%)
3.2%
(2.1% -4. 3%)
1 .6%
(1 % - 2.2%)
2.6%
(1 .7% - 3.6%)
1 .8%
(1 .2% - 2.5%)
2.8%
(1 .8% - 3.8%)
3.1%
(2% - 4.2%)
2.1%
(1 .4% - 2.8%)
2.8%
(1 .8% - 3.8%)
0.6%
(0.4% - 0.8%)
3.4%
(2.2% - 4.5%)
1 .6%
(1%-2.1%)
13/35
3%
(1 .9% - 4%)
3.1%
(2% - 4.2%)
2.3%
(1.5% -3.1%)
3%
(1 .9% - 4%)
2.8%
(1 .8% - 3.8%)
1 .6%
(1 % - 2.2%)
2.3%
(1.5% -3.1%)
1 .8%
(1 .2% - 2.5%)
2.7%
(1 .7% - 3.6%)
2.8%
(1 .8% - 3.8%)
2.1%
(1 .4% - 2.8%)
2.6%
(1 .7% - 3.5%)
0.6%
(0.4% - 0.8%)
3%
(1 .9% - 4%)
1 .6%
(1%-2.1%)
12/35
2.6%
(1 .7% - 3.5%)
2.7%
(1 .8% - 3.7%)
2%
(1 .3% - 2.6%)
2.6%
(1 .7% - 3.5%)
2.4%
(1 .6% - 3.3%)
1 .6%
(1 % - 2.2%)
1 .9%
(1 .2% - 2.6%)
1 .6%
(1 % - 2.2%)
2.3%
(1.5% -3.1%)
2.4%
(1 .5% - 3.2%)
1 .9%
(1 .2% - 2.5%)
2.3%
(1.5% -3.1%)
0.6%
(0.4% - 0.8%)
2.5%
(1 .6% - 3.4%)
1 .6%
(1%-2.1%)
13/30
3%
(1 .9% - 4%)
3%
(2% -4.1%)
2.3%
(1.5% -3.1%)
3%
(1 .9% - 4%)
2.5%
(1 .6% - 3.3%)
1.1%
(0.7% - 1 .4%)
2.3%
(1.5% -3.1%)
1 .3%
(0.8% - 1 .7%)
2.1%
(1 .3% - 2.8%)
2.4%
(1 .5% - 3.2%)
1 .8%
(1 .2% - 2.5%)
2.1%
(1 .3% - 2.8%)
0.2%
(0.1% -0.3%)
2.9%
(1 .9% - 3.9%)
1.1%
(0.7% - 1 .4%)
12/25
2.5%
(1 .6% - 3.4%)
2.1%
(1 .4% - 2.9%)
1 .6%
(1.1% -2. 2%)
2.6%
(1 .7% - 3.5%)
1 .7%
(1.1% -2. 3%)
0.5%
(0.3% - 0.7%)
1 .9%
(1 .2% - 2.6%)
0.7%
(0.4% - 0.9%)
1 .4%
(0.9% - 1 .8%)
1 .6%
(1 % - 2.2%)
1.1%
(0.7% - 1 .5%)
1 .4%
(0.9% - 1 .8%)
0%
(0% - 0%)
2%
(1 .3% - 2.7%)
0.5%
(0.3% - 0.7%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
''The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-5
-------
Table E-5. Estimated Percent of Total Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2 s Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.s Concentrations in a Recent Year
and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2.5
Concentrations
4.3%
(2.8% - 5.8%)
3.4%
(2.2% - 4.6%)
4%
(2.6% - 5.4%)
2.2%
(1 .4% - 2.9%)
3.2%
(2.1% -4. 4%)
4.7%
(3% - 6.3%)
3.1%
(2% -4.1%)
3.6%
(2.3% - 4.8%)
2.9%
(1 .9% - 3.9%)
3.2%
(2.1% -4. 4%)
2.1%
(1 .4% - 2.9%)
3.4%
(2.2% - 4.6%)
1 .7%
(1.1% -2. 3%)
#DIV/0!
#DIV/0!
3.3%
(2.1% -4. 4%)
15/353
3.8%
(2.5% -5.1%)
3.1%
(2% - 4.2%)
2.8%
(1 .8% - 3.8%)
2.2%
(1 .4% - 2.9%)
2.2%
(1 .4% - 3%)
1 .7%
(1.1% -2.2%)
2.8%
(1 .8% - 3.8%)
1 .5%
(1 % - 2%)
2.1%
(1 .3% - 2.8%)
2.8%
(1 .8% - 3.8%)
2.1%
(1 .4% - 2.9%)
2.1%
(1 .4% - 2.8%)
0.3%
(0.2% - 0.4%)
2.7%
(1 .8% - 3.7%)
0.9%
(0.6% - 1 .2%)
14/35
3.4%
(2.2% - 4.6%)
2.8%
(1 .8% - 3.8%)
2.4%
(1 .6% - 3.3%)
2.2%
(1 .4% - 2.9%)
2.2%
(1 .4% - 3%)
1 .7%
(1.1% -2. 2%)
2.5%
(1 .6% - 3.3%)
1 .5%
(1%-2%)
2.1%
(1 .3% - 2.8%)
2.8%
(1 .8% - 3.8%)
2.1%
(1 .4% - 2.9%)
2.1%
(1 .4% - 2.8%)
0.3%
(0.2% - 0.4%)
2.4%
(1 .6% - 3.3%)
0.9%
(0.6% - 1 .2%)
13/35
3%
(1 .9% - 4%)
2.5%
(1 .6% - 3.3%)
2.1%
(1 .4% - 2.8%)
2.2%
(1 .4% - 2.9%)
1 .9%
(1 .2% - 2.5%)
1 .7%
(1.1% -2.2%)
2.1%
(1 .4% - 2.9%)
1 .5%
(1 % - 2%)
2%
(1 .3% - 2.6%)
2.5%
(1 .6% - 3.4%)
2.1%
(1 .4% - 2.9%)
1 .9%
(1 .3% - 2.6%)
0.3%
(0.2% - 0.4%)
2.1%
(1 .3% - 2.8%)
0.9%
(0.6% - 1 .2%)
12/35
2.6%
(1 .7% - 3.5%)
2.1%
(1 .3% - 2.8%)
1 .8%
(1.1% -2. 4%)
1 .9%
(1 .2% - 2.5%)
1 .5%
(1%-2.1%)
1 .7%
(1.1% -2. 2%)
1 .8%
(1.1% -2. 4%)
1 .3%
(0.8% - 1 .8%)
1 .6%
(1 % - 2.2%)
2.1%
(1 .4% - 2.9%)
1 .9%
(1 .2% - 2.6%)
1 .7%
(1.1% -2. 3%)
0.3%
(0.2% - 0.4%)
1 .7%
(1.1% -2. 4%)
0.9%
(0.6% - 1 .2%)
13/30
3%
(1 .9% - 4%)
2.4%
(1 .5% - 3.2%)
2.1%
(1 .4% - 2.8%)
2.2%
(1 .4% - 2.9%)
1 .6%
(1%-2.2%)
1.1%
(0.7% - 1 .5%)
2.1%
(1 .4% - 2.9%)
1%
(0.6% - 1 .3%)
1 .4%
(0.9% - 2%)
2.1%
(1 .4% - 2.8%)
1 .9%
(1 .2% - 2.5%)
1 .5%
(1 % - 2%)
0%
(0% - 0%)
2%
(1 .3% - 2.7%)
0.5%
(0.3% - 0.6%)
12/25
2.5%
(1 .6% - 3.4%)
1 .6%
(1 % - 2.2%)
1 .5%
(0.9% - 2%)
1 .9%
(1 .2% - 2.5%)
1%
(0.6% - 1 .3%)
0.6%
(0.4% - 0.8%)
1 .8%
(1.1% -2.4%)
0.5%
(0.3% - 0.6%)
0.8%
(0.5% -1.1%)
1 .4%
(0.9% - 1 .9%)
1 .2%
(0.8% - 1 .6%)
0.9%
(0.6% - 1 .2%)
0%
(0% - 0%)
1 .3%
(0.8% - 1 .8%)
0%
(0%-0.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
-------
Table E-6. Estimated Percent of Total Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year
and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2 5
Concentrations
4%
(2.6% - 5.4%)
3.4%
(2.2% - 4.6%)
4.2%
(2.7% - 5.6%)
2.4%
(1 .5% - 3.2%)
3.4%
(2.2% - 4.6%)
4.9%
(3.2% - 6.5%)
3.1%
(2% - 4.2%)
3.7%
(2.4% - 5%)
3.4%
(2.2% - 4.6%)
3.2%
(2.1% -4.3%)
1 .8%
(1.1% -2.4%)
3.8%
(2.5% - 5.2%)
2.4%
(1 .5% - 3.2%)
3.6%
(2.3% - 4.8%)
1 .7%
(1.1% -2.2%)
15/353
3.6%
(2.3% - 4.8%)
3.1%
(2% - 4.2%)
2.9%
(1 .9% - 3.9%)
2.4%
(1 .5% - 3.2%)
2.4%
(1 .5% - 3.2%)
1 .8%
(1.1% -2.4%)
2.9%
(1 .8% - 3.9%)
1 .6%
(1%-2.1%)
2.5%
(1 .6% - 3.3%)
2.8%
(1 .8% - 3.8%)
1 .8%
(1.1% -2.4%)
2.5%
(1 .6% - 3.3%)
0.7%
(0.5% - 1 %)
3%
(1 .9% - 4%)
1%
(0.6% - 1 .3%)
14/35
3.2%
(2% - 4.3%)
2.8%
(1 .8% - 3.8%)
2.6%
(1 .7% - 3.5%)
2.4%
(1 .5% - 3.2%)
2.4%
(1 .5% - 3.2%)
1 .8%
(1.1% -2.4%)
2.5%
(1 .6% - 3.4%)
1 .6%
(1%-2.1%)
2.5%
(1 .6% - 3.3%)
2.8%
(1 .8% - 3.8%)
1 .8%
(1.1% -2.4%)
2.5%
(1 .6% - 3.3%)
0.7%
(0.5% - 1 %)
2.7%
(1 .7% - 3.6%)
1%
(0.6% - 1 .3%)
13/35
2.8%
(1 .8% - 3.7%)
2.5%
(1 .6% - 3.3%)
2.2%
(1 .4% - 3%)
2.4%
(1 .5% - 3.2%)
2%
(1 .3% - 2.7%)
1 .8%
(1.1% -2. 4%)
2.2%
(1 .4% - 2.9%)
1 .6%
(1%-2.1%)
2.3%
(1 .5% - 3.2%)
2.5%
(1 .6% - 3.3%)
1 .8%
(1.1% -2. 4%)
2.3%
(1.5% -3.1%)
0.7%
(0.5% - 1 %)
2.3%
(1.5% -3.1%)
1%
(0.6% - 1 .3%)
12/35
2.4%
(1 .5% - 3.2%)
2.1%
(1 .3% - 2.8%)
1 .9%
(1 .2% - 2.5%)
2.1%
(1 .3% - 2.8%)
1 .7%
(1.1% -2. 3%)
1 .8%
(1.1% -2. 4%)
1 .8%
(1 .2% - 2.5%)
1 .4%
(0.9% - 1 .9%)
2%
(1 .3% - 2.7%)
2.1%
(1 .4% - 2.8%)
1 .6%
(1%-2.1%)
2%
(1 .3% - 2.7%)
0.7%
(0.5% - 1 %)
2%
(1 .3% - 2.7%)
1%
(0.6% - 1 .3%)
13/30
2.8%
(1 .8% - 3.7%)
2.4%
(1 .5% - 3.2%)
2.2%
(1 .4% - 3%)
2.4%
(1 .5% - 3.2%)
1 .7%
(1.1% -2. 4%)
1 .2%
(0.8% - 1 .6%)
2.2%
(1 .4% - 2.9%)
1%
(0.7% - 1 .4%)
1 .8%
(1 .2% - 2.4%)
2.1%
(1 .3% - 2.8%)
1 .5%
(1 % - 2%)
1 .8%
(1 .2% - 2.4%)
0.3%
(0.2% - 0.5%)
2.3%
(1.5% -3.1%)
0.5%
(0.3% - 0.7%)
12/25
2.3%
(1.5% -3.1%)
1 .6%
(1%-2.2%)
1 .5%
(1%-2.1%)
2.1%
(1 .3% - 2.8%)
1.1%
(0.7% - 1 .5%)
0.6%
(0.4% - 0.9%)
1 .8%
(1 .2% - 2.5%)
0.5%
(0.3% - 0.7%)
1.1%
(0.7% - 1 .5%)
1 .4%
(0.9% - 1 .8%)
0.9%
(0.6% - 1 .2%)
1.1%
(0.7% - 1 .5%)
0%
(0% - 0%)
1 .5%
(1%-2.1%)
0.1%
(0%-0.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-7
-------
Table E-7. Percent Reduction from the Current Standards: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to
PM2 5 Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily
(m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
-1 3%
(-13% --13%)
-9%
(-9% - -9%)
-43%
(-43% - -43%)
0%
(0% - 0%)
-38%
(-37% - -38%)
-187%
(-185% --188%)
-9%
(-9% - -9%)
-123%
(-122% --124%)
-35%
(-35% - -36%)
-1 4%
(-14% --15%)
0%
(0% - 0%)
-53%
(-53% - -54%)
-255%
(-254% - -256%)
-1 8%
(-18% --18%)
-53%
(-53% - -54%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(1 1 % - 1 1 %)
8%
(8% - 9%)
12%
(1 2% - 1 2%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(1 2% - 1 2%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
22%
(21% -22%)
19%
(19% -19%)
23%
(23% - 24%)
0%
(0% - 0%)
13%
(13% -14%)
0%
(0% - 0%)
24%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(20% - 21 %)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
30%
(30% - 30%)
35%
(35% - 35%)
11%
(11% -11%)
26%
(26% - 26%)
0%
(0% - 0%)
35%
(35% - 35%)
12%
(12% -12%)
18%
(18% -18%)
24%
(23% - 24%)
10%
(10% -10%)
18%
(18% -18%)
0%
(0% - 0%)
31%
(31 % - 32%)
0%
(0% - 0%)
13/30
22%
(21 % - 22%)
22%
(22% - 22%)
23%
(23% - 24%)
0%
(0% - 0%)
24%
(24% - 24%)
34%
(34% - 34%)
24%
(23% - 24%)
31%
(31% -31%)
26%
(25% - 26%)
24%
(24% - 24%)
13%
(13% -13%)
25%
(25% - 26%)
64%
(64% - 64%)
23%
(22% - 23%)
34%
(34% - 34%)
12/25
34%
(34% - 34%)
45%
(45% - 45%)
46%
(46% - 46%)
11%
(11% -11%)
48%
(48% - 48%)
68%
(68% - 68%)
35%
(35% - 35%)
63%
(63% - 63%)
51%
(51% -51%)
49%
(49% - 49%)
46%
(45% - 46%)
51%
(51% -51%)
1 00%
(100% -100%)
45%
(45% - 45%)
67%
(67% - 67%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-S
-------
Table E-8. Percent Reduction from the Current Standards: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to
PM2.s Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily
(m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
-1 3%
(-13% --13%)
-1 0%
(-10% --10%)
-44%
(-44% - -45%)
0%
(0% - 0%)
-45%
(-45% - -45%)
-1 82%
(-180% --184%)
-1 0%
(-10% --10%)
-1 38%
(-137% --139%)
-41%
(-41% --41%)
-1 5%
(-15% --15%)
0%
(0% - 0%)
-61%
(-61% --62%)
-438%
(-437% - -440%)
-20%
(-20% - -21 %)
-76%
(-76% - -76%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(1 1 % - 1 1 %)
9%
(9% - 9%)
12%
(1 2% - 1 2%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
12%
(1 2% - 1 2%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
13/35
22%
(21 % - 22%)
21%
(21% -21%)
24%
(24% - 24%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
24%
(24% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
33%
(33% - 33%)
36%
(36% - 36%)
13%
(13% -13%)
31%
(30% -31%)
0%
(0% - 0%)
36%
(36% - 36%)
14%
(14% -14%)
21%
(21% -21%)
25%
(25% - 25%)
10%
(10% -10%)
20%
(20% - 20%)
0%
(0% - 0%)
36%
(36% - 36%)
0%
(0% - 0%)
13/30
22%
(21% -22%)
24%
(24% - 24%)
24%
(24% - 24%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
24%
(24% - 24%)
35%
(35% - 35%)
30%
(30% - 30%)
25%
(25% - 26%)
13%
(13% -13%)
29%
(29% - 29%)
1 00%
(100% -100%)
26%
(26% - 26%)
48%
(48% - 48%)
12/25
34%
(34% - 34%)
49%
(49% - 49%)
48%
(48% - 48%)
13%
(13% -13%)
57%
(57% - 57%)
66%
(66% - 66%)
36%
(36% - 36%)
70%
(70% - 70%)
59%
(59% - 60%)
51%
(51% -51%)
45%
(45% - 45%)
58%
(58% - 59%)
100%
(100% -100%)
52%
(51% -52%)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-9
-------
Table E-9. Percent Reduction from the Current Standards: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term
Exposure to Ambient PM25 Concentrations, Based on Adjusting 2007 PM25 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM25 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidenceof All Cause Mortality Associated with Long-Term Exposure to
PMzs Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n)and Daily
(m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concent ratbns
-1 3%
(-13% --13%)
-1 0%
(-10% --10%)
-44%
(-43% - -44%)
0%
(0% - 0%)
-43%
(-43% - -44%)
-176%
(-1 75% - -1 78%)
-1 0%
(-9% - -1 0%)
-134%
(-1 33% - -1 35%)
-38%
(-37% - -38%)
-1 5%
(-15% --15%)
0%
(0% - 0%)
-56%
(-55% - -56%)
-21 8%
(-21 7% --21 9%)
-20%
(-19% --20%)
-74%
(-73% - -74%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(1 1 % - 1 1 %)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
22%
(22% - 22%)
21%
(21% -21%)
24%
(24% - 24%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
24%
(24% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
12/35
33%
(33% - 33%)
33%
(33% - 33%)
36%
(36% - 36%)
12%
(12% -12%)
30%
(29% - 30%)
0%
(0% - 0%)
36%
(36% - 36%)
13%
(13% -13%)
20%
(20% - 20%)
25%
(25% - 25%)
11%
(1 1 % - 1 1 %)
19%
(19% -19%)
0%
(0% - 0%)
34%
(34% - 34%)
0%
(0% - 0%)
13/30
22%
(22% -22%)
24%
(24% -24%)
24%
(24% -24%)
0%
(0% - 0% )
27%
(27% -27%)
32%
(32% -32%)
24%
(24% -24%)
34%
(34% -34%)
27%
(27% -27%)
26%
(26% -26%)
14%
(14%- 14%)
27%
(27% -27%)
55%
(55% -55%)
25%
(25% -25%)
45%
(45% -45%)
12/25
35%
(35% - 35%)
49%
(49% - 49%)
47%
(47% - 47%)
12%
(12%- 12%)
55%
(55% - 55%)
64%
(64% - 64%)
36%
(36% - 36%)
68%
(68% - 68%)
54%
(54% - 55%)
51%
(51% -52%)
50%
(50% - 50%)
54%
(54% - 54%)
1 00%
(100%- 100%)
49%
(49% - 50%)
93%
(93% - 93%)
Based on follow-upthrough2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski etal., 2009).
Numbers roundedto the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3and a daily standard set at 35 ug/m3.
E-10
-------
Table E-10. Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in a Recent
Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005 PM2 5
Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
830
(531 -1123)
763
(488 - 1 033)
543
(347 - 734)
486
(310-659)
1021
(653 - 1 380)
325
(208 - 439)
780
(498 - 1 058)
2986
(1910-4042)
2560
(1636-3468)
668
(427 - 905)
620
(394 - 843)
759
(485 - 1 026)
132
(84 - 1 79)
1056
(676 - 1 429)
158
(101 -215)
15/353
736
(470 - 997)
702
(448 - 950)
380
(243-516)
486
(31 0 - 659)
743
(474-1008)
114
(72-155)
713
(455 - 968)
1342
(854-1827)
1893
(1207-2571)
584
(372 - 792)
620
(394 - 843)
497
(31 7 - 674)
37
(24-51)
897
(573-1215)
103
(66- 141)
14/35
657
(420 - 891 )
643
(410-871)
336
(214-457)
486
(310-659)
734
(468 - 996)
114
(72-155)
630
(401 - 856)
1342
(854-1827)
1893
(1207-2571)
584
(372 - 792)
620
(394 - 843)
497
(317-674)
37
(24-51)
813
(519-1102)
103
(66-141)
13/35
578
(369 - 785)
566
(361 - 768)
292
(186-397)
486
(310-659)
643
(410-874)
114
(72-155)
546
(348 - 743)
1342
(854-1827)
1808
(1152-2455)
521
(332 - 707)
620
(394 - 843)
466
(297 - 633)
37
(24-51)
714
(456 - 970)
103
(66- 141)
12/35
499
(318-677)
490
(312-665)
247
(157-336)
431
(275 - 586)
552
(352 - 751 )
114
(72-155)
462
(294 - 629)
1180
(750-1607)
1546
(984-2101)
447
(285 - 607)
557
(354 - 757)
409
(260 - 555)
37
(24-51)
616
(392 - 836)
103
(66-141)
13/30
578
(369 - 785)
546
(348 - 741 )
292
(186-397)
486
(310-659)
567
(361 - 770)
75
(48 - 1 03)
546
(348 - 743)
924
(587 - 1 258)
1412
(898 - 1 920)
442
(282 - 601 )
539
(343 - 734)
371
(236 - 504)
13
(8-18)
696
(443 - 944)
69
(44 - 94)
12/25
487
(310-661)
388
(247 - 528)
205
(130-280)
431
(275 - 586)
389
(247 - 530)
36
(23 - 50)
462
(294 - 629)
502
(318-684)
926
(588-1261)
299
(190-408)
338
(214-460)
244
(155-332)
0
(0-0)
492
(313-669)
34
(21 - 46)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-ll
-------
Table E-ll. Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in a Recent
Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006 PM2.5
Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
854
(547-1156)
619
(395 - 839)
510
(326 - 691 )
364
(232 - 495)
737
(471 -1000)
338
(217-457)
755
(481 -1024)
2631
(1680-3565)
1984
(1265-2693)
604
(386-819)
657
(418-893)
599
(383-813)
107
(68 - 1 46)
792
(506 - 1 075)
107
(68 - 1 45)
15/353
758
(484 - 1 026)
565
(360 - 766)
354
(226 - 481 )
364
(232 - 495)
510
(325 - 694)
121
(77 - 1 64)
689
(439 - 935)
1108
(704 - 1 509)
1407
(895-1913)
525
(335-713)
657
(418-893)
372
(237 - 506)
20
(13-27)
659
(420 - 894)
61
(38 - 83)
14/35
677
(432-918)
513
(327 - 696)
312
(198-423)
364
(232 - 495)
503
(320 - 684)
121
(77-164)
606
(386 - 823)
1108
(704- 1509)
1407
(895-1913)
525
(335-713)
657
(418-893)
372
(237 - 506)
20
(13-27)
588
(374 - 799)
61
(38 - 83)
13/35
595
(380 - 808)
446
(284 - 606)
269
(171 -365)
364
(232 - 495)
429
(273 - 584)
121
(77 - 1 64)
523
(333 -711)
1108
(704 - 1 509)
1333
(848 - 1 81 3)
466
(297 - 633)
657
(418-893)
346
(220-471)
20
(13-27)
506
(322 - 688)
61
(38 - 83)
12/35
513
(327 - 697)
378
(241 -515)
226
(1 44 - 307)
317
(202 - 432)
355
(225 - 483)
121
(77 - 1 64)
439
(279 - 598)
958
(608 - 1 305)
1106
(703 - 1 506)
396
(252 - 538)
591
(376 - 803)
297
(1 89 - 404)
20
(13-27)
423
(269 - 575)
61
(38 - 83)
13/30
595
(380 - 808)
428
(272 - 581 )
269
(171 -365)
364
(232 - 495)
366
(233 - 499)
81
(51 -110)
523
(333 -711)
721
(457 - 983)
990
(629-1349)
392
(249 - 533)
572
(364 - 779)
264
(168-359)
0
(0-0)
490
(312-666)
32
(20 - 43)
12/25
501
(319-681)
289
(184-394)
186
(118-253)
317
(202 - 432)
222
(141 -302)
41
(26 - 55)
439
(279 - 598)
331
(210-451)
571
(362 - 779)
257
(163-350)
361
(229 - 492)
155
(98 -211)
0
(0-0)
319
(203 - 435)
3
(2-3)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-12
-------
Table E-12. Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in a Recent
Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007 PM2 5
Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent Year and PM25
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
821
(525 - 1 1 1 2)
618
(394 - 838)
535
(342 - 724)
407
(259 - 553)
778
(496 - 1 054)
357
(228 - 482)
788
(502 - 1 069)
2732
(1 746 - 3702)
2322
(1482-3148)
598
(381 -811)
556
(354 - 757)
675
(431 -914)
154
(98 - 209)
869
(555-1178)
112
(71 -152)
15/353
726
(464 - 984)
564
(360 - 765)
374
(238 - 507)
407
(259 - 553)
544
(346 - 739)
130
(82 - 1 77)
719
(459 - 977)
1170
(744 - 1 593)
1689
(1076-2295)
519
(331 - 704)
556
(354 - 757)
434
(277 - 590)
48
(31 - 66)
728
(464 - 988)
64
(41 - 88)
14/35
647
(413-877)
512
(326 - 695)
330
(210-448)
407
(259 - 553)
536
(341 - 729)
130
(82-177)
634
(404-861)
1170
(744 - 1 593)
1689
(1076-2295)
519
(331 - 704)
556
(354 - 757)
434
(277 - 590)
48
(31 - 66)
653
(416-887)
64
(41 - 88)
13/35
567
(361 - 769)
445
(283 - 605)
285
(182-388)
407
(259 - 553)
460
(293 - 626)
130
(82-177)
548
(349 - 745)
1170
(744-1593)
1607
(1023-2185)
460
(293 - 625)
556
(354 - 757)
406
(258 - 552)
48
(31 - 66)
566
(360 - 769)
64
(41 - 88)
12/35
486
(310-661)
378
(240-514)
241
(153-327)
356
(227 - 485)
384
(244 - 522)
130
(82-177)
461
(293 - 628)
1016
(645-1384)
1359
(864-1848)
391
(249 - 531 )
494
(314-673)
352
(224 - 479)
48
(31 - 66)
478
(304 - 651 )
64
(41 - 88)
13/30
567
(361 - 769)
427
(272 - 580)
285
(182-388)
407
(259 - 553)
396
(252 - 539)
88
(56 - 1 20)
548
(349 - 745)
773
(490 - 1 053)
1232
(783 - 1 676)
386
(246 - 525)
477
(303 - 650)
318
(202 - 432)
22
(14-30)
549
(350 - 747)
35
(22 - 47)
12/25
474
(302 - 644)
289
(184-393)
199
(126-271)
356
(227 - 485)
247
(157-336)
46
(29 - 63)
461
(293 - 628)
372
(236 - 508)
771
(489-1051)
253
(161 -344)
278
(176-379)
200
(127-273)
0
(0-0)
369
(235 - 503)
5
(3-6)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-13
-------
Table E-13. Estimated Percent of Total Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2 s Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2005 PM2 s Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year
and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
5.5%
(3.5% - 7.4%)
5.4%
(3.5% - 7.3%)
5.5%
(3.5% - 7.5%)
3.8%
(2.4% -5.1%)
5.7%
(3.7% - 7.7%)
5.8%
(3.7% - 7.9%)
4.2%
(2.7% - 5.7%)
5.3%
(3.4% -7.1%)
4.9%
(3.1% -6.6%)
4.6%
(2.9% - 6.2%)
2.7%
(1 .7% - 3.7%)
5.5%
(3.5% - 7.4%)
2.8%
(1 .8% - 3.8%)
5.6%
(3.6% - 7.6%)
3.1%
(2% - 4.2%)
15/353
4.9%
(3.1% -6. 6%)
5%
(3.2% - 6.7%)
3.9%
(2.5% - 5.3%)
3.8%
(2.4% -5.1%)
4.2%
(2.7% - 5.6%)
2%
(1 .3% - 2.8%)
3.8%
(2.4% - 5.2%)
2.4%
(1 .5% - 3.2%)
3.6%
(2.3% - 4.9%)
4%
(2.6% - 5.4%)
2.7%
(1 .7% - 3.7%)
3.6%
(2.3% - 4.8%)
0.8%
(0.5% -1.1%)
4.8%
(3% - 6.5%)
2%
(1 .3% - 2.8%)
14/35
4.4%
(2.8% - 5.9%)
4.5%
(2.9% - 6.2%)
3.4%
(2.2% - 4.7%)
3.8%
(2. 4% -5.1%)
4.1%
(2.6% - 5.6%)
2%
(1 .3% - 2.8%)
3.4%
(2.2% - 4.6%)
2.4%
(1 .5% - 3.2%)
3.6%
(2.3% - 4.9%)
4%
(2.6% - 5.4%)
2.7%
(1 .7% - 3.7%)
3.6%
(2.3% - 4.8%)
0.8%
(0.5% -1.1%)
4.3%
(2.8% - 5.9%)
2%
(1 .3% - 2.8%)
13/35
3.8%
(2.4% - 5.2%)
4%
(2.6% - 5.4%)
3%
(1 .9% - 4%)
3.8%
(2. 4% -5.1%)
3.6%
(2.3% - 4.9%)
2%
(1 .3% - 2.8%)
2.9%
(1 .9% - 4%)
2.4%
(1 .5% - 3.2%)
3.4%
(2.2% - 4.7%)
3.6%
(2.3% - 4.9%)
2.7%
(1 .7% - 3.7%)
3.3%
(2.1% -4.5%)
0.8%
(0.5% -1.1%)
3.8%
(2.4% -5.1%)
2%
(1 .3% - 2.8%)
12/35
3.3%
(2.1% -4.5%)
3.5%
(2.2% - 4.7%)
2.5%
(1 .6% - 3.4%)
3.4%
(2.1% -4.6%)
3.1%
(2% - 4.2%)
2%
(1 .3% - 2.8%)
2.5%
(1 .6% - 3.4%)
2.1%
(1 .3% - 2.8%)
2.9%
(1 .9% - 4%)
3.1%
(2% - 4.2%)
2.4%
(1 .5% - 3.3%)
2.9%
(1 .9% - 4%)
0.8%
(0.5% -1.1%)
3.3%
(2.1% -4.4%)
2%
(1 .3% - 2.8%)
13/30
3.8%
(2.4% - 5.2%)
3.9%
(2.5% - 5.2%)
3%
(1 .9% - 4%)
3.8%
(2. 4% -5.1%)
3.2%
(2% - 4.3%)
1 .4%
(0.9% - 1 .8%)
2.9%
(1 .9% - 4%)
1 .6%
(1 % - 2.2%)
2.7%
(1 .7% - 3.6%)
3%
(1.9% -4.1%)
2.3%
(1 .5% - 3.2%)
2.7%
(1 .7% - 3.6%)
0.3%
(0.2% - 0.4%)
3.7%
(2.4% - 5%)
1 .3%
(0.9% - 1 .8%)
12/25
3.2%
(2.1% -4.4%)
2.8%
(1 .8% - 3.7%)
2.1%
(1 .3% - 2.8%)
3.4%
(2.1% -4.6%)
2.2%
(1 .4% - 3%)
0.7%
(0.4% - 0.9%)
2.5%
(1 .6% - 3.4%)
0.9%
(0.6% - 1 .2%)
1 .8%
(1.1% -2.4%)
2.1%
(1 .3% - 2.8%)
1 .5%
(0.9% - 2%)
1 .8%
(1.1% -2.4%)
0%
(0% - 0%)
2.6%
(1 .7% - 3.6%)
0.7%
(0.4% - 0.9%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-14
-------
Table E-14. Estimated Percent of Total Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2 s Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2006 PM2 s Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year
and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
5.5%
(3.5% - 7.4%)
4.4%
(2.8% - 5.9%)
5.1%
(3.3% - 7%)
2.8%
(1 .8% - 3.8%)
4.1%
(2.6% - 5.6%)
6%
(3.8% -8.1%)
3.9%
(2.5% - 5.3%)
4.6%
(2.9% - 6.2%)
3.7%
(2.4% -5.1%)
4.2%
(2.7% - 5.6%)
2.7%
(1 .7% - 3.7%)
4.3%
(2.8% - 5.9%)
2.2%
(1 .4% - 3%)
4.2%
(2.7% - 5.7%)
2.1%
(1 .3% - 2.8%)
15/353
4.9%
(3.1% -6. 6%)
4%
(2.5% - 5.4%)
3.6%
(2.3% - 4.8%)
2.8%
(1 .8% - 3.8%)
2.9%
(1 .8% - 3.9%)
2.1%
(1 .4% - 2.9%)
3.6%
(2.3% - 4.9%)
1 .9%
(1 .2% - 2.6%)
2.6%
(1 .7% - 3.6%)
3.6%
(2.3% - 4.9%)
2.7%
(1 .7% - 3.7%)
2.7%
(1 .7% - 3.7%)
0.4%
(0.3% - 0.6%)
3.5%
(2.2% - 4.7%)
1 .2%
(0.7% - 1 .6%)
14/35
4.4%
(2.8% - 5.9%)
3.6%
(2.3% - 4.9%)
3.1%
(2% - 4.3%)
2.8%
(1 .8% - 3.8%)
2.8%
(1 .8% - 3.8%)
2.1%
(1 .4% - 2.9%)
3.1%
(2% - 4.3%)
1 .9%
(1 .2% - 2.6%)
2.6%
(1 .7% - 3.6%)
3.6%
(2.3% - 4.9%)
2.7%
(1 .7% - 3.7%)
2.7%
(1 .7% - 3.7%)
0.4%
(0.3% - 0.6%)
3.1%
(2% - 4.2%)
1 .2%
(0.7% - 1 .6%)
13/35
3.8%
(2.4% - 5.2%)
3.1%
(2% - 4.3%)
2.7%
(1 .7% - 3.7%)
2.8%
(1 .8% - 3.8%)
2.4%
(1 .5% - 3.3%)
2.1%
(1 .4% - 2.9%)
2.7%
(1 .7% - 3.7%)
1 .9%
(1 .2% - 2.6%)
2.5%
(1 .6% - 3.4%)
3.2%
(2% - 4.4%)
2.7%
(1 .7% - 3.7%)
2.5%
(1 .6% - 3.4%)
0.4%
(0.3% - 0.6%)
2.7%
(1 .7% - 3.6%)
1 .2%
(0.7% - 1 .6%)
12/35
3.3%
(2.1% -4.5%)
2.7%
(1 .7% - 3.6%)
2.3%
(1.4% -3.1%)
2.4%
(1 .5% - 3.3%)
2%
(1 .3% - 2.7%)
2.1%
(1 .4% - 2.9%)
2.3%
(1.4% -3.1%)
1 .7%
(1.1% -2.3%)
2.1%
(1 .3% - 2.8%)
2.7%
(1 .7% - 3.7%)
2.5%
(1 .6% - 3.4%)
2.1%
(1 .4% - 2.9%)
0.4%
(0.3% - 0.6%)
2.2%
(1 .4% - 3%)
1 .2%
(0.7% - 1 .6%)
13/30
3.8%
(2.4% - 5.2%)
3%
(1.9% -4.1%)
2.7%
(1 .7% - 3.7%)
2.8%
(1 .8% - 3.8%)
2.1%
(1 .3% - 2.8%)
1 .4%
(0.9% - 1 .9%)
2.7%
(1 .7% - 3.7%)
1 .3%
(0.8% - 1 .7%)
1 .9%
(1 .2% - 2.5%)
2.7%
(1 .7% - 3.7%)
2.4%
(1 .5% - 3.3%)
1 .9%
(1 .2% - 2.6%)
0%
(0% - 0%)
2.6%
(1 .6% - 3.5%)
0.6%
(0.4% - 0.8%)
12/25
3.2%
(2.1% -4.4%)
2%
(1 .3% - 2.8%)
1 .9%
(1 .2% - 2.5%)
2.4%
(1 .5% - 3.3%)
1 .2%
(0.8% - 1 .7%)
0.7%
(0.5% -1%)
2.3%
(1.4% -3.1%)
0.6%
(0.4% - 0.8%)
1.1%
(0.7% - 1 .5%)
1 .8%
(1.1% -2.4%)
1 .5%
(1%-2.1%)
1.1%
(0.7% - 1 .5%)
0%
(0% - 0%)
1 .7%
(1.1% -2.3%)
0%
(0%-0.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-15
-------
Table E-15. Estimated Percent of Total Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year
and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
5.1%
(3.3% - 7%)
4.4%
(2.8% - 5.9%)
5.3%
(3.4% - 7.2%)
3%
(1.9% -4.1%)
4.4%
(2.8% - 6%)
6.2%
(4% - 8.4%)
4%
(2.6% - 5.4%)
4.8%
(3% - 6.4%)
4.3%
(2.8% - 5.9%)
4.1%
(2.6% - 5.6%)
2.3%
(1.4% -3.1%)
4.9%
(3.1% -6.6%)
3%
(1.9% -4.1%)
4.6%
(2.9% - 6.2%)
2.1%
(1 .3% - 2.9%)
15/353
4.6%
(2.9% - 6.2%)
4%
(2.5% - 5.4%)
3.7%
(2.4% -5.1%)
3%
(1.9% -4.1%)
3.1%
(2% - 4.2%)
2.3%
(1.4% -3.1%)
3.7%
(2.3% - 5%)
2%
(1 .3% - 2.8%)
3.2%
(2% - 4.3%)
3.6%
(2.3% - 4.8%)
2.3%
(1.4% -3.1%)
3.2%
(2% - 4.3%)
1%
(0.6% - 1 .3%)
3.8%
(2.4% - 5.2%)
1 .2%
(0.8% - 1 .7%)
14/35
4.1%
(2.6% - 5.5%)
3.6%
(2.3% - 4.9%)
3.3%
(2.1% -4.5%)
3%
(1.9% -4.1%)
3%
(1.9% -4.1%)
2.3%
(1.4% -3.1%)
3.2%
(2.1% -4.4%)
2%
(1 .3% - 2.8%)
3.2%
(2% - 4.3%)
3.6%
(2.3% - 4.8%)
2.3%
(1.4% -3.1%)
3.2%
(2% - 4.3%)
1%
(0.6% - 1 .3%)
3.4%
(2.2% - 4.7%)
1 .2%
(0.8% - 1 .7%)
13/35
3.6%
(2.3% - 4.8%)
3.2%
(2% - 4.3%)
2.8%
(1 .8% - 3.9%)
3%
(1.9% -4.1%)
2.6%
(1 .7% - 3.5%)
2.3%
(1.4% -3.1%)
2.8%
(1 .8% - 3.8%)
2%
(1 .3% - 2.8%)
3%
(1.9% -4.1%)
3.2%
(2% - 4.3%)
2.3%
(1.4% -3.1%)
2.9%
(1 .9% - 4%)
1%
(0.6% - 1 .3%)
3%
(1.9% -4.1%)
1 .2%
(0.8% - 1 .7%)
12/35
3%
(1.9% -4.1%)
2.7%
(1 .7% - 3.6%)
2.4%
(1 .5% - 3.3%)
2.7%
(1 .7% - 3.6%)
2.2%
(1 .4% - 3%)
2.3%
(1.4% -3.1%)
2.3%
(1 .5% - 3.2%)
1 .8%
(1.1% -2.4%)
2.5%
(1 .6% - 3.4%)
2.7%
(1 .7% - 3.7%)
2%
(1 .3% - 2.7%)
2.6%
(1 .6% - 3.5%)
1%
(0.6% - 1 .3%)
2.5%
(1 .6% - 3.4%)
1 .2%
(0.8% - 1 .7%)
13/30
3.6%
(2.3% - 4.8%)
3%
(1.9% -4.1%)
2.8%
(1 .8% - 3.9%)
3%
(1.9% -4.1%)
2.2%
(1 .4% - 3%)
1 .5%
(1%-2.1%)
2.8%
(1 .8% - 3.8%)
1 .3%
(0.9% - 1 .8%)
2.3%
(1.5% -3.1%)
2.7%
(1 .7% - 3.6%)
1 .9%
(1 .2% - 2.6%)
2.3%
(1.5% -3.1%)
0.4%
(0.3% - 0.6%)
2.9%
(1 .8% - 3.9%)
0.7%
(0.4% - 0.9%)
12/25
3%
(1 .9% - 4%)
2%
(1 .3% - 2.8%)
2%
(1 .3% - 2.7%)
2.7%
(1 .7% - 3.6%)
1 .4%
(0.9% - 1 .9%)
0.8%
(0.5% -1.1%)
2.3%
(1 .5% - 3.2%)
0.6%
(0.4% - 0.9%)
1 .4%
(0.9% - 2%)
1 .7%
(1.1% -2.4%)
1.1%
(0.7% - 1 .5%)
1 .5%
(0.9% - 2%)
0%
(0% - 0%)
1 .9%
(1 .2% - 2.7%)
0.1%
(0.1% -0.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-16
-------
Table E-16. Percent Reduction from the Current Standards: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to
PM2 5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily
(m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-13% --13%)
-9%
(-9% - -9%)
-43%
(-42% - -43%)
0%
(0% - 0%)
-37%
(-37% - -38%)
-185%
(-183% --187%)
-9%
(-9% - -9%)
-122%
(-121% --124%)
-35%
(-35% - -36%)
-14%
(-14% --15%)
0%
(0% - 0%)
-53%
(-52% - -53%)
-254%
(-253% - -256%)
-18%
(-18% --18%)
-53%
(-53% - -53%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(1 1 % - 1 1 %)
8%
(8% - 8%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
21%
(21 % - 22%)
19%
(19% -19%)
23%
(23% - 23%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
23%
(23% - 24%)
0%
(0% - 0%)
5%
(4% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
12/35
32%
(32% - 32%)
30%
(30% - 30%)
35%
(35% - 35%)
11%
(1 1 % - 1 1 %)
26%
(25% - 26%)
0%
(0% - 0%)
35%
(35% - 35%)
12%
(12% -12%)
18%
(18% -18%)
23%
(23% - 24%)
10%
(10% -10%)
18%
(18% -18%)
0%
(0% - 0%)
31%
(31 % - 32%)
0%
(0% - 0%)
13/30
21%
(21 % - 22%)
22%
(22% - 22%)
23%
(23% - 23%)
0%
(0% - 0%)
24%
(24% - 24%)
34%
(34% - 34%)
23%
(23% - 24%)
31%
(31% -31%)
25%
(25% - 26%)
24%
(24% - 24%)
13%
(13% -13%)
25%
(25% - 25%)
64%
(64% - 64%)
22%
(22% - 23%)
34%
(33% - 34%)
12/25
34%
(34% - 34%)
45%
(44% - 45%)
46%
(46% - 46%)
11%
(1 1 % - 1 1 %)
48%
(47% - 48%)
68%
(68% - 68%)
35%
(35% - 35%)
63%
(63% - 63%)
51%
(51% -51%)
49%
(49% - 49%)
45%
(45% - 46%)
51%
(51% -51%)
1 00%
(100% -100%)
45%
(45% - 45%)
67%
(67% - 67%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-17
-------
Table E-17. Percent Reduction from the Current Standards: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to
PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily
(m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-13% --13%)
-10%
(-10% --10%)
-44%
(-44% - -44%)
0%
(0% - 0%)
-45%
(-44% - -45%)
-181%
(-179% --183%)
-10%
(-10% --10%)
-137%
(-136% --139%)
-41 %
(-41 % - -41 %)
-15%
(-15% --15%)
0%
(0% - 0%)
-61 %
(-61 % - -62%)
-437%
(-435% - -439%)
-20%
(-20% - -20%)
-76%
(-76% - -76%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(1 1 % - 1 1 %)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
13/35
21%
(21 % - 22%)
21%
(21 % - 21 %)
24%
(24% - 24%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
24%
(24% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(32% - 32%)
33%
(33% - 33%)
36%
(36% - 36%)
13%
(13% -13%)
30%
(30% - 31 %)
0%
(0% - 0%)
36%
(36% - 36%)
14%
(14% -14%)
21%
(21 % - 21 %)
25%
(24% - 25%)
10%
(10% -10%)
20%
(20% - 20%)
0%
(0% - 0%)
36%
(36% - 36%)
0%
(0% - 0%)
13/30
21%
(21 % - 22%)
24%
(24% - 24%)
24%
(24% - 24%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
24%
(24% - 24%)
35%
(35% - 35%)
30%
(29% - 30%)
25%
(25% - 26%)
13%
(13% -13%)
29%
(29% - 29%)
100%
(100% -100%)
26%
(26% - 26%)
48%
(48% - 48%)
12/25
34%
(34% - 34%)
49%
(49% - 49%)
48%
(47% - 48%)
13%
(13% -13%)
57%
(56% - 57%)
66%
(66% - 66%)
36%
(36% - 36%)
70%
(70% - 70%)
59%
(59% - 60%)
51%
(51% -51%)
45%
(45% - 45%)
58%
(58% - 58%)
1 00%
(100% -100%)
51%
(51 % - 52%)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-18
-------
Table E-18. Percent Reduction from the Current Standards: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term
Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2007 PM2.5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to
PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily
(m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-13% --13%)
-10%
(-10% --10%)
-43%
(-43% - -44%)
0%
(0% - 0%)
-43%
(-43% - -43%)
-175%
(-173% --177%)
-9%
(-9% - -1 0%)
-134%
(-132% --135%)
-37%
(-37% - -38%)
-15%
(-15% --15%)
0%
(0% - 0%)
-55%
(-55% - -56%)
-21 7%
(-21 6% --21 8%)
-19%
(-19% --20%)
-74%
(-73% - -74%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(1 1 % - 1 1 %)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
22%
(22% - 22%)
21%
(21 % - 21 %)
24%
(24% - 24%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
24%
(24% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
12/35
33%
(33% - 33%)
33%
(33% - 33%)
36%
(35% - 36%)
12%
(12% -12%)
29%
(29% - 30%)
0%
(0% - 0%)
36%
(36% - 36%)
13%
(13% -13%)
20%
(19% -20%)
25%
(25% - 25%)
11%
(1 1 % - 1 1 %)
19%
(19% -19%)
0%
(0% - 0%)
34%
(34% - 34%)
0%
(0% - 0%)
13/30
22%
(22% - 22%)
24%
(24% - 24%)
24%
(24% - 24%)
0%
(0% - 0%)
27%
(27% - 27%)
32%
(32% - 32%)
24%
(24% - 24%)
34%
(34% - 34%)
27%
(27% - 27%)
26%
(25% - 26%)
14%
(14% -14%)
27%
(27% - 27%)
55%
(55% - 55%)
25%
(24% - 25%)
46%
(46% - 46%)
12/25
35%
(35% - 35%)
49%
(49% - 49%)
47%
(47% - 47%)
12%
(12% -12%)
55%
(54% - 55%)
64%
(64% - 64%)
36%
(36% - 36%)
68%
(68% - 68%)
54%
(54% - 55%)
51%
(51% -51%)
50%
(50% - 50%)
54%
(54% - 54%)
1 00%
(100% -100%)
49%
(49% - 49%)
93%
(93% - 93%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-19
-------
Table E-19. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year
and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2.5
Concentrations
249
(205-291)
396
(326 - 464)
186
(153-218)
231
(189-272)
689
(567 - 806)
187
(154-219)
370
(304 - 435)
2124
(1 746 - 2489)
2614
(2147-3068)
333
(273-391)
351
(286 - 41 4)
436
(359 -511)
40
(33 - 47)
636
(523 - 744)
93
(76-109)
15/353
222
(182-260)
366
(301 - 429)
133
(109-156)
231
(189-272)
509
(418-599)
68
(56-81)
340
(278 - 400)
984
(802 - 1 1 63)
1959
(1603-2307)
293
(240 - 345)
351
(286-414)
291
(238 - 343)
12
(9-14)
544
(447 - 639)
61
(50 - 72)
14/35
199
(163-234)
337
(276 - 395)
118
(96-139)
231
(189-272)
504
(413-592)
68
(56-81)
302
(247 - 356)
984
(802 - 1 1 63)
1959
(1603-2307)
293
(240 - 345)
351
(286-414)
291
(238 - 343)
12
(9-14)
496
(406 - 583)
61
(50 - 72)
13/35
176
(144-207)
298
(244 - 351 )
103
(84-121)
231
(189-272)
444
(363 - 523)
68
(56-81)
263
(215-310)
984
(802 - 1 1 63)
1874
(1533-2208)
263
(215-309)
351
(286-414)
274
(224 - 323)
12
(9-14)
438
(359-516)
61
(50 - 72)
12/35
153
(125-180)
259
(21 2 - 306)
87
(71 -103)
206
(169-243)
383
(31 3 - 452)
68
(56 - 81 )
223
(182-264)
867
(707-1026)
1610
(1315-1900)
226
(185-267)
316
(258 - 374)
241
(197-285)
12
(9-14)
379
(31 0 - 447)
61
(50 - 72)
13/30
176
(144-207)
288
(236 - 339)
103
(84-121)
231
(189-272)
393
(321 - 463)
45
(37 - 54)
263
(215-310)
682
(555 - 808)
1475
(1204-1742)
224
(183-264)
307
(250 - 362)
219
(179-259)
4
(3-5)
427
(350 - 503)
41
(33 - 48)
12/25
149
(122-176)
207
(169-245)
73
(59 - 86)
206
(169-243)
272
(222 - 322)
22
(18-26)
223
(182-264)
373
(303 - 443)
976
(795 - 1 1 56)
153
(125-181)
194
(158-230)
146
(119-172)
0
(0-0)
305
(249-361)
20
(16-24)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-20
-------
Table E-20. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent Year
and PM2.s Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2.5
Concentrations
256
(21 1 - 300)
325
(266 - 382)
176
(144-206)
175
(143-207)
506
(415-595)
194
(160-227)
359
(294 - 423)
1884
(1546-2212)
2050
(1678-2413)
303
(248 - 356)
372
(303 - 439)
349
(286-410)
33
(27 - 39)
484
(397 - 569)
63
(51 - 75)
15/353
229
(188-268)
297
(244 - 350)
124
(101 -146)
175
(143-207)
355
(290-418)
72
(59 - 85)
329
(269 - 388)
815
(664 - 965)
1470
(1200-1736)
264
(216-311)
372
(303 - 439)
220
(180-260)
6
(5-7)
405
(331 - 477)
36
(29 - 43)
14/35
205
(168-241)
271
(222-319)
110
(90 - 1 29)
175
(143-207)
350
(286-413)
72
(59 - 85)
291
(238 - 343)
815
(664 - 965)
1470
(1200- 1736)
264
(216-311)
372
(303 - 439)
220
(180-260)
6
(5-7)
363
(297 - 428)
36
(29 - 43)
13/35
181
(149-214)
237
(194-279)
95
(77 - 1 1 2)
175
(143-207)
300
(244 - 354)
72
(59 - 85)
252
(206 - 298)
815
(664 - 965)
1394
(1138-1648)
236
(193-278)
372
(303 - 439)
205
(167-242)
6
(5-7)
314
(256 - 370)
36
(29 - 43)
12/35
157
(1 29 - 1 85)
202
(1 65 - 239)
80
(65 - 95)
153
(125-181)
249
(203 - 294)
72
(59 - 85)
213
(1 73 - 252)
707
(575 - 837)
1163
(947-1375)
201
(1 64 - 238)
335
(273 - 396)
177
(1 44 - 209)
6
(5-7)
263
(215-311)
36
(29 - 43)
13/30
181
(149-214)
228
(186-268)
95
(77-112)
175
(143-207)
257
(209 - 304)
49
(40 - 58)
252
(206 - 298)
534
(434 - 633)
1043
(850 - 1 235)
199
(163-235)
325
(265 - 384)
157
(128-186)
0
(0-0)
304
(248 - 359)
19
(15-23)
12/25
154
(126-181)
155
(127-184)
66
(54 - 78)
153
(125-181)
157
(127-186)
25
(20 - 29)
213
(173-252)
247
(200 - 293)
606
(492-719)
132
(107-156)
207
(168-245)
93
(76-110)
0
(0-0)
200
(163-237)
2
(1-2)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-21
-------
Table E-21. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year
and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2.5
Concentrations
247
(203 - 290)
324
(266 - 381 )
184
(151 -216)
195
(1 59 - 230)
532
(436 - 625)
204
(1 69 - 239)
375
(307 - 441 )
1953
(1604-2293)
2384
(1955-2802)
300
(245 - 352)
317
(258 - 374)
390
(321 - 458)
47
(38 - 55)
529
(434 - 621 )
66
(54 - 78)
15/353
220
(180-258)
297
(243 - 349)
131
(107-154)
195
(159-230)
377
(308 - 445)
77
(63 - 92)
344
(281 - 405)
860
(701 -1018)
1755
(1435-2070)
261
(214-308)
317
(258 - 374)
256
(209 - 302)
15
(12-18)
446
(365 - 525)
38
(31 - 46)
14/35
197
(161 -231)
271
(221 -319)
116
(95 - 1 36)
195
(159-230)
372
(304 - 439)
77
(63 - 92)
304
(249 - 358)
860
(701 -1018)
1755
(1435-2070)
261
(214-308)
317
(258 - 374)
256
(209 - 302)
15
(12-18)
402
(329 - 474)
38
(31 - 46)
13/35
173
(142-204)
236
(193-279)
101
(82-119)
195
(159-230)
321
(262 - 379)
77
(63 - 92)
264
(215-312)
860
(701 -1018)
1673
(1367-1974)
233
(190-275)
317
(258 - 374)
239
(196-283)
15
(12-18)
350
(286-413)
38
(31 - 46)
12/35
149
(122-176)
202
(165-238)
85
(70-101)
172
(140-203)
269
(219-318)
77
(63 - 92)
223
(182-264)
749
(610-887)
1421
(1160-1679)
199
(162-235)
282
(230 - 333)
209
(170-246)
15
(12-18)
297
(243 - 351 )
38
(31 - 46)
13/30
173
(142-204)
227
(186-268)
101
(82-119)
195
(159-230)
277
(226 - 327)
53
(43 - 63)
264
(215-312)
572
(465 - 678)
1292
(1053-1527)
197
(160-232)
272
(222 - 322)
189
(154-223)
7
(6-8)
340
(278 - 401 )
21
(17-25)
12/25
146
(119-172)
155
(1 26 - 1 84)
71
(58 - 84)
172
(1 40 - 203)
174
(1 42 - 206)
28
(23 - 33)
223
(1 82 - 264)
278
(225 - 330)
815
(663 - 966)
130
(1 06 - 1 54)
160
(1 30 - 1 89)
120
(98-142)
0
(0-0)
231
(1 88 - 273)
3
(2-3)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-22
-------
Table £-22. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5
from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
15.8%
(13% -18.6%)
15.6%
(12.8% -18.2%)
15.9%
(13.1% -18.6%)
11.1%
(9.1% -13.1%)
16.4%
(13.5% -19.2%)
16.8%
(13.8% -19.6%)
12.2%
(10% -14.4%)
15.2%
(12.5% -17.8%)
14.1%
(11. 6% -16.5%)
1 3.3%
(10.9% -15.6%)
8%
(6.5% - 9.4%)
15.7%
(12.9% -18.4%)
8.2%
(6.7% - 9.7%)
16.1%
(13.3% -18.9%)
9.2%
(7.5% -10.8%)
15/353
14.1%
(11. 6% -16.6%)
14.4%
(11. 8% -16.9%)
1 1 .3%
(9.3% - 1 3.4%)
11.1%
(9.1% -13.1%)
12.2%
(10% -14.3%)
6.1%
(5% - 7.2%)
1 1 .2%
(9.2% -13.2%)
7%
(5.7% - 8.3%)
10.6%
(8.6% - 1 2.4%)
1 1 .7%
(9.6% - 1 3.8%)
8%
(6.5% - 9.4%)
10.5%
(8.6% - 1 2.3%)
2.4%
(1 .9% - 2.8%)
13.8%
(11. 3% -16.2%)
6.1%
(4.9% - 7.2%)
14/35
12.7%
(10.4% -14.9%)
13.2%
(10.9% -15.5%)
10.1%
(8.2% - 1 1 .9%)
11.1%
(9.1% -13.1%)
12%
(9.8% -14.1%)
6.1%
(5% - 7.2%)
9.9%
(8.1% -11. 7%)
7%
(5.7% - 8.3%)
10.6%
(8.6% -12.4%)
1 1 .7%
(9.6% -13.8%)
8%
(6.5% - 9.4%)
10.5%
(8.6% -12.3%)
2.4%
(1 .9% - 2.8%)
12.6%
(10.3% -14.8%)
6.1%
(4.9% - 7.2%)
13/35
1 1 .2%
(9.2% -13.2%)
1 1 .7%
(9.6% -13.8%)
8.8%
(7.2% -10.4%)
11.1%
(9.1% -13.1%)
10.6%
(8.7% -12.5%)
6.1%
(5% - 7.2%)
8.7%
(7.1% -10.2%)
7%
(5.7% - 8.3%)
10.1%
(8.3% - 1 1 .9%)
10.5%
(8.6% -12.4%)
8%
(6.5% - 9.4%)
9.9%
(8.1% -11. 6%)
2.4%
(1 .9% - 2.8%)
11.1%
(9.1% -13.1%)
6.1%
(4.9% - 7.2%)
12/35
9.7%
(8% - 1 1 .5%)
10.2%
(8.3% - 1 2%)
7.5%
(6.1% -8.8%)
9.9%
(8.1% -11. 7%)
9.1%
(7.5% -10.8%)
6.1%
(5% - 7.2%)
7.4%
(6% - 8.7%)
6.2%
(5.1% -7.3%)
8.7%
(7.1% -10.2%)
9.1%
(7.4% -10.7%)
7.2%
(5.9% - 8.5%)
8.7%
(7.1% -10.2%)
2.4%
(1 .9% - 2.8%)
9.6%
(7.9% - 1 1 .4%)
6.1%
(4.9% - 7.2%)
13/30
1 1 .2%
(9.2% -13.2%)
1 1 .3%
(9.3% -13.3%)
8.8%
(7.2% -10.4%)
11.1%
(9.1% -13.1%)
9.4%
(7.7% -11.1%)
4.1%
(3.3% - 4.8%)
8.7%
(7.1% -10.2%)
4.9%
(4% - 5.8%)
7.9%
(6.5% - 9.4%)
9%
(7.3% -10.6%)
7%
(5.7% - 8.2%)
7.9%
(6.4% - 9.3%)
0.8%
(0.7% -1%)
10.8%
(8.9% -12.8%)
4.1%
(3.3% - 4.8%)
12/25
9.5%
(7.8% - 1 1 .2%)
8.1%
(6.7% - 9.6%)
6.2%
(5.1% -7.4%)
9.9%
(8.1% -11. 7%)
6.5%
(5.3% - 7.7%)
2%
(1 .6% - 2.4%)
7.4%
(6% - 8.7%)
2.7%
(2.2% - 3.2%)
5.3%
(4.3% - 6.2%)
6.1%
(5% - 7.3%)
4.4%
(3.6% - 5.2%)
5.2%
(4.3% - 6.2%)
0%
(0% - 0%)
7.7%
(6.3% - 9.2%)
2%
(1 .6% - 2.4%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-23
-------
Table E-23. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5
from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
15.8%
(13% -18.5%)
12.7%
(10.5% -15%)
14.8%
(12.2% -17.4%)
8.2%
(6.7% - 9.7%)
12.1%
(9.9% -14.2%)
17.2%
(14.1% -20.1%)
1 1 .5%
(9.4% -13.5%)
13.4%
(11% -15.7%)
1 0.9%
(9% -12.9%)
12.1%
(9.9% -14.3%)
8.1%
(6.6% - 9.6%)
1 2.6%
(10.4% -14.8%)
6.5%
(5.3% - 7.7%)
12.2%
(10% -14.4%)
6.1%
(5% - 7.3%)
15/353
14.1%
(11. 6% -16.6%)
1 1 .7%
(9.6% -13.7%)
10.5%
(8.6% - 1 2.3%)
8.2%
(6.7% - 9.7%)
8.5%
(6.9% - 1 0%)
6.4%
(5.2% - 7.5%)
10.5%
(8.6% -12.4%)
5.8%
(4.7% - 6.9%)
7.8%
(6.4% - 9.3%)
10.6%
(8.7% -12.5%)
8.1%
(6.6% - 9.6%)
8%
(6.5% - 9.4%)
1 .2%
(1%-1.5%)
10.2%
(8.4% -12.1%)
3.5%
(2.9% - 4.2%)
14/35
12.7%
(10.4% -14.9%)
10.6%
(8.7% -12.5%)
9.3%
(7.6% -10.9%)
8.2%
(6.7% - 9.7%)
8.4%
(6.8% - 9.9%)
6.4%
(5.2% - 7.5%)
9.3%
(7.6% -10.9%)
5.8%
(4.7% - 6.9%)
7.8%
(6.4% - 9.3%)
10.6%
(8.7% -12.5%)
8.1%
(6.6% - 9.6%)
8%
(6.5% - 9.4%)
1 .2%
(1%-1.5%)
9.2%
(7.5% - 1 0.8%)
3.5%
(2.9% - 4.2%)
13/35
1 1 .2%
(9.2% -13.2%)
9.3%
(7.6% -11%)
8%
(6.5% - 9.5%)
8.2%
(6.7% - 9.7%)
7.2%
(5.8% - 8.5%)
6.4%
(5.2% - 7.5%)
8%
(6.6% - 9.5%)
5.8%
(4.7% - 6.9%)
7.4%
(6.1% -8.8%)
9.4%
(7.7% -11.1%)
8.1%
(6.6% - 9.6%)
7.4%
(6.1% -8.8%)
1 .2%
(1%-1.5%)
7.9%
(6.5% - 9.4%)
3.5%
(2.9% - 4.2%)
12/35
9.7%
(8% - 1 1 .5%)
7.9%
(6.5% - 9.4%)
6.8%
(5.5% - 8%)
7.2%
(5.9% - 8.5%)
5.9%
(4.8% - 7%)
6.4%
(5.2% - 7.5%)
6.8%
(5.5% - 8%)
5%
(4.1% -5.9%)
6.2%
(5.1% -7.3%)
8.1%
(6.6% - 9.5%)
7.3%
(6% - 8.7%)
6.4%
(5.2% - 7.6%)
1 .2%
(1 % - 1 .5%)
6.7%
(5.4% - 7.9%)
3.5%
(2.9% - 4.2%)
13/30
1 1 .2%
(9.2% -13.2%)
8.9%
(7.3% -10.5%)
8%
(6.5% - 9.5%)
8.2%
(6.7% - 9.7%)
6.1%
(5% - 7.3%)
4.3%
(3.5% -5.1%)
8%
(6.6% - 9.5%)
3.8%
(3.1% -4.5%)
5.6%
(4.5% - 6.6%)
8%
(6.5% - 9.4%)
7.1%
(5.8% - 8.4%)
5.7%
(4.6% - 6.7%)
0%
(0% - 0%)
7.7%
(6.3% -9.1%)
1 .8%
(1 .5% - 2.2%)
12/25
9.5%
(7.8% - 1 1 .2%)
6.1%
(5% - 7.2%)
5.6%
(4.5% - 6.6%)
7.2%
(5.9% - 8.5%)
3.7%
(3% - 4.4%)
2.2%
(1 .8% - 2.6%)
6.8%
(5.5% - 8%)
1 .8%
(1.4% -2.1%)
3.2%
(2.6% - 3.8%)
5.3%
(4.3% - 6.3%)
4.5%
(3.7% - 5.4%)
3.4%
(2.7% - 4%)
0%
(0% - 0%)
5.1%
(4.1% -6%)
0.1%
(0.1% -0.2%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-24
-------
Table E-24. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5
from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
14.9%
(12.2% -17.4%)
12.7%
(10.5% -15%)
15.4%
(12.7% -18%)
9%
(7.3% -10.6%)
12.8%
(10.5% -15%)
17.7%
(14.6% -20.7%)
1 1 .7%
(9.6% -13.8%)
1 3.8%
(11. 3% -16.2%)
12.6%
(10.4% -14.8%)
12%
(9.8% -14.1%)
6.7%
(5.5% - 7.9%)
1 4.2%
(11. 7% -16.7%)
9%
(7.4% -10.6%)
13.3%
(10.9% -15.7%)
6.3%
(5.2% - 7.5%)
15/353
13.2%
(10.9% -15.5%)
1 1 .7%
(9.6% -13.7%)
10.9%
(8.9% -12.9%)
9%
(7.3% -10.6%)
9.1%
(7.4% -10.7%)
6.7%
(5.5% - 8%)
10.7%
(8.8% -12.6%)
6.1%
(4.9% - 7.2%)
9.3%
(7.6% - 1 1 %)
10.5%
(8.6% -12.3%)
6.7%
(5.5% - 7.9%)
9.3%
(7.6% - 1 1 %)
2.9%
(2.4% - 3.4%)
1 1 .2%
(9.2% - 1 3.2%)
3.7%
(3% - 4.4%)
14/35
1 1 .8%
(9.7% -13.9%)
10.6%
(8.7% -12.5%)
9.7%
(7.9% - 1 1 .4%)
9%
(7.3% -10.6%)
9%
(7.3% -10.6%)
6.7%
(5.5% - 8%)
9.5%
(7.8% - 1 1 .2%)
6.1%
(4.9% - 7.2%)
9.3%
(7.6% -11%)
10.5%
(8.6% -12.3%)
6.7%
(5.5% - 7.9%)
9.3%
(7.6% -11%)
2.9%
(2.4% - 3.4%)
10.1%
(8.3% - 1 1 .9%)
3.7%
(3% - 4.4%)
13/35
10.4%
(8.5% -12.3%)
9.3%
(7.6% -11%)
8.4%
(6.9% - 9.9%)
9%
(7.3% -10.6%)
7.7%
(6.3% -9.1%)
6.7%
(5.5% - 8%)
8.2%
(6.7% - 9.7%)
6.1%
(4.9% - 7.2%)
8.9%
(7.2% -10.5%)
9.3%
(7.6% -11%)
6.7%
(5.5% - 7.9%)
8.7%
(7.1% -10.3%)
2.9%
(2.4% - 3.4%)
8.8%
(7.2% - 1 0.4%)
3.7%
(3% - 4.4%)
12/35
9%
(7.4% -10.6%)
7.9%
(6.5% - 9.4%)
7.1%
(5.8% - 8.4%)
7.9%
(6.5% - 9.3%)
6.5%
(5.3% - 7.6%)
6.7%
(5.5% - 8%)
7%
(5.7% - 8.3%)
5.3%
(4.3% - 6.3%)
7.5%
(6.1% -8.9%)
8%
(6.5% - 9.4%)
6%
(4.9% -7.1%)
7.6%
(6.2% - 9%)
2.9%
(2.4% - 3.4%)
7.5%
(6.1% -8.9%)
3.7%
(3% - 4.4%)
13/30
10.4%
(8.5% -12.3%)
8.9%
(7.3% -10.5%)
8.4%
(6.9% - 9.9%)
9%
(7.3% -10.6%)
6.7%
(5.4% - 7.9%)
4.6%
(3.7% - 5.5%)
8.2%
(6.7% - 9.7%)
4%
(3.3% - 4.8%)
6.8%
(5.6% -8.1%)
7.9%
(6.4% - 9.3%)
5.8%
(4.7% - 6.8%)
6.9%
(5.6% -8.1%)
1 .3%
(1.1% -1.6%)
8.6%
(7% -10.1%)
2%
(1 .6% - 2.4%)
12/25
8.8%
(7.2% -10.4%)
6.1%
(5% - 7.2%)
5.9%
(4.8% - 7%)
7.9%
(6.5% - 9.3%)
4.2%
(3.4% - 5%)
2.4%
(2% - 2.9%)
7%
(5.7% - 8.3%)
2%
(1 .6% - 2.3%)
4.3%
(3.5% -5.1%)
5.2%
(4.2% - 6.2%)
3.4%
(2.8% - 4%)
4.4%
(3.5% - 5.2%)
0%
(0% - 0%)
5.8%
(4.7% - 6.9%)
0.3%
(0.2% - 0.3%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-25
-------
Table E-25. Percent Reduction from the Current Standards: Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2005 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-
Term Exposure to PM25 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-12% --12%)
-8%
(.30/0 - -8%)
-40%
(-39% - -41 %)
0%
(0% - 0%)
-35%
(-35% - -36%)
-174%
(-171% --177%)
-9%
(-9% - -9%)
-1 1 6%
(-11 4% --11 8%)
-33%
(-33% - -34%)
-14%
(-14% --14%)
0%
(0% - 0%)
-50%
(-49% - -51 %)
-247%
(-245% - -249%)
-17%
(-16% --17%)
-51 %
(-51 % - -52%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -10%)
8%
(8% - 8%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
21%
(20% - 21 %)
18%
(18% -19%)
23%
(22% - 23%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
4%
(4% - 4%)
10%
(10% -10%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(19% -20%)
0%
(0% - 0%)
12/35
31%
(31 % - 31 %)
29%
(29% - 29%)
34%
(34% - 34%)
11%
(1 1 % - 1 1 %)
25%
(25% - 25%)
0%
(0% - 0%)
34%
(34% - 35%)
12%
(12% -12%)
18%
(18% -18%)
23%
(22% - 23%)
10%
(10% -10%)
17%
(17% -17%)
0%
(0% - 0%)
30%
(30% -31%)
0%
(0% - 0%)
13/30
21%
(20% - 21 %)
21%
(21% -22%)
23%
(22% - 23%)
0%
(0% - 0%)
23%
(23% - 23%)
33%
(33% - 34%)
23%
(23% - 23%)
31%
(31 % - 31 %)
25%
(25% - 25%)
23%
(23% - 24%)
13%
(13% -13%)
25%
(24% - 25%)
64%
(64% - 64%)
22%
(21% -22%)
33%
(33% - 33%)
12/25
33%
(32% - 33%)
43%
(43% - 44%)
45%
(45% - 45%)
11%
(11% -11%)
47%
(46% - 47%)
68%
(67% - 68%)
34%
(34% - 35%)
62%
(62% - 62%)
50%
(50% - 50%)
48%
(47% - 48%)
45%
(45% - 45%)
50%
(50% - 50%)
1 00%
(100% -100%)
44%
(44% - 44%)
67%
(67% - 67%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-26
-------
Table E-26. Percent Reduction from the Current Standards: Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2006 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-
Term Exposure to PM25 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-12% --12%)
-9%
(-9% - -9%)
-42%
(-41 % - -42%)
0%
(0% - 0%)
-43%
(-42% - -43%)
-170%
(-167% --173%)
-9%
(-9% - -9%)
-131%
(-129% --133%)
-39%
(-39% - -40%)
-14%
(-14% --15%)
0%
(0% - 0%)
-58%
(-58% - -59%)
-427%
(-425% - -430%)
-19%
(-1 9% - -20%)
-74%
(-74% - -75%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -10%)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
21%
(20% - 21 %)
20%
(20% - 21 %)
23%
(23% - 24%)
0%
(0% - 0%)
16%
(15% -16%)
0%
(0% - 0%)
23%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
23%
(22% - 23%)
0%
(0% - 0%)
12/35
31%
(31 % - 31 %)
32%
(32% - 32%)
35%
(35% - 36%)
13%
(12% -13%)
30%
(30% - 30%)
0%
(0% - 0%)
35%
(35% - 36%)
13%
(13% -13%)
21%
(21% -21%)
24%
(24% - 24%)
10%
(10% -10%)
20%
(20% - 20%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
13/30
21%
(20% - 21 %)
24%
(23% - 24%)
23%
(23% - 24%)
0%
(0% - 0%)
28%
(27% - 28%)
33%
(32% - 33%)
23%
(23% - 24%)
35%
(34% - 35%)
29%
(29% - 29%)
25%
(24% - 25%)
12%
(12% -13%)
28%
(28% - 29%)
100%
(100% -100%)
25%
(25% - 25%)
47%
(47% - 48%)
12/25
33%
(32% - 33%)
48%
(47% - 48%)
47%
(46% - 47%)
13%
(12% -13%)
56%
(56% - 56%)
66%
(66% - 66%)
35%
(35% - 36%)
70%
(70% - 70%)
59%
(59% - 59%)
50%
(50% - 50%)
44%
(44% - 45%)
58%
(58% - 58%)
1 00%
(100% -100%)
51%
(50% -51%)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-27
-------
Table E-27. Percent Reduction from the Current Standards: Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2007 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-
Term Exposure to PM25 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-12% --13%)
-9%
(-9% - -9%)
-41 %
(-40% - -42%)
0%
(0% - 0%)
-41 %
(-41 % - -42%)
-164%
(-161% --167%)
-9%
(-9% - -9%)
-127%
(-125% --129%)
-36%
(-35% - -36%)
-15%
(-14% --15%)
0%
(0% - 0%)
-53%
(-52% - -53%)
-210%
(-209% --21 2%)
-19%
(-18% --19%)
-72%
(-71 % - -72%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -11%)
9%
(9% - 9%)
11%
(11% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(11% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
21%
(21 % - 21 %)
20%
(20% - 21 %)
23%
(23% - 23%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
22%
(21 % - 22%)
0%
(0% - 0%)
12/35
32%
(32% - 32%)
32%
(32% - 32%)
35%
(34% - 35%)
12%
(12% -12%)
29%
(29% - 29%)
0%
(0% - 0%)
35%
(35% - 35%)
13%
(13% -13%)
19%
(19% -19%)
24%
(24% - 24%)
11%
(1 1 % - 1 1 %)
18%
(18% -19%)
0%
(0% - 0%)
33%
(33% - 34%)
0%
(0% - 0%)
13/30
21%
(21 % - 21 %)
24%
(23% - 24%)
23%
(23% - 23%)
0%
(0% - 0%)
27%
(26% - 27%)
31%
(31 % - 32%)
23%
(23% - 23%)
34%
(33% - 34%)
26%
(26% - 27%)
25%
(25% - 25%)
14%
(14% -14%)
26%
(26% - 26%)
55%
(55% - 55%)
24%
(24% - 24%)
46%
(46% - 46%)
12/25
34%
(33% - 34%)
48%
(47% - 48%)
46%
(46% - 46%)
12%
(12% -12%)
54%
(54% - 54%)
64%
(64% - 64%)
35%
(35% - 35%)
68%
(68% - 68%)
54%
(53% - 54%)
50%
(50% - 51 %)
50%
(49% - 50%)
53%
(53% - 53%)
1 00%
(100% -100%)
48%
(48% - 49%)
93%
(93% - 93%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-28
-------
Table E-28. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2 s
Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2005 PM2 s Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM
and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m
Denoted n/m)2:
Recent PM25
Concentrations
312
(257 - 364)
497
(409 - 581 )
233
(192-273)
292
(239 - 344)
862
(711 -1007)
234
(193-273)
467
(383 - 548)
2664
(2192-3117)
3285
(2700 - 3849)
419
(344 - 492)
445
(363 - 526)
547
(450 - 639)
51
(42 - 60)
796
(656 - 930)
117
(96 - 1 38)
15/353
279
(229 - 327)
460
(378 - 538)
168
(137-197)
292
(239 - 344)
642
(526 - 754)
87
(71 -103)
429
(351 - 505)
1249
(1017-1477)
2475
(2024 - 291 4)
369
(303 - 434)
445
(363 - 526)
368
(301 - 433)
15
(12-18)
684
(562 - 802)
78
(63 - 92)
14/35
251
(206 - 295)
423
(348 - 497)
149
(122-176)
292
(239 - 344)
635
(520 - 746)
87
(71 -103)
382
(312-449)
1249
(1017-1477)
2475
(2024-2914)
369
(303 - 434)
445
(363 - 526)
368
(301 - 433)
15
(12-18)
624
(512-733)
78
(63 - 92)
13/35
222
(182-262)
376
(308 - 442)
130
(106-154)
292
(239 - 344)
561
(459 - 660)
87
(71 -103)
333
(272 - 393)
1249
(1017-1477)
2369
(1936-2790)
332
(271 -391)
445
(363 - 526)
346
(283 - 408)
15
(12-18)
553
(453 - 650)
78
(63 - 92)
12/35
193
(158-228)
328
(268 - 386)
111
(90-131)
261
(21 3 - 307)
485
(396 - 572)
87
(71 -103)
284
(231 - 335)
1103
(897 - 1 306)
2040
(1665-2408)
287
(234 - 338)
401
(327 - 475)
305
(249-361)
15
(12-18)
480
(392 - 566)
78
(63 - 92)
25 Concentrations in a Recent Year
Standards (Standard Combination
13/30
222
(182-262)
363
(298 - 427)
130
(106-154)
292
(239 - 344)
497
(406 - 586)
58
(47 - 69)
333
(272 - 393)
869
(705 - 1 030)
1871
(1525-2210)
284
(232 - 335)
389
(317-461)
278
(227 - 329)
5
(4-6)
539
(441 - 635)
52
(42 - 62)
12/25
189
(1 54 - 223)
263
(214-310)
93
(75-110)
261
(213-307)
346
(282-410)
28
(23 - 34)
284
(231 - 335)
477
(386 - 567)
1243
(1010-1474)
195
(159-231)
247
(200 - 293)
185
(151 -220)
0
(0-0)
388
(316-458)
26
(21 -31)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-29
-------
Table E-29. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2 s Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 s from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year
and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
321
(264 - 375)
409
(335 - 480)
221
(181 -258)
222
(181 -262)
638
(523 - 749)
243
(201 - 284)
453
(371 - 533)
2370
(1 945 - 2779)
2588
(2118-3046)
381
(31 3 - 448)
471
(384 - 557)
439
(360 - 51 6)
42
(34 - 50)
610
(500 - 71 6)
80
(65 - 95)
15/353
287
(236 - 336)
375
(307 - 441 )
157
(128-184)
222
(181 -262)
449
(367 - 530)
92
(75-108)
416
(340 - 490)
1038
(843-1229)
1865
(1520-2203)
334
(273 - 393)
471
(384 - 557)
279
(228 - 330)
8
(6-10)
512
(419-603)
46
(37 - 55)
14/35
258
(212-303)
342
(280 - 403)
139
(113-164)
222
(181 -262)
443
(361 - 523)
92
(75-108)
368
(301 - 434)
1038
(843-1229)
1865
(1520-2203)
334
(273 - 393)
471
(384 - 557)
279
(228 - 330)
8
(6-10)
460
(375 - 542)
46
(37 - 55)
13/35
229
(187-269)
300
(245 - 353)
120
(98-142)
222
(181 -262)
380
(310-450)
92
(75-108)
320
(261 - 378)
1038
(843-1229)
1770
(1442-2092)
298
(244 - 352)
471
(384 - 557)
260
(212-308)
8
(6-10)
398
(324 - 470)
46
(37 - 55)
12/35
199
(163-235)
256
(209 - 303)
102
(83-120)
195
(158-230)
316
(257 - 375)
92
(75-108)
270
(220 - 320)
901
(731 -1068)
1478
(1202-1750)
255
(208 - 302)
426
(347 - 503)
225
(183-266)
8
(6-10)
335
(272 - 396)
46
(37 - 55)
13/30
229
(187-269)
288
(235 - 340)
120
(98-142)
222
(181 -262)
327
(266 - 387)
62
(50 - 74)
320
(261 - 378)
682
(553 - 809)
1328
(1079-1573)
253
(206 - 298)
413
(336 - 488)
200
(163-237)
0
(0-0)
386
(31 4 - 456)
24
(20 - 29)
12/25
194
(159-229)
198
(161 -234)
84
(68-100)
195
(158-230)
200
(162-237)
32
(26 - 38)
270
(220 - 320)
316
(255 - 376)
774
(627 - 920)
168
(137-199)
264
(214-313)
119
(96-141)
0
(0-0)
255
(207 - 302)
2
(2-2)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-30
-------
Table E-30. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 s Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 s from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year
and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
310
(255 - 363)
408
(335 - 479)
231
(190-270)
247
(202-291)
670
(549 - 786)
255
(21 1 - 298)
473
(387 - 556)
2456
(201 7 - 2879)
3003
(2462 - 3525)
378
(309 - 444)
402
(327 - 476)
490
(403 - 574)
59
(48 - 70)
665
(546 - 780)
84
(68 - 99)
15/353
277
(227 - 324)
374
(307 - 440)
165
(135-194)
247
(202 - 291 )
478
(390 - 563)
98
(80 - 1 1 6)
434
(355 -511)
1094
(890-1296)
2222
(1814-2620)
330
(270 - 389)
402
(327 - 476)
324
(264 - 382)
19
(16-23)
563
(461 - 662)
49
(40 - 58)
14/35
248
(203 - 291 )
342
(280 - 402)
146
(120-173)
247
(202 - 291 )
471
(385 - 556)
98
(80 - 1 1 6)
385
(314-453)
1094
(890-1296)
2222
(1814-2620)
330
(270 - 389)
402
(327 - 476)
324
(264 - 382)
19
(16-23)
508
(415-599)
49
(40 - 58)
13/35
219
(179-258)
299
(244 - 353)
128
(104-151)
247
(202 - 291 )
407
(332 - 481 )
98
(80 - 1 1 6)
335
(273 - 395)
1094
(890-1296)
2120
(1730-2501)
295
(241 - 347)
402
(327 - 476)
303
(248 - 358)
19
(16-23)
443
(362 - 523)
49
(40 - 58)
12/35
189
(154-223)
256
(209 - 302)
108
(88-128)
218
(178-257)
341
(278 - 404)
98
(80 - 1 1 6)
284
(231 - 335)
954
(775 -1131)
1804
(1469-2132)
252
(205 - 298)
359
(291 - 425)
265
(216-313)
19
(16-23)
377
(307 - 446)
49
(40 - 58)
13/30
219
(179-258)
288
(235 - 339)
128
(104-151)
247
(202-291)
352
(286 - 41 6)
68
(55 - 80)
335
(273 - 395)
730
(592 - 866)
1641
(1336-1941)
249
(203 - 295)
347
(282 - 41 0)
240
(195-284)
9
(7-10)
431
(351 - 509)
27
(21 - 32)
12/25
185
(151 -218)
197
(160-234)
90
(73-107)
218
(178-257)
222
(180-264)
36
(29 - 43)
284
(231 - 335)
355
(287 - 423)
1040
(843-1234)
165
(134-196)
204
(165-242)
153
(124-181)
0
(0-0)
294
(239 - 348)
4
(3-4)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-31
-------
Table E-31. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2005) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5from 1999 - 2001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Ischemic Heart Disease Mortality Associated with Long-Term E
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n
Combination Denoted n/m)2:
Recent PM25
Concentrations
19.9%
(16.4% -23.2%)
19.5%
(16.1% -22.8%)
19.9%
(16.4% -23.3%)
14%
(11. 5% -16.5%)
20.6%
(17% -24%)
21%
(17.3% -24.5%)
15.4%
(12.6% -18.1%)
19%
(15.7% -22.3%)
17.7%
(14.5% -20.7%)
16.8%
(13.8% -19.7%)
10.1%
(8.2% - 1 1 .9%)
19.7%
(16. 2% -23%)
10.4%
(8.5% -12.3%)
20.2%
(16.6% -23.6%)
1 1 .6%
(9.5% -13.7%)
15/353
17.8%
(14.6% -20.8%)
18.1%
(14.8% -21 .2%)
14.3%
(11. 7% -16.8%)
14%
(11. 5% -16.5%)
15.3%
(12. 6% -18%)
7.8%
(6.3% - 9.2%)
14.2%
(11. 6% -16.6%)
8.9%
(7. 3% -10.6%)
13.3%
(10.9% -15.7%)
14.8%
(12.1% -17.4%)
10.1%
(8.2% - 1 1 .9%)
13.2%
(10.8% -15.6%)
3%
(2.4% - 3.6%)
17.4%
(14.3% -20.4%)
7.7%
(6.3% -9.1%)
14/35
16%
(13.1% -18. 8%)
16.6%
(13. 7% -19. 5%)
12.7%
(10. 4% -15%)
14%
(11. 5% -16. 5%)
15.1%
(12. 4% -17. 8%)
7.8%
(6.3% - 9.2%)
12.6%
(10. 3% -14.8%)
8.9%
(7. 3% -10. 6%)
13.3%
(10. 9% -15. 7%)
14.8%
(12.1% -17.4%)
10.1%
(8.2% - 1 1 .9%)
13.2%
(10. 8% -15.6%)
3%
(2.4% - 3.6%)
15.8%
(13% -18.6%)
7.7%
(6. 3% -9.1%)
13/35
14.2%
(11. 6% -16. 7%)
14.8%
(12.1% -17. 4%)
11.1%
(9.1% -13.1%)
14%
(11. 5% -16. 5%)
13.4%
(10.9% -15. 7%)
7.8%
(6.3% - 9.2%)
11%
(9% -13%)
8.9%
(7.3% -10. 6%)
12.8%
(10.4% -15%)
13.3%
(10. 8% -15.6%)
10.1%
(8.2% - 1 1 .9%)
12.5%
(10. 2% -14.7%)
3%
(2.4% - 3.6%)
14%
(11. 5% -16. 5%)
7.7%
(6.3% -9.1%)
12/35
12.3%
(10.1% -14. 5%)
12.9%
(10. 5% -15. 2%)
9.5%
(7.7% - 1 1 .2%)
12.5%
(10. 2% -14. 7%)
1 1 .6%
(9. 4% -13. 6%)
7.8%
(6.3% - 9.2%)
9.4%
(7.6% -11.1%)
7.9%
(6.4% - 9.3%)
11%
(9% - 1 3%)
1 1 .5%
(9. 4% -13. 5%)
9.1%
(7.4% -10. 8%)
11%
(9% -13%)
3%
(2.4% - 3.6%)
12.2%
(10% -14.4%)
7.7%
(6. 3% -9.1%)
xposure to PM25 Concentrations in a
and Daily (m) Standards (Standard
13/30
14.2%
(11. 6% -16.7%)
14.3%
(11. 7% -16.8%)
11.1%
(9.1% -13.1%)
14%
(11. 5% -16. 5%)
1 1 .9%
(9. 7% -14%)
5.2%
(4.2% - 6.2%)
11%
(9% -13%)
6.2%
(5% - 7.4%)
10.1%
(8.2% - 1 1 .9%)
1 1 .4%
(9. 3% -13. 4%)
8.8%
(7.2% -10.5%)
10%
(8.2% - 1 1 .8%)
1.1%
(0.9% - 1 .3%)
13.7%
(11. 2% -16.1%)
5.2%
(4.2% -6.1%)
12/25
12%
(9. 8% -14. 2%)
10.3%
(8. 4% -12. 2%)
7.9%
(6.4% - 9.4%)
12.5%
(10.2% -14. 7%)
8.3%
(6.7% - 9.8%)
2.5%
(2.1% -3%)
9.4%
(7. 6% -11.1%)
3.4%
(2.8% -4.1%)
6.7%
(5.4% - 7.9%)
7.8%
(6.3% - 9.2%)
5.6%
(4.5% - 6.6%)
6.7%
(5.4% - 7.9%)
0%
(0% - 0%)
9.8%
(8% - 1 1 .6%)
2.6%
(2.1% -3.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-32
-------
Table E-32. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2006) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5from 1999 - 2001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Ischemic Heart Disease Mortality Associated with Long-Term E
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n
Combination Denoted n/m)2:
Recent PM25
Concentrations
19.8%
(16.3% -23.2%)
16%
(13.2% -18.8%)
18.6%
(15.3% -21 .8%)
10.4%
(8.5% -12. 3%)
15.2%
(12.5% -17.9%)
21 .5%
(17.7% -25.1%)
14.5%
(11. 8% -17%)
16.8%
(13.8% -19.7%)
13.8%
(11. 3% -16.2%)
15.3%
(12. 5% -18%)
10.3%
(8.4% -12.2%)
15.9%
(13% -18.7%)
8.3%
(6.7% - 9.8%)
15.4%
(12.6% -18.1%)
7.8%
(6.3% - 9.2%)
15/353
17.7%
(14.6% -20.8%)
14.7%
(12.1% -17.3%)
13.2%
(10.8% -15.6%)
10.4%
(8. 5% -12. 3%)
10.7%
(8. 8% -12.7%)
8.1%
(6.6% - 9.6%)
13.3%
(10.8% -15.6%)
7.4%
(6% - 8.7%)
9.9%
(8.1% -11. 7%)
13.4%
(10.9% -15.8%)
10.3%
(8. 4% -12. 2%)
10.1%
(8.2% - 1 1 .9%)
1 .6%
(1 .3% - 1 .9%)
12.9%
(10.6% -15.2%)
4.5%
(3.6% - 5.3%)
14/35
16%
(13.1% -18. 7%)
13.4%
(11% -15. 8%)
1 1 .7%
(9. 6% -13. 8%)
10.4%
(8.5% -12.3%)
10.6%
(8. 6% -12. 5%)
8.1%
(6.6% - 9.6%)
1 1 .7%
(9.6% -13.8%)
7.4%
(6% - 8.7%)
9.9%
(8.1% -11. 7%)
13.4%
(10. 9% -15. 8%)
10.3%
(8.4% -12.2%)
10.1%
(8.2% - 1 1 .9%)
1 .6%
(1 .3% - 1 .9%)
1 1 .6%
(9.5% -13.7%)
4.5%
(3.6% - 5.3%)
13/35
14.2%
(11. 6% -16. 6%)
1 1 .8%
(9.6% -13.9%)
10.2%
(8. 3% -12%)
10.4%
(8. 5% -12. 3%)
9.1%
(7.4% -10. 7%)
8.1%
(6.6% - 9.6%)
10.2%
(8.3% -12%)
7.4%
(6% - 8.7%)
9.4%
(7.7% - 1 1 .2%)
1 1 .9%
(9.8% -14.1%)
10.3%
(8.4% -12.2%)
9.4%
(7. 7% -11.1%)
1 .6%
(1 .3% - 1 .9%)
10%
(8.2% - 1 1 .9%)
4.5%
(3.6% - 5.3%)
12/35
12.3%
(10% -14.5%)
10.1%
(8.2% - 1 1 .9%)
8.6%
(7% -10.2%)
9.1%
(7.4% -10.8%)
7.6%
(6.1% -8.9%)
8.1%
(6.6% - 9.6%)
8.6%
(7% -10.2%)
6.4%
(5.2% - 7.6%)
7.9%
(6.4% - 9.3%)
10.2%
(8. 3% -12.1%)
9.3%
(7.6% - 1 1 %)
8.1%
(6.6% - 9.6%)
1 .6%
(1 .3% - 1 .9%)
8.5%
(6. 9% -10%)
4.5%
(3.6% - 5.3%)
xposure to PM25 Concentrations in a
and Daily (m) Standards (Standard
13/30
14.2%
(11. 6% -16.6%)
1 1 .3%
(9. 2% -13. 3%)
10.2%
(8. 3% -12%)
10.4%
(8.5% -12.3%)
7.8%
(6.3% - 9.2%)
5.5%
(4.4% - 6.5%)
10.2%
(8.3% -12%)
4.8%
(3.9% - 5.8%)
7.1%
(5.8% - 8.4%)
10.1%
(8. 3% -12%)
9%
(7. 3% -10.7%)
7.3%
(5.9% - 8.6%)
0%
(0% - 0%)
9.7%
(7.9% - 1 1 .5%)
2.4%
(1 .9% - 2.8%)
12/25
12%
(9. 8% -14. 2%)
7.8%
(6.3% - 9.2%)
7.1%
(5.8% - 8.4%)
9.1%
(7. 4% -10. 8%)
4.8%
(3.9% - 5.7%)
2.8%
(2.3% - 3.3%)
8.6%
(7% -10.2%)
2.2%
(1 .8% - 2.7%)
4.1%
(3.3% - 4.9%)
6.7%
(5.5% - 8%)
5.8%
(4.7% - 6.8%)
4.3%
(3.5% -5.1%)
0%
(0% - 0%)
6.4%
(5.2% - 7.6%)
0.2%
(0.2% - 0.2%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-33
-------
Table E-33. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2007) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Ischemic Heart Disease Mortality Associated with Long-Term E
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n
Combination Denoted n/m)2:
Recent PM25
Concentrations
18.7%
(15.4% -21 .8%)
16.1%
(13.2% -18.8%)
19.3%
(15.9% -22.6%)
1 1 .4%
(9.3% -13. 4%)
16.1%
(13.2% -18.9%)
22.2%
(18.3% -25.9%)
14.8%
(12.1% -17.4%)
17.3%
(14.2% -20.3%)
15.9%
(13% -18. 7%)
15.1%
(12.4% -17.8%)
8.5%
(6.9% -10.1%)
17.8%
(14.7% -20.9%)
1 1 .4%
(9.3% -13.4%)
16.8%
(13.8% -19. 7%)
8%
(6.5% - 9.5%)
15/353
16.7%
(13.7% -19.5%)
14.7%
(12.1% -17.3%)
13.8%
(11. 3% -16.2%)
1 1 .4%
(9. 3% -13. 4%)
1 1 .5%
(9.4% -13.5%)
8.5%
(7% -10.1%)
13.6%
(11.1% -16%)
7.7%
(6.3% -9.1%)
1 1 .8%
(9. 6% -13. 9%)
13.2%
(10.8% -15.6%)
8.5%
(6. 9% -10.1%)
1 1 .8%
(9. 6% -13. 9%)
3.7%
(3% - 4.4%)
14.2%
(11. 6% -16. 7%)
4.7%
(3.8% - 5.6%)
14/35
14.9%
(12. 2% -17.6%)
13.4%
(11% -15. 8%)
12.2%
(10% -14. 4%)
1 1 .4%
(9.3% -13.4%)
1 1 .3%
(9. 3% -13.4%)
8.5%
(7% -10.1%)
12%
(9.8% -14.2%)
7.7%
(6. 3% -9.1%)
1 1 .8%
(9.6% -13.9%)
13.2%
(10. 8% -15. 6%)
8.5%
(6.9% -10.1%)
1 1 .8%
(9.6% -13.9%)
3.7%
(3% - 4.4%)
12.8%
(10. 5% -15.1%)
4.7%
(3.8% - 5.6%)
13/35
13.2%
(10. 8% -15.5%)
1 1 .8%
(9.6% -13.9%)
10.7%
(8.7% -12.6%)
1 1 .4%
(9. 3% -13. 4%)
9.8%
(8% - 1 1 .6%)
8.5%
(7% -10.1%)
10.5%
(8.5% -12.3%)
7.7%
(6.3% -9.1%)
1 1 .2%
(9.2% -13.2%)
1 1 .8%
(9. 6% -13. 9%)
8.5%
(6.9% -10.1%)
11%
(9% - 1 3%)
3.7%
(3% - 4.4%)
1 1 .2%
(9.1% -13.2%)
4.7%
(3.8% - 5.6%)
12/35
1 1 .4%
(9.3% -13.4%)
10.1%
(8.2% - 1 1 .9%)
9.1%
(7. 4% -10. 7%)
10%
(8.2% - 1 1 .9%)
8.2%
(6.7% - 9.7%)
8.5%
(7% -10.1%)
8.9%
(7. 2% -10. 5%)
6.7%
(5.5% - 8%)
9.6%
(7.8% - 1 1 .3%)
10.1%
(8.2% - 1 1 .9%)
7.6%
(6.2% - 9%)
9.6%
(7.8% - 1 1 .4%)
3.7%
(3% - 4.4%)
9.5%
(7.7% - 1 1 .2%)
4.7%
(3.8% - 5.6%)
xposure to PM25 Concentrations in a
and Daily (m) Standards (Standard
13/30
13.2%
(10.8% -15.5%)
1 1 .3%
(9. 2% -13. 3%)
10.7%
(8. 7% -12. 6%)
1 1 .4%
(9.3% -13.4%)
8.5%
(6. 9% -10%)
5.9%
(4.8% - 7%)
10.5%
(8.5% -12.3%)
5.2%
(4.2% -6.1%)
8.7%
(7.1% -10. 3%)
10%
(8.1% -11. 8%)
7.3%
(6% - 8.7%)
8.7%
(7.1% -10. 3%)
1 .7%
(1 .4% - 2%)
10.9%
(8. 9% -12. 8%)
2.5%
(2.1% -3%)
12/25
11.1%
(9.1% -13.1%)
7.8%
(6.3% - 9.2%)
7.5%
(6.1% -8.9%)
10%
(8.2% - 1 1 .9%)
5.3%
(4.3% - 6.3%)
3.1%
(2.5% - 3.7%)
8.9%
(7. 2% -10. 5%)
2.5%
(2% - 3%)
5.5%
(4.5% - 6.5%)
6.6%
(5.4% - 7.8%)
4.3%
(3.5% -5.1%)
5.6%
(4.5% - 6.6%)
0%
(0% - 0%)
7.4%
(6% - 8.8%)
0.3%
(0.3% - 0.4%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-34
-------
Table E-34. Percent Reduction from the Current Standards: Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated
with Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2005 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-
Term Exposure to PM2.5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-11% --12%)
-8%
(.30/0 - -8%)
-39%
(-38% - -40%)
0%
(0% - 0%)
-34%
(-34% - -35%)
-170%
(-166% --174%)
-9%
(-9% - -9%)
-1 1 3%
(-1 1 1 % - -1 1 6%)
-33%
(-32% - -33%)
-13%
(-13% --14%)
0%
(0% - 0%)
-49%
(-48% - -50%)
-244%
(-242% - -247%)
-16%
(-16% --17%)
-51 %
(-50% - -51 %)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -10%)
8%
(8% - 8%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
20%
(20% - 21 %)
18%
(18% -18%)
22%
(22% - 23%)
0%
(0% - 0%)
13%
(12% -13%)
0%
(0% - 0%)
22%
(22% - 23%)
0%
(0% - 0%)
4%
(4% - 4%)
10%
(10% -10%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
19%
(19% -19%)
0%
(0% - 0%)
12/35
31%
(30% - 31 %)
29%
(28% - 29%)
34%
(33% - 34%)
11%
(1 1 % - 1 1 %)
24%
(24% - 25%)
0%
(0% - 0%)
34%
(34% - 34%)
12%
(12% -12%)
18%
(17% -18%)
22%
(22% - 23%)
10%
(10% -10%)
17%
(17% -17%)
0%
(0% - 0%)
30%
(29% - 30%)
0%
(0% - 0%)
13/30
20%
(20% -21%)
21%
(21% -21%)
22%
(22% - 23%)
0%
(0% - 0%)
23%
(22% - 23%)
33%
(33% - 33%)
22%
(22% - 23%)
30%
(30% -31%)
24%
(24% - 25%)
23%
(23% - 23%)
12%
(12% -13%)
24%
(24% - 25%)
64%
(64% - 64%)
21%
(21 % - 22%)
33%
(33% - 33%)
12/25
32%
(32% - 33%)
43%
(42% - 43%)
45%
(44% - 45%)
11%
(1 1 % - 1 1 %)
46%
(46% - 46%)
67%
(67% - 67%)
34%
(34% - 34%)
62%
(62% - 62%)
50%
(49% - 50%)
47%
(47% - 48%)
45%
(44% - 45%)
50%
(49% - 50%)
1 00%
(100% -100%)
43%
(43% - 44%)
67%
(66% - 67%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-35
-------
Table E-35. Percent Reduction from the Current Standards: Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated
with Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2006 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-
Term Exposure to PM25 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-11% --12%)
-9%
(-9% - -9%)
-41 %
(-40% - -42%)
0%
(0% - 0%)
-42%
(-41 % - -43%)
-165%
(-162% --169%)
-9%
(-9% - -9%)
-128%
(-126% --131%)
-39%
(-38% - -39%)
-14%
(-14% --14%)
0%
(0% - 0%)
-57%
(-56% - -58%)
-423%
(-420% - -427%)
-19%
(-19% --19%)
-74%
(-73% - -74%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -10%)
9%
(9% - 9%)
11%
(11% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(11% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
20%
(20% - 21 %)
20%
(20% - 20%)
23%
(23% - 23%)
0%
(0% - 0%)
15%
(15% -16%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
22%
(22% - 23%)
0%
(0% - 0%)
12/35
31%
(30% - 31 %)
32%
(31 % - 32%)
35%
(35% - 35%)
12%
(12% -13%)
30%
(29% - 30%)
0%
(0% - 0%)
35%
(35% - 35%)
13%
(13% -13%)
21%
(21 % - 21 %)
24%
(23% - 24%)
10%
(10% -10%)
20%
(19% -20%)
0%
(0% - 0%)
35%
(34% - 35%)
0%
(0% - 0%)
13/30
20%
(20% -21%)
23%
(23% - 23%)
23%
(23% - 23%)
0%
(0% - 0%)
27%
(27% - 28%)
32%
(32% - 33%)
23%
(23% - 23%)
34%
(34% - 34%)
29%
(29% - 29%)
24%
(24% - 25%)
12%
(12% -12%)
28%
(28% - 28%)
100%
(100% -100%)
25%
(24% - 25%)
47%
(47% - 47%)
12/25
32%
(32% - 33%)
47%
(47% - 48%)
46%
(46% - 47%)
12%
(12% -13%)
56%
(55% - 56%)
66%
(65% - 66%)
35%
(35% - 35%)
70%
(69% - 70%)
58%
(58% - 59%)
50%
(49% - 50%)
44%
(44% - 44%)
57%
(57% - 58%)
1 00%
(100% -100%)
50%
(50% -51%)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-36
-------
Table E-36. Percent Reduction from the Current Standards: Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated
with Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2007 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-
Term Exposure to PM2.5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-12% --12%)
-9%
(-9% - -9%)
-40%
(-39% - -41 %)
0%
(0% - 0%)
-40%
(-40% - -41 %)
-159%
(-156% --163%)
-9%
(-9% - -9%)
-124%
(-122% --127%)
-35%
(-35% - -36%)
-14%
(-14% --15%)
0%
(0% - 0%)
-51 %
(-51 % - -52%)
-208%
(-205% --210%)
-18%
(-18% --18%)
-71 %
(-71 % - -72%)
15/352
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -10%)
9%
(9% - 9%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
21%
(21 % - 21 %)
20%
(20% - 20%)
23%
(22% - 23%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
21%
(21 % - 22%)
0%
(0% - 0%)
12/35
32%
(31% -32%)
32%
(31 % - 32%)
34%
(34% - 35%)
12%
(12% -12%)
28%
(28% - 29%)
0%
(0% - 0%)
35%
(34% - 35%)
13%
(13% -13%)
19%
(19% -19%)
24%
(23% - 24%)
11%
(1 1 % - 1 1 %)
18%
(18% -18%)
0%
(0% - 0%)
33%
(33% - 33%)
0%
(0% - 0%)
13/30
21%
(21% -21%)
23%
(23% - 23%)
23%
(22% - 23%)
0%
(0% - 0%)
26%
(26% - 27%)
31%
(31 % - 31 %)
23%
(23% - 23%)
33%
(33% - 33%)
26%
(26% - 26%)
25%
(24% - 25%)
14%
(14% -14%)
26%
(26% - 26%)
55%
(55% - 55%)
23%
(23% - 24%)
46%
(46% - 46%)
12/25
33%
(33% - 34%)
47%
(47% - 48%)
45%
(45% - 46%)
12%
(12% -12%)
53%
(53% - 54%)
63%
(63% - 64%)
35%
(34% - 35%)
68%
(67% - 68%)
53%
(53% - 54%)
50%
(50% - 50%)
49%
(49% - 50%)
53%
(52% - 53%)
1 00%
(100% -100%)
48%
(47% - 48%)
93%
(93% - 93%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-37
-------
Table E-37. Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent Year and
PM2.s Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2.5
Concentrations
512
(391 - 630)
507
(387 - 624)
365
(279 - 450)
321
(244 - 396)
748
(572 - 920)
240
(184-295)
499
(380 - 61 6)
2357
(1800-2902)
2205
(1683-2717)
439
(335-541)
406
(309 - 503)
529
(405 - 652)
76
(58 - 94)
758
(580 - 933)
110
(84 - 1 36)
15/353
455
(347 - 561 )
467
(356 - 575)
258
(196-318)
321
(244 - 396)
547
(417-675)
85
(65-105)
457
(348 - 564)
1069
(812-1324)
1637
(1246-2022)
384
(293 - 474)
406
(309 - 503)
349
(265 - 431 )
22
(16-27)
646
(493 - 796)
72
(55 - 89)
14/35
407
(310-502)
428
(326 - 528)
228
(174-282)
321
(244 - 396)
540
(412-667)
85
(65 - 1 05)
404
(307 - 499)
1069
(812-1324)
1637
(1246-2022)
384
(293 - 474)
406
(309 - 503)
349
(265 - 431 )
22
(16-27)
586
(447 - 723)
72
(55 - 89)
13/35
359
(273 - 443)
378
(288 - 466)
198
(151 -245)
321
(244 - 396)
474
(361 - 586)
85
(65- 105)
351
(267 - 434)
1069
(812- 1324)
1564
(1190-1933)
343
(261 - 424)
406
(309 - 503)
328
(249 - 405)
22
(16-27)
516
(393 - 637)
72
(55 - 89)
12/35
310
(236 - 383)
327
(249 - 405)
168
(128-208)
285
(217-352)
408
(310-505)
85
(65-105)
297
(226 - 368)
941
(714-1166)
1339
(1018-1657)
295
(224 - 365)
365
(278 - 453)
287
(219-356)
22
(16-27)
445
(339 - 550)
72
(55 - 89)
13/30
359
(273 - 443)
364
(278 - 450)
198
(151 -245)
321
(244 - 396)
419
(318-518)
56
(43 - 70)
351
(267 - 434)
737
(559-915)
1224
(930 - 1 51 5)
292
(222 - 361 )
354
(269 - 439)
261
(198-323)
8
(6-10)
503
(383 - 621 )
48
(36 - 60)
12/25
302
(230 - 374)
260
(198-322)
140
(106-173)
285
(217-352)
288
(219-357)
27
(21 - 34)
297
(226 - 368)
401
(304 - 499)
805
(61 1 - 998)
198
(151 -246)
222
(169-276)
172
(131 -213)
0
(0-0)
357
(271 - 442)
24
(18-29)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-38
-------
Table E-38. Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 s Concentrations
in a Recent Year (2006) and PM2 s Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 sfrom 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year and
PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
527
(403 - 649)
412
(31 4 - 509)
344
(263 - 424)
241
(183-298)
543
(41 4 - 670)
250
(191 -307)
483
(368 - 597)
2081
(1 587 - 2566)
1715
(1306-2118)
398
(303-491)
431
(327 - 533)
420
(320 - 51 8)
62
(47 - 77)
572
(436 - 705)
74
(56 - 92)
15/353
469
(358 - 577)
377
(287 - 465)
240
(183-297)
241
(183-298)
377
(287 - 466)
90
(68 - 1 1 2)
441
(336 - 545)
884
(671 -1095)
1220
(927-1510)
346
(263 - 427)
431
(327 - 533)
262
(199-324)
12
(9-15)
476
(362 - 588)
42
(32 - 53)
14/35
419
(320-517)
342
(261 - 423)
211
(161 -261)
241
(183-298)
372
(283 - 460)
90
(68 - 1 1 2)
389
(296 - 481 )
884
(671 -1095)
1220
(927-1510)
346
(263 - 427)
431
(327 - 533)
262
(199-324)
12
(9-15)
426
(324 - 526)
42
(32 - 53)
13/35
369
(281 - 456)
298
(227 - 369)
183
(139-226)
241
(183-298)
317
(241 - 393)
90
(68 - 1 1 2)
336
(255-416)
884
(671 -1095)
1156
(878-1431)
307
(234 - 380)
431
(327 - 533)
244
(185-302)
12
(9-15)
366
(278 - 453)
42
(32 - 53)
12/35
319
(243 - 394)
253
(193-314)
154
(117-190)
210
(160-260)
263
(199-326)
90
(68 - 1 1 2)
283
(21 5 - 350)
765
(580 - 948)
961
(729 -1191)
262
(199-324)
388
(294 - 480)
209
(159-259)
12
(9-15)
307
(233 - 380)
42
(32 - 53)
13/30
369
(281 - 456)
286
(21 8 - 354)
183
(139-226)
241
(183-298)
272
(206 - 336)
60
(46 - 75)
336
(255 - 41 6)
576
(436 - 71 5)
861
(653-1067)
259
(197-320)
376
(286 - 466)
186
(141 -231)
0
(0-0)
355
(270 - 440)
22
(17-28)
12/25
311
(237 - 385)
194
(147-241)
127
(96-157)
210
(160-260)
165
(125-204)
30
(23 - 38)
283
(21 5 - 350)
265
(200 - 329)
497
(377 - 61 8)
171
(129-211)
238
(180-295)
110
(83-136)
0
(0-0)
232
(176-288)
2
(1-2)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-39
-------
Table E-39. Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2007) and PM2 s Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM25 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2Sfrom 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent Year and
PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2 5
Concentrations
507
(387 - 625)
412
(314-508)
361
(276 - 444)
269
(205 - 333)
572
(436 - 705)
263
(201 - 323)
504
(384 - 622)
2160
(1648-2663)
2003
(1527-2470)
393
(300 - 485)
366
(278 - 453)
472
(360 - 581 )
89
(68-110)
626
(478 - 772)
78
(59 - 97)
15/353
449
(343 - 554)
376
(286 - 464)
253
(193-313)
269
(205 - 333)
402
(305 - 497)
97
(74 - 1 20)
461
(351 - 569)
933
(708-1156)
1462
(1112-1808)
342
(260 - 423)
366
(278 - 453)
305
(232 - 378)
28
(21 - 35)
526
(400 - 649)
45
(34 - 56)
14/35
401
(305 - 495)
342
(260 - 422)
224
(170-276)
269
(205 - 333)
396
(301 - 490)
97
(74 - 1 20)
407
(309 - 503)
933
(708-1156)
1462
(1112-1808)
342
(260 - 423)
366
(278 - 453)
305
(232 - 378)
28
(21 - 35)
472
(359 - 584)
45
(34 - 56)
13/35
352
(268 - 435)
298
(226 - 368)
194
(147-240)
269
(205 - 333)
340
(259 - 421 )
97
(74 - 1 20)
352
(268 - 436)
933
(708-1156)
1392
(1059-1722)
304
(231 - 375)
366
(278 - 453)
286
(217-353)
28
(21 - 35)
410
(312-507)
45
(34 - 56)
12/35
302
(230 - 374)
253
(192-313)
164
(124-203)
236
(179-292)
284
(216-352)
97
(74 - 1 20)
297
(225 - 368)
811
(615-1 006)
1179
(895 - 1 459)
258
(196-320)
325
(247 - 403)
248
(188-307)
28
(21 - 35)
347
(264 - 430)
45
(34 - 56)
13/30
352
(268 - 435)
286
(217-353)
194
(147-240)
269
(205 - 333)
293
(223 - 363)
66
(50 - 82)
352
(268 - 436)
617
(468 - 766)
1069
(812-1 324)
255
(194-316)
314
(238 - 389)
224
(170-277)
13
(10-16)
398
(302 - 492)
24
(18-30)
12/25
295
(224 - 365)
194
(1 47 - 240)
136
(1 03 - 1 68)
236
(1 79 - 292)
183
(1 39 - 227)
35
(26 - 43)
297
(225 - 368)
298
(226 - 370)
671
(508 - 832)
168
(1 27 - 208)
183
(1 39 - 227)
141
(1 07 - 1 75)
0
(0-0)
268
(203 - 332)
3
(2-4)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-40
-------
Table E-40. Estimated Percent of Total Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
8.8%
(6.7% -10.8%)
8.6%
(6.6% -10.6%)
8.8%
(6.7% -10.8%)
6.1%
(4.6% - 7.5%)
9.1%
(7% - 1 1 .2%)
9.3%
(7.1% -11. 4%)
6.7%
(5.1% -8.3%)
8.4%
(6.4% -10.3%)
7.8%
(5.9% - 9.6%)
7.3%
(5.6% - 9%)
4.3%
(3.3% - 5.3%)
8.7%
(6.6% -10.7%)
4.5%
(3.4% - 5.5%)
8.9%
(6.8% - 1 1 %)
5%
(3.8% - 6.2%)
15/353
7.8%
(5.9% - 9.6%)
7.9%
(6% - 9.8%)
6.2%
(4.7% - 7.7%)
6.1%
(4.6% - 7.5%)
6.7%
(5.1% -8.2%)
3.3%
(2.5% -4.1%)
6.1%
(4.7% - 7.6%)
3.8%
(2.9% - 4.7%)
5.8%
(4.4% -7.1%)
6.4%
(4.9% - 7.9%)
4.3%
(3.3% - 5.3%)
5.7%
(4.4% -7.1%)
1 .3%
(1%-1.6%)
7.6%
(5.8% - 9.4%)
3.3%
(2.5% -4.1%)
14/35
7%
(5.3% - 8.6%)
7.3%
(5.5% - 9%)
5.5%
(4.2% - 6.8%)
6.1%
(4.6% - 7.5%)
6.6%
(5% -8.1%)
3.3%
(2.5% -4.1%)
5.4%
(4.1% -6.7%)
3.8%
(2.9% - 4.7%)
5.8%
(4.4% -7.1%)
6.4%
(4.9% - 7.9%)
4.3%
(3.3% - 5.3%)
5.7%
(4.4% -7.1%)
1 .3%
(1%-1.6%)
6.9%
(5.3% - 8.5%)
3.3%
(2.5% -4.1%)
13/35
6.1%
(4.7% - 7.6%)
6.4%
(4.9% - 7.9%)
4.8%
(3.6% - 5.9%)
6.1%
(4.6% - 7.5%)
5.8%
(4.4% -7.1%)
3.3%
(2.5% -4.1%)
4.7%
(3.6% - 5.8%)
3.8%
(2.9% - 4.7%)
5.5%
(4.2% - 6.8%)
5.7%
(4. 4% -7.1%)
4.3%
(3.3% - 5.3%)
5.4%
(4.1% -6.6%)
1 .3%
(1 % - 1 .6%)
6.1%
(4.6% - 7.5%)
3.3%
(2.5% -4.1%)
12/35
5.3%
(4% - 6.6%)
5.6%
(4.2% - 6.9%)
4%
(3.1% -5%)
5.4%
(4.1% -6. 7%)
5%
(3. 8% -6.1%)
3.3%
(2. 5% -4.1%)
4%
(3% - 4.9%)
3.3%
(2.5% -4.1%)
4.7%
(3.6% - 5.8%)
4.9%
(3.7% -6.1%)
3.9%
(2.9% - 4.8%)
4.7%
(3.6% - 5.8%)
1 .3%
(1 % - 1 .6%)
5.2%
(4% - 6.5%)
3.3%
(2. 5% -4.1%)
13/30
6.1%
(4.7% - 7.6%)
6.2%
(4.7% - 7.6%)
4.8%
(3.6% - 5.9%)
6.1%
(4.6% - 7.5%)
5.1%
(3.9% - 6.3%)
2.2%
(1 .7% - 2.7%)
4.7%
(3.6% - 5.8%)
2.6%
(2% - 3.3%)
4.3%
(3.3% - 5.3%)
4.9%
(3.7% - 6%)
3.8%
(2.9% - 4.7%)
4.3%
(3.3% - 5.3%)
0.5%
(0.3% - 0.6%)
5.9%
(4.5% - 7.3%)
2.2%
(1 .7% - 2.7%)
12/25
5.2%
(3.9% - 6.4%)
4.4%
(3.4% - 5.5%)
3.4%
(2.6% - 4.2%)
5.4%
(4.1% -6.7%)
3.5%
(2.7% - 4.3%)
1.1%
(0.8% - 1 .3%)
4%
(3% - 4.9%)
1 .4%
(1.1% -1.8%)
2.8%
(2.1% -3.5%)
3.3%
(2.5% -4.1%)
2.4%
(1 .8% - 2.9%)
2.8%
(2.1% -3.5%)
0%
(0% - 0%)
4.2%
(3.2% - 5.2%)
1.1%
(0.8% - 1 .3%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-41
-------
Table E-41. Estimated Percent of Total Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
8.8%
(6.7% -10.8%)
7%
(5.3% - 8.6%)
8.2%
(6.3% -10.1%)
4.5%
(3.4% - 5.5%)
6.6%
(5% - 8.2%)
9.5%
(7.3% - 1 1 .7%)
6.3%
(4.8% - 7.7%)
7.4%
(5.6% -9.1%)
6%
(4.5% - 7.4%)
6.6%
(5.1% -8.2%)
4.4%
(3.3% - 5.4%)
6.9%
(5.3% - 8.5%)
3.5%
(2.7% - 4.4%)
6.7%
(5.1% -8.3%)
3.3%
(2.5% -4.1%)
15/353
7.8%
(5.9% - 9.6%)
6.4%
(4.9% - 7.9%)
5.7%
(4.3% -7.1%)
4.5%
(3.4% - 5.5%)
4.6%
(3.5% - 5.7%)
3.4%
(2.6% - 4.3%)
5.7%
(4.4% -7.1%)
3.1%
(2.4% - 3.9%)
4.2%
(3.2% - 5.3%)
5.8%
(4.4% -7.1%)
4.4%
(3.3% - 5.4%)
4.3%
(3.3% - 5.4%)
0.7%
(0.5% - 0.8%)
5.6%
(4.2% - 6.9%)
1 .9%
(1 .4% - 2.3%)
14/35
7%
(5.3% - 8.6%)
5.8%
(4.4% - 7.2%)
5%
(3.8% - 6.2%)
4.5%
(3.4% - 5.5%)
4.5%
(3.4% - 5.6%)
3.4%
(2.6% - 4.3%)
5%
(3.8% - 6.2%)
3.1%
(2.4% - 3.9%)
4.2%
(3.2% - 5.3%)
5.8%
(4.4% -7.1%)
4.4%
(3.3% - 5.4%)
4.3%
(3.3% - 5.4%)
0.7%
(0.5% - 0.8%)
5%
(3.8% - 6.2%)
1 .9%
(1 .4% - 2.3%)
13/35
6.1%
(4.7% - 7.6%)
5.1%
(3.8% - 6.2%)
4.3%
(3.3% - 5.4%)
4.5%
(3.4% - 5.5%)
3.9%
(2.9% - 4.8%)
3.4%
(2.6% - 4.3%)
4.4%
(3.3% - 5.4%)
3.1%
(2.4% - 3.9%)
4%
(3.1% -5%)
5.1%
(3.9% - 6.4%)
4.4%
(3.3% - 5.4%)
4%
(3.1% -5%)
0.7%
(0.5% - 0.8%)
4.3%
(3.3% - 5.3%)
1 .9%
(1 .4% - 2.3%)
12/35
5.3%
(4% - 6.5%)
4.3%
(3.3% - 5.3%)
3.7%
(2.8% - 4.5%)
3.9%
(3% - 4.8%)
3.2%
(2.4% - 4%)
3.4%
(2.6% - 4.3%)
3.7%
(2.8% - 4.5%)
2.7%
(2.1% -3. 4%)
3.3%
(2. 5% -4.1%)
4.4%
(3.3% - 5.4%)
4%
(3% - 4.9%)
3.5%
(2.6% - 4.3%)
0.7%
(0.5% - 0.8%)
3.6%
(2.7% - 4.5%)
1 .9%
(1 .4% - 2.3%)
13/30
6.1%
(4.7% - 7.6%)
4.9%
(3.7% - 6%)
4.3%
(3.3% - 5.4%)
4.5%
(3.4% - 5.5%)
3.3%
(2.5% -4.1%)
2.3%
(1 .7% - 2.9%)
4.4%
(3.3% - 5.4%)
2%
(1 .5% - 2.5%)
3%
(2.3% - 3.7%)
4.3%
(3.3% - 5.4%)
3.8%
(2.9% - 4.8%)
3.1%
(2.3% - 3.8%)
0%
(0% - 0%)
4.2%
(3.2% -5.1%)
1%
(0.7% - 1 .2%)
12/25
5.2%
(3.9% - 6.4%)
3.3%
(2.5% -4.1%)
3%
(2.3% - 3.7%)
3.9%
(3% - 4.8%)
2%
(1 .5% - 2.5%)
1 .2%
(0.9% - 1 .4%)
3.7%
(2.8% - 4.5%)
0.9%
(0.7% - 1 .2%)
1 .7%
(1 .3% - 2.2%)
2.9%
(2.2% - 3.5%)
2.4%
(1 .8% - 3%)
1 .8%
(1 .4% - 2.2%)
0%
(0% - 0%)
2.7%
(2.1% -3.4%)
0.1%
(0.1% -0.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-42
-------
Table E-42. Estimated Percent of Total Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
8.2%
(6.3% -10.1%)
7%
(5.3% - 8.6%)
8.5%
(6.5% -10.5%)
4.9%
(3.7% - 6%)
7%
(5.4% - 8.7%)
9.9%
(7.6% -12.1%)
6.4%
(4.9% - 7.9%)
7.6%
(5.8% - 9.4%)
6.9%
(5.3% - 8.5%)
6.6%
(5% -8.1%)
3.6%
(2.8% - 4.5%)
7.8%
(6% - 9.6%)
4.9%
(3.7% -6.1%)
7.3%
(5.6% - 9%)
3.4%
(2.6% - 4.2%)
15/353
7.3%
(5.5% - 9%)
6.4%
(4.9% - 7.9%)
6%
(4.5% - 7.4%)
4.9%
(3.7% - 6%)
4.9%
(3.8% -6.1%)
3.6%
(2.8% - 4.5%)
5.9%
(4.5% - 7.2%)
3.3%
(2.5% -4.1%)
5.1%
(3.8% - 6.3%)
5.7%
(4.3% -7.1%)
3.6%
(2.8% - 4.5%)
5.1%
(3.8% - 6.3%)
1 .6%
(1 .2% - 1 .9%)
6.1%
(4.7% - 7.6%)
2%
(1 .5% - 2.5%)
14/35
6.5%
(4.9% - 8%)
5.8%
(4.4% - 7.2%)
5.3%
(4% - 6.5%)
4.9%
(3.7% - 6%)
4.9%
(3.7% - 6%)
3.6%
(2.8% - 4.5%)
5.2%
(3.9% - 6.4%)
3.3%
(2.5% -4.1%)
5.1%
(3.8% - 6.3%)
5.7%
(4.3% -7.1%)
3.6%
(2.8% - 4.5%)
5.1%
(3.8% - 6.3%)
1 .6%
(1 .2% - 1 .9%)
5.5%
(4.2% - 6.8%)
2%
(1 .5% - 2.5%)
13/35
5.7%
(4.3% - 7%)
5.1%
(3.8% - 6.3%)
4.6%
(3.5% - 5.7%)
4.9%
(3.7% - 6%)
4.2%
(3.2% - 5.2%)
3.6%
(2.8% - 4.5%)
4.5%
(3.4% - 5.5%)
3.3%
(2.5% -4.1%)
4.8%
(3.7% - 6%)
5.1%
(3.9% - 6.3%)
3.6%
(2.8% - 4.5%)
4.7%
(3.6% - 5.9%)
1 .6%
(1 .2% - 1 .9%)
4.8%
(3.6% - 5.9%)
2%
(1 .5% - 2.5%)
12/35
4.9%
(3.7% -6.1%)
4.3%
(3.3% - 5.3%)
3.9%
(2.9% - 4.8%)
4.3%
(3.3% - 5.3%)
3.5%
(2.6% - 4.3%)
3.6%
(2.8% - 4.5%)
3.8%
(2.9% - 4.7%)
2.8%
(2.2% - 3.5%)
4.1%
(3.1% -5%)
4.3%
(3.3% - 5.3%)
3.2%
(2.4% - 4%)
4.1%
(3.1% -5.1%)
1 .6%
(1 .2% - 1 .9%)
4.1%
(3.1% -5%)
2%
(1 .5% - 2.5%)
13/30
5.7%
(4.3% - 7%)
4.9%
(3.7% - 6%)
4.6%
(3.5% - 5.7%)
4.9%
(3.7% - 6%)
3.6%
(2.7% - 4.5%)
2.5%
(1.9% -3.1%)
4.5%
(3.4% - 5.5%)
2.2%
(1 .6% - 2.7%)
3.7%
(2.8% - 4.6%)
4.3%
(3.2% - 5.3%)
3.1%
(2.4% - 3.9%)
3.7%
(2.8% - 4.6%)
0.7%
(0.5% - 0.9%)
4.7%
(3.5% - 5.8%)
1.1%
(0.8% - 1 .3%)
12/25
4.8%
(3.6% - 5.9%)
3.3%
(2.5% -4.1%)
3.2%
(2.4% - 4%)
4.3%
(3.3% - 5.3%)
2.3%
(1 .7% - 2.8%)
1 .3%
(1%-1.6%)
3.8%
(2.9% - 4.7%)
1%
(0.8% - 1 .3%)
2.3%
(1 .8% - 2.9%)
2.8%
(2.1% -3.5%)
1 .8%
(1 .4% - 2.3%)
2.3%
(1 .8% - 2.9%)
0%
(0% - 0%)
3.1%
(2.4% - 3.9%)
0.1%
(0.1% -0.2%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-43
-------
Table E-43. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiopulmonary Disease Mortality Associated
with Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2005 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with Long-
Term Exposure to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -9%)
-42%
(-41 % - -42%)
0%
(0% - 0%)
-37%
(-36% - -37%)
-182%
(-180% --184%)
-9%
(-9% - -9%)
-120%
(-11 9% --122%)
-35%
(-34% - -35%)
-14%
(-14% --14%)
0%
(0% - 0%)
-52%
(-51 % - -52%)
-252%
(-251 % - -254%)
-17%
(-17% --18%)
-53%
(-52% - -53%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(10% -11%)
8%
(8% - 8%)
12%
(11% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(11% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
21%
(21 % - 21 %)
19%
(19% -19%)
23%
(23% - 23%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
4%
(4% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
12/35
32%
(32% - 32%)
30%
(30% - 30%)
35%
(35% - 35%)
11%
(1 1 % - 1 1 %)
25%
(25% - 26%)
0%
(0% - 0%)
35%
(35% - 35%)
12%
(12% -12%)
18%
(18% -18%)
23%
(23% - 23%)
10%
(10% -10%)
18%
(17% -18%)
0%
(0% - 0%)
31%
(31 % - 31 %)
0%
(0% - 0%)
13/30
21%
(21% -21%)
22%
(22% - 22%)
23%
(23% - 23%)
0%
(0% - 0%)
23%
(23% - 24%)
34%
(34% - 34%)
23%
(23% - 23%)
31%
(31% -31%)
25%
(25% - 25%)
24%
(24% - 24%)
13%
(13% -13%)
25%
(25% - 25%)
64%
(64% - 64%)
22%
(22% - 22%)
33%
(33% - 33%)
12/25
34%
(33% - 34%)
44%
(44o/0 . 44o/0)
46%
(46% - 46%)
11%
(1 1 % - 1 1 %)
47%
(47% - 47%)
68%
(68% - 68%)
35%
(35% - 35%)
62%
(62% - 63%)
51%
(51% -51%)
48%
(48% - 49%)
45%
(45% - 45%)
51%
(50% -51%)
1 00%
(100% -100%)
45%
(45% - 45%)
67%
(67% - 67%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-44
-------
Table E-44. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiopulmonary Disease Mortality Associated
with Long-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with Long-
Term Exposure to PM25 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-12% --12%)
-9%
(-9% - -9%)
-43%
(-42% - -43%)
0%
(0% - 0%)
-43%
(-43% - -44%)
-1 73%
(-171% --176%)
-9%
(-9% - -9%)
-1 33%
(-132% --135%)
-40%
(-40% - -40%)
-15%
(-15% --15%)
0%
(0% - 0%)
-59%
(-59% - -60%)
-431 %
(-428% - -433%)
-20%
(-20% - -20%)
-75%
(-74% - -75%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -11%)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -11%)
0%
(0% - 0%)
13/35
21%
(21% -21%)
21%
(20% - 21 %)
24%
(23% - 24%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
24%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(31 % - 32%)
32%
(32% - 33%)
36%
(35% - 36%)
13%
(13% -13%)
30%
(30% - 30%)
0%
(0% - 0%)
36%
(35% - 36%)
13%
(13% -13%)
21%
(21% -21%)
24%
(24% - 24%)
10%
(10% -10%)
20%
(20% - 20%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
13/30
21%
(21% -21%)
24%
(24% - 24%)
24%
(23% - 24%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
24%
(23% - 24%)
35%
(35% - 35%)
29%
(29% - 29%)
25%
(25% - 25%)
13%
(1 3% - 1 3%)
29%
(29% - 29%)
100%
(100% -100%)
25%
(25% - 25%)
48%
(47% - 48%)
12/25
33%
(33% - 33%)
48%
(48% - 48%)
47%
(47% - 47%)
13%
(13% -13%)
56%
(56% - 56%)
66%
(66% - 66%)
36%
(35% - 36%)
70%
(70% - 70%)
59%
(59% - 59%)
50%
(50% - 51 %)
45%
(44% - 45%)
58%
(58% - 58%)
1 00%
(100% -100%)
51%
(51% -51%)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-45
-------
Table E-45. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiopulmonary Disease Mortality Associated
with Long-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current
Term Exposure to PM2.5 Concentrat
Annual (n
Recent PM2 5
Concentrations
-13%
(-13% --13%)
-10%
(-9% --10%)
-43%
(-42% - -43%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-171%
(-169% --174%)
-9%
(-9% - -9%)
-1 32%
(-130% --133%)
-37%
(-37% - -37%)
-15%
(-15% --15%)
0%
(0% - 0%)
-55%
(-54% - -55%)
-215%
(-21 4% --21 6%)
-19%
(-19% --19%)
-73%
(-73% - -73%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
Standards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with Long-
ons in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative
and Daily (m) Standards (Standard Combination Denoted n/m)2:
14/35
11%
(11% -11%)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
22%
(22% - 22%)
21%
(21% -21%)
23%
(23% - 24%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
24%
(24% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
6%
(6% - 7%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
12/35
33%
(33% - 33%)
33%
(33% - 33%)
35%
(35% - 36%)
12%
(12% -12%)
29%
(29% - 29%)
0%
(0% - 0%)
36%
(35% - 36%)
13%
(13% -13%)
19%
(19% -19%)
24%
(24% - 25%)
11%
(11% -11%)
19%
(19% -19%)
0%
(0% - 0%)
34%
(34% - 34%)
0%
(0% - 0%)
13/30
22%
(22% - 22%)
24%
(24% - 24%)
23%
(23% - 24%)
0%
(0% - 0%)
27%
(27% - 27%)
32%
(32% - 32%)
24%
(24% - 24%)
34%
(34% - 34%)
27%
(27% - 27%)
25%
(25% - 25%)
14%
(14% -14%)
27%
(27% - 27%)
55%
(55% - 55%)
24%
(24% - 24%)
46%
(46% - 46%)
12/25
34%
(34% - 35%)
48%
(48% - 49%)
46%
(46% - 47%)
12%
(12% -12%)
54%
(54% - 55%)
64%
(64% - 64%)
36%
(35% - 36%)
68%
(68% - 68%)
54%
(54% - 54%)
51%
(51% -51%)
50%
(50% - 50%)
54%
(54% - 54%)
100%
(100% -100%)
49%
(49% - 49%)
93%
(93% - 93%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-46
-------
Table E-46. Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in
a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year and
PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM2.5
Concentrations
722
(569 - 872)
715
(563 - 863)
516
(406 - 622)
455
(357 - 552)
1054
(830-1271)
338
(266 - 408)
707
(555 - 856)
3328
(2618-4019)
3117
(2450 - 3768)
621
(488 - 752)
579
(453 - 704)
747
(588 - 902)
109
(85 - 1 32)
1069
(842 - 1 290)
156
(122-190)
15/353
643
(506 - 778)
660
(518-797)
366
(287 - 443)
455
(357 - 552)
775
(608 - 939)
122
(95 - 1 48)
649
(508 - 786)
1526
(1191 -1856)
2326
(1821 -2820)
545
(427 - 660)
579
(453 - 704)
495
(388 - 601 )
31
(24 - 38)
913
(718-1104)
103
(80 - 1 25)
14/35
577
(453 - 698)
606
(476 - 733)
324
(254 - 393)
455
(357 - 552)
766
(601 - 928)
122
(95 - 1 48)
574
(450 - 697)
1526
(1191 - 1856)
2326
(1821 -2820)
545
(427 - 660)
579
(453 - 704)
495
(388 - 601 )
31
(24 - 38)
830
(651 -1005)
103
(80 - 1 25)
13/35
509
(399-617)
536
(420 - 649)
282
(221 - 343)
455
(357 - 552)
674
(528-817)
122
(95-148)
500
(391 - 607)
1526
(1191 -1856)
2223
(1740-2697)
488
(382 - 592)
579
(453 - 704)
466
(364 - 565)
31
(24 - 38)
732
(574 - 888)
103
(80-125)
12/35
441
(345 - 535)
465
(364 - 564)
240
(187-291)
405
(317-492)
581
(454 - 705)
122
(95-148)
424
(331 -516)
1344
(1048-1636)
1907
(1491 -2317)
420
(328 - 51 0)
521
(407 - 634)
409
(320 - 497)
31
(24 - 38)
633
(496 - 769)
103
(80-125)
13/30
509
(399-617)
517
(405 - 627)
282
(221 - 343)
455
(357 - 552)
596
(466 - 723)
81
(63 - 99)
500
(391 - 607)
1055
(822 - 1 286)
1745
(1363-2121)
416
(325 - 505)
506
(395-615)
372
(291 - 452)
11
(9-14)
714
(559 - 865)
69
(54 - 84)
12/25
430
(337 - 522)
371
(290 - 451 )
200
(156-243)
405
(317-492)
412
(321 -501)
39
(31 - 48)
424
(331 -516)
576
(448 - 703)
1151
(897 - 1 403)
283
(221 - 345)
318
(248 - 388)
246
(192-300)
0
(0-0)
509
(397-618)
34
(26 - 42)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-47
-------
Table E-47. Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in
a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent Year and
PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
744
(586 - 898)
584
(459 - 707)
486
(382 - 587)
344
(268 - 41 7)
770
(604 - 932)
352
(277 - 424)
686
(537-831)
2945
(2313-3562)
2435
(1907-2951)
564
(442 - 683)
614
(480 - 746)
595
(467 - 720)
89
(69 - 1 08)
810
(636-981)
106
(83 - 1 29)
15/353
662
(521 -801)
534
(41 9 - 647)
341
(267 - 41 4)
344
(268 - 41 7)
537
(420 - 652)
129
(100-157)
627
(491 -761)
1263
(985-1538)
1739
(1358-2114)
491
(385 - 596)
614
(480 - 746)
373
(292 - 454)
17
(13-21)
677
(530-821)
61
(47 - 74)
14/35
594
(466-719)
486
(381 -590)
301
(235 - 365)
344
(268-417)
530
(414-643)
129
(1 00 - 1 57)
553
(433 - 672)
1263
(985 - 1 538)
1739
(1358-2114)
491
(385 - 596)
614
(480 - 746)
373
(292 - 454)
17
(13-21)
606
(474 - 735)
61
(47 - 74)
13/35
524
(41 1 - 635)
424
(332 - 51 5)
260
(203 - 31 6)
344
(268 - 41 7)
453
(354-551)
129
(100-157)
479
(374 - 582)
1263
(985-1538)
1649
(1 288 - 2005)
437
(342-531)
614
(480 - 746)
348
(271 - 423)
17
(13-21)
522
(408 - 635)
61
(47 - 74)
12/35
454
(355 - 550)
361
(282 - 439)
219
(171 -267)
300
(234 - 365)
375
(293 - 457)
129
(100-157)
404
(315-491)
1094
(852-1333)
1373
(1071 -1671)
373
(291 - 453)
553
(432 - 673)
299
(233 - 364)
17
(13-21)
438
(342 - 533)
61
(47 - 74)
13/30
524
(411 -635)
407
(319-495)
260
(203-316)
344
(268-417)
388
(303 - 472)
87
(67 - 1 06)
479
(374 - 582)
825
(642 - 1 007)
1231
(960 - 1 499)
369
(288 - 449)
536
(419-653)
266
(208 - 324)
0
(0-0)
506
(395-616)
32
(25 - 39)
12/25
443
(346 - 538)
277
(21 6 - 338)
181
(141 -220)
300
(234 - 365)
236
(184-288)
44
(34 - 53)
404
(315-491)
380
(296 - 465)
713
(555 - 870)
244
(190-297)
340
(265 - 41 5)
157
(122- 192)
0
(0-0)
332
(259 - 405)
3
(2-3)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
, 2009).
statistical uncertainty surrounding the PM coefficient.
E-48
-------
Table E-48. Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in
a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year and
PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m)2:
Recent PM25
Concentrations
717
(563 - 865)
583
(458 - 706)
509
(401 -615)
383
(299 - 465)
810
(636 - 980)
370
(292 - 446)
715
(561 - 866)
3056
(2401 - 3695)
2837
(2227 - 3434)
558
(437 - 675)
522
(407 - 635)
667
(524 - 806)
127
(99-154)
887
(696-1072)
111
(87-135)
15/353
636
(500 - 770)
533
(418-646)
359
(282 - 436)
383
(299 - 465)
572
(447 - 694)
138
(108-168)
655
(513-794)
1333
(1040-1623)
2080
(1627-2526)
486
(381 - 589)
522
(407 - 635)
434
(340 - 527)
41
(32 - 50)
746
(585 - 904)
65
(50 - 79)
14/35
568
(446 - 689)
485
(380 - 589)
318
(249 - 386)
383
(299 - 465)
564
(441 - 685)
138
(108-168)
579
(453 - 702)
1333
(1040-1623)
2080
(1627-2526)
486
(381 - 589)
522
(407 - 635)
434
(340 - 527)
41
(32 - 50)
671
(526-814)
65
(50 - 79)
13/35
500
(391 - 606)
423
(331 -514)
276
(216-335)
383
(299 - 465)
485
(379 - 590)
138
(108-168)
501
(392 - 609)
1333
(1040-1623)
1982
(1550-2408)
432
(338 - 525)
522
(407 - 635)
407
(318-494)
41
(32 - 50)
584
(456 - 709)
65
(50 - 79)
12/35
430
(337 - 523)
361
(282 - 438)
234
(182-284)
337
(263 - 409)
406
(317-494)
138
(108-168)
424
(331 -515)
1160
(904-1413)
1681
(1313-2044)
368
(288 - 447)
464
(362 - 565)
354
(276 - 430)
41
(32 - 50)
495
(386 - 602)
65
(50 - 79)
13/30
500
(391 - 606)
407
(318-494)
276
(216-335)
383
(299 - 465)
419
(327 - 509)
95
(74 - 1 1 5)
501
(392 - 609)
884
(688 - 1 079)
1526
(1191 -1857)
364
(284 - 443)
448
(350 - 546)
320
(249 - 389)
18
(14-22)
567
(443 - 688)
35
(27 - 43)
12/25
420
(328-510)
277
(216-337)
194
(151 -236)
337
(263 - 409)
262
(204 - 320)
50
(39-61)
424
(331 -515)
428
(333 - 523)
960
(748-1171)
240
(187-292)
262
(204 - 320)
202
(158-247)
0
(0-0)
383
(299 - 467)
5
(4-6)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-49
-------
Table E-49. Estimated Percent of Total Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM25 Concentrations in a
Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
12.4%
(9.7% -14.9%)
12.2%
(9.6% -14.7%)
12.4%
(9.8% -15%)
8.6%
(6.7% -10.4%)
12.8%
(10.1% -15.5%)
13.1%
(10.3% -15.8%)
9.5%
(7.4% - 1 1 .5%)
1 1 .8%
(9.3% -14.3%)
11%
(8.6% -13.3%)
10.4%
(8.1% -12.5%)
6.2%
(4.8% - 7.5%)
12.3%
(9.6% -14.8%)
6.3%
(5% - 7.7%)
12.6%
(9.9% -15.2%)
7.1%
(5.6% - 8.6%)
15/353
11%
(8.7% -13. 3%)
1 1 .2%
(8. 8% -13. 5%)
8.8%
(6. 9% -10. 7%)
8.6%
(6.7% -10. 4%)
9.4%
(7.4% - 1 1 .4%)
4.7%
(3.7% - 5.7%)
8.7%
(6. 8% -10. 5%)
5.4%
(4.2% - 6.6%)
8.2%
(6.4% - 9.9%)
9.1%
(7.1% -11%)
6.2%
(4.8% - 7.5%)
8.1%
(6.4% - 9.9%)
1 .8%
(1 .4% - 2.2%)
10.8%
(8.4% -13%)
4.7%
(3.6% - 5.7%)
14/35
9.9%
(7.7% - 1 1 .9%)
10.3%
(8.1% -12.4%)
7.8%
(6.1% -9.5%)
8.6%
(6.7% -10.4%)
9.3%
(7.3% - 1 1 .3%)
4.7%
(3.7% - 5.7%)
7.7%
(6% - 9.4%)
5.4%
(4.2% - 6.6%)
8.2%
(6.4% - 9.9%)
9.1%
(7.1% -11%)
6.2%
(4.8% - 7.5%)
8.1%
(6.4% - 9.9%)
1 .8%
(1 .4% - 2.2%)
9.8%
(7.7% - 1 1 .8%)
4.7%
(3.6% - 5.7%)
13/35
8.7%
(6.8% -10. 6%)
9.1%
(7.1% -11%)
6.8%
(5.3% - 8.3%)
8.6%
(6.7% -10. 4%)
8.2%
(6.4% -10%)
4.7%
(3.7% - 5.7%)
6.7%
(5.2% -8.1%)
5.4%
(4.2% - 6.6%)
7.8%
(6.1% -9.5%)
8.1%
(6.4% - 9.9%)
6.2%
(4.8% - 7.5%)
7.6%
(6% - 9.3%)
1 .8%
(1 .4% - 2.2%)
8.6%
(6.8% -10.5%)
4.7%
(3.6% - 5.7%)
12/35
7.5%
(5.9% - 9.2%)
7.9%
(6.2% - 9.6%)
5.8%
(4.5% - 7%)
7.7%
(6% - 9.3%)
7.1%
(5.5% - 8.6%)
4.7%
(3.7% - 5.7%)
5.7%
(4.4% - 6.9%)
4.8%
(3.7% - 5.8%)
6.7%
(5.2% -8.1%)
7%
(5.5% - 8.5%)
5.5%
(4.3% - 6.7%)
6.7%
(5.2% - 8.2%)
1 .8%
(1 .4% - 2.2%)
7.5%
(5.8% - 9%)
4.7%
(3.6% - 5.7%)
13/30
8.7%
(6.8% -10.6%)
8.8%
(6.9% -10.6%)
6.8%
(5.3% - 8.3%)
8.6%
(6.7% -10.4%)
7.3%
(5.7% - 8.8%)
3.1%
(2.4% - 3.8%)
6.7%
(5.2% -8.1%)
3.8%
(2.9% - 4.6%)
6.1%
(4.8% - 7.5%)
6.9%
(5.4% - 8.4%)
5.4%
(4.2% - 6.5%)
6.1%
(4.8% - 7.4%)
0.6%
(0.5% - 0.8%)
8.4%
(6.6% -10.2%)
3.1%
(2.4% - 3.8%)
12/25
7.4%
(5.8% - 8.9%)
6.3%
(4.9% - 7.7%)
4.8%
(3.8% - 5.9%)
7.7%
(6% - 9.3%)
5%
(3.9% -6.1%)
1 .5%
(1 .2% - 1 .9%)
5.7%
(4.4% - 6.9%)
2%
(1 .6% - 2.5%)
4%
(3.2% - 4.9%)
4.7%
(3.7% - 5.8%)
3.4%
(2.6% -4.1%)
4%
(3.1% -4.9%)
0%
(0% - 0%)
6%
(4.7% - 7.3%)
1 .5%
(1 .2% - 1 .9%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-50
-------
Table E-50. Estimated Percent of Total Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiopulmonary Mortality Associated with Long-Term Expc
Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n
Combination Denoted n/m)2:
Recent PM25
Concentrations
12.4%
(9.7% -14.9%)
9.9%
(7.8% -12%)
1 1 .6%
(9.1% -14%)
6.4%
(5% - 7.7%)
9.4%
(7.4% - 1 1 .4%)
13.4%
(10.6% -16. 2%)
8.9%
(7% -10. 8%)
10.4%
(8.2% -12.6%)
8.5%
(6.6% -10.3%)
9.4%
(7.4% - 1 1 .4%)
6.3%
(4.9% - 7.6%)
9.8%
(7.7% - 1 1 .9%)
5%
(3.9% -6.1%)
9.5%
(7.4% - 1 1 .5%)
4.7%
(3.7% - 5.8%)
15/353
11%
(8.6% -13.3%)
9.1%
(7.1% -11%)
8.1%
(6.4% - 9.8%)
6.4%
(5% - 7.7%)
6.5%
(5.1% -8%)
4.9%
(3.8% - 6%)
8.1%
(6.4% - 9.9%)
4.5%
(3.5% - 5.4%)
6.1%
(4.7% - 7.4%)
8.2%
(6.4% -10%)
6.3%
(4.9% - 7.6%)
6.2%
(4.8% - 7.5%)
0.9%
(0.7% - 1 .2%)
7.9%
(6.2% - 9.6%)
2.7%
(2.1% -3. 3%)
14/35
9.9%
(7.7% - 1 1 .9%)
8.2%
(6.5% -10%)
7.2%
(5.6% - 8.7%)
6.4%
(5% - 7.7%)
6.5%
(5% - 7.8%)
4.9%
(3.8% - 6%)
7.2%
(5.6% - 8.7%)
4.5%
(3.5% - 5.4%)
6.1%
(4.7% - 7.4%)
8.2%
(6.4% -10%)
6.3%
(4.9% - 7.6%)
6.2%
(4.8% - 7.5%)
0.9%
(0.7% - 1 .2%)
7.1%
(5.5% - 8.6%)
2.7%
(2.1% -3.3%)
13/35
8.7%
(6.8% -10. 5%)
7.2%
(5.6% - 8.7%)
6.2%
(4.8% - 7.5%)
6.4%
(5% - 7.7%)
5.5%
(4.3% - 6.7%)
4.9%
(3.8% - 6%)
6.2%
(4.9% - 7.6%)
4.5%
(3.5% - 5.4%)
5.7%
(4.5% - 7%)
7.3%
(5.7% - 8.9%)
6.3%
(4.9% - 7.6%)
5.7%
(4.5% - 7%)
0.9%
(0.7% - 1 .2%)
6.1%
(4.8% - 7.4%)
2.7%
(2.1% -3.3%)
12/35
7.5%
(5.9% -9.1%)
6.1%
(4.8% - 7.4%)
5.2%
(4.1% -6. 4%)
5.5%
(4.3% - 6.7%)
4.6%
(3.6% - 5.6%)
4.9%
(3.8% - 6%)
5.2%
(4.1% -6. 4%)
3.9%
(3% - 4.7%)
4.8%
(3.7% - 5.8%)
6.2%
(4.9% - 7.6%)
5.7%
(4.4% - 6.9%)
4.9%
(3.8% - 6%)
0.9%
(0.7% - 1 .2%)
5.1%
(4% - 6.2%)
2.7%
(2.1% -3.3%)
)sure to PM2.s Concentrations in a
and Daily (m) Standards (Standard
13/30
8.7%
(6.8% -10.5%)
6.9%
(5.4% - 8.4%)
6.2%
(4.8% - 7.5%)
6.4%
(5% - 7.7%)
4.7%
(3.7% - 5.8%)
3.3%
(2.6% - 4%)
6.2%
(4.9% - 7.6%)
2.9%
(2.3% - 3.6%)
4.3%
(3.3% - 5.2%)
6.2%
(4.8% - 7.5%)
5.5%
(4.3% - 6.7%)
4.4%
(3.4% - 5.4%)
0%
(0% - 0%)
5.9%
(4.6% - 7.2%)
1 .4%
(1.1% -1.7%)
12/25
7.4%
(5.8% - 8.9%)
4.7%
(3.7% - 5.7%)
4.3%
(3.4% - 5.2%)
5.5%
(4.3% - 6.7%)
2.9%
(2.2% - 3.5%)
1 .7%
(1 .3% - 2%)
5.2%
(4.1% -6.4%)
1 .3%
(1 % - 1 .6%)
2.5%
(1 .9% - 3%)
4.1%
(3.2% - 5%)
3.5%
(2.7% - 4.2%)
2.6%
(2% - 3.2%)
0%
(0% - 0%)
3.9%
(3% - 4.7%)
0.1%
(0.1% -0.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-51
-------
Table E-51. Estimated Percent of Total Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 2001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
1 1 .6%
(9.1% -14%)
9.9%
(7.8% -12%)
12%
(9.5% -14.5%)
7%
(5.4% - 8.4%)
9.9%
(7.8% -12%)
13.9%
(11% -16.7%)
9.1%
(7.1% -11%)
10.7%
(8.4% -13%)
9.8%
(7.7% - 1 1 .9%)
9.3%
(7.3% - 1 1 .3%)
5.2%
(4% - 6.3%)
11.1%
(8.7% -13.4%)
7%
(5.4% - 8.5%)
10.4%
(8.1% -12.5%)
4.9%
(3.8% - 5.9%)
15/353
10.3%
(8.1% -12.5%)
9.1%
(7.1% -11%)
8.5%
(6.6% -10.3%)
7%
(5.4% - 8.4%)
7%
(5.5% - 8.5%)
5.2%
(4.1% -6.3%)
8.3%
(6.5% -10.1%)
4.7%
(3.7% - 5.7%)
7.2%
(5.6% - 8.7%)
8.1%
(6.4% - 9.8%)
5.2%
(4% - 6.3%)
7.2%
(5.6% - 8.7%)
2.2%
(1 .7% - 2.7%)
8.7%
(6.8% -10.6%)
2.8%
(2.2% - 3.5%)
14/35
9.2%
(7.2% -11.1%)
8.2%
(6.5% -10%)
7.5%
(5.9% -9.1%)
7%
(5.4% - 8.4%)
6.9%
(5.4% - 8.4%)
5.2%
(4.1% -6.3%)
7.4%
(5.8% - 8.9%)
4.7%
(3.7% - 5.7%)
7.2%
(5.6% - 8.7%)
8.1%
(6.4% - 9.8%)
5.2%
(4% - 6.3%)
7.2%
(5.6% - 8.7%)
2.2%
(1 .7% - 2.7%)
7.9%
(6.1% -9.5%)
2.8%
(2.2% - 3.5%)
13/35
8.1%
(6.3% - 9.8%)
7.2%
(5.6% - 8.7%)
6.5%
(5.1% -7.9%)
7%
(5.4% - 8.4%)
6%
(4.7% - 7.2%)
5.2%
(4.1% -6.3%)
6.4%
(5% - 7.8%)
4.7%
(3.7% - 5.7%)
6.9%
(5.4% - 8.3%)
7.2%
(5.6% - 8.8%)
5.2%
(4% - 6.3%)
6.7%
(5.3% - 8.2%)
2.2%
(1 .7% - 2.7%)
6.8%
(5.3% - 8.3%)
2.8%
(2.2% - 3.5%)
12/35
7%
(5.4% - 8.5%)
6.1%
(4.8% - 7.5%)
5.5%
(4.3% - 6.7%)
6.1%
(4.8% - 7.4%)
5%
(3.9% -6.1%)
5.2%
(4.1% -6.3%)
5.4%
(4.2% - 6.6%)
4.1%
(3.2% - 5%)
5.8%
(4.5% -7.1%)
6.2%
(4.8% - 7.5%)
4.6%
(3.6% - 5.6%)
5.9%
(4.6% -7.1%)
2.2%
(1 .7% - 2.7%)
5.8%
(4.5% - 7%)
2.8%
(2.2% - 3.5%)
13/30
8.1%
(6.3% - 9.8%)
6.9%
(5.4% - 8.4%)
6.5%
(5.1% -7. 9%)
7%
(5.4% - 8.4%)
5.1%
(4% - 6.3%)
3.5%
(2.8% - 4.3%)
6.4%
(5% - 7.8%)
3.1%
(2.4% - 3.8%)
5.3%
(4.1% -6. 4%)
6.1%
(4.8% - 7.4%)
4.4%
(3.5% - 5.4%)
5.3%
(4.1% -6. 4%)
1%
(0.8% - 1 .2%)
6.6%
(5.2% -8.1%)
1 .5%
(1 .2% - 1 .9%)
12/25
6.8%
(5.3% - 8.3%)
4.7%
(3.7% - 5.7%)
4.6%
(3.6% - 5.6%)
6.1%
(4.8% - 7.4%)
3.2%
(2.5% - 3.9%)
1 .9%
(1 .5% - 2.3%)
5.4%
(4.2% - 6.6%)
1 .5%
(1 .2% - 1 .8%)
3.3%
(2.6% -4.1%)
4%
(3.1% -4.9%)
2.6%
(2% - 3.2%)
3.4%
(2.6% -4.1%)
0%
(0% - 0%)
4.5%
(3.5% - 5.5%)
0.2%
(0.2% - 0.2%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-52
-------
Table E-52. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiopulmonary Disease Mortality Associated with
Long-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with Long-
Term Exposure to PM25 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
-12%
(-12% --12%)
-8%
(-8o/0 - -go/0)
-41%
(-40% - -42%)
0%
(0% - 0%)
-36%
(-35% - -37%)
-178%
(-175% --181%)
-9%
(-9% - -9%)
-1 1 8%
(-11 6% --120%)
-34%
(-34% - -34%)
-1 4%
(-14% --14%)
0%
(0% - 0%)
-51%
(-50% - -52%)
-250%
(-248% --251%)
-1 7%
(-17% --17%)
-52%
(-52% - -52%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -10%)
8%
(8% - 8%)
11%
(11% -11%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(11% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
21%
(21% -21%)
19%
(19% -19%)
23%
(23% - 23%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
4%
(4% - 4%)
10%
(10% -11%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
12/35
32%
(31 % - 32%)
29%
(29% - 30%)
34%
(34% - 35%)
11%
(1 1 % - 1 1 %)
25%
(25% - 25%)
0%
(0% - 0%)
35%
(34% - 35%)
12%
(12% -12%)
18%
(18% -18%)
23%
(23% - 23%)
10%
(10% -10%)
17%
(17% -18%)
0%
(0% - 0%)
31%
(30% - 31 %)
0%
(0% - 0%)
13/30
21%
(21% -21%)
22%
(21 % - 22%)
23%
(23% - 23%)
0%
(0% - 0%)
23%
(23% - 23%)
34%
(33% - 34%)
23%
(23% - 23%)
31%
(31 % - 31 %)
25%
(25% - 25%)
24%
(24% - 24%)
13%
(13% -13%)
25%
(25% - 25%)
64%
(64% - 64%)
22%
(22% - 22%)
33%
(33% - 33%)
12/25
33%
(33% - 33%)
44%
(43% - 44%)
45%
(45% - 46%)
11%
(11% -11%)
47%
(47% - 47%)
68%
(68% - 68%)
35%
(34% - 35%)
62%
(62% - 62%)
50%
(50% -51%)
48%
(48% - 48%)
45%
(45% - 45%)
50%
(50% -51%)
1 00%
(100% -100%)
44%
(44% - 45%)
67%
(67% - 67%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-53
-------
Table E-53. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiopulmonary Disease Mortality Associated with
Long-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with Long-
Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
-12%
(-12% --12%)
-9%
(-9% - -9%)
-43%
(-42% - -43%)
0%
(0% - 0%)
-43%
(-43% - -44%)
-173%
(-171% --176%)
-9%
(-9% - -9%)
-133%
(-132% --135%)
-40%
(-40% - -40%)
-15%
(-15% --15%)
0%
(0% - 0%)
-59%
(-59% - -60%)
-431 %
(-428% - -433%)
-20%
(-20% - -20%)
-75%
(-74% - -75%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(1 0% - 1 1 %)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(1 0% - 1 1 %)
0%
(0% - 0%)
13/35
21%
(21 % - 21 %)
21%
(20% -21%)
24%
(23% - 24%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
24%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(31 % - 32%)
32%
(32% - 33%)
36%
(35% - 36%)
13%
(13% -13%)
30%
(30% - 30%)
0%
(0% - 0%)
36%
(35% - 36%)
13%
(13% -13%)
21%
(21% -21%)
24%
(24% - 24%)
10%
(10% -10%)
20%
(20% - 20%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
13/30
21%
(21% -21%)
24%
(24% - 24%)
24%
(23% - 24%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
24%
(23% - 24%)
35%
(35% - 35%)
29%
(29% - 29%)
25%
(25% - 25%)
13%
(13% -13%)
29%
(29% - 29%)
1 00%
(100% -100%)
25%
(25% - 25%)
48%
(47% - 48%)
12/25
33%
(33% - 33%)
48%
(48% - 48%)
47%
(47% - 47%)
13%
(13% -13%)
56%
(56% - 56%)
66%
(66% - 66%)
36%
(35% - 36%)
70%
(70% - 70%)
59%
(59% - 59%)
50%
(50% -51%)
45%
(44% - 45%)
58%
(58% - 58%)
100%
(100% -100%)
51%
(51% -51%)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-54
-------
Table E-54. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiopulmonary Disease Mortality Associated with
Long-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al.
(2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with Long-
Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -9%)
-42%
(-41% --42%)
0%
(0% - 0%)
-42%
(-41% --42%)
-168%
(-165% --170%)
-9%
(-9% - -9%)
-129%
(-128% --131%)
-36%
(-36% - -37%)
-15%
(-15% --15%)
0%
(0% - 0%)
-54%
(-53% - -54%)
-213%
(-211% --215%)
-19%
(-19% --19%)
-72%
(-72% - -73%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(1 1 % - 1 1 %)
9%
(9% - 9%)
12%
(11% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
21%
(21 % - 22%)
21%
(20% -21%)
23%
(23% - 23%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
23%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
32%
(32% - 33%)
35%
(35% - 35%)
12%
(12% -12%)
29%
(29% - 29%)
0%
(0% - 0%)
35%
(35% - 36%)
13%
(13% -13%)
19%
(19% -19%)
24%
(24% - 24%)
11%
(11% -11%)
19%
(18% -19%)
0%
(0% - 0%)
34%
(33% - 34%)
0%
(0% - 0%)
13/30
21%
(21 % - 22%)
24%
(24% - 24%)
23%
(23% - 23%)
0%
(0% - 0%)
27%
(27% - 27%)
32%
(32% - 32%)
23%
(23% - 24%)
34%
(34% - 34%)
27%
(26% - 27%)
25%
(25% - 25%)
14%
(14% -14%)
26%
(26% - 27%)
55%
(55% - 55%)
24%
(24% - 24%)
46%
(46% - 46%)
12/25
34%
(34% - 34%)
48%
(48% - 48%)
46%
(46% - 46%)
12%
(12% -12%)
54%
(54% - 54%)
64%
(64% - 64%)
35%
(35% - 36%)
68%
(68% - 68%)
54%
(54% - 54%)
51%
(50% -51%)
50%
(50% - 50%)
53%
(53% - 54%)
100%
(100% -100%)
49%
(48% - 49%)
93%
(93% - 93%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-55
-------
Table E-55. Estimated Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year and PM25
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
77
(29 - 1 22)
81
(31 -128)
55
(21 - 87)
50
(19-79)
112
(43 - 1 78)
26
(10-41)
76
(29 - 1 22)
248
(94 - 393)
208
(79 - 331 )
70
(26 - 1 1 1 )
58
(22 - 94)
80
(31 -127)
8
(3-13)
116
(44 - 1 84)
19
(7 - 30)
15/353
68
(26-108)
74
(28 - 1 1 8)
39
(15-62)
50
(19-79)
82
(31 -131)
9
(3-15)
70
(26 - 1 1 2)
112
(42-181)
155
(58 - 247)
61
(23 - 98)
58
(22 - 94)
53
(20 - 84)
2
(1-4)
99
(37-157)
12
(5 - 20)
14/35
61
(23 - 97)
68
(26 - 1 09)
34
(13-55)
50
(19-79)
81
(31 -129)
9
(3-15)
62
(23 - 99)
112
(42 -181)
155
(58 - 247)
61
(23 - 98)
58
(22 - 94)
53
(20 - 84)
2
(1 -4)
90
(34 - 1 43)
12
(5 - 20)
13/35
54
(20 - 86)
60
(23 - 96)
30
(1 1 - 48)
50
(19-79)
71
(27 - 1 1 4)
9
(3-15)
54
(20 - 86)
112
(42-181)
148
(56 - 236)
55
(21 - 87)
58
(22 - 94)
50
(19-79)
2
(1 -4)
79
(30-126)
12
(5 - 20)
12/35
46
(17-74)
52
(20 - 84)
25
(10-41)
44
(17-71)
61
(23 - 98)
9
(3-15)
46
(17-73)
99
(37-160)
126
(48 - 203)
47
(18-75)
53
(20 - 85)
44
(16-70)
2
(1 -4)
68
(26-109)
12
(5 - 20)
13/30
54
(20 - 86)
58
(22 - 93)
30
(11 -48)
50
(19-79)
63
(24-101)
6
(2-10)
54
(20 - 86)
78
(29 - 1 25)
116
(43 - 1 86)
46
(17-75)
51
(19-82)
40
(15-64)
1
(0-1)
77
(29 - 1 23)
8
(3-13)
12/25
45
(17-73)
42
(16-67)
21
(8 - 34)
44
(17-71)
43
(16-70)
3
(1-5)
46
(17-73)
42
(16-68)
76
(28 - 1 23)
32
(12-51)
32
(12-52)
26
(10-42)
0
(0-0)
54
(20 - 88)
4
(1 -6)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
, 2009).
statistical uncertainty surrounding the PM coefficient.
E-56
-------
Table E-56. Estimated Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
79
(30 - 1 25)
66
(25 - 1 05)
52
(20 - 82)
37
(14-60)
82
(31 -130)
27
(10-43)
74
(28-118)
219
(83 - 348)
162
(61 - 259)
63
(24 -101)
62
(23 - 1 00)
64
(24 -101)
6
(2-10)
87
(33 - 1 39)
13
(5-20)
15/353
70
(27-112)
60
(23 - 96)
36
(14-58)
37
(14-60)
57
(21 -91)
10
(4-16)
68
(26-108)
93
(35-150)
115
(43-185)
55
(21 - 88)
62
(23-100)
40
(15-64)
1
(0-2)
73
(27-116)
7
(3-12)
14/35
63
(24-100)
55
(21 - 87)
32
(12-51)
37
(14-60)
56
(21 - 90)
10
(4-16)
60
(22 - 96)
93
(35-150)
115
(43-185)
55
(21 - 88)
62
(23-100)
40
(15-64)
1
(0-2)
65
(24-104)
7
(3-12)
13/35
55
(21 - 88)
48
(18-76)
28
(10-44)
37
(14-60)
48
(18-77)
10
(4-16)
52
(19-83)
93
(35-150)
109
(41 -176)
49
(18-78)
62
(23 - 1 00)
37
(14-59)
1
(0-2)
56
(21 - 90)
7
(3-12)
12/35
48
(18-77)
40
(15-65)
23
(9 - 37)
33
(12-52)
39
(15-64)
10
(4-16)
43
(16-70)
80
(30 - 1 30)
91
(34-146)
42
(16-67)
56
(21 - 90)
32
(12-51)
1
(0-2)
47
(18-75)
7
(3-12)
13/30
55
(21 - 88)
46
(17-73)
28
(10-44)
37
(14-60)
41
(15-66)
7
(2-11)
52
(19-83)
61
(23 - 98)
81
(30-131)
41
(15-66)
54
(20 - 87)
28
(11-45)
0
(0-0)
54
(20 - 87)
4
(1-6)
12/25
47
(18-75)
31
(12-50)
19
(7-31)
33
(12-52)
25
(9 - 40)
3
(1-5)
43
(16-70)
28
(10-45)
47
(17-76)
27
(10-44)
34
(13-55)
17
(6 - 27)
0
(0-0)
35
(13-57)
0
(0-0)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-57
-------
Table E-57. Estimated Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year and PM25
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
76
(29 -121)
66
(25 - 1 05)
54
(21 - 86)
42
(16-67)
86
(33 - 1 37)
29
(11 -45)
77
(29 - 1 23)
227
(86 - 361 )
189
(72 - 301 )
63
(24 - 1 00)
53
(20 - 85)
71
(27-114)
9
(3-15)
96
(36 - 1 52)
13
(5-21)
15/353
67
(26-107)
60
(23 - 96)
38
(14-61)
42
(16-67)
60
(23 - 97)
11
(4-17)
71
(27 - 1 1 3)
98
(37-158)
138
(52-221)
54
(21 - 87)
53
(20 - 85)
46
(17-74)
3
(1 -5)
80
(30-128)
8
(3-12)
14/35
60
(23 - 96)
55
(21 - 87)
34
(13-54)
42
(16-67)
59
(22 - 95)
11
(4-17)
62
(23 - 1 00)
98
(37 - 1 58)
138
(52 - 221 )
54
(21 - 87)
53
(20 - 85)
46
(17-74)
3
(1 -5)
72
(27-116)
8
(3-12)
13/35
53
(20 - 84)
47
(18-76)
29
(1 1 - 47)
42
(16-67)
51
(19-82)
11
(4-17)
54
(20 - 87)
98
(37-158)
131
(49 -211)
48
(18-77)
53
(20 - 85)
43
(16-69)
3
(1-5)
63
(24- 100)
8
(3-12)
12/35
45
(17-73)
40
(15-65)
25
(9 - 40)
37
(14-59)
43
(16-69)
11
(4-17)
46
(17-73)
85
(32-138)
111
(42-179)
41
(15-66)
47
(17-75)
38
(14-60)
3
(1-5)
53
(20 - 85)
8
(3-12)
13/30
53
(20 - 84)
46
(17-73)
29
(11-47)
42
(16-67)
44
(16-71)
7
(3-12)
54
(20 - 87)
65
(24 - 1 05)
101
(38 - 1 63)
41
(15-65)
45
(17-73)
34
(13-55)
1
(0-2)
61
(23 - 98)
4
(2-7)
12/25
44
(17-71)
31
(12-50)
20
(8 - 33)
37
(14-59)
28
(10-44)
4
(1 -6)
46
(17-73)
31
(12-51)
63
(24-102)
27
(10-43)
26
(10-43)
21
(8 - 35)
0
(0-0)
41
(15-66)
1
(0-1)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-58
-------
Table E-58. Estimated Percent of Total Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2005 PM2 s Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from
1979 -1983
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2 5
Concentrations
8.6%
(3.3% -13.6%)
8.4%
(3.2% -13.4%)
8.6%
(3.3% -13.7%)
5.9%
(2.2% - 9.5%)
8.9%
(3. 4% -14.1%)
9.1%
(3. 5% -14.4%)
6.6%
(2. 5% -10.5%)
8.2%
(3.1% -13%)
7.6%
(2. 9% -12.1%)
7.2%
(2.7% - 1 1 .4%)
4.2%
(1 .6% - 6.8%)
8.5%
(3. 2% -13.5%)
4.4%
(1 .6% - 7%)
8.8%
(3. 3% -13.9%)
4.9%
(1 .8% - 7.8%)
15/353
7.6%
(2.9% -12.1%)
7.8%
(3% -12.3%)
6.1%
(2.3% - 9.7%)
5.9%
(2.2% - 9.5%)
6.5%
(2.5% -10.4%)
3.2%
(1 .2% - 5.2%)
6%
(2.3% - 9.6%)
3.7%
(1 .4% - 6%)
5.6%
(2.1% -9%)
6.3%
(2.4% -10%)
4.2%
(1 .6% - 6.8%)
5.6%
(2.1% -9%)
1 .2%
(0.5% - 2%)
7.5%
(2.8% - 1 1 .9%)
3.2%
(1 .2% - 5.2%)
14/35
6.8%
(2.6% -10.9%)
7.1%
(2.7% - 1 1 .3%)
5.4%
(2% - 8.6%)
5.9%
(2.2% - 9.5%)
6.5%
(2.4% -10.3%)
3.2%
(1 .2% - 5.2%)
5.3%
(2% - 8.5%)
3.7%
(1 .4% - 6%)
5.6%
(2.1% -9%)
6.3%
(2.4% -10%)
4.2%
(1 .6% - 6.8%)
5.6%
(2.1% -9%)
1 .2%
(0.5% - 2%)
6.8%
(2.6% -10.8%)
3.2%
(1 .2% - 5.2%)
13/35
6%
(2.3% - 9.6%)
6.3%
(2.4% -10%)
4.7%
(1 .8% - 7.5%)
5.9%
(2.2% - 9.5%)
5.7%
(2.1% -9.1%)
3.2%
(1 .2% - 5.2%)
4.6%
(1 .7% - 7.4%)
3.7%
(1 .4% - 6%)
5.4%
(2% - 8.6%)
5.6%
(2.1% -9%)
4.2%
(1 .6% - 6.8%)
5.3%
(2% - 8.4%)
1 .2%
(0.5% - 2%)
6%
(2.3% - 9.5%)
3.2%
(1 .2% - 5.2%)
12/35
5.2%
(2% - 8.3%)
5.5%
(2.1% -8.7%)
4%
(1 .5% - 6.4%)
5.3%
(2% - 8.5%)
4.9%
(1 .8% - 7.8%)
3.2%
(1 .2% - 5.2%)
3.9%
(1 .5% - 6.3%)
3.3%
(1 .2% - 5.3%)
4.6%
(1 .7% - 7.4%)
4.8%
(1 .8% - 7.7%)
3.8%
(1.4% -6.1%)
4.6%
(1 .7% - 7.4%)
1 .2%
(0.5% - 2%)
5.1%
(1 .9% - 8.2%)
3.2%
(1 .2% - 5.2%)
13/30
6%
(2.3% - 9.6%)
6.1%
(2.3% - 9.7%)
4.7%
(1 .8% - 7.5%)
5.9%
(2.2% - 9.5%)
5%
(1 .9% - 8%)
2.1%
(0.8% - 3.5%)
4.6%
(1 .7% - 7.4%)
2.6%
(1%-4.2%)
4.2%
(1 .6% - 6.8%)
4.8%
(1 .8% - 7.7%)
3.7%
(1 .4% - 5.9%)
4.2%
(1 .6% - 6.7%)
0.4%
(0.2% - 0.7%)
5.8%
(2.2% - 9.3%)
2.1%
(0.8% - 3.5%)
12/25
5.1%
(1.9% -8.1%)
4.3%
(1 .6% - 7%)
3.3%
(1 .2% - 5.3%)
5.3%
(2% - 8.5%)
3.4%
(1 .3% - 5.5%)
1%
(0.4% - 1 .7%)
3.9%
(1 .5% - 6.3%)
1 .4%
(0.5% - 2.3%)
2.8%
(1%-4.5%)
3.2%
(1 .2% - 5.2%)
2.3%
(0.9% - 3.7%)
2.8%
(1%-4.5%)
0%
(0% - 0%)
4.1%
(1 .5% - 6.6%)
1.1%
(0.4% - 1 .7%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-59
-------
Table E-59. Estimated Percent of Total Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure
to Ambient PM2 5 Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current
and Alternative Standards, Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009),
Using Ambient PM2 s from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2 5
Concentrations
8.6%
(3.3% -13.6%)
6.9%
(2.6% -10.9%)
8%
(3.1% -12.7%)
4.4%
(1 .6% - 7%)
6.5%
(2.5% -10.4%)
9.3%
(3.6% -14.8%)
6.1%
(2.3% - 9.8%)
7.2%
(2.7% - 1 1 .5%)
5.9%
(2.2% - 9.4%)
6.5%
(2.5% -10.4%)
4.3%
(1 .6% - 6.9%)
6.8%
(2.6% -10.8%)
3.4%
(1 .3% - 5.6%)
6.6%
(2.5% -10.5%)
3.2%
(1 .2% - 5.2%)
15/353
7.6%
(2.9% -12.1%)
6.3%
(2.4% -10%)
5.6%
(2.1% -9%)
4.4%
(1 .6% - 7%)
4.5%
(1 .7% - 7.2%)
3.4%
(1 .3% - 5.4%)
5.6%
(2.1% -9%)
3.1%
(1.1% -4.9%)
4.2%
(1 .6% - 6.7%)
5.7%
(2.1% -9.1%)
4.3%
(1 .6% - 6.9%)
4.2%
(1 .6% - 6.8%)
0.6%
(0.2% - 1 %)
5.5%
(2.1% -8.8%)
1 .9%
(0.7% - 3%)
14/35
6.8%
(2.6% -10.9%)
5.7%
(2.1% -9.1%)
4.9%
(1 .9% - 7.9%)
4.4%
(1 .6% - 7%)
4.4%
(1.7% -7.1%)
3.4%
(1 .3% - 5.4%)
4.9%
(1 .9% - 7.9%)
3.1%
(1.1% -4.9%)
4.2%
(1 .6% - 6.7%)
5.7%
(2.1% -9.1%)
4.3%
(1 .6% - 6.9%)
4.2%
(1 .6% - 6.8%)
0.6%
(0.2% -1%)
4.9%
(1 .8% - 7.8%)
1 .9%
(0.7% - 3%)
13/35
6%
(2.3% - 9.6%)
5%
(1 .9% - 7.9%)
4.3%
(1 .6% - 6.8%)
4.4%
(1 .6% - 7%)
3.8%
(1.4% -6.1%)
3.4%
(1 .3% - 5.4%)
4.3%
(1 .6% - 6.9%)
3.1%
(1.1% -4.9%)
3.9%
(1 .5% - 6.4%)
5%
(1.9% -8.1%)
4.3%
(1 .6% - 6.9%)
3.9%
(1 .5% - 6.3%)
0.6%
(0.2% - 1 %)
4.2%
(1 .6% - 6.8%)
1 .9%
(0.7% - 3%)
12/35
5.2%
(2% - 8.3%)
4.2%
(1 .6% - 6.8%)
3.6%
(1 .3% - 5.8%)
3.8%
(1.4% -6.1%)
3.1%
(1.2% -5.1%)
3.4%
(1 .3% - 5.4%)
3.6%
(1 .3% - 5.8%)
2.6%
(1%-4.3%)
3.3%
(1 .2% - 5.3%)
4.3%
(1 .6% - 6.9%)
3.9%
(1 .5% - 6.3%)
3.4%
(1 .3% - 5.5%)
0.6%
(0.2% -1%)
3.5%
(1 .3% - 5.7%)
1 .9%
(0.7% - 3%)
13/30
6%
(2.3% - 9.6%)
4.8%
(1 .8% - 7.6%)
4.3%
(1 .6% - 6.8%)
4.4%
(1 .6% - 7%)
3.2%
(1 .2% - 5.2%)
2.3%
(0.8% - 3.7%)
4.3%
(1 .6% - 6.9%)
2%
(0.7% - 3.2%)
2.9%
(1.1% -4.7%)
4.2%
(1 .6% - 6.8%)
3.8%
(1.4% -6.1%)
3%
(1.1% -4.9%)
0%
(0% - 0%)
4.1%
(1 .5% - 6.6%)
1%
(0.4% - 1 .6%)
12/25
5.1%
(1.9% -8.1%)
3.2%
(1 .2% - 5.2%)
3%
(1.1% -4.8%)
3.8%
(1.4% -6.1%)
2%
(0.7% - 3.2%)
1.1%
(0.4% - 1 .9%)
3.6%
(1 .3% - 5.8%)
0.9%
(0.3% - 1 .5%)
1 .7%
(0.6% - 2.8%)
2.8%
(1%-4.5%)
2.4%
(0.9% - 3.9%)
1 .8%
(0.7% - 2.9%)
0%
(0% - 0%)
2.7%
(1%-4.3%)
0.1%
(0%-0.1%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-60
-------
Table E-60. Estimated Percent of Total Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient
PM2.5 Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2007 PM2 s Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from
1979 -1983
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2 5
Concentrations
8%
(3.1% -12.8%)
6.9%
(2.6% -10.9%)
8.3%
(3.2% -13.2%)
4.8%
(1 .8% - 7.7%)
6.9%
(2.6% -11%)
9.7%
(3.7% -15.3%)
6.3%
(2.4% -10%)
7.4%
(2.8% - 1 1 .8%)
6.8%
(2.6% -10.8%)
6.4%
(2.4% -10.3%)
3.6%
(1 .3% - 5.7%)
7.7%
(2.9% -12.2%)
4.8%
(1 .8% - 7.7%)
7.2%
(2.7% - 1 1 .4%)
3.3%
(1 .3% - 5.4%)
15/353
7.1%
(2.7% - 1 1 .4%)
6.3%
(2.4% -10%)
5.9%
(2.2% - 9.4%)
4.8%
(1 .8% - 7.7%)
4.8%
(1 .8% - 7.8%)
3.6%
(1 .3% - 5.7%)
5.7%
(2.2% - 9.2%)
3.2%
(1 .2% - 5.2%)
5%
(1 .9% - 7.9%)
5.6%
(2.1% -9%)
3.6%
(1 .3% - 5.7%)
5%
(1 .9% - 8%)
1 .5%
(0.6% - 2.5%)
6%
(2.3% - 9.6%)
1 .9%
(0.7% -3.1%)
14/35
6.4%
(2.4% -10.1%)
5.7%
(2.1% -9.1%)
5.2%
(1 .9% - 8.3%)
4.8%
(1 .8% - 7.7%)
4.8%
(1 .8% - 7.7%)
3.6%
(1 .3% - 5.7%)
5.1%
(1.9% -8.1%)
3.2%
(1 .2% - 5.2%)
5%
(1 .9% - 7.9%)
5.6%
(2.1% -9%)
3.6%
(1 .3% - 5.7%)
5%
(1 .9% - 8%)
1 .5%
(0.6% - 2.5%)
5.4%
(2% - 8.7%)
1 .9%
(0.7% -3.1%)
13/35
5.6%
(2.1% -8.9%)
5%
(1 .9% - 7.9%)
4.5%
(1 .7% - 7.2%)
4.8%
(1 .8% - 7.7%)
4.1%
(1 .5% - 6.6%)
3.6%
(1 .3% - 5.7%)
4.4%
(1 .6% - 7%)
3.2%
(1 .2% - 5.2%)
4.7%
(1 .8% - 7.6%)
5%
(1 .9% - 8%)
3.6%
(1 .3% - 5.7%)
4.6%
(1 .7% - 7.4%)
1 .5%
(0.6% - 2.5%)
4.7%
(1 .8% - 7.5%)
1 .9%
(0.7% -3.1%)
12/35
4.8%
(1 .8% - 7.7%)
4.2%
(1 .6% - 6.8%)
3.8%
(1.4% -6.1%)
4.2%
(1 .6% - 6.8%)
3.4%
(1 .3% - 5.5%)
3.6%
(1 .3% - 5.7%)
3.7%
(1 .4% - 6%)
2.8%
(1%-4.5%)
4%
(1 .5% - 6.4%)
4.2%
(1 .6% - 6.8%)
3.2%
(1.2% -5.1%)
4%
(1 .5% - 6.5%)
1 .5%
(0.6% - 2.5%)
4%
(1 .5% - 6.4%)
1 .9%
(0.7% -3.1%)
13/30
5.6%
(2.1% -8.9%)
4.8%
(1 .8% - 7.6%)
4.5%
(1 .7% - 7.2%)
4.8%
(1 .8% - 7.7%)
3.5%
(1 .3% - 5.7%)
2.4%
(0.9% - 3.9%)
4.4%
(1 .6% - 7%)
2.1%
(0.8% - 3.4%)
3.6%
(1 .4% - 5.8%)
4.2%
(1 .6% - 6.7%)
3.1%
(1.1% -4.9%)
3.6%
(1 .4% - 5.9%)
0.7%
(0.3% -1.1%)
4.6%
(1 .7% - 7.3%)
1%
(0.4% - 1 .7%)
12/25
4.7%
(1 .8% - 7.5%)
3.2%
(1 .2% - 5.2%)
3.1%
(1.2% -5.1%)
4.2%
(1 .6% - 6.8%)
2.2%
(0.8% - 3.6%)
1 .3%
(0.5% -2.1%)
3.7%
(1 .4% - 6%)
1%
(0.4% - 1 .7%)
2.3%
(0.8% - 3.7%)
2.7%
(1<>/0-4.4o/0)
1 .8%
(0.7% - 2.9%)
2.3%
(0.9% - 3.7%)
0%
(0% - 0%)
3.1%
(1 .2% - 5%)
0.1%
(0.1% -0.2%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-61
-------
Table E-61. Percent Reduction from the Current Standards: Estimated Annual Incidence of Lung Cancer Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2005 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure
to PM2.5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-12% --13%)
-9%
(-8o/0 - -go/0)
-42%
(-41 % - -43%)
0%
(0% - 0%)
-37%
(-36% - -38%)
-182%
(-177% --188%)
-9%
(-9% - -9%)
-121%
(-11 7% --124%)
-35%
(-34% - -36%)
-14%
(-14% --15%)
0%
(0% - 0%)
-52%
(-50% - -53%)
-252%
(-249% - -256%)
-17%
(-17% --18%)
-53%
(-52% - -54%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
1435
11%
(10% -11%)
8%
(8% - 9%)
12%
(11% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(11% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
21%
(21 % - 22%)
19%
(19% -19%)
23%
(23% - 23%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
23%
(23% - 24%)
0%
(0% - 0%)
4%
(4% - 5%)
11%
(10% -11%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(20% - 21 %)
0%
(0% - 0%)
12/35
32%
(31% -32%)
30%
(29% - 30%)
35%
(34% - 35%)
11%
(1 1 % - 1 1 %)
25%
(25% - 26%)
0%
(0% - 0%)
35%
(35% - 35%)
12%
(12% -12%)
18%
(18% -18%)
23%
(23% - 24%)
10%
(10% -10%)
18%
(17% -18%)
0%
(0% - 0%)
31%
(31% -32%)
0%
(0% - 0%)
13/30
21%
(21 % - 22%)
22%
(21 % - 22%)
23%
(23% - 23%)
0%
(0% - 0%)
23%
(23% - 24%)
34%
(33% - 34%)
23%
(23% - 24%)
31%
(31% -31%)
25%
(25% - 26%)
24%
(24% - 24%)
13%
(13% -13%)
25%
(25% - 25%)
64%
(64% - 64%)
22%
(22% - 23%)
33%
(33% - 34%)
12/25
34%
(33% - 34%)
44%
(44% - 45%)
46%
(45% - 46%)
11%
(1 1 % - 1 1 %)
47%
(47% - 48%)
68%
(68% - 68%)
35%
(35% - 35%)
62%
(62% - 63%)
51%
(50% -51%)
48%
(48% - 49%)
45%
(45% - 46%)
51%
(50% -51%)
1 00%
(100% -100%)
45%
(44% - 45%)
67%
(67% - 67%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-62
-------
Table E-62. Percent Reduction from the Current Standards: Estimated Annual Incidence of Lung Cancer Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2006 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure
to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-12% --13%)
-10%
(-9% - -1 0%)
-43%
(-42% - -45%)
0%
(0% - 0%)
-44%
(-43% - -45%)
-177%
(-172% --183%)
-9%
(-9% - -1 0%)
-136%
(-132% --139%)
-41 %
(-40% - -41 %)
-15%
(-15% --15%)
0%
(0% - 0%)
-60%
(-59% - -62%)
-434%
(-430% - -439%)
-20%
(-20% - -21 %)
-75%
(-75% - -76%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
1435
11%
(10% -11%)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
11%
(10% -11%)
0%
(0% - 0%)
13/35
21%
(21 % - 22%)
21%
(21 % - 21 %)
24%
(24% - 24%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
24%
(24% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(31% -32%)
33%
(32% - 33%)
36%
(36% - 36%)
13%
(13% -13%)
30%
(30% - 31 %)
0%
(0% - 0%)
36%
(36% - 36%)
13%
(13% -14%)
21%
(21% -21%)
24%
(24% - 25%)
10%
(10% -10%)
20%
(20% - 20%)
0%
(0% - 0%)
36%
(35% - 36%)
0%
(0% - 0%)
13/30
21%
(21 % - 22%)
24%
(24% - 24%)
24%
(24% - 24%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
24%
(24% - 24%)
35%
(35% - 35%)
29%
(29% - 30%)
25%
(25% - 26%)
13%
(13% -13%)
29%
(29% - 29%)
100%
(100% -100%)
25%
(25% - 26%)
48%
(48% - 48%)
12/25
34%
(33% - 34%)
48%
(48% - 49%)
47%
(47% - 48%)
13%
(13% -13%)
56%
(56% - 57%)
66%
(66% - 66%)
36%
(36% - 36%)
70%
(70% - 70%)
59%
(59% - 60%)
51%
(50% -51%)
45%
(44% - 45%)
58%
(58% - 59%)
1 00%
(100% -100%)
51%
(51 % - 52%)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-63
-------
Table E-63. Percent Reduction from the Current Standards: Estimated Annual Incidence of Lung Cancer Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2007 PM2.5 Concentrations: Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure
to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-13% --13%)
-10%
(-9% - -1 0%)
-43%
(-41 % - -44%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-172%
(-166% --177%)
-9%
(-9% - -1 0%)
-132%
(-128% --135%)
-37%
(-36% - -38%)
-15%
(-15% --15%)
0%
(0% - 0%)
-55%
(-53% - -56%)
-21 5%
(-21 2% --21 9%)
-19%
(-1 9% - -20%)
-73%
(-72% - -74%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
1435
11%
(1 1 % - 1 1 %)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
22%
(21 % - 22%)
21%
(21 % - 21 %)
23%
(23% - 24%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
24%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
6%
(6% - 7%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
12/35
33%
(32% - 33%)
33%
(32% - 33%)
35%
(35% - 36%)
12%
(12% -12%)
29%
(29% - 30%)
0%
(0% - 0%)
36%
(35% - 36%)
13%
(13% -13%)
19%
(19% -20%)
24%
(24% - 25%)
11%
(1 1 % - 1 1 %)
19%
(19% -19%)
0%
(0% - 0%)
34%
(34% - 34%)
0%
(0% - 0%)
13/30
22%
(21 % - 22%)
24%
(24% - 24%)
23%
(23% - 24%)
0%
(0% - 0%)
27%
(27% - 27%)
32%
(32% - 32%)
24%
(23% - 24%)
34%
(34% - 34%)
27%
(27% - 27%)
25%
(25% - 26%)
14%
(14% -14%)
27%
(26% - 27%)
55%
(55% - 55%)
24%
(24% - 25%)
46%
(46% - 46%)
12/25
34%
(34% - 35%)
48%
(48% - 49%)
46%
(46% - 47%)
12%
(12% -12%)
54%
(54% - 55%)
64%
(64% - 64%)
36%
(35% - 36%)
68%
(68% - 68%)
54%
(54% - 55%)
51%
(51% -51%)
50%
(50% - 50%)
54%
(53% - 54%)
1 00%
(100% -100%)
49%
(49% - 49%)
93%
(93% - 93%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-64
-------
Table E-64. Estimated Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in a
Recent Year (2005) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent Year and PM25
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
110
(49 - 1 67)
116
(52 - 1 76)
79
(35 - 1 20)
72
(32-110)
161
(72 - 244)
38
(17-57)
111
(49 - 1 69)
357
(159-541)
300
(133-457)
101
(45 - 1 54)
85
(37 -131)
115
(51 -175)
11
(5-18)
167
(74 - 252)
27
(12-41)
15/353
98
(44 - 1 49)
107
(48 - 1 63)
56
(25 - 86)
72
(32-110)
119
(52-181)
14
(6-21)
101
(45 - 1 55)
164
(71 - 253)
224
(99 - 343)
88
(39 - 1 35)
85
(37-131)
76
(34-117)
3
(1-5)
142
(63-217)
18
(8 - 27)
14/35
88
(39 - 1 34)
99
(44 - 1 50)
50
(22 - 77)
72
(32-110)
117
(52 - 1 79)
14
(6-21)
90
(39 - 1 38)
164
(71 - 253)
224
(99 - 343)
88
(39 - 1 35)
85
(37-131)
76
(34 - 1 1 7)
3
(1-5)
129
(57 - 1 97)
18
(8 - 27)
13/35
78
(34-119)
87
(38 - 1 33)
43
(19-67)
72
(32-110)
103
(45 - 1 58)
14
(6-21)
78
(34 - 1 20)
164
(71 - 253)
214
(94 - 329)
79
(35-121)
85
(37-131)
72
(32 - 1 1 0)
3
(1-5)
114
(50 - 1 75)
18
(8 - 27)
12/35
67
(30 - 1 03)
76
(33-116)
37
(16-57)
64
(28 - 98)
89
(39 - 1 37)
14
(6-21)
66
(29 - 1 02)
144
(63 - 223)
184
(80 - 283)
68
(30 - 1 05)
76
(33-118)
63
(28 - 97)
3
(1-5)
99
(43 - 1 52)
18
(8 - 27)
13/30
78
(34-119)
84
(37 - 1 29)
43
(19-67)
72
(32-110)
91
(40 - 1 40)
9
(4-14)
78
(34 - 1 20)
113
(49 - 1 76)
168
(73 - 259)
67
(30 - 1 04)
74
(32-115)
57
(25 - 89)
1
(1-2)
111
(49 - 1 70)
12
(5-18)
12/25
66
(29-101)
60
(26 - 93)
31
(13-48)
64
(28 - 98)
63
(27 - 98)
4
(2-7)
66
(29-102)
62
(27 - 96)
111
(48-172)
46
(20 - 71 )
47
(20 - 73)
38
(17-59)
0
(0-0)
79
(35-122)
6
(3-9)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-65
-------
Table E-65. Estimated Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in a
Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year and PM25
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
114
(51 -172)
95
(42 - 1 45)
75
(33-113)
54
(24 - 84)
118
(52 - 1 80)
39
(18-59)
107
(47 - 1 64)
316
(140-481)
234
(103-359)
91
(40 - 1 40)
90
(39 - 1 39)
92
(41 -140)
9
(4-14)
126
(56 - 1 93)
18
(8 - 28)
15/353
101
(45-154)
87
(38-133)
52
(23 - 80)
54
(24 - 84)
82
(36-127)
14
(6 - 22)
98
(43-150)
135
(59-210)
167
(73 - 258)
80
(35-122)
90
(39-139)
58
(25 - 89)
2
(1 -3)
105
(46-162)
10
(5-16)
14/35
91
(40-138)
79
(35-121)
46
(20-71)
54
(24 - 84)
81
(35-125)
14
(6 - 22)
87
(38-133)
135
(59 - 21 0)
167
(73 - 258)
80
(35-122)
90
(39-139)
58
(25 - 89)
2
(1-3)
94
(41 -145)
10
(5-16)
13/35
80
(35 - 1 23)
69
(30 - 1 06)
40
(18-62)
54
(24 - 84)
69
(30 - 1 07)
14
(6 - 22)
75
(33 - 1 1 6)
135
(59-210)
159
(69 - 245)
71
(31 -109)
90
(39 - 1 39)
54
(23 - 83)
2
(1-3)
81
(36 - 1 26)
10
(5-16)
12/35
69
(30 - 1 06)
59
(26-91)
34
(15-52)
47
(21 - 73)
58
(25 - 89)
14
(6 - 22)
63
(28 - 98)
117
(51 -182)
132
(58 - 205)
61
(26 - 93)
81
(35 - 1 25)
46
(20-71)
2
(1-3)
68
(30 - 1 06)
10
(5-16)
13/30
80
(35 - 1 23)
66
(29 - 1 02)
40
(18-62)
54
(24 - 84)
59
(26 - 92)
10
(4-15)
75
(33-116)
89
(38 - 1 38)
119
(52 - 1 84)
60
(26 - 92)
79
(34 - 1 22)
41
(18-64)
0
(0-0)
79
(34 - 1 22)
5
(2-9)
12/25
68
(30-104)
45
(20 - 70)
28
(12-43)
47
(21 - 73)
36
(16-56)
5
(2-8)
63
(28 - 98)
41
(18-64)
69
(30-107)
40
(17-61)
50
(22 - 78)
24
(10-38)
0
(0-0)
52
(22 - 80)
0
(0-1)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-66
-------
Table E-66. Estimated Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations in a
Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent Year and PM25
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
109
(49 - 1 66)
95
(42 - 1 45)
78
(35-119)
60
(27 - 93)
124
(55 - 1 89)
41
(18-62)
112
(49-171)
328
(1 45 - 499)
273
(121 -417)
91
(40 - 1 38)
77
(33-118)
103
(46 - 1 57)
13
(6-21)
138
(61 -210)
19
(8 - 30)
15/353
97
(43- 148)
87
(38- 133)
55
(24 - 85)
60
(27 - 93)
88
(38-135)
15
(7 - 24)
102
(45-157)
143
(62 - 222)
200
(88 - 308)
79
(35-121)
77
(33 - 1 1 8)
67
(29-103)
4
(2-7)
116
(51 -178)
11
(5-17)
14/35
87
(38-133)
79
(35-121)
49
(21-75)
60
(27 - 93)
86
(38-133)
15
(7 - 24)
90
(40-139)
143
(62 - 222)
200
(88 - 308)
79
(35-121)
77
(33 - 1 1 8)
67
(29-103)
4
(2-7)
105
(46-161)
11
(5-17)
13/35
76
(34 - 1 1 7)
69
(30- 106)
42
(19-65)
60
(27 - 93)
74
(32 - 1 1 5)
15
(7 - 24)
78
(34- 121)
143
(62 - 222)
191
(84 - 294)
70
(31 -108)
77
(33 - 1 1 8)
63
(28 - 97)
4
(2-7)
91
(40- 140)
11
(5-17)
12/35
66
(29-101)
59
(26-91)
36
(16-56)
53
(23 - 82)
62
(27 - 96)
15
(7 - 24)
66
(29-102)
124
(54-193)
162
(71 - 250)
60
(26 - 92)
68
(30-106)
55
(24 - 84)
4
(2-7)
77
(34-119)
11
(5-17)
13/30
76
(34-117)
66
(29 - 1 02)
42
(19-65)
60
(27 - 93)
64
(28 - 99)
11
(5-16)
78
(34-121)
95
(41 -148)
147
(64 - 227)
59
(26-91)
66
(29 - 1 02)
49
(22 - 76)
2
(1 -3)
88
(39 - 1 36)
6
(3-9)
12/25
64
(28 - 99)
45
(20 - 70)
30
(13-46)
53
(23 - 82)
40
(17-63)
6
(2-9)
66
(29 - 1 02)
46
(20 - 72)
92
(40 - 1 44)
39
(17-60)
38
(17-60)
31
(14-49)
0
(0-0)
60
(26 - 93)
1
(0-1)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-67
-------
Table E-67. Estimated Percent of Total Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient
PM2 s Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2005 PM2 s Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5
from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
12.4%
(5.5% -18.7%)
12.2%
(5.4% -18.4%)
12.4%
(5.5% -18.8%)
8.6%
(3.8% -13.2%)
12.8%
(5.7% -19.4%)
13.1%
(5. 8% -19.8%)
9.5%
(4.2% -14.5%)
1 1 .8%
(5.3% -18%)
11%
(4.9% -16.7%)
10.4%
(4.6% -15.8%)
6.2%
(2.7% - 9.5%)
12.3%
(5.5% -18.6%)
6.3%
(2.8% - 9.8%)
12.6%
(5.6% -19.1%)
7.1%
(3.1% -10.9%)
15/353
11%
(4.9% -16.7%)
1 1 .2%
(5% -17%)
8.8%
(3.9% -13.5%)
8.6%
(3.8% -13.2%)
9.4%
(4.2% -14.4%)
4.7%
(2.1% -7.3%)
8.7%
(3.8% -13.3%)
5.4%
(2.4% - 8.4%)
8.2%
(3.6% -12.5%)
9.1%
(4% -13.9%)
6.2%
(2.7% - 9.5%)
8.1%
(3.6% -12.5%)
1 .8%
(0.8% - 2.8%)
10.8%
(4.8% -16.4%)
4.7%
(2% - 7.2%)
14/35
9.9%
(4.4% -15.1%)
10.3%
(4.6% -15.7%)
7.8%
(3.4% -12%)
8.6%
(3.8% -13.2%)
9.3%
(4.1% -14.3%)
4.7%
(2.1% -7.3%)
7.7%
(3.4% - 1 1 .8%)
5.4%
(2.4% - 8.4%)
8.2%
(3.6% -12.5%)
9.1%
(4% -13.9%)
6.2%
(2.7% - 9.5%)
8.1%
(3.6% -12.5%)
1 .8%
(0.8% - 2.8%)
9.8%
(4.3% -14.9%)
4.7%
(2% - 7.2%)
13/35
8.7%
(3.8% -13.3%)
9.1%
(4% -13.9%)
6.8%
(3% -10.5%)
8.6%
(3.8% -13.2%)
8.2%
(3.6% -12.6%)
4.7%
(2.1% -7.3%)
6.7%
(2.9% -10.3%)
5.4%
(2.4% - 8.4%)
7.8%
(3.4% -12%)
8.1%
(3.6% -12.5%)
6.2%
(2.7% - 9.5%)
7.6%
(3.4% - 1 1 .7%)
1 .8%
(0.8% - 2.8%)
8.6%
(3.8% -13.2%)
4.7%
(2% - 7.2%)
12/35
7.5%
(3.3% - 1 1 .6%)
7.9%
(3.5% -12.1%)
5.8%
(2.5% - 8.9%)
7.7%
(3.4% - 1 1 .8%)
7.1%
(3.1% -10.9%)
4.7%
(2.1% -7.3%)
5.7%
(2.5% - 8.8%)
4.8%
(2.1% -7.4%)
6.7%
(2.9% -10.3%)
7%
(3.1% -10.8%)
5.5%
(2.4% - 8.6%)
6.7%
(2.9% -10.3%)
1 .8%
(0.8% - 2.8%)
7.5%
(3.3% - 1 1 .5%)
4.7%
(2% - 7.2%)
13/30
8.7%
(3.8% -13.3%)
8.8%
(3.9% -13.4%)
6.8%
(3% -10.5%)
8.6%
(3.8% -13.2%)
7.3%
(3.2% - 1 1 .2%)
3.1%
(1 .4% - 4.9%)
6.7%
(2.9% -10.3%)
3.8%
(1 .6% - 5.8%)
6.1%
(2.7% - 9.5%)
6.9%
(3% -10.7%)
5.4%
(2.3% - 8.3%)
6.1%
(2.7% - 9.4%)
0.6%
(0.3% -1%)
8.4%
(3.7% -12.9%)
3.1%
(1 .4% - 4.9%)
12/25
7.4%
(3.2% - 1 1 .3%)
6.3%
(2.8% - 9.7%)
4.8%
(2.1% -7.4%)
7.7%
(3.4% - 1 1 .8%)
5%
(2.2% - 7.8%)
1 .5%
(0.7% - 2.4%)
5.7%
(2.5% - 8.8%)
2%
(0.9% - 3.2%)
4%
(1 .8% - 6.3%)
4.7%
(2.1% -7.3%)
3.4%
(1 .5% - 5.3%)
4%
(1 .8% - 6.3%)
0%
(0% - 0%)
6%
(2.6% - 9.2%)
1 .5%
(0.7% - 2.4%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-68
-------
Table E-68. Estimated Percent of Total Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from
1999 - 2000
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2 5 Concentrations in a Recent
Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
12.4%
(5.5% -18.7%)
9.9%
(4.4% -15.1%)
1 1 .6%
(5.1% -17.6%)
6.4%
(2.8% - 9.8%)
9.4%
(4.1% -14.3%)
13.4%
(6% - 20.3%)
8.9%
(3.9% - 1 3.6%)
10.4%
(4.6% - 1 5.9%)
8.5%
(3.7% -13%)
9.4%
(4.2% - 1 4.4%)
6.3%
(2.7% - 9.7%)
9.8%
(4.3% -15%)
5%
(2.2% - 7.8%)
9.5%
(4.2% -14.5%)
4.7%
(2.1% -7.3%)
15/353
11%
(4.9% -16.7%)
9.1%
(4% -13.9%)
8.1%
(3.6% - 1 2.4%)
6.4%
(2.8% - 9.8%)
6.5%
(2.9% -10.1%)
4.9%
(2.1% -7.6%)
8.1%
(3.6% - 1 2.5%)
4.5%
(1 .9% - 6.9%)
6.1%
(2.6% - 9.3%)
8.2%
(3.6% - 1 2.6%)
6.3%
(2.7% - 9.7%)
6.2%
(2.7% - 9.5%)
0.9%
(0.4% - 1 .5%)
7.9%
(3.5% -12.2%)
2.7%
(1 .2% - 4.2%)
14/35
9.9%
(4.4% -15%)
8.2%
(3.6% -12.6%)
7.2%
(3.1% -11%)
6.4%
(2.8% - 9.8%)
6.5%
(2.8% -10%)
4.9%
(2.1% -7.6%)
7.2%
(3.1% -11%)
4.5%
(1 .9% - 6.9%)
6.1%
(2.6% - 9.3%)
8.2%
(3.6% -12.6%)
6.3%
(2.7% - 9.7%)
6.2%
(2.7% - 9.5%)
0.9%
(0.4% - 1 .5%)
7.1%
(3.1% -10.9%)
2.7%
(1 .2% - 4.2%)
13/35
8.7%
(3.8% -13.3%)
7.2%
(3.2% -11.1%)
6.2%
(2.7% - 9.6%)
6.4%
(2.8% - 9.8%)
5.5%
(2.4% - 8.5%)
4.9%
(2.1% -7.6%)
6.2%
(2.7% - 9.6%)
4.5%
(1 .9% - 6.9%)
5.7%
(2.5% - 8.9%)
7.3%
(3.2% - 1 1 .2%)
6.3%
(2.7% - 9.7%)
5.7%
(2.5% - 8.9%)
0.9%
(0.4% - 1 .5%)
6.1%
(2.7% - 9.4%)
2.7%
(1 .2% - 4.2%)
12/35
7.5%
(3.3% - 1 1 .6%)
6.1%
(2.7% - 9.5%)
5.2%
(2.3% -8.1%)
5.5%
(2.4% - 8.6%)
4.6%
(2% -7.1%)
4.9%
(2.1% -7.6%)
5.2%
(2.3% -8.1%)
3.9%
(1 .7% - 6%)
4.8%
(2.1% -7.4%)
6.2%
(2.7% - 9.6%)
5.7%
(2.5% - 8.7%)
4.9%
(2.1% -7.6%)
0.9%
(0.4% - 1 .5%)
5.1%
(2.2% - 7.9%)
2.7%
(1 .2% - 4.2%)
13/30
8.7%
(3.8% -13.3%)
6.9%
(3% -10.6%)
6.2%
(2.7% - 9.6%)
6.4%
(2.8% - 9.8%)
4.7%
(2.1% -7.3%)
3.3%
(1.4% -5.1%)
6.2%
(2.7% - 9.6%)
2.9%
(1 .3% - 4.5%)
4.3%
(1 .9% - 6.7%)
6.2%
(2.7% - 9.5%)
5.5%
(2.4% - 8.5%)
4.4%
(1 .9% - 6.8%)
0%
(0% - 0%)
5.9%
(2.6% - 9.2%)
1 .4%
(0.6% - 2.2%)
12/25
7.4%
(3.2% - 1 1 .3%)
4.7%
(2% - 7.3%)
4.3%
(1 .9% - 6.7%)
5.5%
(2.4% - 8.6%)
2.9%
(1 .2% - 4.5%)
1 .7%
(0.7% - 2.6%)
5.2%
(2.3% -8.1%)
1 .3%
(0.6% -2.1%)
2.5%
(1.1% -3.9%)
4.1%
(1 .8% - 6.3%)
3.5%
(1 .5% - 5.4%)
2.6%
(1.1% -4%)
0%
(0% - 0%)
3.9%
(1 .7% - 6%)
0.1%
(0% - 0.2%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.,
Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
uncertainty surrounding the PM coefficient.
E-69
-------
Table E-69. Estimated Percent of Total Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2007 PM2.5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.S from
1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM25 Concentrations in a Recent
Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
1 1 .6%
(5. 2% -17.6%)
9.9%
(4. 4% -15.1%)
12%
(5. 3% -18.2%)
7%
(3% -10.7%)
9.9%
(4. 4% -15. 2%)
13.9%
(6.2% - 20.9%)
9.1%
(4% -13.9%)
10.7%
(4. 8% -16.3%)
9.8%
(4.3% -15%)
9.3%
(4.1% -14.2%)
5.2%
(2.3% - 8%)
11.1%
(4. 9% -16. 8%)
7%
(3.1% -10.7%)
10.4%
(4. 6% -15.8%)
4.9%
(2.1% -7.6%)
15/353
10.3%
(4.6% -15.7%)
9.1%
(4% -13.9%)
8.5%
(3.7% -13%)
7%
(3% -10. 7%)
7%
(3.1% -10.8%)
5.2%
(2.3% - 8%)
8.3%
(3.7% -12.8%)
4.7%
(2% - 7.3%)
7.2%
(3.2% -11.1%)
8.1%
(3.6% -12.5%)
5.2%
(2.3% - 8%)
7.2%
(3.2% -11.1%)
2.2%
(1%-3.5%)
8.7%
(3.8% -13.4%)
2.8%
(1 .2% - 4.4%)
14/35
9.2%
(4.1% -14.1%)
8.2%
(3.6% -12.6%)
7.5%
(3.3% - 1 1 .5%)
7%
(3% -10.7%)
6.9%
(3% -10.7%)
5.2%
(2.3% - 8%)
7.4%
(3.2% - 1 1 .3%)
4.7%
(2% - 7.3%)
7.2%
(3.2% -11.1%)
8.1%
(3.6% -12.5%)
5.2%
(2.3% - 8%)
7.2%
(3.2% -11.1%)
2.2%
(1%-3.5%)
7.9%
(3.5% -12.1%)
2.8%
(1 .2% - 4.4%)
13/35
8.1%
(3.6% -12.4%)
7.2%
(3.2% -11.1%)
6.5%
(2.8% -10%)
7%
(3% -10.7%)
6%
(2.6% - 9.2%)
5.2%
(2.3% - 8%)
6.4%
(2.8% - 9.8%)
4.7%
(2% - 7.3%)
6.9%
(3% -10.6%)
7.2%
(3.2% -11.1%)
5.2%
(2.3% - 8%)
6.7%
(3% -10.4%)
2.2%
(1%-3.5%)
6.8%
(3% -10.5%)
2.8%
(1 .2% - 4.4%)
12/35
7%
(3.1% -10.7%)
6.1%
(2.7% - 9.5%)
5.5%
(2.4% - 8.5%)
6.1%
(2.7% - 9.4%)
5%
(2.2% - 7.7%)
5.2%
(2.3% - 8%)
5.4%
(2.4% - 8.3%)
4.1%
(1 .8% - 6.3%)
5.8%
(2.5% - 9%)
6.2%
(2.7% - 9.5%)
4.6%
(2% -7.1%)
5.9%
(2.6% -9.1%)
2.2%
(1 % - 3.5%)
5.8%
(2.5% - 8.9%)
2.8%
(1 .2% - 4.4%)
13/30
8.1%
(3.6% -12.4%)
6.9%
(3% -10.6%)
6.5%
(2.8% -10%)
7%
(3% -10. 7%)
5.1%
(2.2% - 8%)
3.5%
(1 .5% - 5.5%)
6.4%
(2.8% - 9.8%)
3.1%
(1 .3% - 4.8%)
5.3%
(2.3% - 8.2%)
6.1%
(2.7% - 9.4%)
4.4%
(1 .9% - 6.9%)
5.3%
(2.3% - 8.2%)
1%
(0.4% - 1 .6%)
6.6%
(2.9% -10.2%)
1 .5%
(0.7% - 2.4%)
12/25
6.8%
(3% -10. 5%)
4.7%
(2% - 7.3%)
4.6%
(2% -7.1%)
6.1%
(2.7% - 9.4%)
3.2%
(1 .4% - 5%)
1 .9%
(0.8% - 2.9%)
5.4%
(2.4% - 8.3%)
1 .5%
(0.6% - 2.4%)
3.3%
(1 .4% - 5.2%)
4%
(1 .7% - 6.2%)
2.6%
(1.1% -4.1%)
3.4%
(1 .5% - 5.2%)
0%
(0% - 0%)
4.5%
(2% - 6.9%)
0.2%
(0.1% -0.3%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-70
-------
Table E-70. Percent Reduction from the Current Standards: Estimated Annual Incidence of Lung Cancer Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2005 PM2.5 Concentrations ~ Estimates Based
on Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure
to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-12% --13%)
-8%
(-8o/0 - -go/0)
-41 %
(-39% - -43%)
0%
(0% - 0%)
-36%
(-35% - -37%)
-178%
(-171% --185%)
-9%
(-9% - -9%)
-1 1 8%
(-11 4% --122%)
-34%
(-33% - -35%)
-14%
(-14% --14%)
0%
(0% - 0%)
-51 %
(-49% - -53%)
-250%
(-245% - -254%)
-17%
(-16% --18%)
-52%
(-51 % - -53%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -11%)
8%
(8% - 8%)
11%
(11% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(11% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
9%
(9% - 9%)
0%
(0% - 0%)
13/35
21%
(20% -21%)
19%
(18% -19%)
23%
(22% - 23%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
4%
(4% - 5%)
10%
(10% -11%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(19% -20%)
0%
(0% - 0%)
12/35
32%
(31 % - 32%)
29%
(29% - 30%)
34%
(34% - 35%)
11%
(11% -11%)
25%
(25% - 26%)
0%
(0% - 0%)
35%
(34% - 35%)
12%
(12% -12%)
18%
(18% -18%)
23%
(22% - 23%)
10%
(10% -10%)
17%
(17% -18%)
0%
(0% - 0%)
31%
(30% -31%)
0%
(0% - 0%)
13/30
21%
(20% -21%)
22%
(21% -22%)
23%
(22% - 23%)
0%
(0% - 0%)
23%
(23% - 24%)
34%
(33% - 34%)
23%
(23% - 23%)
31%
(31% -31%)
25%
(25% - 25%)
24%
(23% - 24%)
13%
(13% -13%)
25%
(24% - 25%)
64%
(64% - 64%)
22%
(21% -22%)
33%
(33% - 34%)
12/25
33%
(32% - 34%)
44%
(43% - 45%)
45%
(45% - 46%)
11%
(11% -11%)
47%
(46% - 48%)
68%
(67% - 68%)
35%
(34% - 35%)
62%
(62% - 63%)
50%
(50% - 51 %)
48%
(47% - 49%)
45%
(45% - 45%)
50%
(50% - 51 %)
1 00%
(100% -100%)
44%
(44% - 45%)
67%
(67% - 67%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-71
-------
Table E-71. Percent Reduction from the Current Standards: Estimated Annual Incidence of Lung Cancer Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2006 PM2.5 Concentrations ~ Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure
to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-12%
(-12% --13%)
-9%
(-9% - -1 0%)
-43%
(-41 % - -44%)
0%
(0% - 0%)
-43%
(-42% - -45%)
-173%
(-167% --181%)
-9%
(-9% - -1 0%)
-133%
(-129% --137%)
-40%
(-39% - -41 %)
-15%
(-14% --15%)
0%
(0% - 0%)
-59%
(-58% - -61 %)
-431 %
(-425% - -437%)
-20%
(-19% --20%)
-75%
(-74% - -76%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
10%
(10% -11%)
9%
(9% - 9%)
12%
(12% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(12% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -11%)
0%
(0% - 0%)
13/35
21%
(20% -21%)
21%
(20% -21%)
24%
(23% - 24%)
0%
(0% - 0%)
16%
(15% -16%)
0%
(0% - 0%)
24%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
7%
(7% - 7%)
0%
(0% - 0%)
23%
(22% - 23%)
0%
(0% - 0%)
12/35
32%
(31 % - 32%)
32%
(32% - 33%)
36%
(35% - 36%)
13%
(12% -13%)
30%
(30% - 30%)
0%
(0% - 0%)
36%
(35% - 36%)
13%
(13% -14%)
21%
(21% -21%)
24%
(24% - 25%)
10%
(10% -10%)
20%
(20% - 20%)
0%
(0% - 0%)
35%
(35% - 36%)
0%
(0% - 0%)
13/30
21%
(20% -21%)
24%
(23% - 24%)
24%
(23% - 24%)
0%
(0% - 0%)
28%
(27% - 28%)
33%
(32% - 33%)
24%
(23% - 24%)
35%
(34% - 35%)
29%
(29% - 30%)
25%
(24% - 25%)
13%
(12% -13%)
29%
(28% - 29%)
100%
(100% -100%)
25%
(25% - 26%)
48%
(47% - 48%)
12/25
33%
(32% - 34%)
48%
(47% - 49%)
47%
(46% - 48%)
13%
(12% -13%)
56%
(56% - 57%)
66%
(66% - 66%)
36%
(35% - 36%)
70%
(70% - 70%)
59%
(59% - 59%)
50%
(50% - 51 %)
45%
(44% - 45%)
58%
(57% - 58%)
1 00%
(100% -100%)
51%
(50% - 51 %)
96%
(96% - 96%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-72
-------
Table E-72. Percent Reduction from the Current Standards: Estimated Annual Incidence of Lung Cancer Mortality Associated with
Long-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2007 PM2.5 Concentrations ~ Estimates Based on
Krewski et al. (2009), Using Ambient PM2 5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure
to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -1 0%)
-42%
(-40% - -43%)
0%
(0% - 0%)
-42%
(-41 % - -43%)
-168%
(-161% --175%)
-9%
(-9% - -9%)
-129%
(-125% --134%)
-36%
(-35% - -37%)
-15%
(-14% --15%)
0%
(0% - 0%)
-54%
(-52% - -55%)
-21 3%
(-209% --21 7%)
-19%
(-18% --19%)
-72%
(-71 % - -74%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
11%
(10% -11%)
9%
(9% - 9%)
12%
(11% -12%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
12%
(11% -12%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
10%
(10% -10%)
0%
(0% - 0%)
13/35
21%
(21 % - 22%)
21%
(20% -21%)
23%
(23% - 24%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
23%
(23% - 24%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
6%
(6% - 7%)
0%
(0% - 0%)
22%
(21 % - 22%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
32%
(32% - 33%)
35%
(34% - 36%)
12%
(12% -12%)
29%
(29% - 29%)
0%
(0% - 0%)
35%
(35% - 36%)
13%
(13% -13%)
19%
(19% -20%)
24%
(24% - 25%)
11%
(11% -11%)
19%
(18% -19%)
0%
(0% - 0%)
34%
(33% - 34%)
0%
(0% - 0%)
13/30
21%
(21 % - 22%)
24%
(23% - 24%)
23%
(23% - 24%)
0%
(0% - 0%)
27%
(26% - 27%)
32%
(31 % - 32%)
23%
(23% - 24%)
34%
(33% - 34%)
27%
(26% - 27%)
25%
(25% - 26%)
14%
(14% -14%)
26%
(26% - 27%)
55%
(55% - 55%)
24%
(24% - 25%)
46%
(46% - 46%)
12/25
34%
(33% - 35%)
48%
(47% - 49%)
46%
(45% - 47%)
12%
(12% -12%)
54%
(54% - 55%)
64%
(64% - 64%)
35%
(35% - 36%)
68%
(68% - 68%)
54%
(53% - 54%)
51%
(50% - 51 %)
50%
(49% - 50%)
53%
(53% - 54%)
1 00%
(100% -100%)
49%
(48% - 49%)
93%
(93% - 93%)
1Based on follow-up through 2000, using models with 44 individual and
2Numbers rounded to the nearest percent. Numbers in parentheses are
''The current primary PM25 standards include an annual standard set at
7 ecological covariates (see Table 33 in Krewski et al., 2009).
95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
15 ug/m3 and a daily standard set at 35 ug/m3.
E-73
-------
Table E-73. Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2005) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 5 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
193
(37 - 347)
271
(110-430)
44
(-68 - 1 54)
156
(37 - 273)
181
(-32 - 390)
79
(11 -145)
227
(46 - 405)
129
(-185-441)
939
(552 - 1 323)
234
(86 - 380)
242
(40 - 442)
224
(66 - 380)
48
(10-85)
290
(84 - 494)
59
(10-107)
15/353
177
(34-319)
256
(104-406)
34
(-53 -121)
156
(37 - 273)
147
(-26-317)
44
(6-81)
214
(44 - 383)
81
(-117-278)
781
(459-1102)
216
(79 - 350)
242
(40 - 442)
159
(47 - 270)
30
(6 - 54)
260
(75 - 443)
48
(8 - 87)
14/35
164
(31 - 295)
242
(98 - 384)
32
(-49-112)
156
(37 - 273)
146
(-26-315)
44
(6-81)
198
(40 - 354)
81
(-117-278)
781
(459-1102)
216
(79 - 350)
242
(40 - 442)
159
(47 - 270)
30
(6 - 54)
244
(71 -416)
48
(8 - 87)
13/35
151
(29 - 272)
224
(91 - 356)
29
(-45-103)
156
(37 - 273)
135
(-24 - 291 )
44
(6-81)
182
(37 - 326)
81
(-117-278)
761
(447 - 1 073)
202
(74 - 328)
242
(40 - 442)
155
(45 - 263)
30
(6 - 54)
226
(65 - 385)
48
(8 - 87)
12/35
137
(26 - 248)
206
(83 - 327)
27
(-41 - 94)
145
(35 - 253)
124
(-22 - 267)
44
(6-81)
166
(34 - 297)
77
(-110-263)
700
(41 1 - 987)
185
(68-301)
230
(38 - 420)
147
(43 - 249)
30
(6 - 54)
207
(60 - 354)
48
(8 - 87)
13/30
151
(29 - 272)
219
(89 - 348)
29
(-45-103)
156
(37 - 273)
125
(-22 - 271 )
37
(5 - 69)
182
(37 - 326)
69
(-1 00 - 238)
668
(392 - 943)
184
(68 - 300)
227
(38-414)
136
(40-231)
26
(5 - 46)
222
(64 - 379)
41
(7 - 74)
12/25
135
(26 - 244)
182
(74 - 289)
24
(-38 - 85)
145
(35 - 253)
104
(-18-225)
31
(4 - 57)
166
(34 - 297)
58
(-82 - 1 97)
555
(325 - 783)
153
(56 - 249)
188
(31 - 344)
112
(33 -191)
21
(4 - 38)
184
(53-315)
34
(6 - 62)
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-74
-------
Table E-74. Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 5 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
196
(37 - 353)
237
(96 - 376)
42
(-66-148)
130
(31 - 228)
145
(-25 - 31 4)
84
(12-155)
221
(45 - 395)
119
(-1 71 - 407)
807
(474-1137)
222
(82 - 359)
254
(42 - 463)
194
(57 - 329)
44
(9 - 78)
240
(69 - 409)
50
(9 - 90)
15/353
180
(34 - 324)
224
(91 - 356)
33
(-51 -116)
130
(31 - 228)
118
(-21 - 255)
47
(7 - 86)
208
(42 - 373)
75
(-108-257)
671
(394 - 946)
204
(75 - 331 )
254
(42 - 463)
136
(40 - 232)
27
(6 - 49)
215
(62 - 367)
40
(7 - 73)
14/35
166
(32 - 300)
212
(86 - 336)
30
(-47-108)
130
(31 - 228)
117
(-20 - 253)
47
(7 - 86)
193
(39 - 345)
75
(-1 08 - 257)
671
(394 - 946)
204
(75-331)
254
(42 - 463)
136
(40 - 232)
27
(6 - 49)
202
(58 - 345)
40
(7 - 73)
13/35
153
(29 - 276)
196
(79 -311)
28
(-44 - 99)
130
(31 - 228)
108
(-1 9 - 234)
47
(7 - 86)
177
(36-317)
75
(-108-257)
654
(383 - 922)
191
(70-310)
254
(42 - 463)
133
(39 - 226)
27
(6 - 49)
187
(54-319)
40
(7 - 73)
12/35
139
(26-251)
180
(73 - 286)
26
(-40 - 90)
121
(29 - 21 2)
99
(-17-215)
47
(7 - 86)
162
(33 - 289)
71
(-1 01 - 242)
601
(352 - 847)
175
(65 - 285)
241
(40 - 440)
126
(37 - 21 5)
27
(6 - 49)
171
(49 - 293)
40
(7 - 73)
13/30
153
(29 - 276)
192
(78 - 305)
28
(-44 - 99)
130
(31 - 228)
101
(-18-218)
40
(6 - 74)
177
(36-317)
64
(-92-219)
574
(336 - 809)
174
(64 - 283)
238
(39 - 434)
116
(34-198)
23
(5 - 42)
184
(53-314)
34
(6 - 62)
12/25
137
(26 - 248)
159
(64 - 253)
23
(-36 - 82)
121
(29 - 21 2)
83
(-15-181)
33
(5-61)
162
(33 - 289)
53
(-76-182)
476
(279 - 672)
145
(53 - 235)
198
(33-361)
96
(28-164)
19
(4 - 35)
152
(44 - 260)
28
(5 - 52)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-75
-------
Table E-75. Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 5 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
193
(37 - 348)
240
(97 - 380)
44
(-68-154)
139
(33 - 243)
150
(-26 - 323)
87
(12-160)
224
(46 - 401 )
121
(-1 74 - 41 5)
882
(518-1243)
226
(83 - 367)
242
(40 - 442)
204
(60 - 346)
54
(11 -96)
251
(73 - 428)
52
(9 - 94)
15/353
177
(34-319)
227
(92 - 360)
34
(-53 - 1 20)
139
(33 - 243)
121
(-21 - 262)
48
(7 - 89)
212
(43 - 378)
77
(-110-262)
734
(431 -1035)
208
(77 - 338)
242
(40 - 442)
143
(42 - 244)
34
(7-61)
225
(65 - 384)
42
(7 - 76)
14/35
164
(31 - 296)
214
(87 - 340)
32
(-49 -111)
139
(33 - 243)
120
(-21 -261)
48
(7 - 89)
196
(40 - 350)
77
(-110-262)
734
(431 -1035)
208
(77 - 338)
242
(40 - 442)
143
(42 - 244)
34
(7-61)
211
(61 - 360)
42
(7 - 76)
13/35
151
(29 - 272)
198
(80-315)
29
(-45 - 1 02)
139
(33 - 243)
111
(-19-241)
48
(7 - 89)
180
(37 - 322)
77
(-110-262)
715
(419-1 008)
195
(72-316)
242
(40 - 442)
140
(41 - 237)
34
(7-61)
195
(56 - 333)
42
(7 - 76)
12/35
137
(26 - 248)
182
(74 - 289)
26
(-41 - 93)
129
(31 - 225)
102
(-18-221)
48
(7 - 89)
164
(33 - 294)
72
(-1 04 - 247)
657
(385 - 927)
179
(66-291)
230
(38 - 420)
133
(39 - 226)
34
(7-61)
179
(52 - 306)
42
(7 - 76)
13/30
151
(29 - 272)
194
(79 - 308)
29
(-45 - 1 02)
139
(33 - 243)
104
(-1 8 - 224)
41
(6 - 76)
180
(37 - 322)
65
(-94 - 224)
627
(368 - 885)
178
(66 - 289)
227
(38-414)
122
(36 - 208)
29
(6 - 52)
192
(55 - 328)
36
(6 - 65)
12/25
135
(26 - 244)
161
(65 - 256)
24
(-37 - 85)
129
(31 - 225)
86
(-15-186)
34
(5 - 63)
164
(33 - 294)
54
(-78-186)
521
(305 - 735)
148
(54 - 240)
188
(31 - 344)
102
(30-173)
24
(5 - 43)
160
(46 - 272)
30
(5 - 54)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-76
-------
Table E-76. Estimated Percent of Total Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM25 Concentrations in a Recent
Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2 5
Concentrations
1 .3%
(0.3% - 2.4%)
2%
(0.8% - 3.2%)
0.5%
(-0.7% - 1 .6%)
1 .2%
(0.3% - 2.2%)
1%
(-0.2% - 2.3%)
1 .5%
(0.2% - 2.7%)
1 .3%
(0.3% - 2.3%)
0.2%
(-0.3% - 0.8%)
1 .8%
(1.1% -2. 6%)
1 .7%
(0.6% - 2.7%)
1.1%
(0.2% - 2%)
1 .7%
(0.5% - 2.8%)
1%
(0.2% - 1 .8%)
1 .6%
(0.5% - 2.7%)
1 .2%
(0.2% - 2.2%)
15/353
1 .2%
(0.2% - 2.2%)
1 .9%
(0.8% - 3%)
0.4%
(-0.6% - 1 .3%)
1 .2%
(0.3% - 2.2%)
0.8%
(-0.1% -1.8%)
0.8%
(0.1% -1.5%)
1 .2%
(0.2% -2.1%)
0.1%
(-0.2% - 0.5%)
1 .5%
(0.9% -2.1%)
1 .5%
(0.6% - 2.5%)
1.1%
(0.2% - 2%)
1 .2%
(0.3% - 2%)
0.6%
(0.1% -1.2%)
1 .4%
(0.4% - 2.4%)
1%
(0.2% - 1 .8%)
14/35
1.1%
(0.2% - 2%)
1 .8%
(0.7% - 2.8%)
0.3%
(-0.5% - 1 .2%)
1 .2%
(0.3% - 2.2%)
0.8%
(-0.1% -1.8%)
0.8%
(0.1% -1.5%)
1.1%
(0.2% - 2%)
0.1%
(-0.2% - 0.5%)
1 .5%
(0.9% -2.1%)
1 .5%
(0.6% - 2.5%)
1.1%
(0.2% - 2%)
1 .2%
(0.3% - 2%)
0.6%
(0.1% -1.2%)
1 .3%
(0.4% - 2.3%)
1%
(0.2% - 1 .8%)
13/35
1%
(0.2% - 1 .9%)
1 .7%
(0.7% - 2.6%)
0.3%
(-0.5% -1.1%)
1 .2%
(0.3% - 2.2%)
0.8%
(-0.1% -1.7%)
0.8%
(0.1% -1.5%)
1%
(0.2% - 1 .8%)
0.1%
(-0.2% - 0.5%)
1 .5%
(0.9% -2.1%)
1 .4%
(0.5% - 2.3%)
1.1%
(0.2% - 2%)
1.1%
(0.3% - 1 .9%)
0.6%
(0.1% -1.2%)
1 .2%
(0.4% -2.1%)
1%
(0.2% - 1 .8%)
12/35
1%
(0.2% - 1 .7%)
1 .5%
(0.6% - 2.4%)
0.3%
(-0.4% - 1 %)
1 .2%
(0.3% - 2%)
0.7%
(-0.1% -1.5%)
0.8%
(0.1% -1.5%)
0.9%
(0.2% - 1 .7%)
0.1%
(-0.2% - 0.5%)
1 .4%
(0.8% - 1 .9%)
1 .3%
(0.5% -2.1%)
1.1%
(0.2% - 1 .9%)
1.1%
(0.3% - 1 .8%)
0.6%
(0.1% -1.2%)
1.1%
(0.3% - 1 .9%)
1%
(0.2% - 1 .8%)
13/30
1%
(0.2% - 1 .9%)
1 .6%
(0.7% - 2.6%)
0.3%
(-0.5% -1.1%)
1 .2%
(0.3% - 2.2%)
0.7%
(-0.1% -1.6%)
0.7%
(0.1% -1.3%)
1%
(0.2% - 1 .8%)
0.1%
(-0.2% - 0.4%)
1 .3%
(0.8% - 1 .8%)
1 .3%
(0.5% -2.1%)
1%
(0.2% - 1 .9%)
1%
(0.3% - 1 .7%)
0.6%
(0.1% -1%)
1 .2%
(0.4% -2.1%)
0.8%
(0.1% -1.5%)
12/25
0.9%
(0.2% - 1 .7%)
1 .3%
(0.5% -2.1%)
0.3%
(-0.4% - 0.9%)
1 .2%
(0.3% - 2%)
0.6%
(-0.1% -1.3%)
0.6%
(0.1% -1.1%)
0.9%
(0.2% - 1 .7%)
0.1%
(-0.1% -0.4%)
1.1%
(0.6% - 1 .5%)
1.1%
(0.4% - 1 .8%)
0.9%
(0.1% -1.6%)
0.8%
(0.2% - 1 .4%)
0.5%
(0.1% -0.8%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.3%)
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-77
-------
Table E-77. Estimated Percent of Total Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
1 .3%
(0.3% - 2.4%)
1 .7%
(0.7% - 2.8%)
0.4%
(-0.7% - 1 .6%)
1%
(0.2% - 1 .8%)
0.8%
(-0.1% -1.8%)
1 .5%
(0.2% - 2.8%)
1 .2%
(0.2% -2.1%)
0.2%
(-0.3% - 0.7%)
1 .6%
(0.9% - 2.2%)
1 .6%
(0.6% - 2.6%)
1.1%
(0.2% - 2%)
1 .4%
(0.4% - 2.5%)
0.9%
(0.2% - 1 .6%)
1 .3%
(0.4% - 2.2%)
1%
(0.2% - 1 .8%)
15/353
1 .2%
(0.2% - 2.2%)
1 .6%
(0.7% - 2.6%)
0.3%
(-0.5% - 1 .2%)
1%
(0.2% - 1 .8%)
0.7%
(-0.1% -1.5%)
0.8%
(0.1% -1.6%)
1.1%
(0.2% - 2%)
0.1%
(-0.2% - 0.5%)
1 .3%
(0.8% - 1 .8%)
1 .5%
(0.5% - 2.4%)
1.1%
(0.2% - 2%)
1%
(0.3% - 1 .7%)
0.6%
(0.1% -1%)
1 .2%
(0.3% - 2%)
0.8%
(0.1% -1.5%)
14/35
1.1%
(0.2% - 2%)
1 .6%
(0.6% - 2.5%)
0.3%
(-0.5% -1.1%)
1%
(0.2% - 1 .8%)
0.7%
(-0.1% -1.5%)
0.8%
(0.1% -1.6%)
1%
(0.2% - 1 .9%)
0.1%
(-0.2% - 0.5%)
1 .3%
(0.8% - 1 .8%)
1 .5%
(0.5% - 2.4%)
1.1%
(0.2% - 2%)
1%
(0.3% - 1 .7%)
0.6%
(0.1% -1%)
1.1%
(0.3% - 1 .9%)
0.8%
(0.1% -1.5%)
13/35
1%
(0.2% - 1 .9%)
1 .4%
(0.6% - 2.3%)
0.3%
(-0.5% - 1 %)
1%
(0.2% - 1 .8%)
0.6%
(-0.1% -1.4%)
0.8%
(0.1% -1.6%)
1%
(0.2% - 1 .7%)
0.1%
(-0.2% - 0.5%)
1 .3%
(0.7% - 1 .8%)
1 .4%
(0.5% - 2.2%)
1.1%
(0.2% - 2%)
1%
(0.3% - 1 .7%)
0.6%
(0.1% -1%)
1%
(0.3% - 1 .7%)
0.8%
(0.1% -1.5%)
12/35
0.9%
(0.2% - 1 .7%)
1 .3%
(0.5% -2.1%)
0.3%
(-0.4% - 0.9%)
0.9%
(0.2% - 1 .7%)
0.6%
(-0.1% -1.3%)
0.8%
(0.1% -1.6%)
0.9%
(0.2% - 1 .6%)
0.1%
(-0.2% - 0.4%)
1 .2%
(0.7% - 1 .6%)
1 .2%
(0.5% - 2%)
1.1%
(0.2% - 1 .9%)
0.9%
(0.3% - 1 .6%)
0.6%
(0.1% -1%)
0.9%
(0.3% - 1 .6%)
0.8%
(0.1% -1.5%)
13/30
1%
(0.2% - 1 .9%)
1 .4%
(0.6% - 2.2%)
0.3%
(-0.5% - 1 %)
1%
(0.2% - 1 .8%)
0.6%
(-0.1% -1.3%)
0.7%
(0.1% -1.3%)
1%
(0.2% - 1 .7%)
0.1%
(-0.2% - 0.4%)
1.1%
(0.6% - 1 .6%)
1 .2%
(0.5% - 2%)
1.1%
(0.2% - 1 .9%)
0.9%
(0.3% - 1 .5%)
0.5%
(0.1% -0.9%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.2%)
12/25
0.9%
(0.2% - 1 .7%)
1 .2%
(0.5% - 1 .9%)
0.2%
(-0.4% - 0.9%)
0.9%
(0.2% - 1 .7%)
0.5%
(-0.1% -1.1%)
0.6%
(0.1% -1.1%)
0.9%
(0.2% - 1 .6%)
0.1%
(-0.1% -0.3%)
0.9%
(0.5% - 1 .3%)
1%
(0.4% - 1 .7%)
0.9%
(0.1% -1.6%)
0.7%
(0.2% - 1 .2%)
0.4%
(0.1% -0.7%)
0.8%
(0.2% - 1 .4%)
0.6%
(0.1% -1%)
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-78
-------
Table E-78. Estimated Percent of Total Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
1 .3%
(0.2% - 2.3%)
1 .8%
(0.7% - 2.8%)
0.5%
(-0.7% - 1 .6%)
1.1%
(0.3% - 1 .9%)
0.9%
(-0.2% - 1 .9%)
1 .6%
(0.2% - 2.9%)
1 .2%
(0.2% -2.1%)
0.2%
(-0.3% - 0.7%)
1 .7%
(1%-2.4%)
1 .6%
(0.6% - 2.6%)
1%
(0.2% - 1 .9%)
1 .5%
(0.4% - 2.6%)
1.1%
(0.2% - 2%)
1 .4%
(0.4% - 2.4%)
1%
(0.2% - 1 .9%)
15/353
1 .2%
(0.2% -2.1%)
1 .7%
(0.7% - 2.6%)
0.4%
(-0.6% - 1 .2%)
1.1%
(0.3% - 1 .9%)
0.7%
(-0.1% -1.6%)
0.9%
(0.1% -1.6%)
1.1%
(0.2% - 2%)
0.1%
(-0.2% - 0.5%)
1 .4%
(0.8% - 2%)
1 .5%
(0.5% - 2.4%)
1%
(0.2% - 1 .9%)
1.1%
(0.3% - 1 .8%)
0.7%
(0.1% -1.3%)
1 .2%
(0.4% -2.1%)
0.8%
(0.1% -1.5%)
14/35
1.1%
(0.2% - 1 .9%)
1 .6%
(0.6% - 2.5%)
0.3%
(-0.5% - 1 .2%)
1.1%
(0.3% - 1 .9%)
0.7%
(-0.1% -1.5%)
0.9%
(0.1% -1.6%)
1%
(0.2% - 1 .9%)
0.1%
(-0.2% - 0.5%)
1 .4%
(0.8% - 2%)
1 .5%
(0.5% - 2.4%)
1%
(0.2% - 1 .9%)
1.1%
(0.3% - 1 .8%)
0.7%
(0.1% -1.3%)
1 .2%
(0.3% - 2%)
0.8%
(0.1% -1.5%)
13/35
1%
(0.2% - 1 .8%)
1 .5%
(0.6% - 2.3%)
0.3%
(-0.5% -1.1%)
1.1%
(0.3% - 1 .9%)
0.7%
(-0.1% -1.4%)
0.9%
(0.1% -1.6%)
1%
(0.2% - 1 .7%)
0.1%
(-0.2% - 0.5%)
1 .4%
(0.8% - 1 .9%)
1 .4%
(0.5% - 2.3%)
1%
(0.2% - 1 .9%)
1%
(0.3% - 1 .8%)
0.7%
(0.1% -1.3%)
1.1%
(0.3% - 1 .8%)
0.8%
(0.1% -1.5%)
12/35
0.9%
(0.2% - 1 .6%)
1 .3%
(0.5% -2.1%)
0.3%
(-0.4% - 1 %)
1%
(0.2% - 1 .8%)
0.6%
(-0.1% -1.3%)
0.9%
(0.1% -1.6%)
0.9%
(0.2% - 1 .6%)
0.1%
(-0.2% - 0.4%)
1 .3%
(0.7% - 1 .8%)
1 .3%
(0.5% -2.1%)
1%
(0.2% - 1 .8%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.3%)
1%
(0.3% - 1 .7%)
0.8%
(0.1% -1.5%)
13/30
1%
(0.2% - 1 .8%)
1 .4%
(0.6% - 2.3%)
0.3%
(-0.5% -1.1%)
1.1%
(0.3% - 1 .9%)
0.6%
(-0.1% -1.3%)
0.7%
(0.1% -1.4%)
1%
(0.2% - 1 .7%)
0.1%
(-0.2% - 0.4%)
1 .2%
(0.7% - 1 .7%)
1 .3%
(0.5% -2.1%)
1%
(0.2% - 1 .8%)
0.9%
(0.3% - 1 .6%)
0.6%
(0.1% -1.1%)
1.1%
(0.3% - 1 .8%)
0.7%
(0.1% -1.3%)
12/25
0.9%
(0.2% - 1 .6%)
1 .2%
(0.5% - 1 .9%)
0.2%
(-0.4% - 0.9%)
1%
(0.2% - 1 .8%)
0.5%
(-0.1% -1.1%)
0.6%
(0.1% -1.1%)
0.9%
(0.2% - 1 .6%)
0.1%
(-0.1% -0.3%)
1%
(0.6% - 1 .4%)
1.1%
(0.4% - 1 .7%)
0.8%
(0.1% -1.5%)
0.8%
(0.2% - 1 .3%)
0.5%
(0.1% -0.9%)
0.9%
(0.3% - 1 .5%)
0.6%
(0.1% -1.1%)
1 Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-79
-------
Table E-79. Percent Reduction from the Current Standards: Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Percent Reduction From the Current Standards to Several Alternative Standards in Non-Accidental Mortality
Associated with Short-Term Exposure to PM2 5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the
Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2 5
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-41%
(-41 % --41%)
-58%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -23%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
5%
(5% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(10% -10%)
14%
(140/0- 140/0)
5%
(5% - 5%)
8%
(8% - 8%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(14% -15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(14% -15%)
15%
(15% -15%)
15%
(14% -15%)
15%
(15% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-80
-------
Table E-80. Percent Reduction from the Current Standards: Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Percent Reduction From the Current Standards to Several Alternative Standards in Non-Accidental Mortality
Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the
Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81 %
(-80% - -82%)
-6%
(-6% - -6%)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -23%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
6%
(5% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(10% -11%)
14%
(1 4o/0. 14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
1 Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-81
-------
Table E-81. Percent Reduction from the Current Standards: Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Percent Reduction From the Current Standards to Several Alternative Standards in Non-Accidental Mortality
Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the
Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81 %
(-80% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -23%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
6%
(5% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(13% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(1 0% - 1 1 %)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
15%
(15% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
1 Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-82
-------
Table E-82. Estimated Annual Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM25 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
35
(-36-104)
74
(-5 -151)
-1
(-55 - 52)
32
(-21 - 85)
89
(-11 -188)
20
(-14-54)
50
(-34-131)
-50
(-223-121)
605
(353 - 853)
94
(25 - 1 63)
84
(-4 - 1 70)
67
(-13-145)
13
(-3 - 28)
136
(30 - 240)
15
(-8 - 38)
15/353
32
(-33 - 95)
70
(-5-143)
-1
(-43 - 40)
32
(-21 - 85)
73
(-9-153)
11
(-8 - 30)
47
(-32 - 1 24)
-31
(-1 40 - 76)
504
(294 -711)
87
(23 - 1 50)
84
(-4-170)
47
(-9-103)
8
(-2-18)
122
(27-215)
12
(-7-31)
14/35
30
(-30 - 88)
66
(-4 - 1 35)
-1
(-39 - 37)
32
(-21 - 85)
72
(-9 - 1 52)
11
(-8 - 30)
43
(-29 - 1 1 4)
-31
(-140-76)
504
(294 - 71 1 )
87
(23-150)
84
(-4 - 1 70)
47
(-9 - 1 03)
8
(-2-18)
115
(26 - 203)
12
(-7-31)
13/35
27
(-28 - 81 )
61
(-4-125)
-1
(-36 - 34)
32
(-21 - 85)
67
(-8-140)
11
(-8 - 30)
40
(-27 - 1 05)
-31
(-1 40 - 76)
491
(286 - 693)
81
(21 -140)
84
(-4-170)
46
(-9-101)
8
(-2-18)
106
(24 - 1 87)
12
(-7-31)
12/35
25
(-25 - 74)
56
(-4-115)
-1
(-33 - 31 )
30
(-20 - 79)
61
(-8 - 1 29)
11
(-8 - 30)
36
(-25 - 96)
-30
(-132-72)
451
(263 - 637)
75
(19-129)
80
(-3 -161)
44
(-9 - 96)
8
(-2-18)
98
(22-172)
12
(-7-31)
13/30
27
(-28-81)
60
(-4-122)
-1
(-36 - 34)
32
(-21 - 85)
62
(-8-131)
10
(-7 - 26)
40
(-27 - 1 05)
-27
(-119-65)
431
(251 - 609)
74
(19-129)
79
(-3-159)
41
(-8 - 88)
7
(-2-15)
105
(23-185)
11
(-6 - 27)
12/25
24
(-25 - 73)
50
(-3 - 1 02)
0
(-30 - 29)
30
(-20 - 79)
51
(-6 - 1 09)
8
(-6-21)
36
(-25 - 96)
-22
(-99 - 54)
358
(208 - 506)
62
(16-107)
65
(-3 - 1 32)
34
(-7 - 73)
6
(-1-12)
87
(19-153)
9
(-5 - 22)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-83
-------
Table E-83. Estimated Annual Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM25 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
35
(-36-105)
65
(-4 - 1 32)
-1
(-52 - 50)
27
(-18-71)
72
(-9 - 1 52)
22
(-15-58)
48
(-33-128)
-46
(-205 - 1 1 2)
519
(303 - 733)
89
(23 - 1 54)
88
(-4-178)
58
(-12-126)
11
(-3 - 25)
113
(25 - 1 99)
13
(-7 - 32)
15/353
32
(-33 - 97)
61
(-4-125)
-1
(-41 - 39)
27
(-18-71)
58
(-7-123)
12
(-8 - 32)
45
(-31 -120)
-29
(-1 29 - 70)
432
(252 -611)
82
(21 -142)
88
(-4-178)
41
(-8 - 89)
7
(-2-16)
101
(23 - 1 79)
10
(-6 - 26)
14/35
30
(-31 - 90)
58
(-4-118)
-1
(-38 - 36)
27
(-18-71)
58
(-7 - 1 22)
12
(-8 - 32)
42
(-29 - 1 1 2)
-29
(-129-70)
432
(252 - 61 1 )
82
(21 -142)
88
(-4-178)
41
(-8 - 89)
7
(-2-16)
95
(21 -168)
10
(-6 - 26)
13/35
28
(-28 - 82)
53
(-4-109)
-1
(-35 - 33)
27
(-18-71)
54
(-7-113)
12
(-8 - 32)
39
(-26 - 1 03)
-29
(-1 29 - 70)
421
(246 - 595)
77
(20 - 1 33)
88
(-4-178)
40
(-8 - 86)
7
(-2-16)
88
(20 - 1 55)
10
(-6 - 26)
12/35
25
(-26 - 75)
49
(-3 -101)
0
(-32 - 30)
25
(-16-66)
49
(-6 - 1 04)
12
(-8 - 32)
35
(-24 - 94)
-27
(-122-66)
387
(226 - 548)
71
(18-122)
84
(-4 - 1 69)
38
(-8 - 82)
7
(-2-16)
81
(18-143)
10
(-6 - 26)
13/30
28
(-28 - 82)
52
(-4 - 1 07)
-1
(-35 - 33)
27
(-18-71)
50
(-6 - 1 05)
10
(-7 - 27)
39
(-26 - 1 03)
-25
(-110-60)
370
(216-523)
70
(18-122)
82
(-4-167)
35
(-7 - 76)
6
(-1-14)
87
(19-153)
9
(-5 - 22)
12/25
25
(-25 - 74)
43
(-3 - 89)
0
(-29 - 27)
25
(-16-66)
41
(-5 - 87)
9
(-6 - 23)
35
(-24 - 94)
-20
(-91 - 50)
307
(179-435)
58
(15-101)
69
(-3 - 1 39)
29
(-6 - 63)
5
(-1-11)
72
(16-127)
7
(-4-19)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-84
-------
Table E-84. Estimated Annual Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM25 Concentrations in a Recent Year and PM2 5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM25
Concentrations
35
(-36-104)
65
(-4 - 1 33)
-1
(-54 - 51 )
29
(-19-76)
74
(-9 - 1 56)
23
(-16-59)
49
(-33-130)
-47
(-209 - 1 1 4)
568
(332 - 802)
91
(24 - 1 57)
84
(-4 - 1 70)
61
(-12-132)
14
(-3-31)
118
(26 - 208)
14
(-7 - 34)
15/353
32
(-33 - 95)
62
(-4-126)
-1
(-42 - 40)
29
(-19-76)
60
(-8-127)
12
(-9 - 33)
46
(-31 -122)
-30
(-1 32 - 72)
473
(276 - 668)
84
(22 - 1 45)
84
(-4-170)
43
(-9 - 93)
9
(-2 - 20)
106
(24 - 1 87)
11
(-6 - 27)
14/35
30
(-30 - 88)
58
(-4-119)
-1
(-39 - 37)
29
(-1 9 - 76)
60
(-7 - 1 26)
12
(-9 - 33)
43
(-29 - 1 1 3)
-30
(-132-72)
473
(276 - 668)
84
(22-145)
84
(-4 - 1 70)
43
(-9 - 93)
9
(-2 - 20)
99
(22-176)
11
(-6 - 27)
13/35
27
(-28 - 81 )
54
(-4-111)
-1
(-36 - 34)
29
(-19-76)
55
(-7-116)
12
(-9 - 33)
39
(-27 - 1 04)
-30
(-1 32 - 72)
461
(269-651)
79
(20 - 1 36)
84
(-4-170)
42
(-8-91)
9
(-2 - 20)
92
(20 - 1 62)
11
(-6 - 27)
12/35
25
(-25 - 74)
50
(-3 - 1 02)
-1
(-33 - 31 )
27
(-17-70)
51
(-6 - 1 07)
12
(-9 - 33)
36
(-24 - 95)
-28
(-124-68)
424
(247 - 599)
72
(19-125)
80
(-3 - 1 62)
40
(-8 - 87)
9
(-2 - 20)
84
(19-149)
11
(-6 - 27)
13/30
27
(-28-81)
53
(-4 - 1 08)
-1
(-36 - 34)
29
(-19-76)
51
(-6 - 1 08)
11
(-7 - 28)
39
(-27 - 1 04)
-25
(-112-61)
405
(236 - 572)
72
(19-124)
79
(-3-159)
37
(-7 - 80)
8
(-2-17)
91
(20-160)
9
(-5 - 23)
12/25
24
(-25 - 73)
44
(-3 - 90)
0
(-30 - 28)
27
(-17-70)
43
(-5 - 90)
9
(-6 - 24)
36
(-24 - 95)
-21
(-93 - 51 )
336
(196-476)
60
(15-103)
65
(-3 - 1 33)
30
(-6 - 66)
6
(-1-14)
75
(17-133)
8
(-4-19)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-85
-------
Table E-85. Estimated Percent of Total Annual Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2 5
Concentrations
0.9%
(-1%-2.8%)
1 .9%
(-0.1% -3. 9%)
0%
(-2% - 1 .9%)
1%
(-0.6% - 2.5%)
1 .5%
(-0.2% -3.1%)
1 .2%
(-0.9% - 3.3%)
1%
(-0.7% - 2.7%)
-0.3%
(-1 .2% - 0.6%)
2.7%
(1 .6% - 3.8%)
2.3%
(0.6% - 4%)
1 .4%
(-0.1% -2. 9%)
1 .6%
(-0.3% - 3.6%)
1.1%
(-0.3% - 2.5%)
2.4%
(0.5% - 4.2%)
1.1%
(-0.6% - 2.7%)
15/353
0.9%
(-0.9% - 2.6%)
1 .8%
(-0.1% -3. 7%)
0%
(-1 .6% - 1 .5%)
1%
(-0.6% - 2.5%)
1 .2%
(-0.2% - 2.5%)
0.7%
(-0.5% - 1 .8%)
1%
(-0.7% - 2.5%)
-0.2%
(-0.7% - 0.4%)
2.2%
(1 .3% - 3.2%)
2.2%
(0.6% - 3.7%)
1 .4%
(-0.1% -2. 9%)
1 .2%
(-0.2% - 2.5%)
0.7%
(-0.2% - 1 .6%)
2.2%
(0.5% - 3.8%)
0.9%
(-0.5% - 2.2%)
14/35
0.8%
(-0.8% - 2.4%)
1 .7%
(-0.1% -3.5%)
0%
(-1 .5% - 1 .4%)
1%
(-0.6% - 2.5%)
1 .2%
(-0.2% - 2.5%)
0.7%
(-0.5% - 1 .8%)
0.9%
(-0.6% - 2.3%)
-0.2%
(-0.7% - 0.4%)
2.2%
(1 .3% - 3.2%)
2.2%
(0.6% - 3.7%)
1 .4%
(-0.1% -2.9%)
1 .2%
(-0.2% - 2.5%)
0.7%
(-0.2% - 1 .6%)
2%
(0.5% - 3.6%)
0.9%
(-0.5% - 2.2%)
13/35
0.7%
(-0.7% - 2.2%)
1 .6%
(-0.1% -3.2%)
0%
(-1 .3% - 1 .3%)
1%
(-0.6% - 2.5%)
1.1%
(-0.1% -2.3%)
0.7%
(-0.5% - 1 .8%)
0.8%
(-0.6% - 2.2%)
-0.2%
(-0.7% - 0.4%)
2.2%
(1.3% -3.1%)
2%
(0.5% - 3.5%)
1 .4%
(-0.1% -2.9%)
1.1%
(-0.2% - 2.5%)
0.7%
(-0.2% - 1 .6%)
1 .9%
(0.4% - 3.3%)
0.9%
(-0.5% - 2.2%)
12/35
0.7%
(-0.7% - 2%)
1 .4%
(-0.1% -3%)
0%
(-1 .2% - 1 .2%)
0.9%
(-0.6% - 2.3%)
1%
(-0.1% -2.1%)
0.7%
(-0.5% - 1 .8%)
0.7%
(-0.5% - 2%)
-0.2%
(-0.7% - 0.4%)
2%
(1 .2% - 2.8%)
1 .9%
(0.5% - 3.2%)
1 .4%
(-0.1% -2.7%)
1.1%
(-0.2% - 2.3%)
0.7%
(-0.2% - 1 .6%)
1 .7%
(0.4% - 3%)
0.9%
(-0.5% - 2.2%)
13/30
0.7%
(-0.7% - 2.2%)
1 .5%
(-0.1% -3.1%)
0%
(-1 .3% - 1 .3%)
1%
(-0.6% - 2.5%)
1%
(-0.1% -2.2%)
0.6%
(-0.4% - 1 .6%)
0.8%
(-0.6% - 2.2%)
-0.1%
(-0.6% - 0.3%)
1 .9%
(1.1% -2.7%)
1 .8%
(0.5% - 3.2%)
1 .3%
(-0.1% -2.7%)
1%
(-0.2% - 2.2%)
0.6%
(-0.1% -1.4%)
1 .8%
(0.4% - 3.2%)
0.7%
(-0.4% - 1 .8%)
12/25
0.7%
(-0.7% - 2%)
1 .3%
(-0.1% -2.6%)
0%
(-1.1% -1.1%)
0.9%
(-0.6% - 2.3%)
0.9%
(-0.1% -1.8%)
0.5%
(-0.3% - 1 .3%)
0.7%
(-0.5% - 2%)
-0.1%
(-0.5% - 0.3%)
1 .6%
(0.9% - 2.3%)
1 .5%
(0.4% - 2.7%)
1.1%
(0% - 2.3%)
0.8%
(-0.2% - 1 .8%)
0.5%
(-0.1% -1.1%)
1 .5%
(0.3% - 2.7%)
0.6%
(-0.3% - 1 .5%)
1 Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-86
-------
Table E-86. Estimated Percent of Total Annual Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2 5 Concentrations in a Recent
Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
0.9%
(-0.9% - 2.8%)
1 .7%
(-0.1% -3.4%)
0%
(-1 .9% - 1 .8%)
0.8%
(-0.5% -2.1%)
1 .2%
(-0.2% - 2.5%)
1 .3%
(-0.9% - 3.5%)
1%
(-0.6% - 2.5%)
-0.2%
(-1.1% -0.6%)
2.3%
(1 .3% - 3.3%)
2.2%
(0.6% - 3.8%)
1 .4%
(-0.1% -2.9%)
1 .4%
(-0.3% -3.1%)
1%
(-0.2% - 2.2%)
2%
(0.4% - 3.5%)
0.9%
(-0.5% - 2.2%)
15/353
0.8%
(-0.9% - 2.5%)
1 .6%
(-0.1% -3. 2%)
0%
(-1 .5% - 1 .4%)
0.8%
(-0.5% -2.1%)
1%
(-0.1% -2.1%)
0.7%
(-0.5% - 1 .9%)
0.9%
(-0.6% - 2.4%)
-0.2%
(-0.7% - 0.4%)
1 .9%
(1.1% -2. 7%)
2.1%
(0.5% - 3.5%)
1 .4%
(-0.1% -2. 9%)
1%
(-0.2% - 2.2%)
0.6%
(-0.1% -1.4%)
1 .8%
(0.4% -3.1%)
0.7%
(-0.4% - 1 .8%)
14/35
0.8%
(-0.8% - 2.3%)
1 .5%
(-0.1% -3%)
0%
(-1 .4% - 1 .3%)
0.8%
(-0.5% -2.1%)
1%
(-0.1% -2.1%)
0.7%
(-0.5% - 1 .9%)
0.8%
(-0.6% - 2.2%)
-0.2%
(-0.7% - 0.4%)
1 .9%
(1.1% -2.7%)
2.1%
(0.5% - 3.5%)
1 .4%
(-0.1% -2.9%)
1%
(-0.2% - 2.2%)
0.6%
(-0.1% -1.4%)
1 .7%
(0.4% - 3%)
0.7%
(-0.4% - 1 .8%)
13/35
0.7%
(-0.7% - 2.2%)
1 .4%
(-0.1% -2.8%)
0%
(-1 .3% - 1 .2%)
0.8%
(-0.5% -2.1%)
0.9%
(-0.1% -1.9%)
0.7%
(-0.5% - 1 .9%)
0.8%
(-0.5% - 2%)
-0.2%
(-0.7% - 0.4%)
1 .9%
(1.1% -2.6%)
1 .9%
(0.5% - 3.3%)
1 .4%
(-0.1% -2.9%)
1%
(-0.2% -2.1%)
0.6%
(-0.1% -1.4%)
1 .5%
(0.3% - 2.7%)
0.7%
(-0.4% - 1 .8%)
12/35
0.7%
(-0.7% - 2%)
1 .3%
(-0.1% -2.6%)
0%
(-1.2% -1.1%)
0.7%
(-0.5% - 1 .9%)
0.8%
(-0.1% -1.7%)
0.7%
(-0.5% - 1 .9%)
0.7%
(-0.5% - 1 .9%)
-0.1%
(-0.6% - 0.4%)
1 .7%
(1%-2.4%)
1 .8%
(0.5% -3.1%)
1 .4%
(-0.1% -2.8%)
0.9%
(-0.2% - 2%)
0.6%
(-0.1% -1.4%)
1 .4%
(0.3% - 2.5%)
0.7%
(-0.4% - 1 .8%)
13/30
0.7%
(-0.7% - 2.2%)
1 .3%
(-0.1% -2.7%)
0%
(-1 .3% - 1 .2%)
0.8%
(-0.5% -2.1%)
0.8%
(-0.1% -1.8%)
0.6%
(-0.4% - 1 .6%)
0.8%
(-0.5% - 2%)
-0.1%
(-0.6% - 0.3%)
1 .6%
(1 % - 2.3%)
1 .8%
(0.5% - 3%)
1 .4%
(-0.1% -2.7%)
0.9%
(-0.2% - 1 .9%)
0.5%
(-0.1% -1.2%)
1 .5%
(0.3% - 2.7%)
0.6%
(-0.3% - 1 .5%)
12/25
0.6%
(-0.7% - 1 .9%)
1.1%
(-0.1% -2. 3%)
0%
(-1.1% -1%)
0.7%
(-0.5% - 1 .9%)
0.7%
(-0.1% -1.5%)
0.5%
(-0.4% - 1 .4%)
0.7%
(-0.5% - 1 .9%)
-0.1%
(-0.5% - 0.3%)
1 .4%
(0.8% - 1 .9%)
1 .5%
(0.4% - 2.5%)
1.1%
(0% - 2.3%)
0.7%
(-0.1% -1.6%)
0.5%
(-0.1% -1%)
1 .3%
(0.3% - 2.2%)
0.5%
(-0.3% - 1 .3%)
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-87
-------
Table E-87. Estimated Percent of Total Annual Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
0.9%
(-0.9% - 2.7%)
1 .7%
(-0.1% -3.4%)
0%
(-2% - 1 .9%)
0.8%
(-0.5% - 2.2%)
1 .3%
(-0.2% - 2.7%)
1 .3%
(-0.9% - 3.5%)
1%
(-0.6% - 2.5%)
-0.2%
(-1.1% -0.6%)
2.5%
(1 .5% - 3.5%)
2.3%
(0.6% - 3.9%)
1 .3%
(-0.1% -2.7%)
1 .5%
(-0.3% - 3.3%)
1 .2%
(-0.3% - 2.7%)
2.1%
(0.5% - 3.7%)
0.9%
(-0.5% - 2.3%)
15/353
0.8%
(-0.8% - 2.4%)
1 .6%
(-0.1% -3. 2%)
0%
(-1 .5% - 1 .5%)
0.8%
(-0.5% - 2.2%)
1%
(-0.1% -2. 2%)
0.7%
(-0.5% - 2%)
0.9%
(-0.6% - 2.4%)
-0.2%
(-0.7% - 0.4%)
2.1%
(1 .2% - 3%)
2.1%
(0.5% - 3.6%)
1 .3%
(-0.1% -2. 7%)
1.1%
(-0.2% - 2.3%)
0.8%
(-0.2% - 1 .7%)
1 .9%
(0.4% - 3.3%)
0.7%
(-0.4% - 1 .8%)
14/35
0.8%
(-0.8% - 2.3%)
1 .5%
(-0.1% -3.1%)
0%
(-1 .4% - 1 .4%)
0.8%
(-0.5% - 2.2%)
1%
(-0.1% -2.1%)
0.7%
(-0.5% - 2%)
0.8%
(-0.6% - 2.2%)
-0.2%
(-0.7% - 0.4%)
2.1%
(1 .2% - 3%)
2.1%
(0.5% - 3.6%)
1 .3%
(-0.1% -2.7%)
1.1%
(-0.2% - 2.3%)
0.8%
(-0.2% - 1 .7%)
1 .7%
(0.4% -3.1%)
0.7%
(-0.4% - 1 .8%)
13/35
0.7%
(-0.7% -2.1%)
1 .4%
(-0.1% -2.8%)
0%
(-1 .3% - 1 .2%)
0.8%
(-0.5% - 2.2%)
0.9%
(-0.1% -2%)
0.7%
(-0.5% - 2%)
0.8%
(-0.5% - 2%)
-0.2%
(-0.7% - 0.4%)
2%
(1 .2% - 2.9%)
2%
(0.5% - 3.4%)
1 .3%
(-0.1% -2.7%)
1%
(-0.2% - 2.3%)
0.8%
(-0.2% - 1 .7%)
1 .6%
(0.4% - 2.9%)
0.7%
(-0.4% - 1 .8%)
12/35
0.6%
(-0.6% - 1 .9%)
1 .3%
(-0.1% -2.6%)
0%
(-1.2% -1.1%)
0.8%
(-0.5% - 2%)
0.9%
(-0.1% -1.8%)
0.7%
(-0.5% - 2%)
0.7%
(-0.5% - 1 .9%)
-0.1%
(-0.7% - 0.4%)
1 .9%
(1.1% -2.6%)
1 .8%
(0.5% -3.1%)
1 .3%
(-0.1% -2.6%)
1%
(-0.2% -2.1%)
0.8%
(-0.2% - 1 .7%)
1 .5%
(0.3% - 2.6%)
0.7%
(-0.4% - 1 .8%)
13/30
0.7%
(-0.7% -2.1%)
1 .4%
(-0.1% -2.8%)
0%
(-1 .3% - 1 .2%)
0.8%
(-0.5% - 2.2%)
0.9%
(-0.1% -1.8%)
0.6%
(-0.4% - 1 .7%)
0.8%
(-0.5% - 2%)
-0.1%
(-0.6% - 0.3%)
1 .8%
(1%-2.5%)
1 .8%
(0.5% -3.1%)
1 .3%
(-0.1% -2. 5%)
0.9%
(-0.2% - 2%)
0.7%
(-0.2% - 1 .5%)
1 .6%
(0.4% - 2.8%)
0.6%
(-0.3% - 1 .6%)
12/25
0.6%
(-0.6% - 1 .9%)
1.1%
(-0.1% -2. 3%)
0%
(-1.1% -1%)
0.8%
(-0.5% - 2%)
0.7%
(-0.1% -1.5%)
0.5%
(-0.4% - 1 .4%)
0.7%
(-0.5% - 1 .9%)
-0.1%
(-0.5% - 0.3%)
1 .5%
(0.9% -2.1%)
1 .5%
(0.4% - 2.6%)
1%
(0%-2.1%)
0.8%
(-0.1% -1.6%)
0.6%
(-0.1% -1.2%)
1 .3%
(0.3% - 2.3%)
0.5%
(-0.3% - 1 .3%)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-88
-------
Table E-88. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiovascular Mortality Associated with Short-Term
Exposure to Ambient PM2 s Concentrations, Based on Adjusting 2005 PM2 s Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiovascular Mortality Associated with Short-Term
Exposure to PM2 5 Concentrations in a Recent Year and PM2.s Concentrations that Just Meet the Current and Alternative Annual
(n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -83%)
-6%
(-60/0 - -60/o)
-59%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-41%
(-41 % - -42%)
-58%
(-58% - -59%)
-1 2%
(-11% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
6%
(5% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(19% -20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(10% -10%)
14%
(14% -14%)
5%
(5% - 5%)
8%
(8% - 8%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(14%- 15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -14%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(14% -15%)
15%
(15% -15%)
14%
(14% -15%)
15%
(14% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-89
-------
Table E-89. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiovascular Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiovascular Mortality Associated with Short-Term
Exposure to PM2 5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual
(n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -83%)
-6%
(-60/0 - -60/o)
-59%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-58%
(-58% - -59%)
-1 2%
(-11% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
6%
(5% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(1 9% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(1 6% - 1 6%)
0%
(0% - 0%)
23%
(22% - 23%)
6%
(6% - 6%)
10%
(1 0% - 1 0%)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(14% -15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(14% -15%)
15%
(15% -15%)
15%
(14% -15%)
15%
(14% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
1 Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-90
-------
Table E-90. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiovascular Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiovascular Mortality Associated with Short-Term
Exposure to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual
(n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81 %
(-79% - -83%)
-6%
(-60/0 - -60/o)
-59%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-58%
(-57% - -59%)
-1 2%
(-1 1 % - -1 2%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
6%
(5% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(19% -20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
23%
(22% - 23%)
6%
(6% - 6%)
10%
(10% -10%)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(14% -15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(14% -15%)
15%
(14% -15%)
15%
(14% -15%)
15%
(14% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-91
-------
Table E-91. Estimated Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations in a
Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
21
(-9-51)
38
(7 - 67)
12
(-10-33)
11
(-10-32)
35
(2 - 67)
15
(0 - 30)
36
(6 - 65)
90
(9-171)
128
(45 - 208)
25
(-2 - 52)
47
(4 - 90)
28
(-3 - 58)
8
(1 -15)
35
(-9 - 78)
9
(0-18)
15/353
20
(-8 - 47)
36
(7 - 64)
9
(-7 - 26)
11
(-10-32)
28
(1 - 55)
9
(0-17)
34
(5-61)
57
(6-108)
106
(37 - 1 74)
23
(-2 - 48)
47
(4 - 90)
20
(-2 - 42)
5
(1-10)
31
(-8 - 70)
7
(0-15)
14/35
18
(-7 - 43)
34
(6 - 60)
9
(-7 - 24)
11
(-1 0 - 32)
28
(1 - 54)
9
(0-17)
31
(5 - 57)
57
(6 - 1 08)
106
(37-174)
23
(-2 - 48)
47
(4 - 90)
20
(-2 - 42)
5
(1-10)
29
(-8 - 65)
7
(0-15)
13/35
17
(-7 - 40)
31
(6 - 56)
8
(-6 - 22)
11
(-10-32)
26
(1 - 50)
9
(0-17)
29
(5 - 52)
57
(6-108)
104
(37 - 1 69)
22
(-2 - 45)
47
(4 - 90)
20
(-2 - 40)
5
(1-10)
27
(-7-61)
7
(0-15)
12/35
15
(-6 - 36)
29
(5-51)
7
(-6 - 20)
10
(-10-30)
24
(1 - 46)
9
(0-17)
26
(4 - 48)
54
(5 - 1 02)
95
(34- 156)
20
(-2-41)
45
(4 - 85)
19
(-2 - 38)
5
(1-10)
25
(-7 - 56)
7
(0-15)
13/30
17
(-7 - 40)
30
(6 - 55)
8
(-6 - 22)
11
(-10-32)
24
(1 - 47)
7
(0-14)
29
(5 - 52)
49
(5 - 93)
91
(32-149)
20
(-2-41)
44
(4 - 84)
17
(-2 - 36)
4
(1 -8)
27
(-7 - 60)
6
(0-13)
12/25
15
(-6 - 36)
25
(5 - 45)
7
(-5-18)
10
(-10-30)
20
(1 - 39)
6
(0-12)
26
(4 - 48)
41
(4 - 77)
76
(27-124)
16
(-2 - 34)
37
(3 - 70)
14
(-1 - 30)
4
(0-7)
22
(-6 - 50)
5
(0-10)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-92
-------
Table E-92. Estimated Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations in a
Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
22
(-9-51)
33
(6 - 59)
11
(-9-31)
9
(-9 - 27)
28
(1 - 54)
16
(1 - 32)
35
(6 - 63)
84
(8 - 1 58)
110
(39 - 1 79)
24
(-2 - 49)
50
(4 - 94)
24
(-2-51)
8
(1 -14)
29
(-8 - 64)
8
(0-15)
15/353
20
(-8 - 47)
31
(6 - 56)
9
(-7 - 25)
9
(-9 - 27)
23
(1 - 44)
9
(0-18)
33
(5 - 60)
53
(5-100)
91
(32 - 1 49)
22
(-2 - 45)
50
(4 - 94)
17
(-2 - 36)
5
(1 -9)
26
(-7 - 58)
6
(0-12)
14/35
18
(-7 - 44)
29
(5 - 53)
8
(-7 - 23)
9
(-9 - 27)
23
(1 - 44)
9
(0-18)
30
(5 - 55)
53
(5 - 1 00)
91
(32-149)
22
(-2 - 45)
50
(4 - 94)
17
(-2 - 36)
5
(1 -9)
24
(-6 - 54)
6
(0-12)
13/35
17
(-7 - 40)
27
(5 - 49)
8
(-6-21)
9
(-9 - 27)
21
(1-41)
9
(0-18)
28
(4-51)
53
(5-100)
89
(31 -146)
20
(-2 - 42)
50
(4 - 94)
17
(-2 - 35)
5
(1-9)
22
(-6 - 50)
6
(0-12)
12/35
16
(-6 - 37)
25
(5 - 45)
7
(-6-19)
9
(-8 - 25)
19
(1 - 37)
9
(0-18)
25
(4 - 46)
50
(5 - 95)
82
(29- 134)
19
(-2 - 39)
47
(4 - 90)
16
(-2 - 33)
5
(1-9)
21
(-5 - 46)
6
(0-12)
13/30
17
(-7 - 40)
27
(5 - 48)
8
(-6-21)
9
(-9 - 27)
20
(1 - 38)
8
(0-15)
28
(4-51)
45
(4 - 86)
78
(27-128)
19
(-2 - 39)
46
(4 - 88)
15
(-1-31)
4
(1 -8)
22
(-6 - 49)
5
(0-11)
12/25
15
(-6 - 36)
22
(4 - 40)
6
(-5-17)
9
(-8 - 25)
16
(1 -31)
6
(0-13)
25
(4 - 46)
37
(4-71)
65
(23-107)
15
(-1 - 32)
39
(3 - 74)
12
(-1 - 25)
3
(0-6)
18
(-5-41)
4
(0-9)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-93
-------
Table E-93. Estimated Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure to Ambient PM2 5 Concentrations in a
Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted
n/m)2:
Recent PM2.5
Concentrations
21
(-9-51)
33
(6 - 60)
12
(-10-32)
10
(-9 - 29)
29
(1 - 56)
17
(1 - 33)
35
(6 - 64)
85
(8-161)
120
(42 - 1 96)
24
(-2 - 50)
47
(4 - 90)
26
(-3 - 53)
9
(1 -17)
30
(-8 - 67)
8
(0-16)
15/353
20
(-8 - 47)
31
(6 - 56)
9
(-7 - 25)
10
(-9 - 29)
24
(1 - 45)
9
(0-18)
33
(5-61)
54
(5-102)
100
(35 - 1 63)
22
(-2 - 46)
47
(4 - 90)
18
(-2 - 38)
6
(1 -11)
27
(-7 - 60)
6
(0-13)
14/35
18
(-7 - 43)
30
(6 - 53)
9
(-7 - 24)
10
(-9 - 29)
23
(1 - 45)
9
(0-18)
31
(5 - 56)
54
(5 - 1 02)
100
(35-163)
22
(-2 - 46)
47
(4 - 90)
18
(-2 - 38)
6
(1 -11)
25
(-7 - 57)
6
(0-13)
13/35
17
(-7 - 40)
28
(5 - 49)
8
(-6 - 22)
10
(-9 - 29)
22
(1 - 42)
9
(0-18)
28
(5 - 52)
54
(5-102)
97
(34 - 1 59)
21
(-2 - 43)
47
(4 - 90)
18
(-2 - 37)
6
(1-11)
23
(-6 - 53)
6
(0-13)
12/35
15
(-6 - 36)
25
(5 - 45)
7
(-6 - 20)
9
(-8 - 27)
20
(1 - 38)
9
(0-18)
26
(4 - 47)
51
(5 - 96)
89
(31 -147)
19
(-2 - 40)
45
(4 - 85)
17
(-2 - 35)
6
(1-11)
22
(-6 - 48)
6
(0-13)
13/30
17
(-7 - 40)
27
(5 - 48)
8
(-6 - 22)
10
(-9 - 29)
20
(1 - 39)
8
(0-16)
28
(5 - 52)
46
(4 - 87)
85
(30-140)
19
(-2 - 39)
44
(4 - 84)
15
(-2 - 32)
5
(1 -9)
23
(-6 - 52)
5
(0-11)
12/25
15
(-6 - 36)
22
(4 - 40)
7
(-5-18)
9
(-8 - 27)
17
(1 - 32)
7
(0-13)
26
(4 - 47)
38
(4 - 73)
71
(25 - 1 1 7)
16
(-1 - 33)
37
(3 - 70)
13
(-1 - 27)
4
(1 -8)
19
(-5 - 43)
5
(0-9)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-94
-------
Table E-94. Estimated Percent of Total Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.s Concentrations in a Recent
Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
1 .7%
(-0.7% -4.1%)
3.1%
(0.6% - 5.6%)
1 .4%
(-1.1% -3.7%)
1%
(-0.9% - 3%)
2.6%
(0.1% -5%)
2.6%
(0.1% -5.1%)
2.5%
(0.4% - 4.6%)
1 .6%
(0.2% -3.1%)
3%
(1%-4.8%)
2.1%
(-0.2% - 4.3%)
1 .9%
(0.2% - 3.7%)
2.4%
(-0.2% - 4.9%)
1 .9%
(0.2% - 3.5%)
2%
(-0.5% - 4.5%)
1 .8%
(0% - 3.6%)
15/353
1 .6%
(-0.6% - 3.7%)
2.9%
(0.5% - 5.3%)
1.1%
(-0.9% - 2.9%)
1%
(-0.9% - 3%)
2.1%
(0.1% -4.1%)
1 .5%
(0% - 2.8%)
2.4%
(0.4% - 4.4%)
1%
(0.1% -1.9%)
2.5%
(0.9% - 4%)
1 .9%
(-0.2% - 3.9%)
1 .9%
(0.2% - 3.7%)
1 .7%
(-0.2% - 3.5%)
1 .2%
(0.1% -2.2%)
1 .8%
(-0.5% - 4%)
1 .5%
(0% - 3%)
14/35
1 .5%
(-0.6% - 3.5%)
2.8%
(0.5% - 5%)
1%
(-0.8% - 2.7%)
1%
(-0.9% - 3%)
2.1%
(0.1% -4%)
1 .5%
(0% - 2.8%)
2.2%
(0.4% - 4%)
1%
(0.1% -1.9%)
2.5%
(0.9% - 4%)
1 .9%
(-0.2% - 3.9%)
1 .9%
(0.2% - 3.7%)
1 .7%
(-0.2% - 3.5%)
1 .2%
(0.1% -2.2%)
1 .7%
(-0.4% - 3.8%)
1 .5%
(0% - 3%)
13/35
1 .3%
(-0.5% - 3.2%)
2.6%
(0.5% - 4.6%)
0.9%
(-0.7% - 2.5%)
1%
(-0.9% - 3%)
1 .9%
(0.1% -3.7%)
1 .5%
(0% - 2.8%)
2%
(0.3% - 3.7%)
1%
(0.1% -1.9%)
2.4%
(0.8% - 3.9%)
1 .8%
(-0.2% - 3.7%)
1 .9%
(0.2% - 3.7%)
1 .6%
(-0.2% - 3.4%)
1 .2%
(0.1% -2.2%)
1 .6%
(-0.4% - 3.5%)
1 .5%
(0% - 3%)
12/35
1 .2%
(-0.5% - 2.9%)
2.4%
(0.4% - 4.2%)
0.8%
(-0.7% - 2.3%)
1%
(-0.9% - 2.7%)
1 .8%
(0.1% -3.4%)
1 .5%
(0% - 2.8%)
1 .9%
(0.3% - 3.4%)
1%
(0.1% -1.8%)
2.2%
(0.8% - 3.6%)
1 .6%
(-0.2% - 3.4%)
1 .8%
(0.1% -3. 5%)
1 .6%
(-0.2% - 3.2%)
1 .2%
(0.1% -2. 2%)
1 .4%
(-0.4% - 3.2%)
1 .5%
(0% - 3%)
13/30
1 .3%
(-0.5% - 3.2%)
2.5%
(0.5% - 4.5%)
0.9%
(-0.7% - 2.5%)
1%
(-0.9% - 3%)
1 .8%
(0.1% -3. 5%)
1 .2%
(0% - 2.4%)
2%
(0.3% - 3.7%)
0.9%
(0.1% -1.7%)
2.1%
(0.7% - 3.5%)
1 .6%
(-0.2% - 3.4%)
1 .8%
(0.1% -3.5%)
1 .4%
(-0.1% -3%)
1%
(0.1% -1.9%)
1 .5%
(-0.4% - 3.4%)
1 .3%
(0% - 2.5%)
12/25
1 .2%
(-0.5% - 2.9%)
2.1%
(0.4% - 3.8%)
0.8%
(-0.6% -2.1%)
1%
(-0.9% - 2.7%)
1 .5%
(0.1% -2.9%)
1%
(0% - 2%)
1 .9%
(0.3% - 3.4%)
0.7%
(0.1% -1.4%)
1 .8%
(0.6% - 2.9%)
1 .3%
(-0.1% -2.8%)
1 .5%
(0.1% -2.9%)
1 .2%
(-0.1% -2.5%)
0.8%
(0.1% -1.6%)
1 .3%
(-0.3% - 2.9%)
1.1%
(0%-2.1%)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-95
-------
Table E-95. Estimated Percent of Total Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.s Concentrations in a Recent
Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
1 .7%
(-0.7% - 4%)
2.7%
(0.5% - 4.9%)
1 .3%
(-1%-3.6%)
0.8%
(-0.8% - 2.4%)
2.1%
(0.1% -4.1%)
2.8%
(0.1% -5.3%)
2.4%
(0.4% - 4.4%)
1 .5%
(0.1% -2.8%)
2.5%
(0.9% -4.1%)
2%
(-0.2% - 4%)
2%
(0.2% - 3.7%)
2.1%
(-0.2% - 4.3%)
1 .7%
(0.2% -3.1%)
1 .7%
(-0.4% - 3.7%)
1 .5%
(0% - 3%)
15/353
1 .6%
(-0.6% - 3.7%)
2.6%
(0.5% - 4.6%)
1%
(-0.8% - 2.8%)
0.8%
(-0.8% - 2.4%)
1 .7%
(0.1% -3.3%)
1 .5%
(0% - 3%)
2.3%
(0.4% -4.1%)
0.9%
(0.1% -1.8%)
2.1%
(0.7% - 3.5%)
1 .8%
(-0.2% - 3.7%)
2%
(0.2% - 3.7%)
1 .5%
(-0.1% -3%)
1.1%
(0.1% -2%)
1 .5%
(-0.4% - 3.3%)
1 .2%
(0% - 2.5%)
14/35
1 .4%
(-0.6% - 3.4%)
2.4%
(0.4% - 4.3%)
0.9%
(-0.8% - 2.6%)
0.8%
(-0.8% - 2.4%)
1 .7%
(0.1% -3.3%)
1 .5%
(0% - 3%)
2.1%
(0.3% - 3.8%)
0.9%
(0.1% -1.8%)
2.1%
(0.7% - 3.5%)
1 .8%
(-0.2% - 3.7%)
2%
(0.2% - 3.7%)
1 .5%
(-0.1% -3%)
1.1%
(0.1% -2%)
1 .4%
(-0.4% -3.1%)
1 .2%
(0% - 2.5%)
13/35
1 .3%
(-0.5% -3.1%)
2.2%
(0.4% - 4%)
0.9%
(-0.7% - 2.4%)
0.8%
(-0.8% - 2.4%)
1 .6%
(0.1% -3%)
1 .5%
(0% - 3%)
1 .9%
(0.3% - 3.5%)
0.9%
(0.1% -1.8%)
2.1%
(0.7% - 3.4%)
1 .7%
(-0.2% - 3.5%)
2%
(0.2% - 3.7%)
1 .4%
(-0.1% -3%)
1.1%
(0.1% -2%)
1 .3%
(-0.3% - 2.9%)
1 .2%
(0% - 2.5%)
12/35
1 .2%
(-0.5% - 2.9%)
2.1%
(0.4% - 3.7%)
0.8%
(-0.6% - 2.2%)
0.8%
(-0.7% - 2.3%)
1 .4%
(0.1% -2. 8%)
1 .5%
(0% - 3%)
1 .8%
(0.3% - 3.2%)
0.9%
(0.1% -1.7%)
1 .9%
(0.7% -3.1%)
1 .6%
(-0.1% -3.2%)
1 .9%
(0.1% -3. 6%)
1 .4%
(-0.1% -2.8%)
1.1%
(0.1% -2%)
1 .2%
(-0.3% - 2.7%)
1 .2%
(0% - 2.5%)
13/30
1 .3%
(-0.5% -3.1%)
2.2%
(0.4% - 3.9%)
0.9%
(-0.7% - 2.4%)
0.8%
(-0.8% - 2.4%)
1 .5%
(0.1% -2.8%)
1 .3%
(0% - 2.6%)
1 .9%
(0.3% - 3.5%)
0.8%
(0.1% -1.5%)
1 .8%
(0.6% - 3%)
1 .5%
(-0.1% -3. 2%)
1 .8%
(0.1% -3.5%)
1 .2%
(-0.1% -2. 6%)
0.9%
(0.1% -1.7%)
1 .3%
(-0.3% - 2.8%)
1%
(0%-2.1%)
12/25
1 .2%
(-0.5% - 2.8%)
1 .8%
(0.3% - 3.3%)
0.7%
(-0.6% - 2%)
0.8%
(-0.7% - 2.3%)
1 .2%
(0.1% -2.3%)
1.1%
(0%-2.1%)
1 .8%
(0.3% - 3.2%)
0.7%
(0.1% -1.3%)
1 .5%
(0.5% - 2.5%)
1 .3%
(-0.1% -2.7%)
1 .5%
(0.1% -2.9%)
1%
(-0.1% -2.2%)
0.8%
(0.1% -1.4%)
1.1%
(-0.3% - 2.4%)
0.9%
(0% - 1 .7%)
1 Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-96
-------
Table E-96. Estimated Percent of Total Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure to Ambient PM2 5
Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on
Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.s Concentrations in a Recent
Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM25
Concentrations
1 .6%
(-0.7% - 3.9%)
2.7%
(0.5% - 4.9%)
1 .3%
(-1.1% -3.7%)
0.9%
(-0.8% - 2.6%)
2.2%
(0.1% -4.2%)
2.8%
(0.1% -5.4%)
2.4%
(0.4% - 4.4%)
1 .5%
(0.1% -2.9%)
2.8%
(1%-4.5%)
2%
(-0.2% -4.1%)
1 .8%
(0.1% -3.5%)
2.2%
(-0.2% - 4.5%)
2%
(0.3% - 3.8%)
1 .7%
(-0.5% - 3.9%)
1 .6%
(0%-3.1%)
15/353
1 .5%
(-0.6% - 3.6%)
2.6%
(0.5% - 4.7%)
1%
(-0.8% - 2.9%)
0.9%
(-0.8% - 2.6%)
1 .8%
(0.1% -3.4%)
1 .6%
(0.1% -3%)
2.3%
(0.4% -4.1%)
1%
(0.1% -1.8%)
2.3%
(0.8% - 3.8%)
1 .8%
(-0.2% - 3.8%)
1 .8%
(0.1% -3.5%)
1 .5%
(-0.2% - 3.2%)
1 .3%
(0.2% - 2.4%)
1 .6%
(-0.4% - 3.5%)
1 .3%
(0% - 2.5%)
14/35
1 .4%
(-0.6% - 3.3%)
2.5%
(0.5% - 4.4%)
1%
(-0.8% - 2.7%)
0.9%
(-0.8% - 2.6%)
1 .8%
(0.1% -3.4%)
1 .6%
(0.1% -3%)
2.1%
(0.3% - 3.8%)
1%
(0.1% -1.8%)
2.3%
(0.8% - 3.8%)
1 .8%
(-0.2% - 3.8%)
1 .8%
(0.1% -3.5%)
1 .5%
(-0.2% - 3.2%)
1 .3%
(0.2% - 2.4%)
1 .5%
(-0.4% - 3.3%)
1 .3%
(0% - 2.5%)
13/35
1 .3%
(-0.5% - 3%)
2.3%
(0.4% -4.1%)
0.9%
(-0.7% - 2.5%)
0.9%
(-0.8% - 2.6%)
1 .6%
(0.1% -3.2%)
1 .6%
(0.1% -3%)
1 .9%
(0.3% - 3.5%)
1%
(0.1% -1.8%)
2.2%
(0.8% - 3.7%)
1 .7%
(-0.2% - 3.6%)
1 .8%
(0.1% -3.5%)
1 .5%
(-0.1% -3.1%)
1 .3%
(0.2% - 2.4%)
1 .3%
(-0.4% - 3%)
1 .3%
(0% - 2.5%)
12/35
1 .2%
(-0.5% - 2.8%)
2.1%
(0.4% - 3.8%)
0.8%
(-0.7% - 2.2%)
0.8%
(-0.8% - 2.4%)
1 .5%
(0.1% -2. 9%)
1 .6%
(0.1% -3%)
1 .8%
(0.3% - 3.2%)
0.9%
(0.1% -1.7%)
2.1%
(0.7% - 3.4%)
1 .6%
(-0.1% -3.3%)
1 .7%
(0.1% -3. 3%)
1 .4%
(-0.1% -3%)
1 .3%
(0.2% - 2.4%)
1 .2%
(-0.3% - 2.8%)
1 .3%
(0% - 2.5%)
13/30
1 .3%
(-0.5% - 3%)
2.2%
(0.4% - 4%)
0.9%
(-0.7% - 2.5%)
0.9%
(-0.8% - 2.6%)
1 .5%
(0.1% -2.9%)
1 .3%
(0% - 2.6%)
1 .9%
(0.3% - 3.5%)
0.8%
(0.1% -1.6%)
2%
(0.7% - 3.2%)
1 .6%
(-0.1% -3. 3%)
1 .7%
(0.1% -3.3%)
1 .3%
(-0.1% -2. 7%)
1.1%
(0.1% -2%)
1 .3%
(-0.3% - 3%)
1.1%
(0% - 2.2%)
12/25
1 .2%
(-0.5% - 2.7%)
1 .8%
(0.3% - 3.3%)
0.7%
(-0.6% - 2%)
0.8%
(-0.8% - 2.4%)
1 .3%
(0.1% -2.4%)
1.1%
(0% - 2.2%)
1 .8%
(0.3% - 3.2%)
0.7%
(0.1% -1.3%)
1 .6%
(0.6% - 2.7%)
1 .3%
(-0.1% -2.7%)
1 .4%
(0.1% -2.7%)
1.1%
(-0.1% -2.3%)
0.9%
(0.1% -1.7%)
1.1%
(-0.3% - 2.5%)
0.9%
(0% - 1 .8%)
1 Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-97
-------
Table E-97. Percent Reduction from the Current Standards: Estimated Annual Incidence of Respiratory Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure
to PM2 5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-80%
(-78% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-57% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-41%
(-40% - -42%)
-58%
(-57% - -59%)
-1 2%
(-11% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
5%
(5% - 6%)
7%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
12%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 7%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(1 9% - 20%)
22%
(22% - 23%)
7%
(7% - 7%)
16%
(1 5% - 1 6%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(1 0% - 1 0%)
14%
(14% -14%)
5%
(5% - 5%)
8%
(8% - 8%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
14%
(14% -15%)
14%
(14% -15%)
15%
(15% -15%)
15%
(14% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(14% -15%)
15%
(14% -15%)
15%
(14% -15%)
15%
(14% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
29%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-98
-------
Table E-98. Percent Reduction from the Current Standards: Estimated Annual Incidence of Respiratory Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure
to PM2.5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-80%
(-79% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-58%
(-58% - -59%)
-1 2%
(-11% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
5%
(5% - 6%)
7%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
12%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 7%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(1 9% - 20%)
22%
(22% - 23%)
7%
(7% - 7%)
16%
(1 6% - 1 6%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(1 0% - 1 1 %)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(14% -15%)
14%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(14% -15%)
15%
(15% -15%)
15%
(14% -15%)
15%
(14% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
29%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-99
-------
Table E-99. Percent Reduction from the Current Standards: Estimated Annual Incidence of Respiratory Mortality Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Mortality Associated with Short-Term Exposure
to PM2.5 Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-80%
(-78% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-57% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -43%)
-58%
(-57% - -59%)
-1 2%
(-11% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
7%
(7% - 8%)
5%
(5% - 6%)
7%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
7%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(12% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 7%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(1 9% - 20%)
22%
(22% - 23%)
7%
(7% - 7%)
16%
(1 6% - 1 6%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(1 0% - 1 1 %)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(14% -15%)
14%
(14% -15%)
15%
(15% -15%)
15%
(14% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(14% -15%)
15%
(14% -15%)
15%
(14% -15%)
15%
(14% -15%)
12/25
24%
(23% - 24%)
29%
(29% - 29%)
29%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. "Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-100
-------
Table E-100. Estimated Annual Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to Ambient
PM2.5 Concentrations in a Recent Year (2005) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Total Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to PM25 Concentrations
in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
43
(-28 - 1 1 5)
262
(192-331)
21
(-1 4 - 56)
31
(-20-81)
345
(253 - 435)
38
(0 - 75)
60
(-39 - 1 58)
418
(5 - 827)
952
(700 - 1 204)
233
(171 -294)
108
(1 -213)
222
(163-280)
13
(0 - 25)
231
(170-293)
26
(-65 - 1 1 3)
15/353
40
(-26 - 1 05)
247
(182-313)
17
(-1 1 - 44)
31
(-20-81)
280
(206 - 354)
21
(0-41)
56
(-37 - 1 49)
264
(3 - 523)
792
(582 - 1 002)
214
(157-271)
108
(1 -213)
157
(115-199)
8
(0-16)
207
(152-262)
21
(-52 - 92)
14/35
37
(-24 - 98)
234
(172-295)
15
(-10-41)
31
(-20-81)
278
(204 - 351 )
21
(0-41)
52
(-34 - 1 38)
264
(3 - 523)
792
(582 - 1 002)
214
(157-271)
108
(1 -213)
157
(115-199)
8
(0-16)
195
(143-246)
21
(-52 - 92)
13/35
34
(-22 - 90)
216
(159-273)
14
(-9-37)
31
(-20-81)
257
(189-325)
21
(0-41)
48
(-31 -127)
264
(3 - 523)
772
(567 - 976)
200
(147-253)
108
(1 -213)
153
(112-193)
8
(0-16)
180
(132-228)
21
(-52 - 92)
12/35
31
(-20 - 82)
199
(146-251)
13
(-8-34)
28
(-19-75)
236
(173-298)
21
(0-41)
44
(-29-115)
249
(3 - 494)
709
(521 - 897)
184
(135-233)
102
(1 - 203)
145
(106-183)
8
(0-16)
165
(121 -209)
21
(-52 - 92)
13/30
34
(-22 - 90)
212
(155-267)
14
(-9-37)
31
(-20-81)
239
(176-302)
18
(0 - 35)
48
(-31 -127)
225
(3 - 447)
677
(497 - 857)
183
(134-232)
101
(1 - 200)
134
(98-170)
7
(0-13)
177
(130-224)
18
(-44 - 79)
12/25
30
(-20 - 81 )
176
(129-222)
12
(-8-31)
28
(-19-75)
198
(146-251)
15
(0 - 29)
44
(-29 - 1 1 5)
187
(2-371)
562
(413-711)
152
(112-192)
84
(1 -166)
111
(82-141)
6
(0-11)
147
(108-186)
15
(-37 - 65)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function
estimates were not available from this study.
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-101
-------
Table E-101. Estimated Annual Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to Ambient
PM2.5 Concentrations in a Recent Year (2006) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Total Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to PM25 Concentrations
in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
44
(-29 - 1 1 7)
227
(167-287)
20
(-1 3 - 53)
26
(-1 7 - 68)
278
(204-351)
40
(0 - 80)
58
(-38 - 1 54)
392
(5 - 776)
822
(604 - 1 040)
218
(160-276)
113
(1 - 224)
190
(140-240)
12
(0 - 23)
191
(140-241)
22
(-54 - 95)
15/353
41
(-27 - 1 08)
214
(157-271)
16
(-10-42)
26
(-17-68)
225
(165-285)
22
(0 - 44)
55
(-36 - 1 45)
248
(3-491)
684
(502 - 865)
201
(147-254)
113
(1 - 224)
134
(98 - 1 69)
7
(0-15)
171
(126-216)
18
(-44 - 78)
14/35
38
(-25 - 99)
203
(149-256)
15
(-10-38)
26
(-17-68)
224
(164-283)
22
(0 - 44)
51
(-33 - 1 34)
248
(3-491)
684
(502 - 865)
201
(147-254)
113
(1 - 224)
134
(98 - 1 69)
7
(0-15)
160
(118-203)
18
(-44 - 78)
13/35
35
(-23 - 91 )
187
(138-237)
13
(-9 - 35)
26
(-17-68)
207
(152-261)
22
(0 - 44)
47
(-31 -123)
248
(3-491)
666
(489 - 843)
188
(138-237)
113
(1 - 224)
130
(96 - 1 65)
7
(0-15)
148
(109-188)
18
(-44 - 78)
12/35
31
(-21 - 83)
172
(126-218)
12
(-8 - 32)
24
(-16-63)
190
(139-240)
22
(0 - 44)
43
(-28-113)
234
(3 - 463)
612
(449 - 774)
173
(127-218)
107
(1 -212)
124
(91 -157)
7
(0-15)
136
(100-172)
18
(-44 - 78)
13/30
35
(-23 - 91 )
183
(135-232)
13
(-9 - 35)
26
(-17-68)
192
(141 -243)
19
(0 - 38)
47
(-31 -123)
211
(3-419)
585
(429 - 740)
172
(126-217)
106
(1 - 209)
114
(84 - 1 44)
6
(0-12)
146
(107-185)
15
(-37 - 66)
12/25
31
(-20 - 82)
152
(112-192)
11
(-7-29)
24
(-16-63)
160
(117-202)
16
(0-31)
43
(-28 - 1 1 3)
175
(2 - 348)
485
(356-614)
142
(105-180)
88
(1-174)
95
(69 - 1 20)
5
(0-10)
121
(89-153)
13
(-31 - 55)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function
estimates were not available from this study.
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-102
-------
Table E-102. Estimated Annual Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to Ambient
PM2.5 Concentrations in a Recent Year (2007) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Total Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to PM25 Concentrations
in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m)2:
Recent Ambient
PM2.5
Concentrations
45
(-29 - 1 1 9)
229
(168-289)
21
(-14-54)
28
(-18-73)
288
(211 -364)
42
(1 -83)
60
(-39 - 1 58)
408
(5 - 807)
905
(665 - 1 1 44)
221
(162-279)
108
(1-215)
199
(146-251)
15
(0 - 29)
199
(146-251)
23
(-57-101)
15/353
41
(-27 - 1 09)
216
(159-273)
16
(-11-43)
28
(-18-73)
233
(171 -295)
23
(0 - 46)
56
(-37 - 1 49)
258
(3-511)
752
(552 - 951 )
203
(149-257)
108
(1-215)
140
(103-177)
9
(0-18)
178
(131 -225)
19
(-46 - 82)
14/35
38
(-25-101)
204
(150-258)
15
(-10-39)
28
(-18-73)
232
(170-293)
23
(0 - 46)
52
(-34 - 1 38)
258
(3-511)
752
(552 - 951 )
203
(149-257)
108
(1-215)
140
(103-177)
9
(0-18)
167
(123-212)
19
(-46 - 82)
13/35
35
(-23 - 92)
189
(139-239)
14
(-9 - 36)
28
(-18-73)
214
(157-271)
23
(0 - 46)
48
(-31 -127)
258
(3-511)
733
(538 - 927)
190
(140-240)
108
(1-215)
136
(100-172)
9
(0-18)
155
(114-196)
19
(-46 - 82)
12/35
32
(-21 - 84)
174
(127-220)
12
(-8 - 33)
26
(-17-68)
197
(144-249)
23
(0 - 46)
44
(-29-116)
243
(3 - 482)
673
(494 - 852)
175
(128-221)
103
(1 - 204)
129
(95 - 1 64)
9
(0-18)
142
(104-180)
19
(-46 - 82)
13/30
35
(-23 - 92)
185
(136-234)
14
(-9 - 36)
28
(-18-73)
199
(146-252)
20
(0 - 39)
48
(-31 -127)
220
(3 - 436)
643
(472-814)
174
(128-220)
102
(1 -201)
119
(88-151)
8
(0-16)
152
(112-193)
16
(-39 - 70)
12/25
31
(-21 - 83)
153
(113-194)
11
(-7-30)
26
(-17-68)
165
(121 -209)
16
(0 - 32)
44
(-29 - 1 1 6)
182
(2 - 362)
534
(392 - 676)
144
(106-183)
84
(1-167)
99
(73-125)
7
(0-13)
126
(93-160)
13
(-33 - 58)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function
estimates were not available from this study.
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-103
-------
Table E-103. Estimated Percent of Total Annual Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and
Alternative Standards, Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Ex
Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual
Standard
Recent PM25
Concentrations
0.41 %
(-0.27% - 1 .09%)
1 .59%
(1.1 7% -2.01%)
0.42%
(-0.28% -1.1 2%)
0.32%
(-0.21 % - 0.85%)
1 .65%
(1 .22% - 2.09%)
0.81 %
(0.01% -1.59%)
0.35%
(-0.23% - 0.93%)
0.77%
(0.01 % - 1 .52%)
1 .49%
(1 .09% - 1 .88%)
1 .41 %
(1 .04% - 1 .79%)
0.53%
(0.01% -1.05%)
1 .72%
(1.26% -2. 17%)
0.52%
(0.01% -1.03%)
1 .64%
(1.21% -2.08%)
0.76%
(-1 .86% - 3.26%)
15/353
0.4%
(-0.2% -1%)
1 .5%
(1.1% -1.9%)
0.3%
(-0.2% - 0.9%)
0.3%
(-0.2% - 0.9%)
1 .3%
(1 % - 1 .7%)
0.4%
(0% - 0.9%)
0.3%
(-0.2% - 0.9%)
0.5%
(0% - 1 %)
1 .2%
(0.9% - 1 .6%)
1 .3%
(1 % - 1 .6%)
0.5%
(0%-1.1%)
1 .2%
(0.9% - 1 .5%)
0.3%
(0% - 0.7%)
1 .5%
(1.1% -1.9%)
0.6%
(-1 .5% - 2.7%)
14/35
0.35%
(-0.23% - 0.93%)
1 .42%
(1 .05% - 1 .8%)
0.31 %
(-0.2% -0.81%)
0.32%
(-0.21 % - 0.85%)
1 .33%
(0.98% - 1 .68%)
0.44%
(0.01 % - 0.88%)
0.31 %
(-0.2% - 0.82%)
0.49%
(0.01% -0.96%)
1 .24%
(0.91% -1.57%)
1 .3%
(0.96% - 1 .64%)
0.53%
(0.01 % - 1 .05%)
1 .22%
(0.89% - 1 .54%)
0.33%
(0% - 0.65%)
1 .38%
(1 .02% - 1 .75%)
0.62%
(-1 .5% - 2.65%)
13/35
0.32%
(-0.21 % - 0.85%)
1 .32%
(0.97% - 1 .67%)
0.28%
(-0.1 8% -0.75%)
0.32%
(-0.21% -0.85%)
1 .23%
(0.91% -1.56%)
0.44%
(0.01% -0.88%)
0.28%
(-0.1 9% -0.75%)
0.49%
(0.01% -0.96%)
1.21%
(0.89% - 1 .53%)
1 .22%
(0.89% - 1 .54%)
0.53%
(0.01% -1.05%)
1.19%
(0.87% - 1 .5%)
0.33%
(0% - 0.65%)
1 .28%
(0.94% - 1 .62%)
0.62%
(-1 .5% - 2.65%)
12/35
0.29%
(-0.1 9% -0.78%)
1 .21 %
(0.89% - 1 .53%)
0.26%
(-0.1 7% -0.68%)
0.3%
(-0.2% - 0.79%)
1.13%
(0.83% - 1 .43%)
0.44%
(0.01 % - 0.88%)
0.26%
(-0.1 7% -0.68%)
0.46%
(0.01% -0.91%)
1.11%
(0.81% -1.4%)
1.12%
(0.82% - 1 .41 %)
0.51 %
(0.01% -1%)
1.13%
(0.83% - 1 .42%)
0.33%
(0% - 0.65%)
1.17%
(0.86% - 1 .48%)
0.62%
(-1 .5% - 2.65%)
13/30
0.32%
(-0.21% -0.85%)
1 .29%
(0.95% - 1 .63%)
0.28%
(-0.1 8% -0.75%)
0.32%
(-0.21% -0.85%)
1.15%
(0.84% - 1 .45%)
0.38%
(0% - 0.75%)
0.28%
(-0.1 9% -0.75%)
0.41%
(0% - 0.82%)
1 .06%
(0.78% - 1 .34%)
1.11%
(0.82% -1.41%)
0.5%
(0.01 % - 0.99%)
1 .04%
(0.76% - 1 .32%)
0.28%
(0% - 0.56%)
1 .26%
(0.92% - 1 .59%)
0.53%
(-1 .28% - 2.27%)
josure to PM2.s
n) and Daily (m)
12/25
0.29%
(-0.1 9% -0.77%)
1 .07%
(0.79% - 1 .35%)
0.23%
(-0.1 5% -0.62%)
0.3%
(-0.2% - 0.79%)
0.95%
(0.7% - 1 .2%)
0.31 %
(0% - 0.62%)
0.26%
(-0.1 7% -0.68%)
0.34%
(0% - 0.68%)
0.88%
(0.65% -1.11%)
0.92%
(0.68% -1.1 7%)
0.41 %
(0% - 0.82%)
0.86%
(0.63% - 1 .09%)
0.23%
(0% - 0.46%)
1 .04%
(0.77% - 1 .32%)
0.44%
(-1 .05% - 1 .89%)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function
estimates were not available from this study.
2 Percents rounded to the nearest hundredth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-104
-------
Table E-104. Estimated Percent of Total Annual Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and
Alternative Standards, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Ex
Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual
Standard
Recent PM25
Concentrations
0.41 %
(-0.27% - 1 .08%)
1 .39%
(1 .02% - 1 .75%)
0.4%
(-0.26% - 1 .06%)
0.27%
(-0.1 7% -0.7%)
1 .34%
(0.98% - 1 .69%)
0.85%
(0.01% -1.68%)
0.33%
(-0.22% - 0.88%)
0.71 %
(0.01% -1.41%)
1 .27%
(0.93% - 1 .61 %)
1 .34%
(0.99% - 1 .7%)
0.54%
(0.01 % - 1 .07%)
1 .5%
(1.1% -1.89%)
0.46%
(0.01% -0.92%)
1 .36%
(1%-1.72%)
0.63%
(-1 .53% - 2.69%)
15/353
0.4%
(-0.2% -1%)
1 .3%
(1 % - 1 .7%)
0.3%
(-0.2% - 0.8%)
0.3%
(-0.2% - 0.7%)
1.1%
(0.8% - 1 .4%)
0.5%
(0% - 0.9%)
0.3%
(-0.2% - 0.8%)
0.4%
(0% - 0.9%)
1.1%
(0.8% - 1 .3%)
1 .2%
(0.9% - 1 .6%)
0.5%
(0%-1.1%)
1.1%
(0.8% - 1 .3%)
0.3%
(0% - 0.6%)
1 .2%
(0.9% - 1 .5%)
0.5%
(-1 .2% - 2.2%)
14/35
0.35%
(-0.23% - 0.92%)
1 .24%
(0.91 % - 1 .57%)
0.29%
(-0.1 9% -0.77%)
0.27%
(-0.1 7% -0.7%)
1 .08%
(0.79% - 1 .37%)
0.47%
(0.01 % - 0.93%)
0.29%
(-0.1 9% -0.77%)
0.45%
(0.01 % - 0.89%)
1 .06%
(0.78% - 1 .34%)
1 .24%
(0.91% -1.56%)
0.54%
(0.01 % - 1 .07%)
1 .05%
(0.77% - 1 .33%)
0.29%
(0% - 0.58%)
1.14%
(0.84% - 1 .45%)
0.51 %
(-1.23% -2. 19%)
13/35
0.32%
(-0.21 % - 0.84%)
1.15%
(0.84% - 1 .45%)
0.27%
(-0.1 7% -0.71%)
0.27%
(-0.1 7% -0.7%)
1%
(0.73% - 1 .26%)
0.47%
(0.01% -0.93%)
0.27%
(-0.1 7% -0.71%)
0.45%
(0.01% -0.89%)
1 .03%
(0.76% - 1 .3%)
1.16%
(0.85% - 1 .46%)
0.54%
(0.01% -1.07%)
1 .03%
(0.75% - 1 .3%)
0.29%
(0% - 0.58%)
1 .06%
(0.78% - 1 .34%)
0.51%
(-1.23% -2. 19%)
12/35
0.29%
(-0.1 9% -0.77%)
1 .05%
(0.77% - 1 .33%)
0.24%
(-0.1 6% -0.64%)
0.25%
(-0.1 6% -0.65%)
0.92%
(0.67% -1.1 6%)
0.47%
(0.01 % - 0.93%)
0.24%
(-0.1 6% -0.64%)
0.42%
(0.01 % - 0.84%)
0.95%
(0.7% - 1 .2%)
1 .06%
(0.78% - 1 .34%)
0.51 %
(0.01 % - 1 .02%)
0.98%
(0.72% - 1 .23%)
0.29%
(0% - 0.58%)
0.97%
(0.71% -1.23%)
0.51 %
(-1.23% -2. 19%)
13/30
0.32%
(-0.21% -0.84%)
1.12%
(0.82% - 1 .42%)
0.27%
(-0.1 7% -0.71%)
0.27%
(-0.1 7% -0.7%)
0.93%
(0.68% -1.1 8%)
0.4%
(0% - 0.79%)
0.27%
(-0.1 7% -0.71%)
0.38%
(0% - 0.76%)
0.9%
(0.66% -1.1 4%)
1 .06%
(0.78% - 1 .34%)
0.5%
(0.01 %-1%)
0.9%
(0.66% -1.1 4%)
0.25%
(0% - 0.49%)
1 .04%
(0.76% - 1 .32%)
0.43%
(-1 .05% - 1 .87%)
josure to PM2.s
n) and Daily (m)
12/25
0.29%
(-0.1 9% -0.76%)
0.93%
(0.68% -1.1 8%)
0.22%
(-0.1 4% -0.59%)
0.25%
(-0.1 6% -0.65%)
0.77%
(0.57% - 0.97%)
0.33%
(0% - 0.66%)
0.24%
(-0.1 6% -0.64%)
0.32%
(0% - 0.63%)
0.75%
(0.55% - 0.95%)
0.88%
(0.64% -1.11%)
0.42%
(0.01 % - 0.83%)
0.75%
(0.55% - 0.94%)
0.21 %
(0% - 0.41 %)
0.86%
(0.63% - 1 .09%)
0.36%
(-0.87% - 1 .56%)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function
estimates were not available from this study.
2 Percents rounded to the nearest hundredth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-105
-------
Table E-105. Estimated Percent of Total Annual Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term
Exposure to Ambient PM2 5 Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and
Alternative Standards, Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Ex
Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual
Standard
Recent PM25
Concentrations
0.4%
(-0.26% - 1 .06%)
1 .41 %
(1 .03% - 1 .78%)
0.41 %
(-0.27% - 1 .09%)
0.28%
(-0.1 8% -0.74%)
1 .4%
(1 .03% - 1 .77%)
0.86%
(0.01% -1.7%)
0.33%
(-0.22% - 0.88%)
0.72%
(0.01% -1.43%)
1 .39%
(1 .02% - 1 .75%)
1 .38%
(1.01% -1.74%)
0.5%
(0.01 % - 0.99%)
1 .58%
(1.1 6% -2%)
0.56%
(0.01% -1.11%)
1 .42%
(1 .04% - 1 .79%)
0.65%
(-1 .58% - 2.77%)
15/353
0.4%
(-0.2% -1%)
1 .3%
(1 % - 1 .7%)
0.3%
(-0.2% - 0.9%)
0.3%
(-0.2% - 0.7%)
1.1%
(0.8% - 1 .4%)
0.5%
(0% - 0.9%)
0.3%
(-0.2% - 0.8%)
0.5%
(0% - 0.9%)
1 .2%
(0.8% - 1 .5%)
1 .3%
(0.9% - 1 .6%)
0.5%
(0% - 1 %)
1.1%
(0.8% - 1 .4%)
0.4%
(0% - 0.7%)
1 .3%
(0.9% - 1 .6%)
0.5%
(-1 .3% - 2.3%)
14/35
0.34%
(-0.22% - 0.9%)
1 .26%
(0.92% - 1 .59%)
0.3%
(-0.1 9% -0.79%)
0.28%
(-0.1 8% -0.74%)
1.13%
(0.83% - 1 .42%)
0.48%
(0.01 % - 0.94%)
0.29%
(-0.1 9% -0.77%)
0.46%
(0.01% -0.91%)
1.15%
(0.85% - 1 .46%)
1 .27%
(0.93% - 1 .6%)
0.5%
(0.01% -0.99%)
1.11%
(0.82% - 1 .41 %)
0.36%
(0% - 0.7%)
1.19%
(0.88% - 1 .51 %)
0.52%
(-1 .28% - 2.26%)
13/35
0.31%
(-0.2% - 0.83%)
1.16%
(0.85% - 1 .47%)
0.27%
(-0.1 8% -0.72%)
0.28%
(-0.1 8% -0.74%)
1 .04%
(0.76% - 1 .32%)
0.48%
(0.01% -0.94%)
0.27%
(-0.1 7% -0.71%)
0.46%
(0.01% -0.91%)
1.12%
(0.83% - 1 .42%)
1.18%
(0.87% - 1 .5%)
0.5%
(0.01 % - 0.99%)
1 .08%
(0.8% - 1 .37%)
0.36%
(0% - 0.7%)
1.1%
(0.81 % - 1 .4%)
0.52%
(-1 .28% - 2.26%)
12/35
0.28%
(-0.1 9% -0.75%)
1 .07%
(0.78% - 1 .35%)
0.25%
(-0.1 6% -0.66%)
0.26%
(-0.1 7% -0.68%)
0.96%
(0.7% -1.21%)
0.48%
(0.01% -0.94%)
0.24%
(-0.1 6% -0.64%)
0.43%
(0.01 % - 0.86%)
1 .03%
(0.76% - 1 .31 %)
1 .09%
(0.8% - 1 .38%)
0.47%
(0.01% -0.94%)
1 .03%
(0.76% - 1 .3%)
0.36%
(0% - 0.7%)
1 .01 %
(0.74% - 1 .28%)
0.52%
(-1 .28% - 2.26%)
13/30
0.31%
(-0.2% - 0.83%)
1.14%
(0.83% - 1 .44%)
0.27%
(-0.1 8% -0.72%)
0.28%
(-0.1 8% -0.74%)
0.97%
(0.71 % - 1 .23%)
0.41%
(0%-0.81%)
0.27%
(-0.1 7% -0.71%)
0.39%
(0% - 0.78%)
0.99%
(0.72% - 1 .25%)
1 .08%
(0.79% - 1 .37%)
0.47%
(0.01 % - 0.93%)
0.95%
(0.7% - 1 .2%)
0.3%
(0% - 0.6%)
1 .09%
(0.8% - 1 .37%)
0.45%
(-1 .09% - 1 .93%)
josure to PM2.s
n) and Daily (m)
12/25
0.28%
(-0.1 8% -0.74%)
0.94%
(0.69% -1.1 9%)
0.23%
(-0.1 5% -0.6%)
0.26%
(-0.1 7% -0.68%)
0.8%
(0.59% - 1 .02%)
0.34%
(0% - 0.67%)
0.24%
(-0.1 6% -0.64%)
0.32%
(0% - 0.64%)
0.82%
(0.6% - 1 .04%)
0.9%
(0.66% -1.1 4%)
0.39%
(0% - 0.77%)
0.79%
(0.58% - 1 %)
0.25%
(0% - 0.5%)
0.9%
(0.66% -1.1 4%)
0.37%
(-0.9% - 1 .6%)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function
estimates were not available from this study.
2 Percents rounded to the nearest hundredth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-106
-------
Table E-106. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiovascular Hospital Admissions Associated with
Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current J
Term Exposure to PM2.5Concentrat
Annual (n
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-81 % - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-41%
(-41 % --41%)
-59%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
Standards: Annual Incidence of Cardiovascular Hospital Admissions Associated with Short-
ens in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative
and Daily (m) Standards (Standard Combination Denoted n/m)2:
14/35
8%
(7% - 8%)
6%
(5% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
8%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(13% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
23%
(23% - 23%)
6%
(6% - 6%)
10%
(10% -10%)
14%
(14% -14%)
5%
(5% - 5%)
8%
(8% - 8%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
14%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
12/25
24%
(24% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(23% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function estimates
were not available from this study.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-107
-------
Table E-107. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiovascular Hospital Admissions Associated with
Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiovascular Hospital Admissions Associated with Short-
Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-81 % - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -42%)
-59%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
8%
(7% - 8%)
6%
(6% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
8%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(13% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(20% - 20%)
23%
(23% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
23%
(23% - 23%)
6%
(6% - 6%)
10%
(10% -11%)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
12/25
24%
(24% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(23% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function estimates
were not available from this study.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-108
-------
Table E-108. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiovascular Hospital Admissions Associated with
Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current J
Term Exposure to PM2.5Concentrat
Annual (n
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-81 % - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-42%
(-42% - -42%)
-59%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
Standards: Annual Incidence of Cardiovascular Hospital Admissions Associated with Short-
ens in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative
and Daily (m) Standards (Standard Combination Denoted n/m)2:
14/35
8%
(7% - 8%)
6%
(6% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1 % - 1 %)
0%
(0% - 0%)
8%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(13% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 6%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
23%
(23% - 23%)
6%
(6% - 6%)
10%
(10% -11%)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
14%
(14% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
14%
(14% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
12/25
24%
(24% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(23% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-specific C-R function estimates
were not available from this study.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-109
-------
Table E-109. Estimated Annual Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
19
(-23 - 60)
21
(-12-54)
9
(-11-29)
15
(-18-47)
31
(-18-79)
25
(6 - 44)
27
(-34 - 86)
269
(63 - 473)
79
(-46 - 203)
19
(-1 1 - 48)
61
(14-107)
18
(-11-47)
9
(2-16)
28
(-16-72)
2
(-34 - 37)
15/353
17
(-22 - 55)
20
(-12-51)
7
(-9 - 23)
15
(-18-47)
25
(-15-64)
14
(3 - 25)
25
(-32 - 81 )
170
(40 - 300)
65
(-38 - 1 69)
17
(-10-44)
61
(14-107)
13
(-8 - 33)
6
(1-10)
25
(-15-64)
2
(-27 - 30)
14/35
16
(-20 - 51 )
19
(-1 1 - 48)
7
(-8-21)
15
(-18-47)
25
(-1 5 - 64)
14
(3 - 25)
23
(-29 - 75)
170
(40 - 300)
65
(-38-169)
17
(-1 0 - 44)
61
(14-107)
13
(-8 - 33)
6
(1-10)
23
(-1 4 - 60)
2
(-27 - 30)
13/35
15
(-18-47)
17
(-10-45)
6
(-8 - 20)
15
(-18-47)
23
(-13-59)
14
(3 - 25)
21
(-27 - 69)
170
(40 - 300)
64
(-37 - 1 64)
16
(-9-41)
61
(14-107)
12
(-7 - 32)
6
(1-10)
22
(-13-56)
2
(-27 - 30)
12/35
13
(-17-43)
16
(-9-41)
6
(-7-18)
14
(-17-44)
21
(-12-54)
14
(3 - 25)
19
(-24 - 63)
161
(37 - 283)
58
(-34-151)
15
(-9 - 38)
58
(14- 102)
12
(-7 - 30)
6
(1-10)
20
(-12-51)
2
(-27 - 30)
13/30
15
(-18-47)
17
(-10-44)
6
(-8 - 20)
15
(-18-47)
21
(-13-55)
12
(3-21)
21
(-27 - 69)
145
(34 - 256)
56
(-33 - 1 44)
15
(-9 - 38)
57
(13-101)
11
(-6 - 28)
5
(1 -9)
21
(-13-55)
2
(-23 - 26)
12/25
13
(-16-42)
14
(-8 - 36)
5
(-6-16)
14
(-17-44)
18
(-10-46)
10
(2-17)
19
(-24 - 63)
121
(28 - 21 3)
46
(-27-120)
12
(-7-31)
47
(11-84)
9
(-5 - 23)
4
(1-7)
18
(-10-46)
1
(-19-21)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in
Table 2 of Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-110
-------
Table E-110. Estimated Annual Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2006 PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
19
(-24 - 61 )
18
(-11-47)
9
(-1 1 - 28)
12
(-15-40)
25
(-15-64)
27
(6 - 47)
26
(-33 - 84)
253
(59 - 444)
68
(-40-175)
17
(-10-45)
64
(15-112)
16
(-9 - 40)
8
(2-15)
23
(-14-59)
2
(-28 - 31 )
15/353
17
(-22 - 56)
17
(-10-44)
7
(-8 - 22)
12
(-15-40)
20
(-12-52)
15
(3 - 26)
25
(-31 - 79)
160
(37 - 281 )
56
(-33 - 1 45)
16
(-9-41)
64
(15-112)
11
(-6 - 28)
5
(1 -9)
21
(-12-53)
2
(-23 - 25)
14/35
16
(-20 - 52)
16
(-1 0 - 42)
6
(-8 - 20)
12
(-1 5 - 40)
20
(-12-51)
15
(3 - 26)
23
(-29 - 73)
160
(37-281)
56
(-33-145)
16
(-9-41)
64
(15-112)
11
(-6 - 28)
5
(1 -9)
19
(-1 1 - 50)
2
(-23 - 25)
13/35
15
(-19-48)
15
(-9 - 39)
6
(-7-19)
12
(-15-40)
18
(-11 -47)
15
(3 - 26)
21
(-26 - 68)
160
(37 - 281 )
55
(-32 - 1 42)
15
(-9 - 39)
64
(15-112)
11
(-6 - 27)
5
(1-9)
18
(-10-46)
2
(-23 - 25)
12/35
13
(-17-44)
14
(-8 - 36)
5
(-7-17)
11
(-14-37)
17
(-10-44)
15
(3 - 26)
19
(-24 - 62)
151
(35 - 265)
50
(-30- 130)
14
(-8 - 36)
61
(14- 107)
10
(-6 - 26)
5
(1-9)
16
(-10-42)
2
(-23 - 25)
13/30
15
(-19-48)
15
(-9 - 38)
6
(-7-19)
12
(-15-40)
17
(-10-44)
13
(3 - 22)
21
(-26 - 68)
136
(32 - 240)
48
(-28 - 1 24)
14
(-8 - 35)
60
(14-105)
9
(-5 - 24)
5
(1 -8)
18
(-10-45)
1
(-19-22)
12/25
13
(-17-43)
12
(-7-31)
5
(-6-15)
11
(-14-37)
14
(-8 - 37)
11
(2-19)
19
(-24 - 62)
113
(26-199)
40
(-24-103)
11
(-7 - 29)
50
(12-87)
8
(-5 - 20)
4
(1-7)
15
(-9 - 38)
1
(-16-18)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in
Table 2 of Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-lll
-------
Table E-lll. Estimated Annual Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards,
Based on Adjusting 2007 PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m)2:
Recent PM2.5
Concentrations
19
(-24 - 62)
18
(-11-47)
9
(-11-29)
13
(-17-43)
26
(-15-66)
28
(7 - 49)
27
(-34 - 87)
263
(61 -461)
75
(-44-193)
18
(-10-46)
61
(14-108)
16
(-10-42)
11
(2-19)
24
(-14-62)
2
(-30 - 33)
15/353
18
(-22 - 57)
17
(-10-45)
7
(-9 - 22)
13
(-17-43)
21
(-12-54)
15
(4 - 27)
25
(-32 - 82)
166
(39 - 293)
62
(-37 - 1 60)
16
(-10-42)
61
(14-108)
11
(-7 - 29)
7
(2-12)
21
(-13-55)
2
(-24 - 27)
14/35
16
(-21 - 53)
16
(-1 0 - 42)
6
(-8-21)
13
(-17-43)
21
(-1 2 - 53)
15
(4 - 27)
23
(-29 - 76)
166
(39 - 293)
62
(-37-160)
16
(-1 0 - 42)
61
(14-108)
11
(-7 - 29)
7
(2-12)
20
(-1 2 - 52)
2
(-24 - 27)
13/35
15
(-19-48)
15
(-9 - 39)
6
(-7-19)
13
(-17-43)
19
(-1 1 - 49)
15
(4 - 27)
21
(-27 - 69)
166
(39 - 293)
60
(-36 - 1 56)
15
(-9 - 39)
61
(14-108)
11
(-7 - 29)
7
(2-12)
19
(-1 1 - 48)
2
(-24 - 27)
12/35
14
(-17-44)
14
(-8 - 36)
5
(-7-17)
12
(-15-40)
18
(-10-45)
15
(4 - 27)
20
(-25 - 63)
157
(37 - 276)
56
(-33- 143)
14
(-8 - 36)
58
(14- 103)
11
(-6 - 27)
7
(2-12)
17
(-10-44)
2
(-24 - 27)
13/30
15
(-19-48)
15
(-9 - 38)
6
(-7-19)
13
(-17-43)
18
(-10-46)
13
(3 - 23)
21
(-27 - 69)
142
(33 - 250)
53
(-31 -137)
14
(-8 - 36)
57
(13-101)
10
(-6 - 25)
6
(1-10)
18
(-11 -47)
2
(-21 - 23)
12/25
13
(-17-44)
12
(-7 - 32)
5
(-6-16)
12
(-15-40)
15
(-9 - 38)
11
(3-19)
20
(-25 - 63)
118
(27 - 207)
44
(-26 - 1 1 3)
12
(-7 - 30)
48
(11-84)
8
(-5-21)
5
(1 -8)
15
(-9 - 39)
1
(-17-19)
Incidence estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in
Table 2 of Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2 Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-112
-------
Table E-112. Estimated Percent of Total Annual Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to
Ambient PM2.5 Concentrations in a Recent Year (2005) and PM2.5 Concentrations that Just Meet the Current and Alternative
Standards, Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Expc
Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual
Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
0.5%
(-0.63% - 1 .61 %)
0.42%
(-0.25% - 1 .08%)
0.51 %
(-0.65% - 1 .65%)
0.39%
(-0.49% - 1 .26%)
0.43%
(-0.26% -1.1 2%)
1 .42%
(0.33% - 2.49%)
0.43%
(-0.54% - 1 .38%)
1 .36%
(0.32% - 2.38%)
0.39%
(-0.23% - 1 .01 %)
0.37%
(-0.22% - 0.95%)
0.94%
(0.22% - 1 .65%)
0.45%
(-0.27% -1.1 6%)
0.92%
(0.21 % - 1 .61 %)
0.43%
(-0.25% -1.11%)
0.2%
(-2.72% - 2.96%)
15/353
0.46%
(-0.58% - 1 .48%)
0.39%
(-0.23% - 1 .02%)
0.4%
(-0.51 % - 1 .3%)
0.39%
(-0.49% - 1 .26%)
0.35%
(-0.21 % - 0.91 %)
0.78%
(0.1 8% -1.38%)
0.4%
(-0.51 % - 1 .3%)
0.86%
(0.2% -1.51%)
0.32%
(-0.1 9% -0.84%)
0.34%
(-0.2% - 0.88%)
0.94%
(0.22% - 1 .65%)
0.32%
(-0.1 9% -0.82%)
0.58%
(0.1 4% -1.02%)
0.39%
(-0.23% - 0.99%)
0.16%
(-2. 19% -2. 41%)
14/35
0.42%
(-0.53% - 1 .37%)
0.37%
(-0.22% - 0.96%)
0.37%
(-0.47% - 1 .2%)
0.39%
(-0.49% - 1 .26%)
0.35%
(-0.21 % - 0.9%)
0.78%
(0.1 8% -1.38%)
0.37%
(-0.47% - 1 .2%)
0.86%
(0.2% - 1 .51 %)
0.32%
(-0.1 9% -0.84%)
0.34%
(-0.2% - 0.88%)
0.94%
(0.22% - 1 .65%)
0.32%
(-0.1 9% -0.82%)
0.58%
(0.1 4% -1.02%)
0.36%
(-0.21% -0.93%)
0.16%
(-2. 19% -2. 41%)
13/35
0.39%
(-0.49% - 1 .26%)
0.34%
(-0.2% - 0.89%)
0.34%
(-0.43% -1.1%)
0.39%
(-0.49% - 1 .26%)
0.32%
(-0.1 9% -0.83%)
0.78%
(0.1 8% -1.38%)
0.34%
(-0.43% -1.11%)
0.86%
(0.2% - 1 .51 %)
0.32%
(-0.1 9% -0.81%)
0.32%
(-0.1 9% -0.82%)
0.94%
(0.22% - 1 .65%)
0.31 %
(-0.1 8% -0.8%)
0.58%
(0.1 4% -1.02%)
0.33%
(-0.2% - 0.86%)
0.16%
(-2. 19% -2.41%)
12/35
0.35%
(-0.45% -1.1 5%)
0.32%
(-0.1 9% -0.82%)
0.31%
(-0.39% -1.01%)
0.36%
(-0.45% -1.1 7%)
0.3%
(-0.1 7% -0.76%)
0.78%
(0.1 8% -1.38%)
0.31%
(-0.39% -1.01%)
0.81%
(0.1 9% -1.42%)
0.29%
(-0.1 7% -0.75%)
0.29%
(-0.1 7% -0.75%)
0.89%
(0.21 % - 1 .57%)
0.29%
(-0.1 7% -0.76%)
0.58%
(0.1 4% -1.02%)
0.31%
(-0.1 8% -0.79%)
0.16%
(-2. 19% -2. 41%)
13/30
0.39%
(-0.49% - 1 .26%)
0.34%
(-0.2% - 0.87%)
0.34%
(-0.43% -1.1%)
0.39%
(-0.49% - 1 .26%)
0.3%
(-0.1 8% -0.77%)
0.67%
(0.1 6% -1.1 8%)
0.34%
(-0.43% -1.11%)
0.73%
(0.1 7% -1.29%)
0.28%
(-0.1 6% -0.71%)
0.29%
(-0.1 7% -0.75%)
0.88%
(0.21% -1.55%)
0.27%
(-0.1 6% -0.7%)
0.49%
(0.1 2% -0.87%)
0.33%
(-0.1 9% -0.85%)
0.14%
(-1 .87% - 2.06%)
>sureto PM25
n) and Daily (m)
12/25
0.35%
(-0.44% -1.1 3%)
0.28%
(-0.1 6% -0.72%)
0.28%
(-0.36% - 0.92%)
0.36%
(-0.45% -1.1 7%)
0.25%
(-0.1 5% -0.64%)
0.56%
(0.1 3% -0.98%)
0.31 %
(-0.39% - 1 .01 %)
0.61%
(0.1 4% -1.07%)
0.23%
(-0.1 4% -0.59%)
0.24%
(-0.1 4% -0.62%)
0.73%
(0.1 7% -1.29%)
0.23%
(-0.1 3% -0.58%)
0.41%
(0.1% -0.72%)
0.27%
(-0.1 6% -0.7%)
0.11%
(-1 .54% - 1 .71 %)
Estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in Table 2 of
Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2 Percents rounded to the nearest hundredth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-113
-------
Table E-113. Estimated Percent of Total Annual Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to
Ambient PM2 5 Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative
Standards, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM2.5
Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m)
Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
0.49%
(-0.62% - 1 .59%)
0.36%
(-0.21 % - 0.94%)
0.48%
(-0.61 % - 1 .57%)
0.32%
(-0.4% - 1 .04%)
0.35%
(-0.21% -0.9%)
1 .49%
(0.35% - 2.62%)
0.4%
(-0.51% -1.3%)
1 .25%
(0.29% - 2.2%)
0.33%
(-0.2% - 0.86%)
0.35%
(-0.21% -0.91%)
0.95%
(0.22% - 1 .67%)
0.39%
(-0.23% -1.01%)
0.82%
(0.1 9% -1.43%)
0.36%
(-0.21 % - 0.92%)
0.16%
(-2.23% - 2.45%)
15/353
0.45%
(-0.57% - 1 .46%)
0.34%
(-0.2% - 0.89%)
0.38%
(-0.48% - 1 .23%)
0.32%
(-0.4% - 1 .04%)
0.28%
(-0.1 7% -0.73%)
0.83%
(0.1 9% -1.45%)
0.38%
(-0.48% - 1 .23%)
0.79%
(0.1 8% -1.4%)
0.28%
(-0.1 6% -0.71%)
0.32%
(-0.1 9% -0.83%)
0.95%
(0.22% - 1 .67%)
0.28%
(-0.1 6% -0.71%)
0.51%
(0.1 2% -0.91%)
0.32%
(-0.1 9% -0.82%)
0.13%
(-1 .8% - 1 .99%)
14/35
0.42%
(-0.53% - 1 .35%)
0.32%
(-0.1 9% -0.84%)
0.35%
(-0.44% -1.1 4%)
0.32%
(-0.4% - 1 .04%)
0.28%
(-0.1 7% -0.73%)
0.83%
(0.1 9% -1.45%)
0.35%
(-0.44% -1.1 3%)
0.79%
(0.1 8% -1.4%)
0.28%
(-0.1 6% -0.71%)
0.32%
(-0.1 9% -0.83%)
0.95%
(0.22% - 1 .67%)
0.28%
(-0.1 6% -0.71%)
0.51 %
(0.1 2% -0.91%)
0.3%
(-0.1 8% -0.77%)
0.13%
(-1 .8% - 1 .99%)
13/35
0.38%
(-0.48% - 1 .24%)
0.3%
(-0.1 8% -0.77%)
0.32%
(-0.41 % - 1 .04%)
0.32%
(-0.4% - 1 .04%)
0.26%
(-0.1 5% -0.67%)
0.83%
(0.1 9% -1.45%)
0.32%
(-0.4% - 1 .04%)
0.79%
(0.1 8% -1.4%)
0.27%
(-0.1 6% -0.7%)
0.3%
(-0.1 8% -0.78%)
0.95%
(0.22% - 1 .67%)
0.27%
(-0.1 6% -0.69%)
0.51%
(0.1 2% -0.91%)
0.28%
(-0.1 6% -0.71%)
0.13%
(-1 .8% - 1 .99%)
12/35
0.35%
(-0.44% -1.1 3%)
0.28%
(-0.1 6% -0.71%)
0.29%
(-0.37% - 0.95%)
0.3%
(-0.37% - 0.96%)
0.24%
(-0.1 4% -0.62%)
0.83%
(0.1 9% -1.45%)
0.29%
(-0.37% - 0.95%)
0.75%
(0.1 7% -1.32%)
0.25%
(-0.1 5% -0.64%)
0.28%
(-0.1 6% -0.72%)
0.9%
(0.21% -1.59%)
0.26%
(-0.1 5% -0.66%)
0.51 %
(0.1 2% -0.91%)
0.25%
(-0.1 5% -0.65%)
0.13%
(-1 .8% - 1 .99%)
13/30
0.38%
(-0.48% - 1 .24%)
0.29%
(-0.1 7% -0.76%)
0.32%
(-0.41 % - 1 .04%)
0.32%
(-0.4% - 1 .04%)
0.24%
(-0.1 4% -0.63%)
0.71%
(0.1 6% -1.24%)
0.32%
(-0.4% - 1 .04%)
0.68%
(0.1 6% -1.1 9%)
0.24%
(-0.1 4% -0.61%)
0.28%
(-0.1 6% -0.71%)
0.89%
(0.21 % - 1 .57%)
0.24%
(-0.1 4% -0.61%)
0.44%
(0.1% -0.77%)
0.27%
(-0.1 6% -0.7%)
0.11%
(-1 .53% - 1 .7%)
12/25
0.34%
(-0.43% -1.1 2%)
0.24%
(-0.1 4% -0.63%)
0.27%
(-0.34% - 0.87%)
0.3%
(-0.37% - 0.96%)
0.2%
(-0.1 2% -0.52%)
0.58%
(0.1 4% -1.03%)
0.29%
(-0.37% - 0.95%)
0.56%
(0.1 3% -0.99%)
0.2%
(-0.1 2% -0.51%)
0.23%
(-0.1 4% -0.59%)
0.74%
(0.1 7% -1.3%)
0.19%
(-0.11% -0.5%)
0.36%
(0.08% - 0.64%)
0.23%
(-0.1 3% -0.58%)
0.09%
(-1 .26% - 1 .41 %)
Estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in Table 2 of
Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2 Percents rounded to the nearest hundredth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-114
-------
Table E-114. Estimated Percent of Total Annual Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to
Ambient PM2 5 Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative
Standards, Based on Adjusting 2007 PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent of Total Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM2 5
Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m)
Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
0.48%
(-0.61 % - 1 .56%)
0.37%
(-0.22% - 0.95%)
0.5%
(-0.63% - 1 .6%)
0.34%
(-0.42% - 1 .09%)
0.37%
(-0.22% - 0.94%)
1 .52%
(0.36% - 2.66%)
0.4%
(-0.51% -1.3%)
1 .28%
(0.3% - 2.25%)
0.36%
(-0.21 % - 0.94%)
0.36%
(-0.21 % - 0.93%)
0.88%
(0.21% -1.55%)
0.42%
(-0.24% - 1 .07%)
0.99%
(0.23% - 1 .74%)
0.37%
(-0.22% - 0.96%)
0.17%
(-2.31 % - 2.52%)
15/353
0.44%
(-0.56% - 1 .43%)
0.35%
(-0.2% - 0.9%)
0.39%
(-0.49% - 1 .26%)
0.34%
(-0.42% - 1 .09%)
0.3%
(-0.1 7% -0.77%)
0.84%
(0.2% - 1 .48%)
0.38%
(-0.48% - 1 .23%)
0.81%
(0.1 9% -1.42%)
0.3%
(-0.1 8% -0.78%)
0.33%
(-0.2% - 0.85%)
0.88%
(0.21 % - 1 .55%)
0.29%
(-0.1 7% -0.75%)
0.63%
(0.1 5% -1.1%)
0.33%
(-0.2% - 0.86%)
0.14%
(-1 .86% - 2.05%)
14/35
0.41 %
(-0.52% - 1 .32%)
0.33%
(-0.1 9% -0.85%)
0.36%
(-0.45% -1.1 6%)
0.34%
(-0.42% - 1 .09%)
0.29%
(-0.1 7% -0.76%)
0.84%
(0.2% - 1 .48%)
0.35%
(-0.44% -1.1 3%)
0.81 %
(0.1 9% -1.42%)
0.3%
(-0.1 8% -0.78%)
0.33%
(-0.2% - 0.85%)
0.88%
(0.21% -1.55%)
0.29%
(-0.1 7% -0.75%)
0.63%
(0.1 5% -1.1%)
0.31 %
(-0.1 8% -0.81%)
0.14%
(-1 .86% - 2.05%)
13/35
0.38%
(-0.47% - 1 .22%)
0.3%
(-0.1 8% -0.78%)
0.33%
(-0.42% - 1 .07%)
0.34%
(-0.42% - 1 .09%)
0.27%
(-0.1 6% -0.7%)
0.84%
(0.2% - 1 .48%)
0.32%
(-0.41 % - 1 .04%)
0.81%
(0.1 9% -1.42%)
0.29%
(-0.1 7% -0.76%)
0.31%
(-0.1 8% -0.8%)
0.88%
(0.21% -1.55%)
0.28%
(-0.1 7% -0.73%)
0.63%
(0.1 5% -1.1%)
0.29%
(-0.1 7% -0.74%)
0.14%
(-1 .86% - 2.05%)
12/35
0.34%
(-0.43% -1.11%)
0.28%
(-0.1 6% -0.72%)
0.3%
(-0.38% - 0.97%)
0.31 %
(-0.39% - 1 .01 %)
0.25%
(-0.1 5% -0.64%)
0.84%
(0.2% - 1 .48%)
0.29%
(-0.37% - 0.95%)
0.76%
(0.1 8% -1.34%)
0.27%
(-0.1 6% -0.7%)
0.28%
(-0.1 7% -0.73%)
0.84%
(0.2% - 1 .48%)
0.27%
(-0.1 6% -0.7%)
0.63%
(0.1 5% -1.1%)
0.26%
(-0.1 6% -0.68%)
0.14%
(-1 .86% - 2.05%)
13/30
0.38%
(-0.47% - 1 .22%)
0.3%
(-0.1 8% -0.77%)
0.33%
(-0.42% - 1 .07%)
0.34%
(-0.42% - 1 .09%)
0.25%
(-0.1 5% -0.65%)
0.72%
(0.1 7% -1.26%)
0.32%
(-0.41% -1.04%)
0.69%
(0.1 6% -1.22%)
0.26%
(-0.1 5% -0.67%)
0.28%
(-0.1 7% -0.73%)
0.83%
(0.1 9% -1.46%)
0.25%
(-0.1 5% -0.64%)
0.54%
(0.1 2% -0.94%)
0.28%
(-0.1 7% -0.73%)
0.12%
(-1 .59% - 1 .76%)
12/25
0.34%
(-0.43% - 1 .09%)
0.25%
(-0.1 5% -0.64%)
0.27%
(-0.34% - 0.89%)
0.31 %
(-0.39% - 1 .01 %)
0.21 %
(-0.1 2% -0.54%)
0.6%
(0.1 4% -1.05%)
0.29%
(-0.37% - 0.95%)
0.57%
(0.1 3% -1.01%)
0.21 %
(-0.1 3% -0.55%)
0.23%
(-0.1 4% -0.61%)
0.69%
(0.1 6% -1.21%)
0.21 %
(-0.1 2% -0.53%)
0.44%
(0.1% -0.78%)
0.24%
(-0.1 4% -0.61%)
0.1%
(-1.31% -1.46%)
Estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in Table 2 of
Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2 Percents rounded to the nearest hundredth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-115
-------
Table E-115. Percent Reduction from the Current Standards: Estimated Annual Incidence of Respiratory Hospital Admissions Associated with
Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2005 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Hospital Admissions Associated with Short-
Term Exposure to PM2 5 Concentrations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative
Annual (n and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-41%
(-41 % --42%)
-58%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
8%
(7% - 8%)
6%
(6% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
8%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(13% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 7%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
23%
(22% - 23%)
6%
(6% - 6%)
11%
(1 1 % - 1 1 %)
14%
(14% -14%)
5%
(5% - 5%)
8%
(8% - 8%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
12/25
24%
(24% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
Estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in Table 2 of
Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-116
-------
Table E-116. Percent Reduction from the Current Standards: Estimated Annual Incidence of Respiratory Hospital Admissions Associated with
Short-Term Exposure to Ambient PM2.5 Concentrations, Based on Adjusting 2006 PM2.5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Hospital Admissions Associated with Short-
Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative
Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81 %
(-80% - -82%)
-6%
(-60/0 - -60/6)
-58%
(-58% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-43%
(-42% - -43%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
8%
(7% - 8%)
6%
(6% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
8%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(13% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 7%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
23%
(22% - 23%)
6%
(6% - 6%)
11%
(1 1 % - 1 1 %)
14%
(1 4% - 1 4%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
12/25
24%
(24% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
Estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in Table 2 of
Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
''The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-117
-------
Table E-117. Percent Reduction from the Current Standards: Estimated Annual Incidence of Respiratory Hospital Admissions Associated with
Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Hospital Admissions Associated with Short-
Term Exposure to PM25 Concentrations in a Recent Year and PM2.s Concentrations that Just Meet the Current and Alternative
Annual (n and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-43%
(-42% - -43%)
-58%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -24%)
15/353
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
14/35
8%
(7% - 8%)
6%
(6% - 6%)
8%
(7% - 8%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
8%
(7% - 8%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
13/35
15%
(15% -15%)
13%
(13% -13%)
15%
(15% -15%)
0%
(0% - 0%)
8%
(8% - 8%)
0%
(0% - 0%)
15%
(15% -15%)
0%
(0% - 0%)
3%
(3% - 3%)
6%
(6% - 7%)
0%
(0% - 0%)
3%
(3% - 3%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
23%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
23%
(22% - 23%)
6%
(6% - 6%)
11%
(1 1 % - 1 1 %)
14%
(14% -14%)
5%
(5% - 5%)
7%
(7% - 7%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
13/30
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
0%
(0% - 0%)
15%
(15% -15%)
15%
(14% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
6%
(6% - 6%)
15%
(15% -15%)
15%
(15% -15%)
15%
(15% -15%)
15%
(14% -15%)
12/25
24%
(24% - 24%)
29%
(29% - 29%)
30%
(29% - 30%)
7%
(7% - 7%)
29%
(29% - 29%)
29%
(29% - 29%)
23%
(22% - 23%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
22%
(22% - 22%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 29%)
29%
(29% - 30%)
Estimates were calculated using the appropriate regional concentration-response function estimates from models with a 2-day lag for respiratory hospital admissions reported in Table 2 of
Bell et al. (2008). Location-specific C-R function estimates were not available from this study.
2Numbers rounded to the nearest percent. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-118
-------
Table E-118. Estimated Annual Incidence of Emergency Room (ER) Visits Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2005) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005
PM2.5 Concentrations1
Study
Tolbertetal. (2007)
Tolbertetal. (2007)
Ito et al. (2007)
Location
Atlanta, GA
Atlanta, GA
New York, NY
ER Visit for:
Cardiovascular
illness
Respiratory
illness
Asthma
Incidence of ER Visits Associated with Short-Term Exposure to PM2.s Concentrations in a Recent Year and
PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2 5
Concentrations
216
(-304 - 727)
814
(-816-2419)
5235
(3346-7071)
15/352
198
(-279 - 668)
746
(-748 - 2220)
4375
(2790 - 5923)
14/35
183
(-258-618)
690
(-691 - 2055)
4375
(2790 - 5923)
13/35
169
(-237 - 568)
634
(-635-1889)
4265
(2719-5776)
12/35
154
(-216-518)
578
(-578 - 1 723)
3927
(2501 - 5323)
13/30
169
(-237 - 568)
634
(-635 - 1 889)
3754
(2390-5091)
12/25
151
(-212-511)
570
(-570 - 1 698)
3127
(1 987 - 4248)
1Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-119
-------
Table E-119. Estimated Annual Incidence of Emergency Room (ER) Visits Associated with Short-Term Exposure to Ambient PM2 5 Concentrations
in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006
PM2 5 Concentrations1
Study
Tolbertetal. (2007)
Tolbertetal. (2007)
Ito et al. (2007)
Location
Atlanta, GA
Atlanta, GA
New York, NY
ER Visit for:
Cardiovascular
illness
Respiratory
illness
Asthma
Incidence of ER Visits Associated with Short-Term Exposure to PM2 5 Concentrations in a Recent Year and
PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m):
Recent PM2.5
Concentrations
220
(-310-741)
829
(-831 - 2465)
4506
(2876 - 6095)
15/352
202
(-284 - 681 )
761
(-762 - 2263)
3764
(2397 - 51 02)
14/35
187
(-263 - 630)
704
(-705 - 2094)
3764
(2397 - 51 02)
13/35
172
(-241 - 579)
647
(-647 - 1 925)
3669
(2336 - 4974)
12/35
157
(-220 - 528)
589
(-590-1756)
3377
(2149-4582)
13/30
172
(-241 - 579)
647
(-647-1925)
3228
(2053 - 4382)
12/25
154
(-216-521)
581
(-581 - 1 730)
2688
(1707-3654)
1Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-120
-------
Table E-120. Estimated Annual Incidence of Emergency Room (ER) Visits Associated with Short-Term Exposure to Ambient PM25 Concentrations
in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007
PM2 5 Concentrations1
Study
Tolbertetal. (2007)
Tolbertetal. (2007)
Ito et al. (2007)
Location
Atlanta, GA
Atlanta, GA
New York, NY
ER Visit for:
Cardiovascular
illness
Respiratory
illness
Asthma
Incidence of ER Visits Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and
PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
219
(-308 - 738)
825
(-827 - 2453)
4926
(3145-6660)
15/352
201
(-283 - 677)
757
(-758 - 2251 )
4115
(2622 - 5575)
14/35
186
(-261 - 627)
700
(-701 - 2084)
4115
(2622 - 5575)
13/35
171
(-240 - 576)
643
(-644- 1915)
4011
(2555 - 5436)
12/35
156
(-219-526)
586
(-587 - 1 747)
3692
(2350 - 5008)
13/30
171
(-240 - 576)
643
(-644 - 1 91 5)
3529
(2245 - 4790)
12/25
154
(-215-518)
578
(-578-1721)
2939
(1 867 - 3995)
1Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-121
-------
Table E-121. Estimated Percent of Total Annual Incidence of Emergency Room (ER) Visits Associated with Short-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2005 PM2 5 Concentrations1
Study
Tolbertetal. (2007)
Tolbertetal. (2007)
Itoetal. (2007)
Location
Atlanta, GA
Atlanta, GA
New York, NY
ER Visit for:
Cardiovascular
illness
Respiratory
illness
Asthma
Percent of Total Incidence of ER Visits Associated with Short -Term Exposure to PM2.s Concentrations in a
Recent Year and PM2 5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m)
Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
0.6%
(-0.9% - 2.2%)
0.6%
(-0.6% - 1 .9%)
6.1%
(3.9% - 8.2%)
15/352
0.6%
(-0.8% - 2%)
0.6%
(-0.6% - 1 .8%)
5.1%
(3.3% - 6.9%)
14/35
0.5%
(-0.8% - 1 .9%)
0.5%
(-0.6% - 1 .6%)
5.1%
(3.3% - 6.9%)
13/35
0.5%
(-0.7% - 1 .7%)
0.5%
(-0.5% - 1 .5%)
5%
(3.2% - 6.7%)
12/35
0.5%
(-0.6% - 1 .6%)
0.5%
(-0.5% - 1 .4%)
4.6%
(2.9% - 6.2%)
13/30
0.5%
(-0.7% - 1 .7%)
0.5%
(-0.5% - 1 .5%)
4.4%
(2.8% - 5.9%)
12/25
0.5%
(-0.6% - 1 .5%)
0.5%
(-0.5% - 1 .4%)
3.6%
(2.3% - 5%)
1Percents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-122
-------
Table E-122. Estimated Percent of Total Annual Incidence of Emergency Room (ER) Visits Associated with Short-Term Exposure to Ambient
PM2.5 Concentrations in a Recent Year (2006) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2006 PM2.5 Concentrations1
Study
Tolbertetal. (2007)
Tolbertetal. (2007)
Itoetal. (2007)
Location
Atlanta, GA
Atlanta, GA
New York, NY
ER Visit for:
Cardiovascular
illness
Respiratory
illness
Asthma
Percent of Total Incidence of ER Visits Associated with Short-Term Exposure to PM2.s Concentrations in a
Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m)
Standards (Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
0.6%
(-0.9% -2.1%)
0.6%
(-0.6% - 1 .9%)
5.2%
(3.3% -7.1%)
15/352
0.6%
(-0.8% - 2%)
0.6%
(-0.6% - 1 .7%)
4.4%
(2.8% - 5.9%)
14/35
0.5%
(-0.8% - 1 .8%)
0.5%
(-0.5% - 1 .6%)
4.4%
(2.8% - 5.9%)
13/35
0.5%
(-0.7% - 1 .7%)
0.5%
(-0.5% - 1 .5%)
4.3%
(2.7% - 5.8%)
12/35
0.5%
(-0.6% - 1 .5%)
0.5%
(-0.5% - 1 .4%)
3.9%
(2.5% - 5.3%)
13/30
0.5%
(-0.7% - 1 .7%)
0.5%
(-0.5% - 1 .5%)
3.7%
(2.4% -5.1%)
12/25
0.4%
(-0.6% - 1 .5%)
0.4%
(-0.4% - 1 .3%)
3.1%
(2% - 4.2%)
Vercents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-123
-------
Table E-123. Estimated Percent of Total Annual Incidence of Emergency Room (ER) Visits Associated with Short-Term Exposure to Ambient
PM2 5 Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
on Adjusting 2007 PM2 5 Concentrations1
Study
Tolbertetal. (2007)
Tolbertetal. (2007)
Ito et al. (2007)
Location
Atlanta, GA
Atlanta, GA
New York, NY
ER Visit for:
Cardiovascular
illness
Respiratory
illness
Asthma
Percent of Total Incidence of ER Visits Associated with Short-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m)
Standards (Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
0.6%
(-0.9% -2.1%)
0.6%
(-0.6% - 1 .8%)
5.7%
(3.6% - 7.7%)
15/352
0.6%
(-0.8% - 1 .9%)
0.6%
(-0.6% - 1 .7%)
4.8%
(3% - 6.5%)
14/35
0.5%
(-0.7% - 1 .8%)
0.5%
(-0.5% - 1 .6%)
4.8%
(3% - 6.5%)
13/35
0.5%
(-0.7% - 1 .6%)
0.5%
(-0.5% - 1 .4%)
4.6%
(3% - 6.3%)
12/35
0.4%
(-0.6% - 1 .5%)
0.4%
(-0.4% - 1 .3%)
4.3%
(2.7% - 5.8%)
13/30
0.5%
(-0.7% - 1 .6%)
0.5%
(-0.5% - 1 .4%)
4.1%
(2.6% - 5.5%)
12/25
0.4%
(-0.6% - 1 .5%)
0.4%
(-0.4% - 1 .3%)
3.4%
(2.2% - 4.6%)
Vercents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
E-124
-------
APPENDIX F: SENSITIVITY ANALYSIS RESULTS
F-l
-------
1 Appendix F. Sensitivity Analysis Results
2
3 This Appendix provides detailed results of the single- and multi-factor sensitivity
4 analyses completed as part of this risk analysis. For additional detail on the sensitivity analysis
5 results completed for this analysis, as well as the types of results generated, see section 4.3.
6
F-l
-------
Table F-l. Sensitivity Analysis: Impact of Using Different Model Choices to Estimate the Incidence of Mortality Associated with Long-Term
Exposure to PM2.5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM2.5 Concentrations1
Health Endpoint
Incidence of Mortality Associated with Long-Term Exposure to
PM25 Concentrations Using:2
Standard Fixed
Effects Log-Linear
(Cox Proportional
Hazard) Model3
Random Effects Log-
Linear Model 4
Random Effects Log-
Log Model5
Percent Difference 6
Fixed Effects vs.
Random Effects Log-
Linear Models
Fixed Effects vs.
Random Effects Log-
Log Models
Los Angeles, CA
All Cause Mortality
Cardbpulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
1342
(854-1827)
1526
(1191 -1856)
1249
(1017-1477)
164
(71 -253)
1656
(772 - 2527)
7
1397
(847- 1924)
3360
(2075-4615)
2569
(1709-3400)
2535
(1793-3232)
307
(160-446)
23%
12%
150%
68%
103%
87%
Philadelphia, PA
All Cause Mortality
Cardbpulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
584
(372 - 792)
545
(427 - 660)
369
(303-434)
88
(39-135)
719
(337- 1090)
411
(253-558)
1254
(779- 1713)
790
(530- 1038)
639
(458-803)
142
(75 - 204)
23%
11%
115%
45%
73%
61%
1The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2 Mortality incidence was estimated for PlVhs concent rations down to the lowest measured level in the study (5.8 ug/m3). Numbers rounded to the nearest whole number. Numbers in
parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecobgical covariates (see Table 33 in Krewski et al., 2009).
4Estimates based on Table 9. Autocorrelation at MSA and ZCA levels; MSA & DIFF, in Krewski et al. (2009) - exposure period from 1999 - 2000.
Estimates based on Table 11, "MSA and DIFF" rows, in Krewski etal. (2009) - exposure period from 1999 - 2000.
Calculated as (core analysis model estimate - alternative model estimate)/(core analysis model estimate.)
Estimates for cardiopulmonary mortality and lung cancer mortality were not available for the random effects log-linear model.
F-l
-------
Table F-2. Sensitivity Analysis: Impact of Using Different Model Choices to Estimate the Incidence of Mortality Associated with Long-Term
Exposure to PMZs Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM2.5 Concentrations1
Health Endpoint
Incidence of Mortality Associated with Long-Term Exposure to
PM2.s Concentrations Using:2
Standard Fixed
Effects Log-Linear
(Cox Proportional
Hazard) Model3
Random Effects Log-
Linear Model4
Random Effects Log-
Log Model5
Percent Difference 6
Fixed Effects vs.
Random Effects Log-
Linear Models
Fixed Effects vs.
Random Effects Log-
Log Models
Los Angeles, CA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
1108
(704 - 1 509)
1263
(985 - 1 538)
1038
(843 - 1 229)
135
(59-210)
1368
(637 - 2090)
7
1162
(702- 1605)
2904
(1790-3995)
2225
(1477-2953)
2212
(1558-2833)
266
(138-388)
23%
12%
1 62%
76%
113%
97%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
525
(335-713)
491
(385 - 596)
334
(273 - 393)
80
(35-122)
647
(303 - 982)
372
(228 - 507)
1166
(723 - 1 595)
736
(493 - 969)
598
(428 - 755)
133
(70- 191)
23%
11%
1 22%
50%
79%
66%
The current primary PM2 5 standards include an annual standard set at 15 ug/m and a daily standard set at 35 ug/m .
2Mortality incidence was estimated for PM2.5 concentrations down to the lowest measured level in the study (5.8 ug/m3). Numbers rounded to the nearest whole number. Numbers in
parentheses are 95% confidence or credible intervals based on statistical uncertainty surroundhg the PM coefficient.
Estimates Based on Krewski et al. (2009), exposure period from 1 999 - 2000, using models with 44 individual and 7 ecobgical covariates (see Table 33 in Krewski et al., 2009).
4Estimates based on Table 9. Autocorrelation at MSA and ZCA levels; MSA & DIFF, in Krewski et al. (2009) - exposure period from 1999 - 2000.
Estimates based on Table 11 , "MSA and DIFF" rows, in Krewski eta I. (2009) -- exposure period from 1999 - 2000.
Calculated as (core analysis model estimate - alternative model estimate)/(core analysis model estimate.)
Estimates for Cardiopulmonary mortality and lung cancer mortality were not available for the random effects log-linear model.
F-2
-------
Table F-3. Sensitivity Analysis: Impact of Using Different Model Choices to Estimate the Incidence of Mortality Associated with Long-Term
Exposure to PM2.S Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2.5 Concentrations1
Health Endpoint
Incidence of Mortality Associated with Long-Term Exposure to
PM25 Concentrations Using:2
Standard Fixed
Effects Log-Linear
(Cox Proportional
Hazard) Model3
Random Effects Log-
Linear Model 4
Random Effects Log-
Log Model5
Percent Difference 6
Fixed Effects vs.
Random Effects Log-
Linear Models
Fixed Effects vs.
Random Effects Log-
Log Models
Los Angeles, CA
All Cause Mortality
Cardbpulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
1170
(744-1593)
1333
(1040- 1623)
1094
(890-1296)
143
(62 - 222)
1444
(672 - 2206)
7
1225
(741 - 1691)
3034
(1871 -4173)
2324
(1544-3082)
2306
(1626-2950)
278
(145-405)
23%
12%
159%
74%
111%
94%
Philadelphia, PA
All Cause Mortality
Cardbpulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
519
(331 - 704)
486
(381 - 589)
330
(270 - 389)
79
(35-121)
639
(299-971)
368
(226 - 502)
1157
(718- 1583)
731
(489 - 962)
594
(424-750)
132
(69- 190)
23%
12%
123%
50%
80%
67%
1The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2 Mortality incidence was estimated for PM2.5 concent rations down to the lowest measured level in the study (5.8 ug/m3). Numbers rounded to the nearest whole number. Numbers in
parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecobgical covariates (see Table 33 in Krewski et al., 2009).
4Estimates based on Table 9. Autocorrelation at MSA and ZCA levels; MSA & DIFF, in Krewski et al. (2009) - exposure period from 1999 - 2000.
Estimates based on Table 11, "MSA and DIFF" rows, in Krewski etal. (2009) - exposure period from 1999 - 2000.
Calculated as (core analysis model estimate - alternative model estimate)/(core analysis model estimate.)
7Estimates for cardiopulmonary mortality and lung cancer mortality were not available for the random effects log-linear model.
-------
Table F-4. Sensitivity Analysis: Impact of Limiting Estimated Annual Incidence of All-Cause Mortality
Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the Current
Standards to the Lowest Measured Level in the Study vs. to PRB, Based on Adjusting 2005
PM2 5 Concentrations1'2
Risk Assessment Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All Cause Mortality Associated with Long-
Term Exposure to PM2.s Concentrations Measured
Down to:
Lowest Measured Level in
Study (5.8 ug/m3)
736
(470 - 997)
702
(448 - 950)
380
(243-516)
486
(310-659)
743
(474-1008)
114
(72-155)
713
(455 - 968)
1342
(854-1827)
1893
(1207-2571)
584
(372 - 792)
620
(394 - 843)
497
(317-674)
37
(24-51)
897
(573-1215)
103
(66-141)
Estimated PRB
1057
(678-1426)
1073
(689-1446)
592
(379 - 800)
762
(488-1030)
1205
(773-1626)
262
(1 67 - 355)
1114
(713-1506)
2845
(1819-3853)
3299
(2113-4456)
971
(622-1310)
1255
(803-1698)
859
(550-1161)
161
(102-218)
1381
(887-1862)
234
(149-317)
Percent Difference3
44%
53%
56%
57%
62%
1 30%
56%
112%
74%
66%
1 02%
73%
335%
54%
1 27%
Estimates based on Table 33 in Krewski et al. (2009) - exposure period from 1999 - 2000, follow-up through 2000, models with 44 individual
and 7 ecological covariates. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals
based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality estimated down to PRB - mortality estimated down to LML)/(mortality estimated down to LML).
F-4
-------
Table F-5. Sensitivity Analysis: Impact of Limiting Estimated Annual Incidence of Ischemic Heart Disease
Mortality Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the Current
Standards to the Lowest Measured Level in the Study vs. to PRB, Based on Adjusting 2006 PM2 5
Concentrations1'2
Risk Assessment Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Ischemic Heart Disease Mortality
Associated with Long-Term Exposure to PM2.s
Concentrations Measured Down to:
Lowest Measured Level in
Study (5.8 ug/m3)
287
(236 - 336)
375
(307 - 441 )
157
(128-184)
222
(1 81 - 262)
449
(367 - 530)
92
(75-108)
416
(340 - 490)
1038
(843-1229)
1865
(1520-2203)
334
(273 - 393)
471
(384 - 557)
279
(228 - 330)
8
(6-10)
512
(419-603)
46
(37 - 55)
Estimated PRB
400
(331 - 465)
601
(497 - 699)
244
(201 - 285)
384
(315-450)
829
(683 - 969)
198
(162-232)
646
(533 - 755)
2366
(1943-2775)
3618
(2979 - 4232)
559
(461 - 651 )
907
(747-1061)
531
(437 - 621 )
57
(47 - 67)
862
(712-1006)
143
(117-168)
Percent Difference3
39%
60%
55%
73%
85%
115%
55%
128%
94%
67%
93%
90%
613%
68%
21 1 %
Estimates based on Table 33 in Krewski et al. (2009) -- exposure period from 1999 - 2000, follow-up through 2000, models with 44 individual
and 7 ecological covariates. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals
based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (IHD mortality estimated down to PRB - IHD mortality estimated down to LML)/(IHD mortality estimated down to LML).
F-5
-------
Table F-6. Sensitivity Analysis: Impact of Limiting Estimated Annual Incidence of All-Cause Mortality
Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the Current
Standards to the Lowest Measured Level in the Study vs. to PRB, Based on Adjusting 2007
PM2 5 Concentrations1'2
Risk Assessment Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All Cause Mortality Associated with Long-
Term Exposure to PM2.s Concentrations Measured
Down to:
Lowest Measured Level in
Study (5.8 ug/m3)
726
(464 - 984)
564
(360 - 765)
374
(238 - 507)
407
(259 - 553)
544
(346 - 739)
130
(82-177)
719
(459 - 977)
1170
(744-1593)
1689
(1076-2295)
519
(331 - 704)
556
(354 - 757)
434
(277 - 590)
48
(31 - 66)
728
(464 - 988)
64
(41 - 88)
Estimated PRB
1067
(684-1440)
938
(602-1267)
590
(377 - 797)
696
(445 - 942)
1007
(644-1362)
282
(180-382)
1143
(732-1545)
2697
(1723-3654)
3124
(2000 - 4224)
907
(581 -1225)
1240
(792-1678)
795
(509-1074)
179
(114-244)
1220
(782-1648)
201
(128-272)
Percent Difference3
47%
66%
58%
71%
85%
117%
59%
131%
85%
75%
123%
83%
273%
68%
214%
Estimates based on Table 33 in Krewski et al. (2009) -- exposure period from 1999 - 2000, follow-up through 2000, models with 44 individual
and 7 ecological covariates. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals
based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality estimated down to PRB - mortality estimated down to LML)/(mortality estimated down to LML).
F-6
-------
Table F-7. Sensitivity Analysis: Impact of Using a Different Study to Estimate the Incidence of Mortality
Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the Current
Standards, Based on Adjusting 2005 PM2.5 Concentrations1
Health Endpoint
Incidence of Mortality Associated with Long-Term
Exposure to PM2.5 Concentrations Using:2
Krewski et al. (2009)3
Krewski et al. (2000)4
Percent Difference5
Los Angeles, CA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
1342
(854-1827)
1526
(1191 -1856)
164
(71 - 253)
2965
(1005-4855)
1981
(693 - 3207)
212
(-152-535)
121%
30%
29%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
584
(372 - 792)
545
(427 - 660)
88
(39-135)
1276
(438 - 2064)
704
(250-1121)
114
(-85 - 276)
1 1 8%
29%
30%
1The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2Mortality incidence was estimated for PM2.5 concentrations down to the lowest measured level in Krewski et al., 2009 (5.8 ug/m3). Numbers
rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
3Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecological covariates
(see Table 33 in Krewski et al., 2009).
Estimates based on Table 21 b in Krewski et al. (2000) [reanalysis of Six Cities Study].
5Calculated as (Krewski et al. (2000) estimate - Krewski et al. (2009) estimate)/(Krewski et al. (2009) estimate).
F-7
-------
Table F-8. Sensitivity Analysis: Impact of Using a Different Study to Estimate the Incidence of Mortality
Associated with Long-Term Exposure to PM2 s Concentrations that Just Meet the Current
Standards, Based on Adjusting 2006 PM2 s Concentrations1
Health Endpoint
Incidence of Mortality Associated with Long-Term
Exposure to PM2.s Concentrations Using:2
Krewski et al. (2009)3
Krewski et al. (2000)4
Percent Difference5
Los Angeles, CA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
1108
(704-1509)
1263
(985-1538)
135
(59-210)
2454
(829-4031)
1642
(572-2671)
176
(-124-448)
121%
30%
30%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
525
(335 - 71 3)
491
(385 - 596)
80
(35-122)
1150
(394 - 1 866)
635
(225-1016)
103
(-76-251)
1 1 9%
29%
29%
1The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2Mortality incidence was estimated for PM2.5 concentrations down to the lowest measured level in Krewski et al., 2009 (5.8 ug/m3). Numbers
rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
3Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecological covariates
(see Table 33 in Krewski et al., 2009).
4Estimates based on Table 21 b in Krewski et al. (2000) [reanalysis of Six Cities Study].
5Calculated as (Krewski et al. (2000) estimate - Krewski et al. (2009) estimate)/(Krewski et al. (2009) estimate).
F-8
-------
Table F-9. Sensitivity Analysis: Impact of Using a Different Study to Estimate the Incidence of Mortality
Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the Current
Standards, Based on Adjusting 2007 PM2 5 Concentrations1
Health Endpoint
Incidence of Mortality Associated with Long-Term
Exposure to PM2.5 Concentrations Using:2
Krewski et al. (2009)3
Krewski et al. (2000)4
Percent Difference5
Los Angeles, CA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
1170
(744-1593)
1333
(1040-1623)
143
(62 - 222)
2590
(876 - 4252)
1732
(604-2815)
186
(-131 -472)
121%
30%
30%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
519
(331 - 704)
486
(381 - 589)
79
(35-121)
1137
(389 - 1 846)
628
(223 - 1 005)
102
(-75 - 249)
1 1 9%
29%
29%
1The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2Mortality incidence was estimated for PM2.5 concentrations down to the lowest measured level in Krewski et al., 2009 (5.8 ug/m3). Numbers
rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
3Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecological covariates
(see Table 33 in Krewski et al., 2009).
Estimates based on Table 21 b in Krewski et al. (2000) [reanalysis of Six Cities Study].
5Calculated as (Krewski et al. (2000) estimate - Krewski et al. (2009) estimate)/(Krewski et al. (2009) estimate).
F-9
-------
Table F-10. Sensitivity Analysis: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2 s
Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005 PM2 5 Concentrations:
Comparison of Proportional and Hybrid Rollback Methods1
Risk Assessment
Location
Baltimore, MD
Birmingham, AL
Detroit, Ml
Los Angeles, CA
New York, NY
St. Louis, MO
Type of Rollback
Proportional
Hybrid
Percent Difference 3
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the
Current and Alternative Combinations of Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m):
15/352
702
(448 - 950)
691
(442 - 936)
-2%
380
(243-516)
461
(294 - 624)
21%
743
(474-1008)
773
(493-1048)
4%
1342
(854-1827)
1675
(1066-2276)
25%
1893
(1207-2571)
1950
(1244-2648)
3%
897
(573-1215)
956
(611 -1294)
7%
14/35
643
(410-871)
667
(426 - 904)
4%
336
(214-457)
411
(262 - 557)
22%
734
(468 - 996)
773
(493-1048)
5%
1342
(854 - 1 827)
1675
(1066-2276)
25%
1893
(1207-2571)
1950
(1244-2648)
3%
813
(519-1102)
855
(546-1159)
5%
13/35
566
(361 - 768)
589
(376 - 799)
4%
292
(186-397)
360
(230 - 489)
23%
643
(410-874)
750
(479-1018)
17%
1342
(854-1827)
1599
(1018-2175)
19%
1808
(1152-2455)
1806
(1151 -2452)
0%
714
(456 - 970)
754
(481 -1022)
6%
12/35
490
(312-665)
511
(326 - 694)
4%
247
(1 57 - 336)
310
(197-421)
26%
552
(352-751)
651
(415-884)
18%
1180
(750-1607)
1344
(855-1830)
14%
1546
(984-2101)
1544
(983 - 2099)
0%
616
(392 - 836)
652
(415-885)
6%
13/30
546
(348-741)
537
(342 - 729)
-2%
292
(186-397)
360
(230 - 489)
23%
567
(361 - 770)
593
(378 - 805)
5%
924
(587-1258)
1209
(769-1647)
31%
1412
(898-1920)
1461
(930-1987)
3%
696
(443 - 944)
754
(481 -1022)
8%
12/25
388
(247 - 528)
381
(242-518)
-2%
205
(130-280)
274
(174-372)
34%
389
(247 - 530)
411
(261 - 559)
6%
502
(318-684)
740
(470-1010)
47%
926
(588-1261)
967
(614-1317)
4%
492
(313-669)
548
(349 - 745)
11%
1Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009). Numbers rounded to
the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
F-10
-------
Table F-ll. Sensitivity Analysis: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2.j
Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006 PM2 5 Concentrations:
Comparison of Proportional and Hybrid Rollback Methods
Risk Assessment
Location
Baltimore, MD
Birmingham, AL
Detroit, Ml
Los Angeles, CA
New York, NY
St. Louis, MO
Type of Rollback
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the
Current and Alternative Combinations of Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m):
15/352
565
(360 - 766)
554
(354 - 752)
-2%
354
(226-481)
430
(275 - 584)
21%
510
(325 - 694)
534
(340 - 725)
5%
1108
(704-1509)
1414
(899-1923)
28%
1407
(895-1913)
1453
(924-1975)
3%
659
(420 - 894)
704
(449 - 956)
7%
14/35
513
(327 - 696)
533
(340 - 724)
4%
312
(1 98 - 423)
382
(244-519)
22%
503
(320 - 684)
534
(340 - 725)
6%
1108
(704-1509)
1414
(899-1923)
28%
1407
(895-1913)
1453
(924-1975)
3%
588
(374 - 799)
620
(395 - 842)
5%
13/35
446
(284 - 606)
465
(296 - 632)
4%
269
(171 -365)
334
(213-454)
24%
429
(273 - 584)
515
(328 - 700)
20%
1108
(704-1509)
1344
(855-1829)
21%
1333
(848-1813)
1327
(844-1806)
0%
506
(322 - 688)
535
(341 - 727)
6%
12/35
378
(241 -515)
396
(252 - 539)
5%
226
(144-307)
286
(182-388)
27%
355
(225 - 483)
434
(276-591)
22%
958
(608 - 1 305)
1108
(704-1510)
16%
1106
(703 - 1 506)
1101
(700 - 1 499)
0%
423
(269 - 575)
450
(286-612)
6%
13/30
428
(272-581)
419
(267 - 569)
-2%
269
(171 -365)
334
(213-454)
24%
366
(233 - 499)
387
(246 - 526)
6%
721
(457 - 983)
984
(625-1340)
36%
990
(629-1349)
1030
(654-1403)
4%
490
(312-666)
535
(341 - 727)
9%
12/25
289
(1 84 - 394)
282
(1 79 - 384)
-2%
186
(118-253)
251
(159-341)
35%
222
(141 -302)
238
(151 -325)
7%
331
(210-451)
550
(349 - 750)
66%
571
(362 - 779)
604
(383 - 823)
6%
319
(203 - 435)
363
(231 - 495)
14%
1Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009). Numbers rounded to
the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
F-ll
-------
Table F-12. Sensitivity Analysis: Estimated Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2.j
Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007 PM2.j Concentrations:
Comparison of Proportional and Hybrid Rollback Methods
Risk Assessment
Location
Baltimore, MD
Birmingham, AL
Detroit, Ml
Los Angeles, CA
New York, NY
St. Louis, MO
Type of Rollback
Proportional
Hybrid
Percent Difference 3
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2 5 Concentrations that Just Meet the
Current and Alternative Combinations of Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m):
15/352
564
(360 - 765)
553
(353-751)
-2%
374
(238 - 507)
454
(290-615)
21%
544
(346 - 739)
568
(362 - 772)
4%
1170
(744-1593)
1484
(944-2019)
27%
1689
(1 076 - 2295)
1741
(1109-2366)
3%
728
(464 - 988)
787
(503-1068)
8%
14/35
512
(326 - 695)
532
(339 - 722)
4%
330
(210-448)
404
(258 - 548)
22%
536
(341 - 729)
568
(362 - 772)
6%
1170
(744-1593)
1484
(944-2019)
27%
1689
(1076-2295)
1741
(1 1 09 - 2366)
3%
653
(416-887)
698
(445 - 947)
7%
13/35
445
(283 - 605)
464
(296 - 630)
4%
285
(182-388)
354
(226-481)
24%
460
(293 - 626)
549
(350 - 747)
19%
1170
(744-1593)
1413
(899-1922)
21%
1607
(1023-2185)
1604
(1021 -2180)
0%
566
(360 - 769)
607
(387 - 825)
7%
12/35
378
(240-514)
395
(252 - 537)
4%
241
(153-327)
304
(193-413)
26%
384
(244 - 522)
466
(297 - 634)
21%
1016
(645 - 1 384)
1171
(744 - 1 594)
15%
1359
(864 - 1 848)
1355
(862 - 1 844)
0%
478
(304-651)
516
(329 - 702)
8%
13/30
427
(272 - 580)
418
(266 - 568)
-2%
285
(182-388)
354
(226-481)
24%
396
(252 - 539)
417
(265 - 568)
5%
773
(490-1053)
1043
(662-1420)
35%
1232
(783-1676)
1277
(812-1738)
4%
549
(350 - 747)
607
(387 - 825)
11%
12/25
289
(1 84 - 393)
281
(1 79 - 383)
-3%
199
(126-271)
268
(1 70 - 364)
35%
247
(157-336)
265
(168-361)
7%
372
(236 - 508)
598
(379-815)
61%
771
(489-1051)
809
(513-1102)
5%
369
(235 - 503)
424
(269 - 577)
15%
Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009). Numbers rounded to
the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
F-12
-------
Table F-13. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 s Concentrations
that Just Meet the Current Standards, Based on Adjusting 2005 PM2 5 Concentrations *'z
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ 5
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5
Concentrations that Just Meet the Current Standards
Winter
55
(8-102)
66
(9- 122)
18
(-4-41)
30
(-5 - 64)
-6
(-83 - 69)
0
(-33 - 33)
45
(-5 - 94)
17
(-84- 117)
279
(102-453)
93
(20 - 1 65)
43
(-4 - 90)
16
(-2 - 32)
37
(-37- 109)
1
(-53 - 53)
Spring
53
(3-101)
46
(1 -91)
25
(-3-51)
30
(-9 - 68)
77
(19-134)
16
(-1 - 32)
61
(5-116)
66
(-35- 166)
159
(1-315)
28
(-33 - 89)
65
(12-117)
6
(-2-14)
75
(14-136)
9
(-7 - 25)
Summer
43
(-15-99)
60
(-7-126)
17
(-21-55)
43
(-3 - 88)
54
(-32-137)
3
(-14-20)
51
(-13-113)
-104
(-257 - 48)
136
(-55 - 323)
34
(-48- 114)
44
(-23-109)
6
(-5-17)
66
(-6- 136)
4
(-9-17)
Fall
33
(-17-83)
50
(7 - 92)
10
(-21 - 40)
46
(6 - 84)
34
(-31 - 98)
11
(-12-34)
55
(-9-117)
-2
(-90 - 85)
206
(89-321)
65
(16- 112)
23
(-28 - 73)
8
(-3-19)
73
(13-133)
14
(-10-37)
Sum of Four
Seasons
184
4
222
70
149
159
30
212
-23
780
220
175
36
251
28
All Year
177
(34-319)
256
(104-406)
34
(-53- 121)
156
(37 - 273)
147
(-26-317)
44
(6-81)
214
(44 - 383)
81
(-1 1 7 - 278)
781
(459 - 1 1 02)
216
(79 - 350)
242
(40 - 442)
159
(47 - 270)
30
(6 - 54)
260
(75 - 443)
48
(8 - 87)
Percent
Difference 3
4%
-13%
106%
-4%
8%
-32%
-1%
-128%
0%
2%
10%
20%
-3%
-42%
1Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-specific
coefficient estimators was not available.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
F-13
-------
Table F-14. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 s Concentrations
that Just Meet the Current Standards, Based on Adjusting 2006 PM2.5 Concentrations *'2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ 5
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM25
Concentrations that Just Meet the Current Standards
Winter
48
(7 - 89)
54
(7-101)
16
(-4 - 36)
24
(-4 - 52)
-5
(-64 - 54)
1
(-36 - 36)
40
(-4 - 84)
17
(-86-120)
242
(89 - 394)
79
(17- 140)
39
(-4-81)
12
(-1 - 25)
26
(-27 - 79)
1
(-38 - 38)
Spring
57
(3-109)
41
(1-81)
27
(-3 - 56)
28
(-8 - 64)
77
(19-134)
14
(-1 - 30)
68
(5 - 1 30)
57
(-30-143)
141
(1 - 279)
26
(-31 - 83)
58
(10-104)
7
(-2-15)
67
(12- 120)
10
(-8 - 26)
Summer
51
(-18- 119)
52
(-6- 109)
18
(-23 - 58)
36
(-3 - 75)
39
(-23-100)
4
(-16-24)
51
(-13- 115)
-97
(-239 - 45)
111
(-44 - 263)
33
(-46 - 1 09)
40
(-21 -100)
7
(-6-19)
58
(-5- 120)
4
(-10-1 9)
Fall
29
(-15-72)
46
(6 - 86)
8
(-16-32)
34
(5 - 63)
26
(-24 - 75)
12
(-12-35)
48
(-8- 102)
-2
(-78 - 74)
183
(79 - 286)
70
(18- 121)
17
(-20 - 53)
7
(-3-17)
60
(10- 110)
12
(-8 - 30)
Sum of Four
Seasons
185
4
193
69
122
137
31
207
-25
677
208
154
33
211
27
All Year
180
(34 - 324)
224
(91 - 356)
33
(-51 - 116)
130
(31 - 228)
118
(-21 - 255)
47
(7 - 86)
208
(42 - 373)
75
(-108-257)
671
(394 - 946)
204
(75-331)
254
(42 - 463)
136
(40 - 232)
27
(6 - 49)
215
(62 - 367)
40
(7 - 73)
Percent
Difference 3
3%
-14%
109%
-6%
16%
-34%
0%
-133%
1%
2%
13%
22%
-2%
-33%
1Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate
regional means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
2The current primary PM2.s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
F-14
-------
Table F-15. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2S Concentrations
that Just Meet the Current Standards, Based on Adjusting 2007 PM2 5 Concentrations lj 2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ 5
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5
Concentrations that Just Meet the Current Standards
Winter
48
(7 - 88)
56
(7 - 1 04)
20
(-5 - 45)
28
(-5 - 60)
-6
(-79 - 66)
1
(-78 - 76)
49
(-5- 102)
23
(-115-160)
319
(117-517)
88
(19-156)
:::
54
(-5- 113)
28
(-3 - 57)
32
(-32 - 95)
1
(-47 - 47)
Spring
70
(4-134)
46
(1 - 90)
44
(-5 - 92)
25
(-8 - 58)
84
(21 - 145)
25
(-2 - 53)
63
(5-120)
112
(-59 - 280)
177
(1 - 350)
32
(-37 - 99)
:::
84
(15- 152)
11
(-4 - 26)
83
(15- 150)
12
(-9 - 32)
Summer
53
(-18-122)
56
(-6-11 8)
21
(-27 - 67)
39
(-3 - 80)
46
(-28- 119)
6
(-23 - 33)
57
(-14-127)
-144
(-359 - 66)
150
(-60 - 355)
34
(-48-114)
:::
57
(-30-140)
11
(-10-32)
63
(-6- 130)
4
(-9-16)
Fall
30
(-16-76)
49
(7-91)
10
(-21 - 39)
39
(5 - 73)
42
(-39- 121)
22
(-23 - 64)
52
(-9-112)
-3
(-140-131)
241
(105-376)
78
(20-134)
:::
30
(-35 - 93)
13
(-5 - 32)
70
(12- 127)
20
(-14-52)
Sum of Four
Seasons
201
4
207
95
131
166
54
221
-12
887
232
:::
225
63
248
37
All Year
177
(34-319)
227
(92 - 360)
34
(-53- 120)
139
(33 - 243)
121
(-21 - 262)
48
(7 - 89)
212
(43 - 378)
77
(-110-262)
734
(431 -1035)
208
(77 - 338)
242
(40 - 442)
143
(42 - 244)
34
(7-61)
225
(65 - 384)
42
(7 - 76)
Percent
Difference 3
14%
-9%
179%
-6%
37%
13%
4%
-1 1 6%
21%
12%
57%
85%
10%
-12%
1Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate
regional means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
2The current primary PM2.s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
F-15
-------
Table F-16. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2 s Concentrations
that Just Meet the Current Standards, Based on Adjusting 2005 PM2.5 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ 5
Pittsburgh, PA
Salt Lake City, UT5
St. Louis, MO
Tacoma, WA
Estimated Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM25
Concentrations that Just Meet the Current Standards
Winter
14
(-21 - 49)
16
(-30 - 59)
4
(-16-24)
10
(-18-36)
-1
(-48 - 43)
-2
(-13-8)
8
(-34 - 49)
-7
(-54 - 39)
149
(35-261)
28
(-6 - 60)
14
(-10-38)
-3
(-68 - 59)
0
(-12-13)
Spring
9
(-32 - 48)
10
(-31 - 48)
1
(-23 - 25)
11
(-21 - 42)
25
(-7-57)
1
(-3 - 5)
2
(-46 - 47)
3
(-45 - 49)
130
(29 - 228)
16
(-14-46)
30
(3 - 56)
48
(-2-95)
0
(-3 - 4)
Summer
9
(-31 - 46)
11
(-45 - 64)
0
(-29 - 27)
13
(-25 - 49)
28
(-21 - 74)
0
(-3 - 4)
27
(-21 - 73)
-43
(-105-17)
160
(30 - 286)
27
(-13-65)
13
(-23 - 47)
38
(-17-90)
0
(-2 - 3)
Fall
-2
(-37-31)
32
(-2 - 65)
-15
(-40-8)
-2
(-33 - 27)
36
(0-72)
3
(-4-9)
7
(-41 - 54)
0
(-43 - 43)
100
(23-174)
27
(4 - 50)
5
(-20 - 29)
43
(-4-88)
2
(-4 - 7)
Sum of Four
Seasons
30
4
69
-10
32
88
2
44
-47
539
98
62
126
2
All Year
32
(-33 - 95)
70
(-5-143)
-1
(-43 - 40)
32
(-21 - 85)
73
(-9-153)
11
(-8 - 30)
47
(-32-124)
-31
(-140-76)
504
(294-711)
87
(23-150)
84
(-4-170)
47
(-9-103)
8
(-2-18)
122
(27-215)
12
(-7-31)
Percent
Difference 3
-6%
-1%
900%
0%
21%
-82%
-6%
52%
7%
13%
32%
3%
-83%
1 Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-specific
coefficient estimators was not available.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
F-16
-------
Table F-17. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2 s Concentrations
that Just Meet the Current Standards, Based on Adjusting 2006 PM2 s Concentrations *'2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ 5
Pittsburgh, PA
Salt Lake City, UT5
St. Louis, MO
Tacoma, WA
Estimated Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2.5
Concentrations that Just Meet the Current Standards
Winter
13
(-18-42)
13
(-25 - 49)
4
(-14-21)
8
(-14-29)
-1
(-37-34)
-2
(-14-9)
7
(-30 - 44)
-7
(-56 - 40)
130
(31 - 227)
24
(-5-51)
13
(-9-35)
-2
(-49 - 43)
0
(-9 - 9)
Spring
9
(-35 - 52)
8
(-28 - 43)
1
(-25 - 27)
11
(-19-40)
25
(-7 - 57)
1
(-3-5)
2
(-51 - 53)
2
(-38 - 42)
115
(25 - 202)
15
(-13-42)
27
(2 - 50)
42
(-2-85)
0
(-3 - 4)
Summer
10
(-38 - 56)
10
(-39 - 56)
0
(-31 - 29)
11
(-21-41)
20
(-15-55)
0
(-4-5)
27
(-22 - 75)
-41
(-98-16)
130
(25 - 233)
26
(-12-62)
12
(-21 - 43)
33
(-15-80)
0
(-2 - 3)
Fall
-2
(-32-27)
30
(-2-61)
-12
(-31 - 7)
-1
(-24-21)
28
(0-55)
3
(-4 - 9)
6
(-35 - 47)
0
(-38 - 37)
88
(21 - 155)
30
(4-54)
4
(-14-21)
36
(-3-73)
1
(-3 - 6)
Sum of Four
Seasons
30
4
61
-7
29
72
2
42
-46
463
95
56
109
1
All Year
32
(-33 - 97)
61
(-4-125)
-1
(-41 - 39)
27
(-18-71)
58
(-7-123)
12
(-8 - 32)
45
(-31 - 120)
-29
(-129-70)
432
(252-611)
82
(21 -142)
88
(-4-178)
41
(-8-89)
7
(-2-16)
101
(23-179)
10
(-6 - 26)
Percent
Difference 3
-6%
0%
600%
7%
24%
-83%
-7%
59%
7%
16%
37%
8%
-90%
1Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate
regional means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
F-17
-------
Table F-18. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2.5 Concentrations
that Just Meet the Current Standards, Based on Adjusting 2007 PM25 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ 5
Pittsburgh, PA
Salt Lake City, UT5
St. Louis, MO
Tacoma, WA
Estimated Incidence of Cardiovascular Mortality Associated with Short-Term Exposure to PM2 5
Concentrations that Just Meet the Current Standards
Winter
11
(-17-39)
13
(-24 - 48)
4
(-14-21)
9
(-16-34)
-1
(-37 - 34)
-3
(-16-11)
8
(-35 - 50)
-6
(-47 - 34)
142
(34 - 248)
24
(-5 - 52)
13
(-9 - 34)
:::
-2
(-53 - 46)
0
(-9 - 9)
Spring
11
(-40 - 58)
9
(-29 - 45)
2
(-33 - 35)
10
(-18-36)
22
(-6 - 50)
1
(-3 - 5)
2
(-45 - 46)
3
(-48 - 53)
120
(26-212)
16
(-15-47)
27
(2-51)
:::
47
(-2 - 94)
0
(-3 - 4)
Summer
10
(-35 - 52)
10
(-39 - 57)
0
(-28 - 26)
11
(-23 - 44)
19
(-14-52)
0
(-3 - 4)
29
(-23 - 78)
-38
(-92- 15)
147
(28 - 262)
25
(-12-60)
12
(-21 - 43)
:::
32
(-15-78)
0
(-2 - 2)
Fall
-2
(-31 - 26)
30
(-2-61)
-12
(-31 - 6)
-2
(-28 - 24)
36
(0 - 72)
3
(-4 - 9)
7
(-37 - 48)
0
(-42 - 42)
97
(23 - 1 70)
30
(5 - 55)
4
(-18-26)
:::
37
(-3 - 76)
2
(-5 - 8)
Sum of Four
Seasons
30
4
62
-6
28
76
1
46
-41
506
95
56
:::
114
2
All Year
32
(-33 - 95)
62
(-4- 126)
-1
(-42 - 40)
29
(-1 9 - 76)
60
(-8- 127)
12
(-9 - 33)
46
(-31 - 122)
-30
(-132-72)
473
(276 - 668)
84
(22-145)
84
(-4- 170)
43
(-9 - 93)
9
(-2 - 20)
106
(24- 187)
11
(-6 - 27)
Percent
Difference 3
-6%
0%
500%
-3%
27%
-92%
0%
37%
7%
13%
30%
8%
-82%
Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate
regional means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
2The current primary PM2.s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
F-18
-------
Table F-19. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2 5 Concentrations
that Just Meet the Current Standards, Based on Adjusting 2005 PM2 5 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM25 Concentrations
that Just Meet the Current Standards
Winter
4
(-6-14)
5
(-6-15)
1
(-4 - 5)
1
(-6 - 9)
5
(-7-16)
-1
(-11-9)
5
(-4-14)
27
(-3 - 56)
51
(19-82)
10
(-1-21)
27
(-29 - 79)
4
(-3-11)
4
(-1 - 9)
1
(-15- 17)
0
(-15- 13)
Spring
1
(-8-11)
6
(-4-15)
2
(-4 - 7)
3
(-4-10)
9
(-1-18)
4
(-1 - 9)
5
(-4-13)
27
(-2 - 56)
18
(-6-41)
7
(-1-15)
30
(-8 - 66)
7
(-1-15)
2
(-2-6)
7
(-6 - 20)
2
(-2 - 6)
Summer
3
(-7-13)
6
(-6-17)
-1
(-8 - 7)
2
(-6 - 9)
10
(0-19)
1
(-2-4)
4
(-5-13)
-15
(-58 - 26)
22
(-10-53)
7
(-3-16)
21
(-3 - 45)
8
(-2-17)
-2
(-6-3)
4
(-10-17)
1
(-2 - 3)
Fall
3
(-5-11)
3
(-4-11)
3
(-2-9)
1
(-6 - 7)
9
(0-18)
1
(-6-8)
4
(-7-15)
0
(-23-21)
22
(1-42)
5
(-2-11)
41
(14-67)
7
(0-14)
-1
(-5-3)
7
(-6-18)
1
(-4 - 6)
Sum of Four
Seasons
11
4
20
5
7
33
5
18
39
113
29
119
26
3
19
4
All Year
20
(-8-47)
36
(7-64)
9
(-7-26)
11
(-10-32)
28
(1 - 55)
9
(0-17)
34
(5-61)
57
(6-108)
106
(37-174)
23
(-2 - 48)
47
(4-90)
20
(-2 - 42)
5
(1 - 10)
31
(-8 - 70)
7
(0-15)
Percent
Difference 3
-45%
-44%
-44%
-36%
18%
-44%
-47%
-32%
7%
26%
153%
30%
-40%
-39%
-43%
1 Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means.
Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-specific
coefficient estimators was not available.
F-19
-------
Table F-20. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2 5 Concentrations
that Just Meet the Current Standards, Based on Adjusting 2006 PM2 5 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.5 Concentrations
that Just Meet the Current Standards
Winter
3
(-5-12)
4
(-5-13)
1
(-3 - 5)
1
(-5 - 7)
4
(-5-13)
-1
(-11-10)
5
(-4-13)
28
(-3 - 57)
44
(16-72)
9
(-1 - 18)
31
(-33 - 90)
4
(-3-10)
3
(-1 - 7)
1
(-10-12)
0
(-11 -10)
Spring
2
(-9-12)
5
(-3-13)
2
(-4-8)
3
(-4-10)
9
(-1-18)
4
(-1 - 9)
5
(-5-15)
24
(-2 - 48)
16
(-5-37)
7
(-1-14)
30
(-8 - 65)
6
(-1-13)
2
(-2 - 6)
6
(-6-18)
2
(-2 - 6)
Summer
4
(-8-16)
5
(-5-15)
-1
(-9 - 7)
2
(-5 - 8)
7
(0-14)
1
(-2 - 5)
4
(-5-13)
-14
(-54 - 24)
18
(-8 - 43)
7
(-3-16)
22
(-3 - 46)
7
(-2-15)
-2
(-7 - 3)
3
(-9-15)
1
(-2 - 4)
Fall
3
(-4-10)
3
(-4-10)
3
(-2 - 7)
1
(-4-6)
7
(0-14)
1
(-6 - 8)
4
(-6-13)
0
(-19- 18)
20
(1 - 37)
5
(-2-12)
41
(14-66)
5
(0-10)
-1
(-5 - 3)
5
(-5-15)
1
(-3 - 5)
Sum of Four
Seasons
12
4
17
5
7
27
5
18
38
98
28
124
22
2
15
4
All Year
20
(-8 - 47)
31
(6-56)
9
(-7 - 25)
9
(-9 - 27)
23
(1 - 44)
9
(0-18)
33
(5-60)
53
(5-100)
91
(32-149)
22
(-2 - 45)
50
(4 - 94)
17
(-2 - 36)
5
(1-9)
26
(-7 - 58)
6
(0-12)
Percent
Difference 3
-40%
-45%
-44%
-22%
17%
-44%
-45%
-28%
8%
27%
148%
29%
-60%
-42%
-33%
Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate
regional means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
season-specific coefficient estimators was not available.
F-20
-------
Table F-21. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2 5 Concentrations
that Just Meet the Current Standards, Based on Adjusting 2007 PM2 s Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Respiratory Mortality Associated with Short-Term Exposure to PM2.5 Concentrations
that Just Meet the Current Standards
Winter
3
(-5-11)
4
(-5-12)
1
(-3 - 4)
1
(-6 - 8)
4
(-5-13)
-1
(-14-11)
5
(-4-15)
23
(-2 - 48)
49
(18-79)
9
(-1-19)
24
(-24 - 68)
4
(-3-10)
5
(-1 - 10)
1
(-11 -13)
0
(-11 -10)
Spring
2
(-10- 13)
5
(-4-14)
2
(-5-10)
3
(-3 - 9)
8
(-1-16)
4
(-1 - 8)
4
(-4-13)
29
(-2 - 60)
17
(-6 - 38)
7
(-1-15)
29
(-8 - 63)
6
(-1-13)
2
(-2 - 6)
7
(-6 - 20)
2
(-2 - 6)
Summer
4
(-8-15)
5
(-5-15)
-1
(-8 - 7)
2
(-5 - 8)
7
(0-14)
1
(-2-4)
4
(-6-14)
-13
(-51 - 23)
20
(-9 - 48)
7
(-3-15)
25
(-4-51)
7
(-2-15)
-2
(-7-3)
3
(-9-14)
1
(-1 - 3)
Fall
3
(-4-9)
3
(-4-10)
3
(-2 - 7)
1
(-5 - 6)
9
(0-18)
1
(-6-8)
4
(-6-13)
0
(-22-21)
21
(1-41)
5
(-2-12)
45
(15-73)
6
(0-13)
-1
(-5-4)
6
(-5-16)
1
(-5 - 7)
Sum of Four
Seasons
12
4
17
5
7
28
5
17
39
107
28
123
23
4
17
4
All Year
20
(-8 - 47)
31
(6 - 56)
9
(-7 - 25)
10
(-9 - 29)
24
(1-45)
9
(0-18)
33
(5-61)
54
(5-102)
100
(35-163)
22
(-2 - 46)
47
(4 - 90)
18
(-2 - 38)
6
(1 -11)
27
(-7 - 60)
6
(0-13)
Percent
Difference 3
-40%
-45%
-44%
-30%
17%
-44%
-48%
-28%
7%
27%
162%
28%
-33%
-37%
-33%
1 Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate
regional means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty
surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
F-21
-------
Table F-22. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to PM2 5
Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM25 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure
to PM2.s Concentrations that Just Meet the Current Standards
Winter
29
(-2 - 60)
129
(90-168)
10
(-1-21)
24
(-2 - 50)
153
(107- 198)
14
(-5-31)
44
(-3-91)
104
(-35-241)
391
(273 - 509)
118
(82-153)
58
(-20-135)
59
(41 - 77)
5
(-2-13)
103
(72-134)
12
(-65 - 82)
Spring
24
(-9 - 57)
45
(16-75)
10
(-3 - 23)
19
(-7 - 45)
54
(18-89)
11
(-6 - 27)
34
(-12-80)
194
(-98 - 479)
161
(55 - 266)
39
(13-64)
81
(-41 - 200)
31
(11-51)
4
(-2-10)
44
(15-73)
0
(-61 - 55)
Summer
-28
(-68-11)
40
(6 - 74)
-13
(-32-5)
-21
(-50-8)
40
(6 - 74)
-7
(-28-14)
-35
(-84-14)
-144
(-613-307)
131
(19-241)
35
(5 - 65)
-47
(-198-99)
28
(4-51)
-3
(-15-7)
30
(4 - 55)
-5
(-56 - 42)
Fall
6
(-27 - 39)
48
(23 - 73)
3
(-12-17)
5
(-19-28)
57
(27 - 87)
3
(-11-17)
10
(-41 - 59)
42
(-138-218)
145
(68-221)
39
(18-59)
14
(-47 - 75)
36
(17-55)
1
(-3 - 5)
44
(20 - 66)
-5
(-53 - 40)
Sum of Four
Seasons
31
4
262
10
27
304
21
53
196
828
231
106
154
7
221
2
All Year
40
(-26-105)
247
(182-313)
17
(-1 1 - 44)
31
(-20-81)
280
(206 - 354)
21
(0-41)
56
(-37-149)
264
(3 - 523)
792
(582-1002)
214
(157-271)
108
(1 -213)
157
(115- 199)
8
(0-16)
207
(152-262)
21
(-52 - 92)
Percent
Difference 3
-23%
6%
-41%
-13%
9%
0%
-5%
-26%
5%
8%
-2%
-2%
-13%
7%
-90%
'incidence estimates were calculated using the appropriate season-specific or all-year regional concentration-response function estimates reported in Table 2 of Bell et al. (2008). Location-
specific C-R function estimates were not available from this study. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals
based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons hospital admissions - all-year hospital admissions)/(all-year hospital admissions).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-specific
coefficient estimators was not available.
F-22
-------
Table F-23. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to PM2 5
Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM2 5 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure
to PM2.5 Concentrations that Just Meet the Current Standards
Winter
27
(-2 - 55)
106
(74-137)
9
(-1-19)
19
(-1 - 40)
119
(83-155)
15
(-5 - 34)
39
(-3-81)
108
(-37 - 252)
342
(239 - 445)
99
(69-129)
66
(-23-154)
53
(37 - 69)
4
(-1-10)
74
(52 - 96)
9
(-48 - 60)
Spring
26
(-9 - 60)
40
(14-66)
10
(-4 - 24)
18
(-6 - 43)
54
(19-90)
10
(-5-25)
38
(-14-90)
170
(-86-419)
143
(49 - 237)
35
(12-59)
80
(-40-198)
27
(9 - 45)
4
(-2-11)
39
(13-64)
0
(-64 - 59)
Summer
-34
(-81 -13)
35
(5 - 64)
-14
(-33 - 5)
-18
(-43 - 7)
29
(4 - 54)
-8
(-34-17)
-36
(-86-14)
-137
(-580-291)
107
(16- 197)
33
(5 - 62)
-48
(-202-101)
25
(4 - 46)
-4
(-17-9)
26
(4 - 48)
-6
(-61 - 46)
Fall
6
(-24 - 35)
44
(21 - 68)
2
(-9-13)
3
(-15-21)
44
(20 - 66)
3
(-11-18)
8
(-36 - 52)
37
(-121 -192)
129
(61 -198)
41
(19-63)
14
(-47 - 74)
26
(12-40)
1
(-3-4)
36
(17-54)
-4
(-43 - 33)
Sum of Four
Seasons
25
4
225
7
22
246
20
49
178
721
208
112
131
5
175
-1
All Year
41
(-27-108)
214
(157-271)
16
(-10-42)
26
(-17-68)
225
(165-285)
22
(0 - 44)
55
(-36-145)
248
(3-491)
684
(502 - 865)
201
(147-254)
113
(1 - 224)
134
(98-169)
7
(0-15)
171
(126-216)
18
(-44 - 78)
Percent
Difference 3
-39%
5%
-56%
-15%
9%
-9%
-11%
-28%
5%
3%
-1%
-2%
-29%
2%
-106%
'incidence estimates were calculated using the appropriate season-specific or all-year regional concentration-response function estimates reported in Table 2 of Bell et al.
(2008). Location-specific C-R function estimates were not available from this study. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95%
confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons hospital admissions - all-year hospital admissions)/(all-year hospital admissions).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
F-23
-------
Table F-24. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure to PM2 5
Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2 5 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Hospital Admissions for Cardiovascular Illness Associated with Short-Term Exposure
to PM2.s Concentrations that Just Meet the Current Standards
Winter
24
(-2 - 50)
103
(72 - 134)
9
(-1-18)
23
(-2 - 47)
121
(84-157)
17
(-6 - 40)
46
(-3 - 95)
93
(-31 -216)
377
(263 - 490)
101
(71 -131)
50
(-17-117)
51
(36 - 67)
6
(-2-15)
80
(56-104)
9
(-48 - 60)
Spring
30
(-11-70)
42
(14-70)
13
(-5-31)
17
(-6 - 39)
48
(16-79)
10
(-5 - 25)
34
(-12-79)
215
(-109-531)
151
(51 - 249)
39
(13-64)
78
(-39- 193)
28
(9-46)
5
(-2 - 12)
43
(15-71)
0
(-64 - 59)
Summer
-33
(-80-13)
35
(5 - 65)
-12
(-30 - 5)
-19
(-46 - 7)
28
(4 - 52)
-6
(-26-13)
-38
(-91 -15)
-131
(-556 - 279)
121
(18-223)
32
(5-59)
-54
(-230-115)
25
(4 - 46)
-4
(-18-9)
25
(4-47)
-4
(-43 - 33)
Fall
5
(-23 - 34)
44
(21-67)
2
(-9-13)
4
(-17-25)
58
(27 - 88)
3
(-11-18)
9
(-37 - 54)
42
(-139-220)
143
(67-218)
42
(20 - 64)
16
(-52 - 83)
32
(15-49)
1
(-3-5)
37
(17-56)
-6
(-63 - 47)
Sum of Four
Seasons
26
4
224
12
25
255
24
51
219
792
214
90
136
8
185
-1
All Year
41
(-27-109)
216
(159-273)
16
(-1 1 - 43)
28
(-18-73)
233
(171 -295)
23
(0 - 46)
56
(-37-149)
258
(3-511)
752
(552-951)
203
(149-257)
108
(1-215)
140
(103-177)
9
(0-18)
178
(131 -225)
19
(-46 - 82)
Percent
Difference 3
-37%
4%
-25%
-11%
9%
4%
-9%
-15%
5%
5%
-17%
-3%
-11%
4%
-105%
'incidence estimates were calculated using the appropriate season-specific or all-year regional concentration-response function estimates reported in Table 2 of Bell et al.
(2008). Location-specific C-R function estimates were not available from this study. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95%
confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons hospital admissions - all-year hospital admissions)/(all-year hospital admissions).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix forthe
season-specific coefficient estimators was not available.
F-24
-------
Table F-25. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM2 5
Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM25 Concentrations 1>2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to
PM2.s Concentrations that Just Meet the Current Standards
Winter
6
(-21 -31)
18
(-5-40)
2
(-7-11)
4
(-17-25)
24
(-6-53)
10
(-1 - 20)
7
(-27 - 42)
71
(-6-148)
57
(-15-129)
16
(-4 - 37)
35
(-3 - 72)
8
(-2-18)
4
(0-8)
23
(-6-53)
0
(-50 - 43)
Spring
9
(-10-28)
1
(-14- 15)
4
(-4-11)
7
(-8 - 23)
1
(-18-20)
3
(-6-11)
12
(-14-38)
45
(-97-183)
2
(-52 - 56)
1
(-12-12)
17
(-36 - 68)
0
(-10-11)
1
(-2 - 5)
1
(-20-21)
4
(-27 - 32)
Summer
-6
(-25-12)
14
(0 - 28)
-3
(-12-6)
-5
(-21 -10)
17
(0 - 35)
4
(-4-11)
-8
(-33-16)
86
(-98-261)
49
(-1 - 99)
12
(0 - 25)
23
(-26 - 70)
9
(0-19)
2
(-3 - 7)
15
(0 - 29)
1
(-19-19)
Fall
1
(-13-15)
2
(-12 - 15)
1
(-5 - 6)
1
(-13-16)
2
(-14- 18)
3
(-5-11)
3
(-27 - 32)
42
(-60-140)
5
(-33 - 42)
1
(-9-12)
13
(-18-43)
1
(-9-12)
1
(-2-4)
2
(-15- 19)
-2
(-24-18)
Sum of Four
Seasons
10
4
35
4
7
44
20
14
244
113
30
88
18
8
41
3
All Year
17
(-22 - 55)
20
(-12-51)
7
(-9 - 23)
15
(-18-47)
25
(-15-64)
14
(3 - 25)
25
(-32-81)
170
(40 - 300)
65
(-38-169)
17
(-10-44)
61
(14-107)
13
(-8 - 33)
6
(1-10)
25
(-15-64)
2
(-27 - 30)
Percent
Difference 3
-41%
75%
-43%
-53%
76%
43%
-44%
44%
74%
76%
44%
38%
33%
64%
50%
'incidence estimates were calculated using the appropriate season-specific or all-year regional concentration-response function estimates from models with a 2-day lag for respiratory
hospital admissions reported in Table 2 of Bell et al. (2008). Location-specific C-R function estimates were not available from this study. Numbers are rounded to the nearest whole number.
Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons hospital admissions - all-year hospital admissions)/(all-year hospital admissions).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-specific
coefficient estimators was not available.
F-25
-------
Table F-26. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM25
Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM2.5 Concentrations *'2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to
PM2.5 Concentrations that Just Meet the Current Standards
Winter
5
(-19-28)
15
(-4 - 33)
2
(-6-10)
4
(-13-20)
18
(-5-41)
11
(-1 - 22)
7
(-24 - 37)
75
(-6-155)
50
(-13-113)
14
(-4-31)
40
(-3 - 82)
7
(-2-16)
3
(0-6)
17
(-5 - 38)
0
(-37-31)
Spring
10
(-1 1 - 30)
1
(-12-1 3)
4
(-4-12)
7
(-8 - 22)
1
(-19-20)
3
(-5-10)
14
(-15-42)
40
(-85 - 1 60)
2
(-46 - 50)
0
(-11-12)
17
(-35 - 67)
0
(-9-10)
1
(-3 - 5)
1
(-17-18)
4
(-29 - 34)
Summer
-7
(-30- 14)
12
(0 - 24)
-3
(-12-6)
-4
(-18-9)
13
(0 - 25)
5
(-5-14)
-8
(-34-17)
82
(-93 - 248)
40
(-1 -81)
12
(0 - 23)
24
(-27 - 72)
8
(0-17)
3
(-3 - 8)
13
(0 - 26)
1
(-21 - 20)
Fall
1
(-11-13)
2
(-11-14)
0
(-4 - 5)
1
(-10-12)
2
(-11-13)
3
(-5-12)
3
(-24 - 28)
37
(-53 - 1 24)
4
(-30 - 38)
1
(-10-13)
13
(-1 8 - 43)
1
(-7 - 8)
1
(-1 - 3)
2
(-13-1 6)
-1
(-20 - 1 5)
Sum of Four
Seasons
9
4
30
3
8
34
22
16
234
96
27
94
16
8
33
4
All Year
17
(-22 - 56)
17
(-1 0 - 44)
7
(-8 - 22)
12
(-1 5 - 40)
20
(-12-52)
15
(3 - 26)
25
(-31 - 79)
160
(37-281)
56
(-33-145)
16
(-9-41)
64
(15-112)
11
(-6 - 28)
5
(1-9)
21
(-12-53)
2
(-23 - 25)
Percent
Difference 3
-47%
76%
-57%
-33%
70%
47%
-36%
46%
71%
69%
47%
45%
60%
57%
100%
Incidence estimates were calculated using the appropriate season-specific or all-year regional concentration-response function estimates from models with a 2-day lag for
respiratory hospital admissions reported in Table 2 of Bell et al. (2008). Location-specific C-R function estimates were not available from this study. Numbers are rounded
to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons hospital admissions - all-year hospital admissions)/(all-year hospital admissions).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
F-26
-------
Table F-27. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
Function to Estimate the Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to PM2.5
Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2.5 Concentrations *'2
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Estimated Incidence of Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to
PM2.5 Concentrations that Just Meet the Current Standards
Winter
5
(-17-26)
14
(-4 - 32)
2
(-6 - 9)
4
(-16-24)
19
(-5 - 42)
13
(-1 - 26)
8
(-28 - 43)
64
(-5-133)
55
(-15- 124)
14
(-4 - 32)
30
(-3 - 63)
7
(-2-16)
5
(0-9)
18
(-5-41)
0
(-37-31)
Spring
11
(-13-35)
1
(-13-14)
5
(-6-16)
6
(-7 - 20)
1
(-16-18)
2
(-5-10)
12
(-14-38)
50
(-108-203)
2
(-48 - 52)
1
(-12-13)
16
(-34 - 65)
0
(-9-10)
1
(-3 - 5)
1
(-19-21)
4
(-29 - 34)
Summer
-7
(-29-14)
12
(0 - 25)
-3
(-11-5)
-5
(-19-10)
12
(0 - 25)
3
(-4-11)
-9
(-36-18)
78
(-89 - 239)
46
(-1-91)
11
(0 - 22)
27
(-31 - 82)
8
(0-17)
3
(-3-9)
12
(0 - 25)
1
(-15-15)
Fall
1
(-11-13)
2
(-11-14)
0
(-4 - 5)
1
(-12-14)
2
(-14-18)
4
(-5-12)
3
(-25 - 29)
42
(-61 -142)
5
(-33 - 42)
1
(-10-13)
14
(-20 - 48)
1
(-8-10)
1
(-2-4)
2
(-13-17)
-2
(-29 - 22)
Sum of Four
Seasons
10
4
29
4
6
34
22
14
234
108
27
87
16
10
33
3
All Year
18
(-22 - 57)
17
(-10-45)
7
(-9 - 22)
13
(-17-43)
21
(-12-54)
15
(4 - 27)
25
(-32 - 82)
166
(39 - 293)
62
(-37-160)
16
(-10-42)
61
(14- 108)
11
(-7 - 29)
7
(2-12)
21
(-13-55)
2
(-24 - 27)
Percent
Difference 3
-44%
71%
-43%
-54%
62%
47%
-44%
41%
74%
69%
43%
45%
43%
57%
50%
Incidence estimates were calculated using the appropriate season-specific or all-year regional concentration-response function estimates from models with a 2-day lag for
respiratory hospital admissions reported in Table 2 of Bell et al. (2008). Location-specific C-R function estimates were not available from this study. Numbers are rounded
to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2.s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (sum-of-4-seasons hospital admissions - all-year hospital admissions)/(all-year hospital admissions).
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
season-specific coefficient estimators was not available.
F-27
-------
Table F-28. Sensitivity Analysis: Impact of Using an Annual Concentration-Response Function vs. a Seasonal Function (for April - August)
Applied Only to that Period to Estimate the Incidence of Emergency Room Visits for Asthma Associated with Short-Term
Exposure to Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative
Standards in New York City, Based on Adjusting 2005 PM2 5 Concentrations1
Concentration-Response (C-R) Function
and Period to Which Applied:
Annual C-R Function Applied to the Whole
Year
Seasonal C-R Function for April -August
Applied Only to that Period:
Incidence of ER Visits Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m):
Recent PM2.5
Concentrations
5235
(3346-7071)
3136
(2058-4162)
15/352
4375
(2790 - 5923)
2634
(1722-3509)
14/35
4375
(2790 - 5923)
2634
(1722-3509)
13/35
4265
(2719-5776)
2569
(1678-3425)
12/35
3927
(2501 - 5323)
2370
(1546-3164)
13/30
3754
(2390-5091)
2268
(1478-3031)
12/25
3127
(1987-4248)
1896
(1232-2541)
1Based on Ito et al. (2007). New York City in this study consisted only of Manhattan. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible
intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
F-28
-------
Table F-29. Sensitivity Analysis: Impact of Using an Annual Concentration-Response Function vs. a Seasonal Function (for April - August)
Applied Only to that Period to Estimate the Incidence of Emergency Room Visits for Asthma Associated with Short-Term Exposure
to Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards in
New York City, Based on Adjusting 2006 PM2 5 Concentrations1
Concentration-Response (C-R) Function
and Period to Which Applied:
Annual C-R Function Applied to the Whole
Year
Seasonal C-R Function for April - August
Applied Only to that Period:
Incidence of ER Visits Associated with Short-Term Exposure to PM2.5 Concentrations in a Recent Year and PT
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m):
Recent PM2.5
Concentrations
4506
(2876 - 6095)
2732
(1791 -3631)
15/352
3764
(2397-5102)
2293
(1497-3059)
14/35
3764
(2397-5102)
2293
(1497-3059)
13/35
3669
(2336 - 4974)
2237
(1460-2985)
12/35
3377
(2149-4582)
2063
(1344-2757)
13/30
3228
(2053 - 4382)
1974
(1285-2640)
12/25
2688
(1707-36J
1649
(1071 -22'
1 Based on Ito et al. (2007). New York City in this study consisted only of Manhattan. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible
intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
F-29
-------
Table F-30. Sensitivity Analysis: Impact of Using an Annual Concentration-Response Function vs. a Seasonal Function (for April - August)
Applied Only to that Period to Estimate the Incidence of Emergency Room Visits for Asthma Associated with Short-Term
Exposure to Concentrations in a Recent Year (2007) and PM2 5 Concentrations that Just Meet the Current and Alternative
Standards in New York City, Based on Adjusting 2007 PM2 5 Concentrations1
Concentration-Response (C-R) Function
and Period to Which Applied:
Annual C-R Function Applied to the Whole
Year
Seasonal C-R Function for April -August
Applied Only to that Period:
Incidence of ER Visits Associated with Short-Term Exposure to PM2.s Concentrations in a Recent Year and PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m):
Recent PM2.5
Concentrations
4926
(3145-6660)
2908
(1906-3864)
15/352
4115
(2622 - 5575)
2441
(1593-3256)
14/35
4115
(2622 - 5575)
2441
(1593-3256)
13/35
4011
(2555 - 5436)
2380
(1553-3177)
12/35
3692
(2350 - 5008)
2195
(1431 -2934)
13/30
3529
(2245 - 4790)
2101
(1368-2810)
12/25
2939
(1867-3995)
1755
(1140-2354)
'Based on Ito et al. (2007). New York City in this study consisted only of Manhattan. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible
intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
F-30
-------
Table F-31. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM2 5 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM25Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impact of Changing the Lag Structure:
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
0 day
1 day
2 day
3 day
4 day
5 day
none
none
none
none
none
none
275
(-35 - 584)
301
(0 - 600)
194
(-97 - 483)
-77
(-373-218)
-46
(-329 - 235)
-287
(-592 - 1 5)
Max. positive est. =
301
Min. positive est. =
194
Percent d iff. =
55%
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impact of Changing the Typeof Model, with a 0-Day Lac
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GLM, 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
0 day
0 day
0 day
0 day
none
none
none
none
275
(-35 - 584)
204
(-174-579)
163
(-115-441)
153
(-218-522)
Max. positive est. =
275
Min. positive est. =
153
Percent d iff. =
80%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2n - Impactof Changing the Typeof Model, with a 1-Day Lac
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GLM, 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
1 day
none
none
none
none
301
(0 - 600)
281
(-86 - 644)
51
(-236 - 336)
-5
(-509 - 494)
Max. positive est. =
301
Min. positive est. =
51
Percent d iff. =
490%
81
(-117-278)
240%
272%
140%
-195%
-157%
-454%
F
81
(-117-278)
240%
152%
101%
89%
1
81
(-117-278)
272%
247%
-37%
-106%
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impactof a Copollutant Model
Mortality, short-term non-accidental
Mortality, short-term non- accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
CO
CO
CO
-272
(-676-128)
-169
(-540-198)
-169
(-603 - 260)
81
(-117- 278)
-436%
-309%
-309%
F-31
-------
Table F-31 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 s Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM25 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM2 5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Cardiovascular Mortality Associated with Short-Term Exposure to PMZs -Impactof Changing the Type of Model, with a 0-Day Lac,
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
Cardiovascular Mortality Associated with Short-Term Exposure to f
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
0 day
none
none
none
171
(17-324)
168
(24-310)
168
(-4-337)
Max. positive est. =
171
Min. positive est. =
168
Pencentdiff. =
2%
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
1
-31
(-140-76)
111%
107%
107%
TW25 -Impactof Changing the Type of Model, witha 1-Day Lag
1 day
1 day
1 day
none
none
none
178
(26 - 328)
139
(-6-282)
120
(-56 - 293)
Max. positive est. =
178
Min. positive est. =
120
Pencentdiff. =
48%
-31
(-140-76)
120%
72%
48%
Cardiovascular Mortality Associated with Short-Term Exposure to PMzs - Impactof a Copollutant Model
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 100df
log-linear, GLM, 100df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
1 day
1 day
Respiratory Mortality Associated with Short-Term Exposure to PM2n - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100 df
log-lhear, GLM, 100df
0 day
0 day
0 day
CO
CO
CO
CO
307
(130-481)
324
(116-529)
158
(-22 - 335)
158
(-60 - 372)
Max. positive est. =
324
Min. positive est. =
158
Pencentdiff. =
105%
-31
(-140-76)
279%
300%
95%
95%
pact of Changing the Type of Model, with a 0-Day Lag
none
none
none
-15
(-80-49)
-37
(-102-25)
-32
(-109-43)
5
Respiratory Mortality Associated with Short-Term Exposure to PM2S - Impactof Changing the Typeof Model, witha 1-Day Lag
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
none
none
none
10
(-56 - 74)
22
(-42 -85)
5
(-75-83)
Max. positive est. =
22
Min. positive est. =
5
Pencentdiff. =
340%
F-32
-------
Table F-31 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 s Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM25 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM2 5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changing the Type of Model, with a 0-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100df
log-lhear, G LM, 100df
0 day
0 day
0 day
none
none
none
794
(457-1128)
584
(254-912)
634
(226-1038)
Max. positive est. =
794
Min. positive est. =
584
Pencentdiff. =
36%
35
(-60-130)
880%
621%
683%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changing the Type of Model, with a 1-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100df
log-lhear, GLM, 100df
1 day
1 day
1 day
none
none
none
699
(347-1048)
569
(234 - 902)
604
(194-1011)
Max. positive est. =
699
Min. positive est. =
569
Pencentdiff. =
23%
35
(-60-130)
763%
602%
646%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of a Copollutant Model
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-lhear, GAM (stringent), 100df
log-lhear, GLM, 100df
log-lhear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
1 day
1 day
CO
CO
CO
CO
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Im
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-linear, GLM, 100df
0 day
0 day
0 day
none
none
none
197
(-224-615)
293
(-208 - 788)
122
(-330-568)
137
(-381 -648)
oact of Changing the T\
336
(138-531)
278
(104-450)
300
(83-514)
Max. positive est. =
293
Min. positive est. =
122
Pencentdiff. =
140%
35
(-60-130)
143%
262%
51%
69%
fpe of Model, with a 0-Day Lag
Max. positive est. =
336
Min. positive est. =
278
Pencentdiff. =
21%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changing the Type of Model, with a 1-Day Lag
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-linear, GLM, 100df
1 day
1 day
1 day
none
none
none
240
(45 - 432)
152
(-22 - 324)
156
(-55 - 364)
Max. positive est. =
240
Min. positive est. =
152
Pencentdiff. =
58%
F-33
-------
Table F-31 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM25 Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM2S Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Im
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
2 day
2 day
2 day
none
none
none
Incidence
Associated with
PM2 5 Above Policy
Relevant
Background
oact of Changng the T\
371
(166-574)
230
(43-414)
208
(-24 - 436)
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
fpe of Model, with a 2-Day Lag
Max. positive est. =
371
Min. positive est. =
208
Pencentdiff. =
78%
Respiratory Hospital Admissions Associated with Short-Term Exposire to PM 2.5 - Impact of Changing the Lag Structure, with a Copollutant Model
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
0 day
1 day
2 day
3 day
NO2
NO2
NO2
NO2
85
(-185-351)
-8
(-329-307)
71
(-209-346)
-223
(-491 - 41)
Max. positive est. =
85
Min. positive est. =
71
Pencentdiff. =
20%
1The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3. Results are based on Moolgavkar (2003) [reanalysis of MooIgavkar (2000a, 2000b, and
2000c)]. Numbers rounded to the nearest whole number. Numbers in parentheses a re 95% confidence or credible intervals based onstatstical uncertainty surrounding the PM coefficient.
2The core analysis estimates for non-accidental mortality and cardiovascuter mortality associated with short-term exposure to PM2.5 are from Zanobetti and Schwartz (2009). The core analysis estimates for
cardbvascular hospital admissions associated with short-term exposure to PM2.5 are from Bell et al. (2008).
Calculated as (maximum positive estimate - minimum positive estimate)/(minimum positive estimate).
"Calculated as (Moolgavkar (2003) estimate - core analysis estimate)/(core analysis estimate).
Because "respiratory illness" was much more broadty defined in both Zanobetti and Schwartz (2009) and Bell et al. (2008) than in Moolgavkar (2003), a comparison betweenthe Moolgavkar (2003) estimates
and the corresponding core analysis estimates is not shown.
F-34
-------
Table F-32. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM2 5 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM25Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impact of Changing the Lag Structure:
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
0 day
1 day
2 day
3 day
4 day
5 day
none
none
none
none
none
none
254
(-32 - 539)
278
(0 - 554)
179
(-89 - 445)
-71
(-344-201)
-42
(-304-217)
-265
(-546 - 1 4)
Max. positive est. =
278
Min. positive est. =
179
Percent d iff. =
55%
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impact of Changing the Typeof Model, with a 0-Day Lac
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GLM, 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
0 day
0 day
0 day
0 day
none
none
none
none
254
(-32 - 539)
188
(-161 -535)
151
(-106-407)
141
(-201 - 482)
Max. positive est. =
254
Min. positive est. =
141
Percent d iff. =
80%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2n - Impactof Changing the Typeof Model, with a 1-Day Lac
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GLM, 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
1 day
none
none
none
none
278
(0 - 554)
259
(-80 - 595)
47
(-218-310)
-5
(-469 - 455)
Max. positive est. =
278
Min. positive est. =
47
Percent d iff. =
491%
75
(-108-257)
239%
271 %
139%
-195%
-156%
-453%
F
75
(-108-257)
239%
1 51 %
1 01 %
88%
1
75
(-108-257)
271 %
245%
-37%
-107%
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impactof a Copollutant Model
Mortality, short-term non-accidental
Mortality, short-term non- accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
CO
CO
CO
-251
(-623 - 1 1 8)
-156
(-497-183)
-156
(-555 - 240)
75
(-1 08 - 257)
-435%
-308%
-308%
F-35
-------
Table F-32 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2S Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM25 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM2 5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Cardiovascular Mortality Associated with Short-Term Exposure to PMZs -Impactof Changing the Type of Model, with a 0-Day Lac,
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
Cardiovascular Mortality Associated with Short-Term Exposure to f
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
0 day
none
none
none
158
(15-299)
155
(22 - 286)
155
(-3-311)
Max. positive est. =
158
Min. positive est. =
155
Pencentdiff. =
2%
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
1
-29
(-129-70)
111%
107%
107%
TW25 -Impactof Changing the Type of Model, witha 1-Day Lag
1 day
1 day
1 day
none
none
none
164
(24 - 303)
128
(-5-260)
110
(-51 - 270)
Max. positive est. =
164
Min. positive est. =
110
Pencentdiff. =
49%
-29
(-129-70)
119%
71%
47%
Cardiovascular Mortality Associated with Short-Term Exposure to PMzs - Impactof a Copollutant Model
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 100df
log-linear, GLM, 100df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
1 day
1 day
Respiratory Mortality Associated with Short-Term Exposure to PM2n - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100 df
log-lhear, GLM, 100df
0 day
0 day
0 day
CO
CO
CO
CO
283
(120-444)
299
(107-489)
145
(-20 - 309)
145
(-56 - 344)
Max. positive est. =
299
Min. positive est. =
145
Pencentdiff. =
106%
-29
(-129-70)
277%
299%
93%
93%
pact of Changing the Type of Model, with a 0-Day Lag
none
none
none
-14
(-74-45)
-35
(-94-23)
-29
(-100-39)
5
Respiratory Mortality Associated with Short-Term Exposure to PM2S - Impactof Changing the Typeof Model, witha 1-Day Lag
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
none
none
none
9
(-51 -68)
21
(-39 - 78)
5
(-69 - 76)
Max. positive est. =
21
Min. positive est. =
5
Pencentdiff. =
320%
F-36
-------
Table F-32 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2S Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM25 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM2 5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changing the Type of Model, with a 0-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100df
log-lhear, G LM, 100df
0 day
0 day
0 day
none
none
none
745
(428-1060)
548
(238 - 856)
595
(212-975)
Max. positive est. =
745
Min. positive est. =
548
Pencentdiff. =
36%
248
(3-491)
893%
631%
693%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changing the Type of Model, with a 1-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100df
log-lhear, GLM, 100df
1 day
1 day
1 day
none
none
none
656
(326 - 984)
534
(220 - 847)
567
(182-949)
Max. positive est. =
656
Min. positive est. =
534
Pencentdiff. =
23%
248
(3-491)
775%
612%
656%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of a Copollutant Model
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-lhear, GAM (stringent), 100df
log-lhear, GLM, 100df
log-lhear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
1 day
1 day
CO
CO
CO
CO
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Im
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-linear, GLM, 100df
0 day
0 day
0 day
none
none
none
185
(-210-577)
275
(-195-740)
114
(-309-533)
128
(-357-608)
oact of Changing the T\
310
(127-491)
256
(96-415)
277
(76 - 475)
Max. positive est. =
275
Min. positive est. =
114
Pencentdiff. =
141%
248
(3-491)
147%
267%
52%
71%
fpe of Model, with a 0-Day Lag
Max. positive est. =
310
Min. positive est. =
256
Pencentdiff. =
21%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changing the Type of Model, with a 1-Day Lag
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-linear, GLM, 100df
1 day
1 day
1 day
none
none
none
221
(42 - 399)
140
(-21 - 299)
144
(-51 - 336)
Max. positive est. =
221
Min. positive est. =
140
Pencentdiff. =
58%
F-37
-------
Table F-32 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 s Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM25 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Im
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
2 day
2 day
2 day
none
none
none
Respiratory Hospital Admissions Associated with Short-Term Exposire to PM 2.5 - Im
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
0 day
1 day
2 day
3 day
NO2
NO2
NO2
NO2
Incidence
Associated with
PM2.5 Above Policy
Relevant
Background
oact of Changing the T\
343
(153-531)
212
(40 - 383)
192
(-22 - 403)
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
fpe of Model, with a 2-Day Lag
Max. positive est. =
343
Min. positive est. =
192
Percentdiff. =
79%
oact of Changjng the Lag Structure, with a Copollutant Model
78
(-171 -324)
-7
(-303-284)
65
(-192 -319)
-205
(-452 - 38)
Max. positive est. =
78
Min. positive est. =
65
Percentdiff. =
20%
The current primary PM2.5 standards include an annual standard set at 15 ug/m and a daily standard set at 35 ug/m . Results are based on Moolgavkar (2003) [reanalysis of MooIgavkar (2000a, 2000b, and
2000c)].
2The core analysis estimates for non-accidental mortality and cardiovascuter mortality associated with short-term exposure to PM2.5 are from Zanobetti and Schwartz (2009). The core analysis estimates for
cardbvascular hospital admissions associated with short-term exposure to PM2.5 are from Bell et al. (2008). Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or
credible intervals based onstatstical uncertainty surrounding the PM coefficient.
Calculated as (maximum positive estimate - minimum positive estimate)/(minimum positive estimate).
"Calculated as (Moolgavkar (2003) estimate - core analysis estimate)/(core analysis estimate).
Because "respiratory illness" was much more broadty defined in both Zanobetti and Schwartz (2009) and Bell et al. (2008) than in Moolgavkar (2003), a comparison betweenthe Moolgavkar (2003) estimates
and the corresponding core analysis estimates is not shown.
F-38
-------
Table F-33. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2 5 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM25Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impact of Changing the Lag Structure:
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 30 df
0 day
1 day
2 day
3 day
4 day
5 day
none
none
none
none
none
none
259
(-33 - 550)
283
(0 - 565)
183
(-91 - 455)
-72
(-351 - 205)
-43
(-310-222)
-271
(-558 - 1 4)
Max. positive est. =
283
Min. positive est. =
183
Percent d iff. =
55%
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impact of Changing the Typeof Model, with a 0-Day Lac
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GLM, 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
0 day
0 day
0 day
0 day
none
none
none
none
259
(-33 - 550)
192
(-164-546)
154
(-109-415)
144
(-206 - 492)
Max. positive est. =
259
Min. positive est. =
144
Percent d iff. =
80%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2n - Impactof Changing the Typeof Model, with a 1-Day Lac
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GLM, 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
1 day
none
none
none
none
283
(0 - 565)
264
(-81 - 607)
48
(-222 - 31 7)
-5
(-480 - 465)
Max. positive est. =
283
Min. positive est. =
48
Percent d iff. =
490%
77
(-110-262)
236%
268%
138%
-194%
-156%
-452%
F
77
(-110-262)
236%
149%
100%
87%
1
77
(-110-262)
268%
243%
-38%
-106%
Non-Accidental Mortality Associated with Short-Term Exposure to PM25 - Impactof a Copollutant Model
Mortality, short-term non-accidental
Mortality, short-term non- accidental
Mortality, short-term non-accidental
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
CO
CO
CO
-256
(-636-121)
-159
(-508-187)
-159
(-567 - 245)
77
(-110- 262)
-432%
-306%
-306%
F-39
-------
Table F-33 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 s Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM25 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM2 5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Cardiovascular Mortality Associated with Short-Term Exposure to PMi5 -Impactof Changing the Type of Model, with a 0-Day Lac,
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
Cardiovascular Mortality Associated with Short-Term Exposure to f
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
0 day
none
none
none
161
(16-306)
158
(23 - 292)
158
(-3-318)
Max. positive est. =
161
Min. positive est. =
158
Percentdiff. =
2%
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
1
-30
(-132 -72)
109%
105%
105%
TW25 -Impactof Changing the Type of Model, witha 1-Day Lag
1 day
1 day
1 day
none
none
none
168
(25 - 309)
130
(-6-265)
113
(-52 - 276)
Max. positive est. =
168
Min. positive est. =
113
Percentdiff. =
49%
-30
(-132 -72)
118%
69%
47%
Cardiovascular Mortality Associated with Short-Term Exposure to PMzs - Impactof a Copollutant Model
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 100df
log-linear, GLM, 100 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
0 day
0 day
1 day
1 day
Respiratory Mortality Associated with Short-Term Exposure to PM 2^ Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100 df
log-lhear, GLM, 100df
0 day
0 day
0 day
CO
CO
CO
CO
289
(123-453)
305
(109-498)
148
(-21 -316)
148
(-57-351)
Max. positive est. =
305
Min. positive est. =
148
Percentdiff. =
106%
-30
(-132 -72)
275%
296%
92%
92%
pact of Changing the Type of Model, with a 0-Day Lag
none
none
none
-14
(-75-46)
-35
(-96-24)
-30
(-103-40)
5
Respiratory Mortality Associated with Short-Term Exposure to PM2S -Impact of Changing the Typeof Model, witha 1-Day Lag
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-lhear, GAM (stringent), 100 df
log-lhear, GLM, 100df
1 day
1 day
1 day
none
none
none
9
(-52 - 70)
21
(-39-80)
5
(-71 -78)
Max. positive est. =
21
Min. positive est. =
5
Percentdiff. =
320%
F-40
-------
Table F-33 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2 5 Concentrations
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM2.5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum 2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.s Impact of Changing the Type of Model, with a 0-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
Oday
Oday
Oday
none
none
none
775
(446-1102)
570
(248 - 890)
619
(221 -1014)
Max. positive est. =
775
Min. positive est. =
570
Percent d iff. =
36%
258
(3-511)
906%
640%
704%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.s Impact of Changing the Type of Model, with a 1-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
1 day
1 day
1 day
none
none
none
682
(339-1023)
556
(228 - 880)
590
(189-987)
Max. positive est. =
682
Min. positive est. =
556
Percent d iff. =
23%
258
(3-511)
786%
622%
666%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2S - Impact of a Copollutant Model
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
Oday
Oday
1 day
1 day
CO
CO
CO
CO
193
(-219-600)
286
(-203 - 769)
119
(-321 - 554)
133
(-371 - 633)
Max. positive est. =
286
Min. positive est. =
119
Percent d iff. =
140%
258
(3-511)
151%
271%
55%
73%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2S - Impact of Changing the Type of Model, with a 0-Day Lag
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
Oday
Oday
Oday
none
none
none
316
(130-501)
262
(98 - 424)
282
(78 - 485)
Max. positive est. =
316
Min. positive est. =
262
Percent d iff. =
21%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2S - Impact of Changing the Type of Model, with a 1-Day Lag
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
1 day
1 day
1 day
none
none
none
226
(42 - 407)
143
(-21 - 305)
146
(-52 - 343)
Max. positive est. =
226
Min. positive est. =
143
Percent d iff. =
58%
F-41
-------
Table F-33 cont'd. Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2S Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Im
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
2 day
2 day
2 day
none
none
none
Incidence
Associated with
PM2 5 Above Policy
Relevant
Background
oact of Changng the T\
350
(156-541)
216
(41 -391)
196
(-22-411)
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
fpe of Model, with a 2-Day Lag
Max. positive est. =
350
Min. positive est. =
196
Pencentdiff. =
79%
Respiratory Hospital Admissions Associated with Short-Term Exposire to PM 2.5 - Impact of Changing the Lag Structure, with a Copollutant Model
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
log-linear, GAM (stringent), 100df
0 day
1 day
2 day
3 day
NO2
NO2
NO2
NO2
80
(-174-331)
-8
(-310-290)
67
(-196-326)
-209
(-462 - 39)
Max. positive est. =
80
Min. positive est. =
67
Pencentdiff. =
19%
The current primary PM2.5 standards include an annual standard set at 15 ug/m and a daily standard set at 35 ug/m . Results are based on Moolgavkar (2003) [reanalysis of MooIgavkar (2000a, 2000b, and
2000c)]. Numbers rounded to the nearest whole number. Numbers in parentheses a re 95% confidence or credible intervals based onstatstical uncertainty surrounding the PM coefficient.
2The core analysis estimates for non-accidental mortality and cardiovascuter mortality associated with short-term exposure to PM2.5 are from Zanobetti and Schwartz (2009). The core analysis estimates for
cardbvascular hospital admissions associated with short-term exposure to PM2.5 are from Bell et al. (2008).
Calculated as (maximum positive estimate - minimum positive estimate)/(minimum positive estimate).
"Calculated as (Moolgavkar (2003) estimate - core analysis estimate)/(core analysis estimate).
Because "respiratory illness" was much more broadty defined in both Zanobetti and Schwartz (2009) and Bell et al. (2008) than in Moolgavkar (2003), a comparison betweenthe Moolgavkar (2003) estimates
and the corresponding core analysis estimates is not shown.
F-42
-------
Table F-34. Sensitivity Analysis: Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to
PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2005 PM2 5 Concentrations:
Comparison of Proportional and Hybrid Rollback Methods1
Risk Assessment
Location
Baltimore, MD
Birmingham, AL
Detroit, Ml
Los Angeles, CA
New York, NY
St. Louis, MO
Type of Rollback
Proportional
Hybrid
Percent Difference 3
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM25 Concentrations that Just
Meet the Current and Alternative Combinations of Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m):
15/35 2
256
(104-406)
254
(103-402)
-1%
34
(-53-121)
39
(-61 -137)
15%
147
(-26-317)
151
(-26 - 325)
3%
81
(-117-278)
91
(-130-311)
12%
781
(459-1102)
795
(467-1121)
2%
260
(75 - 443)
271
(78 - 462)
4%
14/35
242
(98 - 384)
248
(101 -393)
2%
32
(-49-112)
36
(-56-127)
13%
146
(-26-315)
151
(-26 - 325)
3%
81
(-117-278)
91
(-130-311)
12%
781
(459-1102)
795
(467- 1121)
2%
244
(71 -416)
252
(73 - 429)
3%
13/35
224
(91 - 356)
229
(93 - 364)
2%
29
(-45-103)
33
(-52-117)
14%
135
(-24-291)
148
(-26-319)
10%
81
(-117-278)
89
(-127-304)
10%
761
(447-1073)
761
(446- 1073)
0%
226
(65 - 385)
233
(67 - 397)
3%
12/35
206
(83 - 327)
211
(86 - 335)
2%
27
(-41 - 94)
30
(-47-107)
11%
124
(-22 - 267)
136
(-24 - 293)
10%
77
(-110-263)
81
(-117-279)
5%
700
(41 1 - 987)
699
(410-986)
0%
207
(60 - 354)
214
(62 - 365)
3%
13/30
219
(89 - 348)
217
(88 - 344)
-1%
29
(-45-103)
33
(-52-117)
14%
125
(-22-271)
129
(-23 - 278)
3%
69
(-100-238)
78
(-1 1 1 - 266)
13%
668
(392 - 943)
680
(399 - 959)
2%
222
(64 - 379)
233
(67 - 397)
5%
12/25
182
(74 - 289)
180
(73 - 286)
-1%
24
(-38 - 85)
28
(-44 - 99)
17%
104
(-18-225)
107
(-19-231)
3%
58
(-82-197)
64
(-92 - 220)
10%
555
(325 - 783)
564
(331 - 797)
2%
184
(53-315)
195
(56 - 332)
6%
1Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional means. "Shrunken"
coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible
intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
F-43
-------
Table F-35. Sensitivity Analysis: Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 5
Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006 PM2 s Concentrations:
Comparison of Proportional and Hybrid Rollback Methods1
Risk Assessment
Location
Baltimore, MD
Birmingham, AL
Detroit, Ml
Los Angeles, CA
New York, NY
St. Louis, MO
Type of Rollback
Proportional
Hybrid
Percent Difference 3
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 5 Concentrations that Just
Meet the Current and Alternative Combinations of Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m):
15/352
224
(91 - 356)
222
(90 - 352)
-1%
33
(-51 -116)
37
(-58 - 1 32)
12%
118
(-21 - 255)
121
(-21 -261)
3%
75
(-1 08 - 257)
84
(-1 20 - 287)
12%
671
(394 - 946)
682
(400-961)
2%
215
(62 - 367)
224
(64-381)
4%
14/35
212
(86 - 336)
217
(88 - 344)
2%
30
(-47 - 1 08)
35
(-54 - 1 22)
17%
117
(-20 - 253)
121
(-21 -261)
3%
75
(-108-257)
84
(-120-287)
12%
671
(394 - 946)
682
(400-961)
2%
202
(58 - 345)
208
(60 - 354)
3%
13/35
196
(79-311)
200
(81 -318)
2%
28
(-44 - 99)
32
(-49-112)
14%
108
(-19-234)
118
(-21 - 256)
9%
75
(-108-257)
82
(-117-280)
9%
654
(383 - 922)
652
(383 - 920)
0%
187
(54 - 31 9)
192
(55 - 328)
3%
12/35
180
(73 - 286)
184
(75 - 293)
2%
26
(-40 - 90)
29
(-45-102)
12%
99
(-17-215)
109
(-19-235)
10%
71
(-101 -242)
75
(-108-257)
6%
601
(352 - 847)
599
(352 - 846)
0%
171
(49 - 293)
176
(51 -301)
3%
13/30
192
(78 - 305)
190
(77-301)
-1%
28
(-44 - 99)
32
(-49 - 1 1 2)
14%
101
(-18-218)
103
(-1 8 - 223)
2%
64
(-92-219)
72
(-102-245)
13%
574
(336 - 809)
583
(342 - 822)
2%
184
(53-314)
192
(55 - 328)
4%
12/25
159
(64 - 253)
157
(64 - 250)
-1%
23
(-36 - 82)
27
(-42 - 95)
17%
83
(-15-181)
85
(-15-1 85)
2%
53
(-76-182)
59
(-85 - 203)
11%
476
(279 - 672)
484
(284 - 683)
2%
152
(44 - 260)
160
(46 - 274)
5%
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been "shrunken1
"Shrunken" coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email. Numbers rounded to the nearest whole number
credible intervals based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
towards the appropriate regional means.
Numbers in parentheses are 95% confidence or
F-44
-------
Table F-36. Sensitivity Analysis: Estimated Annual Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 s
Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2007 PM2 5 Concentrations:
Comparison of Proportional and Hybrid Rollback Methods1
Risk Assessment
Location
Baltimore, MD
Birmingham, AL
Detroit, Ml
Los Angeles, CA
New York, NY
St. Louis, MO
Type of Rollback
Proportional
Hybrid
Percent Difference 3
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Proportional
Hybrid
Percent Difference
Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM25 Concentrations that Just
Meet the Current and Alternative Combinations of Annual (n) and Daily (m) Standards (Standard Combination
Denoted n/m):
15/35 2
227
(92 - 360)
224
(91 - 356)
-1%
34
(-53-120)
39
(-60-137)
15%
121
(-21 - 262)
124
(-22 - 269)
2%
77
(-110-262)
86
(-123-293)
12%
734
(431 - 1035)
746
(438-1052)
2%
225
(65 - 384)
236
(68 - 402)
5%
14/35
214
(87 - 340)
219
(89 - 348)
2%
32
(-49-111)
36
(-56-127)
13%
120
(-21 -261)
124
(-22 - 269)
3%
77
(-110-262)
86
(-123-293)
12%
734
(431 -1035)
746
(438-1052)
2%
211
(61 - 360)
219
(63 - 374)
4%
13/35
198
(80-315)
203
(82 - 322)
3%
29
(-45-102)
33
(-51 -116)
14%
111
(-19-241)
122
(-21 - 264)
10%
77
(-110-262)
83
(-120-286)
8%
715
(419-1008)
714
(419-1007)
0%
195
(56 - 333)
203
(59 - 346)
4%
12/35
182
(74 - 289)
186
(75 - 296)
2%
26
(-41 - 93)
30
(-47-106)
15%
102
(-18-221)
112
(-20 - 242)
10%
72
(-104-247)
77
(-110-262)
7%
657
(385 - 927)
656
(385 - 926)
0%
179
(52 - 306)
186
(54-318)
4%
13/30
194
(79 - 308)
192
(78 - 304)
-1%
29
(-45-102)
33
(-51 -116)
14%
104
(-18-224)
106
(-19-230)
2%
65
(-94 - 224)
73
(-105-250)
12%
627
(368 - 885)
638
(374 - 900)
2%
192
(55 - 328)
203
(59 - 346)
6%
12/25
161
(65 - 256)
159
(64 - 253)
-1%
24
(-37 - 85)
28
(-43 - 99)
17%
86
(-15-186)
88
(-15-191)
2%
54
(-78-186)
61
(-87 - 207)
13%
521
(305 - 735)
530
(310-748)
2%
160
(46 - 272)
169
(49 - 289)
6%
Based on location-specific single pollutant concentration-response function estimates from Zanobetti and Schwartz (2009) that have been
coefficient estimates and their standard errors were sent to EPA by A. Zanobetti via email. Numbers rounded to the nearest whole number
based on statistical uncertainty surrounding the PM coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Calculated as (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
'shrunken" towards the appropriate regional means. "Shrunken"
. Numbers in parentheses are 95% confidence or credible intervals
F-45
-------
Table F-37. Multi-Factor Sensitivity Analysis: Impact of Using a Log-Linear vs. a Log-Log Model, Estimating Incidence Down to the Lowest
Measured Level (LML) in the Study vs. PRB, and Using a Proportional vs. a Hybrid Rollback to Estimate the Incidence of All Cause
and Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 s Concentrations that Just Meet the Current
Standards, Based on Adjusting 2005 PM2 5 Concentrations1
Modeling Choices:
Fixed Effects (FE) Log-linear vs.
Random Effects (RE) log-log model
Down to LML (5.8 ug/m3) vs. PRB
Proportional vs. hybrid rollback
All Cause Mortality
Percent Difference: 3
Ischemic Heart Disease Mortality
Percent Difference:
All Cause Mortality
Percent Difference:
Ischemic Heart Disease Mortality
Percent Difference
Incidence of Mortality Associated with Long-Term Exposure to PM2.5 Concentrations Using:2
FE Log-Linear
LML
Proportional
FE Log-Linear
LML
Hybrid
FE Log-Linear
PRB
Proportional
FE Log-Linear
PRB
Hybrid
RE Log-Log
LML
Proportional
RE Log-Log
LML
Hybrid
RE Log-Log
PRB
Proportional
RE Log-Log
PRB
Hybrid
Los Angeles, CA
1342
(854-1827)
1249
(1017-1477)
1675
(1066-2276)
25%
1545
(1261 -1824)
24%
2845
(1819-3853)
112%
2548
(2095 - 2983)
104%
3169
(2027 - 4286)
136%
2813
(2318-3288)
125%
3360
(2075-4615)
150%
2535
(1793-3232)
103%
3953
(2446-5418)
195%
2947
(2095 - 3738)
136%
13557
(8709-17917)
910%
8269
(6414-9670)
562%
14037
(9035-18516)
946%
8475
(6602 - 9873)
579%
Philadelphia, PA
584
(372 - 792)
369
(303 - 434)
4
859
(550-1161)
47%
591
(489 - 688)
60%
1254
(779-1713)
115%
639
(458 - 803)
73%
3946
(2554-5176)
576%
1612
(1271 -1859)
337%
1The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000. The fixed effects log-linear estimates are from Table 33, using models with 44 individual and 7 ecological covariates; the random
effects log-log estimates are from Table 11, "MSA and DIFF" rows. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical
uncertainty surrounding the PM coefficient.
3 Percent differences are calculated relative to the model selections used in the core analysis (fixed effects log-linear model; LML, and proportional rollbacks). So, for example, the percent difference in estimated
all cause mortality in Los Angeles resulting from changing from the core analysis input selections to instead using (1) a fixed effects log-linear model, (2) PRB, and (3) hybrid rollbacks is (3169 -1342)/1342 =
136%.
4 Philadelphia was not among the risk assessment urban areas for which hybrid rollbacks were calculated.
F-46
-------
Table F-38. Multi-Factor Sensitivity Analysis: Impact of Using a Log-Linear vs. a Log-Log Model, Estimating Incidence Down to the Lowest
Measured Level (LML) in the Study vs. PRB, and Using a Proportional vs. a Hybrid Rollback to Estimate the Incidence of All Cause
and Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 s Concentrations that Just Meet the Current
Standards, Based on Adjusting 2006 PM2 5 Concentrations1
Modeling Choices:
Fixed Effects (FE) Log-linear vs.
Random Effects (RE) log-log model
Down to LML (5.8 ug/m3) vs. PRB
Proportional vs. hybrid rollback
All Cause Mortality
Percent Difference: 3
Ischemic Heart Disease Mortality
Percent Difference:
All Cause Mortality
Percent Difference:
Ischemic Heart Disease Mortality
Percent Difference
Incidence of Mortality Associated with Long-Term Exposure to PM2.5 Concentrations Using:2
FE Log-Linear
LML
Proportional
FE Log-Linear
LML
Hybrid
FE Log-Linear
PRB
Proportional
FE Log-Linear
PRB
Hybrid
RE Log-Log
LML
Proportional
RE Log-Log
LML
Hybrid
RE Log-Log
PRB
Proportional
RE Log-Log
PRB
Hybrid
Los Angeles, CA
1108
(704-1509)
1038
(843-1229)
1414
(899-1923)
28%
1314
(1070-1553)
27%
2627
(1678-3560)
137%
2366
(1943-2775)
128%
2924
(1869-3959)
164%
2614
(2150-3060)
152%
2904
(1790-3995)
162%
2212
(1558-2833)
113%
3498
(2161 -4803)
216%
2633
(1864-3354)
154%
13255
(8501 -17544)
1096%
8151
(6301 -9561)
685%
13736
(8827-18146)
1140%
8361
(6491 - 9770)
705%
Philadelphia, PA
525
(335-713)
334
(273 - 393)
4
912
(585-1233)
74%
559
(461 -651)
67%
1166
(723-1595)
122%
598
(428 - 755)
79%
3869
(2502 - 5082)
637%
1590
(1251 -1837)
376%
1The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000. The fixed effects log-linear estimates are from Table 33, using models with 44 individual and 7 ecological covariates; the random
effects log-log estimates are from Table 11, "MSA and DIFF" rows. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical
uncertainty surrounding the PM coefficient.
3 Percent differences are calculated relative to the model selections used in the core analysis (fixed effects log-linear model; LML, and proportional rollbacks). So, for example, the percent difference in estimated
all cause mortality in Los Angeles resulting from changing from the core analysis input selections to instead using (1) a fixed effects log-linear model, (2) PRB, and (3) hybrid rollbacks is (2924 -1108)/1108 =
164%.
4 Philadelphia was not among the risk assessment urban areas for which hybrid rollbacks were calculated.
F-47
-------
Table F-39. Multi-Factor Sensitivity Analysis: Impact of Using a Log-Linear vs. a Log-Log Model, Estimating Incidence Down to the Lowest
Measured Level (LML) in the Study vs. PRB, and Using a Proportional vs. a Hybrid Rollback to Estimate the Incidence of All Cause
and Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 s Concentrations that Just Meet the Current
Standards, Based on Adjusting 2007 PM2 5 Concentrations1
Modeling Choices:
Fixed Effects (FE) Log-linear vs.
Random Effects (RE) log-log model
Down to LML (5.8 ug/m3) vs. PRB
Proportional vs. hybrid rollback
All Cause Mortality
Percent Difference: 3
Ischemic Heart Disease Mortality
Percent Difference:
All Cause Mortality
Percent Difference:
Ischemic Heart Disease Mortality
Percent Difference
Incidence of Mortality Associated with Long-Term Exposure to PM2.5 Concentrations Using:2
FE Log-Linear
LML
Proportional
FE Log-Linear
LML
Hybrid
FE Log-Linear
PRB
Proportional
FE Log-Linear
PRB
Hybrid
RE Log-Log
LML
Proportional
RE Log-Log
LML
Hybrid
RE Log-Log
PRB
Proportional
RE Log-Log
PRB
Hybrid
Los Angeles, CA
1170
(744-1593)
1094
(890-1296)
1484
(944-2019)
27%
1377
(1122-1627)
26%
2697
(1723-3654)
131%
2426
(1993-2845)
122%
3003
(1920-4064)
157%
2680
(2205-3136)
145%
3034
(1871 -4173)
159%
2306
(1626-2950)
111%
3633
(2245 - 4986)
211%
2728
(1933-3472)
149%
13430
(8616-17770)
1048%
8243
(6377 - 9662)
653%
13914
(8945-18375)
1089%
8454
(6568-9871)
673%
Philadelphia, PA
519
(331 - 704)
330
(270 - 389)
4
907
(581 -1225)
75%
555
(459 - 647)
68%
1157
(718-1583)
123%
594
(424 - 750)
80%
3864
(2498 - 5075)
645%
1589
(1249-1836)
382%
1The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2Estimates Based on Krewski et al. (2009), exposure period from 1999 - 2000. The fixed effects log-linear estimates are from Table 33, using models with 44 individual and 7 ecological covariates; the random
effects log-log estimates are from Table 11, "MSA and DIFF" rows. Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical
uncertainty surrounding the PM coefficient.
3 Percent differences are calculated relative to the model selections used in the core analysis (fixed effects log-linear model; LML, and proportional rollbacks). So, for example, the percent difference in estimated
all cause mortality in Los Angeles resulting from changing from the core analysis input selections to instead using (1) a fixed effects log-linear model, (2) PRB, and (3) hybrid rollbacks is (3003 -1170)/1170 =
157%.
4 Philadelphia was not among the risk assessment urban areas for which hybrid rollbacks were calculated.
F-48
-------
Table F-40. Sensitivity Analysis: Impact of Using Season-Specific vs. Annual Concentration-Response Functions and
Proportional vs. Hybrid Rollbacks to Estimate the Incidence of Non-Accidental Mortality Associated with Short-
Term Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM2 5 Concentrations 1>z
Modeling Choices:
Seasonal C-R Functions vs. an All-Year
Function
Proportional vs. Hybrid Rollback
Baltimore, MD
Percent Difference 3
Birmingham, AL
Percent Difference
Detroit, Ml
Percent Difference
Los Angeles, CA
Percent Difference
New York, NY
Percent Difference
Pittsburgh, PA
Percent Difference
St. Louis, MO
Percent Difference
Estimated Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5
Concentrations that Just Meet the Current Standards
All Year
Proportional
256
(104-406)
34
(-53-121)
147
(-26-317)
._
81
(-1 1 7 - 278)
781
(459-1102)
159
(47 - 270)
._
260
(75 - 443)
All Year
Hybrid
254
(103-402)
-1%
39
(-61 -137)
15%
151
(-26 - 325)
3%
91
(-130-311)
12%
795
(467-1121)
2%
163
(48 - 277)
3%
271
(78 - 462)
4%
Sum of Four Seasons
Proportional
222
4
-1 3%
70
106%
159
8%
-23
-128%
780
0%
175
10%
251
-3%
Sum of Four Seasons
Hybrid
220
-1 4%
79
132%
163
11%
-25
-131%
792
1%
182
14%
261
0%
Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM
coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Percent differences are calculated relative to the model selections used in the core analysis (all-year C-R function and proportional rollback). So, for example, the percent
difference in estimated non-accidental mortality in Baltimore resulting from changing from the core analysis input selections to instead using the sum of four season-specific mortality
estimates and hybrid rollbacks is (192 - 225)7225 = -15%.
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-
specific coefficient estimators was not available.
F-49
-------
Table F-41. Sensitivity Analysis: Impact of Using Season-Specific vs. Annual Concentration-Response Functions and
Proportional vs. Hybrid Rollbacks to Estimate the Incidence of Non-Accidental Mortality Associated with Short-
Term Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM2 5 Concentrations 1>z
Modeling Choices:
Seasonal C-R Functions vs. an All-Year
Function
Proportional vs. Hybrid Rollback
Baltimore, MD
Percent Difference 3
Birmingham, AL
Percent Difference
Detroit, Ml
Percent Difference
Los Angeles, CA
Percent Difference
New York, NY
Percent Difference
Pittsburgh, PA
Percent Difference
St. Louis, MO
Percent Difference
Estimated Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5
Concentrations that Just Meet the Current Standards
All Year
Proportional
224
(91 - 356)
33
(-51 -116)
118
(-21 - 255)
._
75
(-108-257)
671
(394 - 946)
136
(40 - 232)
._
215
(62 - 367)
All Year
Hybrid
222
(90 - 352)
-1%
37
(-58-132)
12%
121
(-21 -261)
3%
84
(-120-287)
12%
682
(400 - 961 )
2%
147
(43 - 249)
8%
224
(64 - 381 )
4%
Sum of Four Seasons
Proportional
193
4
-1 4%
69
109%
137
16%
-25
-133%
677
1%
154
13%
211
-2%
Sum of Four Seasons
Hybrid
193
-1 4%
78
136%
140
19%
-28
-137%
688
3%
164
21%
219
2%
Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM
coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Percent differences are calculated relative to the model selections used in the core analysis (all-year C-R function and proportional rollback). So, for example, the percent
difference in estimated non-accidental mortality in Baltimore resulting from changing from the core analysis input selections to instead using the sum of four season-specific mortality
estimates and hybrid rollbacks is (192 - 225)7225 = -15%.
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-
specific coefficient estimators was not available.
F-50
-------
Table F-42. Sensitivity Analysis: Impact of Using Season-Specific vs. Annual Concentration-Response Functions and
Proportional vs. Hybrid Rollbacks to Estimate the Incidence of Non-Accidental Mortality Associated with Short-
Term Exposure to PM2 5 Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2 5 Concentrations 1>z
Modeling Choices:
Seasonal C-R Functions vs. an All-Year
Function
Proportional vs. Hybrid Rollback
Baltimore, MD
Percent Difference 3
Birmingham, AL
Percent Difference
Detroit, Ml
Percent Difference
Los Angeles, CA
Percent Difference
New York, NY
Percent Difference
Pittsburgh, PA
Percent Difference
St. Louis, MO
Percent Difference
Estimated Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to
PM2.5 Concentrations that Just Meet the Current Standards
All Year
Proportional
227
(92 - 360)
34
(-53-120)
121
(-21 - 262)
._
77
(-1 1 0 - 262)
734
(431 - 1 035)
143
(42 - 244)
._
225
(65 - 384)
All Year
Hybrid
224
(91 - 356)
-1%
39
(-60-137)
15%
124
(-22 - 269)
2%
86
(-123-293)
12%
746
(438 - 1 052)
2%
147
(43 - 250)
3%
236
(68 - 402)
5%
Sum of Four Seasons
Proportional
207
4
-9%
95
179%
166
37%
-12
-116%
887
21%
225
57%
248
10%
Sum of Four Seasons
Hybrid
194
-1 5%
86
153%
137
13%
-8
-110%
750
2%
162
13%
232
3%
Based on season-specific and all-year location-specific coefficient estimates from Zanobetti and Schwartz (2009) that have been "shrunken" towards the appropriate regional
means. Numbers are rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM
coefficient.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 Percent differences are calculated relative to the model selections used in the core analysis (all-year C-R function and proportional rollback). So, for example, the percent
difference in estimated non-accidental mortality in Baltimore resulting from changing from the core analysis input selections to instead using the sum of four season-specific mortality
estimates and hybrid rollbacks is (192 - 225)7225 = -15%.
4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the season-
specific coefficient estimators was not available.
F-51
-------
Table F-43. Sensitivity Analysis: Impact of Copollutant Models in Estimating the Incidence of All Cause
Mortality Associated with Long-Term Exposure to PM25 Concentrations that Just Meet the
Current Standards, Based on Adjusting 2005 PM2 s Concentrations1'2
Copollutant in Model
Incidence
Percent Difference 3
Los Angeles, CA
None
CO
N02
03
SO2
1122
(580- 1713)
1632
(945 - 2341 )
1954
(1034-2782)
1632
(945 - 2341 )
295
(-515-1209)
0%
45%
74%
45%
-74%
Philadelphia, PA
None
CO
NO2
03
S02
489
(253 - 743)
708
(412- 1012)
847
(451 - 1 1 99)
708
(412- 1012)
129
(-227 - 526)
0%
45%
73%
45%
-74%
1The current primary PM2.5 standards include an annual standard setat 15 ug/m3 and a daily standard setat 35 ug/m3.
Estimates based on Krewski et al. (2000) [reanalysis of the ACS study]. Mortality incidence was estimated for PM2 5 concentrations down to
5.8 ug/m (the lowest measured level used for the analyses of long-term exposure). Numbers rounded to the nearest whole number.
Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
Cabulated as (estimate with Copollutant - estimate without Copollutant)/(estimate without Copollutant).
F-52
-------
Table F-44. Sensitivity Analysis: Impact of Copollutant Models in Estimating the Incidence of All Cause
Mortality Associated with Long-Term Exposure to PM25 Concentrations that Just Meet the
Current Standards, Based on Adjusting 2006 PM2 s Concentrations1'2
Copollutant in Model
Incidence
Percent Difference 3
Los Angeles, CA
None
CO
NO2
03
SO2
926
(478- 1415)
1347
(780- 1936)
1615
(853 - 2302)
1347
(780- 1936)
243
(-424 - 998)
0%
45%
74%
45%
-74%
Philadelphia, PA
None
CO
N02
a
S02
439
(228-669)
637
(370-911)
762
(405- 1080)
637
(370-911)
116
(-203 - 473)
0%
45%
74%
45%
-74%
1The current primary PM25 standards include an annual standard setat 15 ug/m3 and a daily standard setat 35 ug/m3.
2 Estimates based on Krewski etal. (2000) [reanalysis of the ACS study]. Mortality incidence was estimated for PM25 concentrations down to
5.8 ug/m3 (the lowest measured level used for the analyses of long-term exposure). Numbers rounded to the nearest whole number.
Numbers in parentheses a re 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3Cabulated as (estimate with copolutant - estimate without copollutant)/(estimate without Copollutant).
F-53
-------
Table F-45. Sensitivity Analysis: Impact of Copollutant Models in Estimating the Incidence of All Cause
Mortality Associated with Long-Term Exposure to PM25 Concentrations that Just Meet the
Current Standards, Based on Adjusting 2007 PM2 5 Concentrations1'2
Copollutant in Model
Incidence
Percent Difference 3
Los Angeles, CA
None
CO
N02
03
SO2
978
(505 - 1 494)
1423
(824 - 2043)
1705
(901 - 2429)
1423
(824 - 2043)
257
(-448 - 1 054)
0%
46%
74%
46%
-74%
Philadelphia, PA
None
CO
N02
a
S02
434
(225-661)
630
(366-901)
753
(400 - 1 068)
630
(366-901)
115
(-201 - 468)
0%
45%
74%
45%
-74%
The current primary PM25 standards include an annual standard setat 15 ug/m and a daily standard setat 35 ug/m .
2 Estimates based on Krewski etal. (2000) [reanalysis of the ACS study]. Mortality incidence was estimated for PM25 concentrations down to
5.8 ug/m3 (the lowest measured level used for the analyses of long-term exposure). Numbers rounded to the nearest whole number.
Numbers in parentheses a re 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficient.
3Cabulated as (estimate with copolutant - estimate without copollutanty(estimate without Copollutant).
F-54
-------
Table F-46. Sensitivity Analysis: Impact of Different Lag Models on Estimated Annual Incidence of Hospital
Admissions Associated with Short-Term Exposure to Ambient PM25 Concentrations that Just Meet the
Current Standards, Based on Adjusting 2005 PM2 5 Concentrations
1,2
Risk Assessment
Location
Los Angeles, CA
Philadelphia, PA
Cardiovascular Hospital Admissions
0-Day Lag
397
(294-501)
159
(118-200)
1-Day Lag
35
(-60-130)
14
(-24 - 52)
2-Day Lag
30
(-58 - 1 1 8)
12
(-23 - 47)
Respiratory Hospital Admissions
0-Day Lag
40
(-22- 102)
13
(-7 - 34)
1-Day Lag
9
(-53-71)
3
(-18-24)
2-Day Lag
75
(16-133)
25
(5-45)
1The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2 Incidence estimates were calculated using the national concentration-responsef unction estimates reported in Table 1 of Bell et al. (2008). Location-
specific C-R function estimates were not availablefrom this study.
F-55
-------
Table F-47. Sensitivity Analysis: Impact of Different Lag Models on Estimated Annual Incidence of Hospital
Admissions Associated with Short-Term Exposure to Ambient PMZ5 Concentrations that Just Meet the
Current Standards, Based on Adjusting 2006 PM2.5 Concentrations
1,2
Risk Assessment
Location
Los Angeles, CA
Philadelphia, PA
Cardiovascular Hospital Admissions
0-Day Lag
373
(276 - 470)
149
(110- 188)
1-Day Lag
33
(-56-122)
13
(-23 - 49)
2-Day Lag
28
(-54 - 1 1 0)
11
(-22 - 44)
Respiratory Hospital Admissions
0-Day Lag
38
(-21 -96)
13
(-7 - 32)
1-Day Lag
9
(-50 - 67)
3
(-17-22)
2-Day Lag
70
(15-125)
24
(5-42)
1The current primary PIVh.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2 Incidence estimates were calculated using the national concentration-responsef unction estimates reported in Table 1 of Bell et al. (2008). Location-
specific C-R function estimates were not availablefrom this study.
F-56
-------
Table F-48. Sensitivity Analysis: Impact of Different Lag Models on Estimated Annual Incidence of Hospital
Admissions Associated with Short-Term Exposure to Ambient PM25 Concentrations that Just Meet the
Current Standards, Based on Adjusting 2007 PM2.5 Concentrations
1,2
Risk Assessment
Location
Los Angeles, CA
Philadelphia, PA
Cardiovascular Hospital Admissions
0-Day Lag
388
(287 - 489)
151
(112- 190)
1-Day Lag
34
(-59-127)
13
(-23 - 49)
2-Day Lag
29
(-56 - 1 1 5)
11
(-22 - 45)
Respiratory Hospital Admissions
0-Day Lag
39
(-21 -99)
13
(-7 - 32)
1-Day Lag
9
(-52 - 69)
3
(-17-23)
2-Day Lag
73
(15-130)
24
(5-43)
1The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2 Incidence estimates were calculated using the national concentration-responsef unction estimates reported in Table 1 of Bell et al. (2008). Location-
specific C-R function estimates were not availablefrom this study.
F-57
-------
Table F-49 Maximum 3yr Monitor-Specific Average and Annual Composite Monitor Value Given Different Rollback Methods (with comparison of
percent reduction in surrogate for long-term mortality risk across rollback methods)
Risk Assessment
Location 1
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Rollback
Method
Proportional
Hybrid 3
Peak Shaving4
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Design Value
Annual
16.2
15.6
18.7
12.8
17.2
17.4
15.8
19.6
15.9
15.0
12.6
24-Hr
35.0
37.0
44.0
26.0
43.0
63.0
31.0
55.0
42.0
38.0
32.0
Recent
Air
Quality
(2007)
2007
CM
15.3
13.9
15.7
11.4
13.9
17.4
13.2
14.6
13.8
13.4
9.9
Maximum Monitor-Specific Avg. of 2005, 2006, 2007 Annual Avgs. (Max. M-S) and 2007
Annual Average at Composite Monitor (2007CM) (in ug/m3)
15/352
Max. M
S
15.0
14.8
14.3
15.2
15.0
15.0
12.8
14.1
13.2
13.9
9.9
9.8
15.0
12.7
13.3
13.9
13.3
13.6
14.2
13.9
15.5
12.6
2007
CM
14.2
13.1
13.0
13.6
12.7
14.2
11.4
11.4
11.7
12.6
9.9
9.9
12.5
9.5
10.5
12.0
11.6
11.8
13.2
12.3
12.9
9.9
14/35
Max. M
S
14.0
14.0
14.0
14.0
14.0
12.8
14.0
13.2
9.9
9.8
14.0
12.7
13.3
13.9
13.3
13.6
14.2
13.9
15.5
12.6
2007
CM
13.3
12.5
12.7
11.8
13.2
11.4
11.4
11.7
9.9
9.9
11.7
9.5
10.5
12.0
11.6
11.8
13.2
12.3
12.9
9.9
13/35
Max. M
S
13.0
13.0
13.0
13.0
13.0
12.8
13.0
13.0
9.9
9.8
13.0
12.7
13.0
13.9
13.0
13.0
13.0
12.6
2007
CM
12.3
11.6
11.8
11.0
12.3
11.4
10.6
11.5
9.9
9.9
10.9
9.5
10.3
12.0
11.3
11.3
11.6
9.9
12/35
Max. M
S
12.0
12.0
12.0
12.0
12.0
12.0
12.0
12.0
9.9
9.8
12.0
12.0
12.0
12.0
12.0
12.0
12.0
2007
CM
11.4
10.7
10.9
10.2
11.4
10.7
9.8
10.6
9.9
9.9
10.1
9.0
9.5
10.4
10.4
10.7
9.4
13/30
Max. M
S
13.0
12.7
12.3
13.0
13.0
13.0
12.8
12.2
11.4
11.9
8.6
8.4
13.0
10.9
11.5
11.8
11.5
11.7
12.1
11.9
14.1
11.8
2007
CM
12.3
11.3
11.2
11.9
11.0
12.3
11.4
9.9
10.1
10.8
8.6
8.5
10.9
8.2
9.1
10.4
10.0
10.2
11.5
10.7
11.2
9.3
12/25
Max. M
S
11.8
13.6
10.7
10.3
10.8
11.1
11.3
11.9
12.0
10.2
9.6
9.8
7.3
6.9
12.0
9.2
9.6
9.8
9.7
9.8
10.1
10.0
11.7
9.9
in 1
2007
CM
11.2
11.31
9.5
9.4
9.8
9.4
10.7
10.9
10.7
8.3
8.5
8.9
7.3
7.0
10.1
7.0
7.7
8.8
8.4
8.5
9.5
9.0
9.3
7.8
R Q
Percent reduction in a surrogate for
long-term exposure-related mortality
(alternative standard compared with
current standard)6
14/35
11%
9%
4%
12%
11%
196%
1%
0%
0%
0%
12%
0%
0%
0%
0%
0%
0%
0%
0%
0%
13/35
22%
21%
16%
24%
22%
0%
16%
3%
0%
0%
24%
0%
5%
0%
5%
8%
12%
0%
12/35
34%
33%
29%
36%
34%
13%
30%
18%
0%
0%
36%
13%
21%
20%
22%
25%
11%
13/30
22%
25%
25%
25%
24%
22%
0%
27%
27%
27%
32%
32%
24%
34%
30%
34%
27%
27%
27%
26%
26%
14%
12/25
35%
35%
49%
50%
49%
47%
42%
47%
13%
55%
54%
55%
64%
64%
36%
68%
60%
68%
55%
54%
55%
52%
52%
50%
=;n%
F-58
-------
Table F-49 (cont'd) Maximum 3yr Monitor-Specific Average and Annual Composite Monitor Value Given Different Rollback Methods (with comparison
of percent reduction in surrogate for long-term mortality risk across rollback methods)
Risk Assessment
Location 1
Pittsburgh, PA5
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Rollback
Method
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Design Value
Annual
19.8
11.6
16.5
10.2
24-Hr
60.0
55.0
39.0
43.0
Recent
Air
Quality
(2007)
2007
CM
14.9
11.4
14.3
9.7
Maximum Monitor-Specific Avg. of 2005, 2006, 2007 Annual Avgs. (Max. M-S) and 2007
Annual Average at Composite Monitor (2007CM) (in ug/m3)
15/352
Max. M
S
13.3
15.6
7.7
10.8
14.9
15.0
16.5
8.4
8.3
2007
CM
11.6
13.1
7.5
9.5
12.9
13.5
14.1
8.0
7.8
14/35
Max. M
S
13.3
15.6
7.7
10.8
14.0
14.0
8.4
8.3
2007
CM
11.6
13.1
7.5
9.5
12.1
12.6
8.0
7.8
13/35
Max. M
S
12.8
15.3
7.7
10.8
13.0
13.0
8.4
8.3
2007
CM
11.2
11.7
7.5
9.5
11.3
11.7
8.0
7.8
12/35
Max. M
S
11.8
15.3
7.7
10.8
12.0
12.0
8.4
8.3
2007
CM
10.5
11.0
7.5
9.5
10.4
10.8
8.0
7.8
13/30
Max. M
S
11.5
15.6
6.7
10.8
12.8
13.0
14.1
7.4
7.1
2007
CM
10.0
11.2
6.6
8.6
11.1
11.7
12.3
7.0
6.7
12/25
Max. M
S
9.7
13.8
5.7
8.9
10.8
11.0
11.7
6.3
5.9
2007
CM
8.4
9.3
5.6
7.4
9.3
9.9
10.2
6.0
5.5
Percent reduction in a surrogate for
long-term exposure-related mortality
(alternative standard compared with
current standard)6
14/35
0%
0%
0%
0%
10%
12%
0%
0%
13/35
7%
20%
0%
0%
23%
23%
0%
0%
12/35
19%
29%
0%
0%
35%
35%
0%
0%
13/30
27%
26%
55%
24%
25%
23%
25%
46%
57%
12/25
54%
52%
110%
58%
50%
47%
50%
93%
114%
1For some locations (e.g., Atlanta) more than one "version" (group of counties) was used in the risk assessment. In this table only the version that was used for mortality associated with
short-term exposure to PM2.s (Zanobetti and Schwartz, 2009) is included.
2 The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 The hybrid rollback method was applied to only a subset of the risk assessment locations. The "" for a given location indicates that the hybrid rollback method was not applied to that
4 The peak shaving method was applied to a location-standard combination only if the daily standard was controlling in that location. The "-" for a given location-standard combination
indicates that, for that set of annual and daily standards in that location, the annual standard was controlling and so the peak shaving method was not applied.
5 The proportional rollback and peak shaving methods were applied to Pittsburgh differently from the way they were applied in the other locations. See Sections 3.2.3.2 and 3.2.3.3 for
6 Percent reduction in composite monitor value (CMV) with consideration for LMLof 5.8 ug/m3. Percent reduction = (CMVcurrent standard - CMVaitemativestandard)/(CMVcurremtstandard-LML). Note
that greyed cells identify instances where percent change differs by >10% across alternative rollback methods (for a given alternative standard level/study area combination).
F-59
-------
Table F-50. Maximum 3yr Monitor-Specific Average and Annual Composite Monitor Value Given Different Rollback Methods (with percent difference in
surrogate for long-term exposure-related mortality across rollback methods)
Risk
Assessment
Location 1
Atlanta, GA
Baltimore, MD
Birmingham,
AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles,
CA
New York, NY
Philadelphia,
PA
Phoenix, AZ
Pittsburgh, PA
5
Rollback
Method
Proportional
Hybrid 3
Peak Shaving4
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Design Value
Annual
16.2
15.6
18.7
12.8
17.2
17.4
15.8
19.6
15.9
15.0
12.6
19.8
24-Hr
35.0
37.0
44.0
26.0
43.0
63.0
31.0
55.0
42.0
38.0
32.0
60.0
Recent Air
Quality
(2007)
2007 CM
15.3
13.9
15.7
11.4
13.9
17.4
13.2
14.6
13.8
13.4
9.9
14.9
Maximum Monitor-Specific Avg. of 2005, 2006, 2007 Annual Avgs. (Max. M-S) and 2007
Annual Average at Composite Monitor (2007CM) (in ug/m3)
15/352
Max. M
S
15.0
14.8
14.3
15.2
15.0
15.0
12.8
14.1
13.2
13.9
9.9
9.8
15.0
12.7
13.3
13.9
13.3
13.6
14.2
13.9
15.5
12.6
13.3
15.6
2007
CM
14.2
13.1
13.0
13.6
12.7
14.2
11.4
11.4
11.7
12.6
9.9
9.9
12.5
9.5
10.5
12.0
11.6
11.8
13.2
12.3
12.9
9.9
11.6
13.1
14/35
Max. M
S
14.0
14.0
14.0
14.0
14.0
12.8
14.0
13.2
9.9
9.8
14.0
12.7
13.3
13.9
13.3
13.6
14.2
13.9
15.5
12.6
13.3
15.6
2007
CM
13.3
12.5
12.7
11.8
13.2
11.4
11.4
11.7
9.9
9.9
11.7
9.5
10.5
12.0
11.6
11.8
13.2
12.3
12.9
9.9
11.6
13.1
13/35
Max. M
S
13.0
13.0
13.0
13.0
13.0
12.8
13.0
13.0
9.9
9.8
13.0
12.7
13.0
13.9
13.0
13.0
13.0
12.6
12.8
15.3
2007
CM
12.3
11.6
11.8
11.0
12.3
11.4
10.6
11.5
9.9
9.9
10.9
9.5
10.3
12.0
11.3
11.3
11.6
9.9
11.2
11.7
12/35
Max. M
S
12.0
12.0
12.0
12.0
12.0
12.0
12.0
12.0
9.9
9.8
12.0
12.0
12.0
12.0
12.0
12.0
12.0
11.8
15.3
2007
CM
11.4
10.7
10.9
10.2
11.4
10.7
9.8
10.6
9.9
9.9
10.1
9.0
9.5
10.4
10.4
10.7
9.4
10.5
11.0
13/30
Max. M
S
13.0
12.7
12.3
13.0
13.0
13.0
12.8
12.2
11.4
11.9
8.6
8.4
13.0
10.9
11.5
11.8
11.5
11.7
12.1
11.9
14.1
11.8
11.5
15.6
2007
CM
12.3
11.3
11.2
11.9
11.0
12.3
11.4
9.9
10.1
10.8
8.6
8.5
10.9
8.2
9.1
10.4
10.0
10.2
11.5
10.7
11.2
9.3
10.0
11.2
12/25
Max. M
S
11.8
13.6
10.7
10.3
10.8
11.1
11.3
11.9
12.0
10.2
9.6
9.8
7.3
6.9
12.0
9.2
9.6
9.8
9.7
9.8
10.1
10.0
11.7
9.9
10.1
9.7
13.8
2007
CM
11.2
11.31
9.5
9.4
9.8
9.4
10.7
10.9
10.7
8.3
8.5
8.9
7.3
7.0
10.1
7.0
7.7
8.8
8.4
8.5
9.5
9.0
9.3
7.8
8.9
8.4
9.3
Percent difference between composite
monitor value with hybrid or peak shaving
compared with proportional (surrogate for
difference in long-term exposure-related
mortality)6
15/35
-2%
6%
18%
4%
17%
0%
21%
40%
3%
22%
8%
21%
14/35
cells USE
cells use
4%
19%
cells USE
cells use
6%
cells USE
0%
cells use
cells USE
21%
40%
cells USE
3%
22%
cells use
8%
cells USE
cells use
21%
13/35
:d as ba
id as ba
4%
20%
:d as ba
;d as ba
16%
:d as ba
0%
id as ba
:d as ba
17%
40%
:d as ba
0%
id as ba
:d as ba
id as ba
8%
12/35
sis for ca
sis for ca
4%
21%
sis for ca
sis for ca
18%
sis for ca
0%
sis for ca
sis for ca
13%
sis for ca
0%
sis for ca
sis for ca
sis for ca
10%
13/30
Iculation
Iculation
-2%
8%
20%
Iculation
Iculation
5%
18%
Iculation
-5%
Iculation
Iculation
26%
47%
Iculation
4%
26%
Iculation
10%
Iculation
Iculation
22%
12/25
1%
-3%
7%
26%
29%
7%
19%
-21%
38%
60%
5%
30%
9%
35%
24%
F-60
-------
Table F-50. (cont'd) Maximum 3yr Monitor-Specific Average and Annual Composite Monitor Value Given Different Rollback Methods (with percent
difference in surrogate for long-term exposure-related mortality across rollback methods)
Risk
Assessment
Location 1
Salt Lake City,
UT
St. Louis, MO
Tacoma, WA
Rollback
Method
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Proportional
Hybrid
Peak Shaving
Design Value
Annual
11.6
16.5
10.2
24-Hr
55.0
39.0
43.0
Recent Air
Quality
(2007)
2007 CM
11.4
14.3
9.7
Maximum Monitor-Specific Avg. of 2005, 2006, 2007 Annual Avgs. (Max. M-S) and 2007
Annual Average at Composite Monitor (2007CM) (in ug/m3)
15/352
Max. M
S
7.7
10.8
14.9
15.0
16.5
8.4
8.3
2007
CM
7.5
9.5
12.9
13.5
14.1
8.0
7.8
14/35
Max. M
S
7.7
10.8
14.0
14.0
8.4
8.3
2007
CM
7.5
9.5
12.1
12.6
8.0
7.8
13/35
Max. M
S
7.7
10.8
13.0
13.0
8.4
8.3
2007
CM
7.5
9.5
11.3
11.7
8.0
7.8
12/35
Max. M
S
7.7
10.8
12.0
12.0
8.4
8.3
2007
CM
7.5
9.5
10.4
10.8
8.0
7.8
13/30
Max. M
S
6.7
10.8
12.8
13.0
14.1
7.4
7.1
2007
CM
6.6
8.6
11.1
11.7
12.3
7.0
6.7
12/25
Max. M
S
5.7
8.9
10.8
11.0
11.7
6.3
5.9
2007
CM
5.6
7.4
9.3
9.9
10.2
6.0
5.5
Percent difference between composite
monitor value with hybrid or peak shaving
compared with proportional (surrogate for
difference in long-term exposure-related
mortality)6
15/35
53%
8%
15%
-9%
14/35
cells USE
53%
cells USE
6%
cells USE
-9%
13/35
d as ba
53%
d as ba
7%
d as ba
-9%
12/35
sis for ca
53%
sis for ca
7%
sis for ca
-9%
13/30
Iculation
72%
Iculation
10%
18%
Iculation
-36%
12/25
111%
13%
19%
157%
1For some locations (e.g., Atlanta) more than one "version" (group of counties) was used in the risk assessment. In this table only the version that was used for mortality associated with
short-term exposure to PM2.5 (Zanobetti and Schwartz, 2009) is included.
2 The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
3 The hybrid rollback method was applied to only a subset of the risk assessment locations. The "" for a given location indicates that the hybrid rollback method was not applied to that
location.
4 The peak shaving method was applied to a location-standard combination only if the daily standard was controlling in that location. The "--" for a given location-standard combination
indicates that, for that set of annual and daily standards in that location, the annual standard was controlling and so the peak shaving method was not applied.
3 The proportional rollback and peak shaving methods were applied to Pittsburgh differently from the way they were applied in the other locations. See Sections 3.2.3.2 and 3.2.3.3 for
details.
6 Percent reduction in composite monitor value (CMV) with consideration for LML of 5.8 ug/m3. Percent reduction = (CMVpeakshaving or hybrjd - CfM^^^^^CfMpg^^^ag^^^-LML). Note
that greyed cells identify instances where two values differ by >25% across alternative rollback methods (for a given alternative standard level/study area combination).
F-61
-------
APPENDIX G: SUPPLEMEMNT TO THE NATIONAL-SCALE
ASSESSMENT OF LONG-TERM MORTALITY RELATED TO PM2 5
EXPOSURE
G-l
-------
1 Appendix G. National-Scale Assessment of Long-Term Mortality Related to PM2.s
2 Exposure (additional technical detail regarding inputs used in the analysis)
3
4
5 This technical appendix includes additional details regarding the inputs to the national-
6 scale current conditions health impact analysis. Below we present air quality modeling, exposure
7 and risk information.
8
9 Air Quality Modeled Inputs
10 The Community Model for Air Quality (CMAQ) model was used to estimate annual
11 PM2.5 concentrations for the year 2005 for the continental US. These data were then combined
12 with ambient monitored PM2.5 measurements to create "fused" spatial surfaces supplied to
13 BenMAP.
14 CMA Q Model Application and Evaluation
15 CMAQ is a non-proprietary computer model that simulates the formation and fate of
16 photochemical oxidants, including PM2.5 and ozone, for given input sets of meteorological
17 conditions and emissions. This analysis employed a version of CMAQ based on the latest
18 publicly released version (i.e. CMAQ version 4.72).
19 Model Domain and Grid Resolution
20 The CMAQ modeling analyses were performed for two domains covering the continental
21 United States, as shown in Figure G-l. These domains consist of a horizontal grid of 36 km
22 covering the entire continental US and a finer-scale 12-km grid covering the Eastern U.S. The
23 model extends vertically from the surface to 100 millibars (approximately 15 km) using a sigma-
24 pressure coordinate system. The 36-km grid was used to establish the incoming air quality
25 concentrations along the boundaries of the 12-km grids. Table G-l provides some basic
26 geographic information regarding the CMAQ domains. The 36-km and both 12-km CMAQ
27 modeling domains were modeled for the entire year of 2005. All 365 model days were used in
28 the annual average levels of PM2.s.
29
2CMAQ version 4.7 was released on December 1, 2008. It is available from the Community Modeling and
Analysis System (CMAS) at: http://www.cmascenter.org.
G-2
-------
Table G-l. Geographic Information for Modeling Domains
Map Projection
Grid Resolution
Coordinate Center
True Latitudes
Dimensions
Vertical Extent
CMAQ Modeling Configuration
National Grid
Eastern
U.S. Fine Grid
Lambert Conformal Projection
36km
12km
97 W, 40 N
33 and 45 N
148x112x24
279
x 240 x 24
24 Layers: Surface to 100 mb level
2
3
4
5
6
7
36km Domain
Specs:
._. _ . x,y: -1008000,-1620000
12km Domain coi.row: 279,240
Figure G-l. Map of the CMAQ Modeling Domain (Note, the black outer box denotes the
36-km national modeling domain; the red inner box is the 12-km Eastern U.S. fine
grid).
G-3
-------
1 CMA Q Model Inputs
2 Emissions:
3 The 2005 emissions inputs to CMAQ included five source sectors: a) Electric Generating
4 Units (EGUs); b) Other Stationary Sources (Point and Nonpoint); c) Onroad and Nonroad
5 Mobile Sources; d) Biogenic Emissions; and e) Fires. The fires portion of the inventory included
6 emissions from wildfires and prescribed burning computed as hour-specific point sources.
7 Electric Generating Units (EGUs)
8 Annual emissions estimates for EGUs for all National Emissions Inventory (NEI) air
9 pollutants for 2005 were developed using data reported to the USEPA's Clean Air Marketing
10 Division's (CAMD) Acid Rain database. The Acid Rain database contains hourly emissions for
11 SO2 and NOx emissions plus hourly heat input amounts. These three values are reported to the
12 database by the largest electric generating facilities, usually based upon Continuous Emissions
13 Monitors (CEMs). For all pollutants except the directly monitored SO2 and NOx, the ratio of the
14 Acid Rain heat input for 2005 to the Acid Rain heat input for 2002 was used as the adjusting
15 ratio to estimate the 2005 emissions.
16 Other Stationary Sources (Point and Nonpoint)
17 Emission estimates for other stationary sources including both point and nonpoint
18 stationary sources were held constant at the level in Version 3 of the 2002 NEI. The only
19 exception to this was that some information on plants that closed after 2002 was incorporated
20 into the emissions modeled. Emissions for plants that closed were set to zero. U.S. EPA, 2008c
21 provides complete documentation on the development of the 2002 NEI.
22 Onroad and Nonroad Mobile Sources
23 Emission estimates for all pollutants were developed using EPA's National Mobile
24 Inventory Model (NMIM), which uses MOBILE6 to calculate onroad emission factors. A full
25 VMT database at the county, roadway type, and vehicle type level of detail was developed from
26 Federal Highway Administration (FHWA) information. However, state and local agencies had
27 the opportunity to provide model inputs (vehicle populations, fuel characteristics, VMT, etc) for
28 2002 and 2005. If the state or local area submitted 2005 VMT estimates, these data were used.
29 However, if the state or local area only provided 2002 VMT estimates that were incorporated in
30 the 2002 NEI, the 2002 NEI VMT data were grown to 2005 using growth factors developed from
G-4
-------
1 the FHWA data, and these grown VMT data replaced the baseline FHWA-based VMT data.
2 Otherwise, the FHWA-based VMT data were used.
3 Emission estimates for NONROAD model engines were developed using EPA's National
4 Mobile Inventory Model (NMIM), which incorporates NONROAD2005. Where states provided
5 alternate nonroad inputs, these data replaced EPA default inputs, as described above. For more
6 information on how NMIM is run, refer to the 2005 NEI documentation posted at
7 ftp://ftp.epa.gov/EmisInventory/2005_nei/mobile/2005_mobile_nei_version_2_report.pdf
8 Fires
9 Fires in the 2005 emissions inventory were modeled with the same methodology as used
10 for the 2002 NEI (U.S. EPA, 2008). However, as described in Raffuse et al., 2008, the wildland
11 fire emission inventories for 2005 were produced using the BlueSky framework for the
12 conterminous United States, which used the Satellite Mapping Automatic Reanalysis Tool for
13 Fire Incident Reconciliation (SMARTFIRE) as the fire information source. SMARTFIRE is an
14 algorithm and database system designed to reconcile these disparate fire information sources to
15 produce daily fire location and size information (Sullivan et al., 2008).
16 Biogenic Emissions
17 Biogenic emissions were computed for CMAQ based on 2005 meteorology data using the
18 BEIS3.13 model (Schwede, et. al, 2005) from the Sparse Matrix Operator Kernel Emissions
19 (SMOKE). The BEIS3.13 model creates gridded, hourly, model-species emissions from
20 vegetation and soils. It estimates CO, VOC, and NOX emissions for the U.S., Mexico, and
21 Canada. The inputs to BEIS include:
22 temperature data at 10 meters which were obtained from the CMAQ
23 meteorological input files, and
24 land-use data from the Biogenic Emissions Landuse Database, version 3
25 (BELD3), which provides data on the 230 vegetation classes at 1 km resolution over most
26 of North America.
27 Meteorological Input Data:
28 The gridded meteorological input data for the entire year of 2005 were derived from
29 simulations of the Pennsylvania State University / National Center for Atmospheric Research
30 Mesoscale Model. This model, commonly referred to as MM5, is a limited-area, nonhydrostatic,
31 terrain-following system that solves for the full set of physical and thermodynamic equations
G-5
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
which govern atmospheric motions (Grell et al., 1994). Meteorological model input fields were
prepared separately for both of the domains shown in Figure G-l using MM5 version 3.7.4. The
MM5 simulations were run on the same map projection as CMAQ.
Both meteorological model runs were configured similarly. The selections for key MM5
physics options are shown below:
Pleim-Xiu PEL and land surface schemes
Kain-Fritsh 2 cumulus parameterization
Reisner 2 mixed phase moisture scheme
RRTM longwave radiation scheme
Dudhia shortwave radiation scheme
Three dimensional analysis nudging for temperature and moisture was applied above the
boundary layer only. Analysis nudging for the wind field was applied above and below the
boundary layer. The 36 km domain nudging weighting factors were 3.0 x 104 for wind fields and
temperatures and 1.0 x 105 for moisture fields. The 12 km domain nudging weighting factors
were 1.0 x 104 for wind fields and temperatures and 1.0 x 105 for moisture fields.
All model runs were conducted in 5.5 day segments with 12 hours of overlap for spin-up
purposes. Both domains contained 34 vertical layers with an approximately 38m deep surface
layer and a 100 millibar top. The MM5 and CMAQ vertical structures are shown in Table G-2
and do not vary by horizontal grid resolution.
Table G-2. Vertical Layer Structure for MM5 and CMAQ (heights are layer top).
CMAQ
Layers
0
1
2
3
4
5
6
7
8
MM5
Layers
0
1
2
O
4
5
6
7
8
9
10
11
Sigma P
1
0.995
0.99
0.985
0.98
0.97
0.96
0.95
0.94
0.93
0.92
0.91
Approximate
Height (m)
0
38
77
115
154
232
310
389
469
550
631
712
Approximate
Pressure (mb)
1000
995
991
987
982
973
964
955
946
937
928
919
G-6
-------
CMAQ
Layers
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
MM5
Layers
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Sigma P
0.9
0.88
0.86
0.84
0.82
0.8
0.77
0.74
0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Approximate
Height (m)
794
961
1,130
1,303
1,478
1,657
1,930
2,212
2,600
3,108
3,644
4,212
4,816
5,461
6,153
6,903
7,720
8,621
9,625
10,764
12,085
13,670
15,674
Approximate
Pressure (mb)
910
892
874
856
838
820
793
766
730
685
640
595
550
505
460
415
370
325
280
235
190
145
100
2 The meteorological outputs from the MM5 sets were processed to create model-ready
3 inputs for CMAQ using the Meteorology-Chemistry Interface Processor (MCIP), version 3.4, to
4 derive the specific inputs to CMAQ.
5 Before initiating the air quality simulations, it was important to identify the biases and
6 errors associated with the meteorological modeling inputs. The 2005 MM5 model performance
7 evaluations used an approach which included a combination of qualitative and quantitative
8 analyses to assess the adequacy of the MM5 simulated fields. The qualitative aspects involved
9 comparisons of the model-estimated synoptic patterns against observed patterns from historical
10 weather chart archives. Additionally, the evaluations compared spatial patterns of monthly
11 average rainfall and monthly maximum planetary boundary layer (PEL) heights. Qualitatively,
12 the model fields closely matched the observed synoptic patterns, which is not unexpected given
13 the use of nudging. The operational evaluation included statistical comparisons of
14 model/observed pairs (e.g., mean normalized bias, mean normalized error, index of agreement,
15 root mean square errors, etc.) for multiple meteorological parameters, including temperature,
16 humidity, shortwave downward radiation, wind speed, and wind direction (Baker and Dolwick,
G-7
-------
1 2009a, Baker and Dolwick, 2009b). It was ultimately determined that the bias and error values
2 associated with the 2005 meteorological data were generally within the range of past
3 meteorological modeling results that have been used for air quality applications.
4 Initial and Boundary Conditions:
5 The lateral boundary and initial species concentrations are provided by a three-
6 dimensional global atmospheric chemistry model, the GEOS-CHEM model (Yantosca, 2004).
7 The global GEOS-CHEM model simulates atmospheric chemical and physical processes driven
8 by assimilated meteorological observations from the NASA's Goddard Earth Observing System
9 (GEOS). This model was run for 2002 with a grid resolution of 2.0 degrees x 2.5 degrees
10 (latitude-longitude) and 24 vertical layers. The 2005 CMAQ 36km simulation used non-year
11 specific GEOS-CHEM data, which was created by taking the median value for each month in
12 each individual grid cell of the 2002 GEOS-CHEM data described above. The predictions were
13 used to provide one-way dynamic boundary conditions and an initial concentration field for the
14 CMAQ simulations. More information is available about the GEOS-CHEM model and other
15 applications using this tool at: http://www-as.harvard.edu/chemistry/trop/geos.
16 CMA Q Model Performance Evaluation
17 An operational model performance evaluation for PM2.5 and its related speciated
18 components was conducted for 2005 using state/local monitoring sites data in order to estimate
19 the ability of the CMAQ modeling system to replicate the concentrations for the 12-km Eastern
20 domain and 36-km domain in the west. The principal evaluation statistics used to evaluate
21 CMAQ performance included two bias metrics, normalized mean bias and fractional bias; and
22 two error metrics, normalized mean error and fractional error. For the 12-km Eastern domain,
23 performance evaluation statistics were computed for the entire domain as well as its subregions.
24 For the 36-km domain, evaluation focuses on the parts of the US not covered by the 12-km
25 Eastern domain by computing performance evaluation statistics for the states included in the
26 Western Regional Air Partnership (WRAP).
27 The PM2.5 evaluation focuses on PM2.5 total mass and its components, including sulfate
28 (SO4), nitrate (NO3), total nitrate (TNO3 = NO3 + HNO3), ammonium (NH4), elemental carbon
29 (EC), and organic carbon (OC). PM2.5 ambient measurements for 2005 were obtained from the
30 following networks for model evaluation: Speciation Trends Network (STN), Interagency
31 Monitoring of PROtected Visual Environments (IMPROVE), and Clean Air Status and Trends
32 Network (CASTNET). For PM2.5 species that are measured by more than one network, we
33 calculated separate sets of statistics for each network. Table G-3 provides annual model
34 performance statistics for PM2 5 and its component species. Based on the bias and error values
G-8
-------
1
2
3
4
5
6
7
associated with the 2005 CMAQ-modeled PM2.5 concentration data, it was determined that the
annual average PM2.5 data were generally within the range of past modeling results used for air
quality applications and are applicable to be used for this national-scale current conditions
analysis.
Table G-3. CMAQ modeled performance evaluation statistics for PM2.5 for 2005.
CMAQ 2005 Annual
PM2.5 Total
Mass
Sulfate
STN
IMPROVE
STN
IMPROVE
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
No. of
Obs.
11622
2795
2318
2960
2523
3082
10534
2464
668
1963
2768
10,122
13317
3247
2495
3499
2944
3450
10164
2393
622
1990
2640
9693
NMB (%)
-2.2
4.2
4.3
-13.0
-2.2
-35.1
-9.4
5.3
-4.6
-20.8
-10.5
-21.0
-17.1
-13.7
-10.9
-19.2
-25.7
-21.9
-21.8
-14.6
-19.0
-25.2
-27.9
-5.2
NME (%)
39.1
41.3
35.2
37.5
43.1
50.7
44.3
48.6
38.2
42.8
42.8
56.0
34.0
32.4
33.9
32.8
38.7
46.4
36.4
35.5
34.5
35.9
38.0
45.2
FB (%)
-4.7
3.4
5.0
-15.9
-8.4
-40.3
-13.8
2.3
-7.3
-25.9
-12.9
-24.4
-13.5
-9.4
-4.4
-16.8
-23.1
-15.0
-13.2
-6.6
-9.4
-22.3
-22.0
9.6
FE (%)
40.3
39.5
34.1
41.1
45.6
57.4
48.6
46.2
40.8
51.3
47.7
57.6
37.0
34.3
34.9
35.8
43.5
46.5
41.1
38.6
36.7
41.1
42.4
47.6
G-9
-------
CMAQ 2005 Annual
Nitrate
Total Nitrate
(NO3+HNO3)
Ammonium
CASTNet
STN
IMPROVE
CASTNet
STN
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
No. of
Obs.
3170
786
615
1099
300
1112
12186
3248
2495
3499
1812
15,533
10157
2388
622
1990
2640
17,452
3170
786
615
1099
300
4065
13317
3247
2495
3499
2944
16,680
NMB (%)
-16.5
-11.7
-13.6
-18.4
-29.4
-12.6
20.1
28.7
20.2
23.5
8.1
15.2
30.1
67.0
14.0
37.4
17.3
33.1
24.6
36.5
23.3
23.6
10.6
37.7
1.8
7.1
7.1
-2.1
-7.6
8.1
NME (%)
22.9
20.5
21.4
22.9
32.5
34.5
67.8
70.2
61.0
84.0
60.2
79.3
85.2
108.9
67.9
104.6
70.8
99.1
39.7
43.0
36.5
42.2
35.5
51.9
41.9
42.9
40.5
40.5
44.0
47.2
FB (%)
-15.6
-9.8
-11.2
-19.6
-30.3
-3.2
-10.1
-3.7
9.2
-25.0
-5.9
-15.6
-32.5
0.5
-24.1
-46.2
-19.3
-41.9
17.8
30.3
23.9
12.8
5.0
24.2
8.3
18.9
16.4
2.9
-4.0
12.8
FE (%)
26.0
22.6
22.2
25.7
36.1
36.7
76.3
74.1
63.0
87.2
72.4
85.9
99.1
93.4
88.9
105.9
89.6
109.9
38.0
40.6
33.2
40.5
35.0
45.1
45.6
45.7
41.4
43.3
51.4
48.9
G-10
-------
CMAQ 2005 Annual
Elemental
Carbon
Organic Carbon
CASTNet
STN
IMPROVE
STN
IMPROVE
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
No. of
Obs.
3170
786
615
1099
300
4065
13460
3230
2502
3495
3107
16,700
10244
2341
696
1995
2626
17,289
12118
3083
2385
3442
2164
15,397
10210
2336
696
1993
2622
17,295
NMB (%)
2.2
9.2
10.9
-9.2
1.5
12.8
19.7
20.8
7.3
10.2
47.6
2.6
-29.0
-17.8
-26.7
-45.6
-22.9
-16.6
-36.5
-29.1
-42.5
-42.6
-30.6
-41.2
-34.7
-21.0
-41.3
-40.4
-34.1
-22.5
NME (%)
35.4
38.1
35.3
33.3
36.9
39.6
63.5
61.9
46.1
60.2
88.2
56.7
49.7
49.2
41.9
53.3
49.2
53.4
53.6
53.1
52.6
53.5
57.7
56.1
53.7
52.2
47.6
53.7
52.8
57.5
FB (%)
3.1
13.3
14.8
-9.7
3.0
13.0
11.9
14.6
10.8
3.0
23.0
2.6
-39.1
-25.6
-39.6
-58.5
-31.3
-23.4
-40.6
-27.6
-41.7
-55.6
-39.6
-45.7
-53.0
-29.2
-55.7
-64.0
-52.7
-40.8
FE (%)
36.5
36.6
33.7
37.6
40.2
40.1
53.9
52.0
44.9
50.6
64.9
55.0
61.3
57.7
55.7
69.8
56.8
60.2
66.5
64.2
65.3
70.2
66.5
69.2
70.0
58.4
63.6
74.2
68.1
67.6
G-ll
-------
1
2
3 "Fused" Spatial Surfaces
4 Spatial surfaces of the 2005 data were created by fusing CMAQ-modeled annual average
5 PM2.5 concentrations with total PM2.5 data from STN, IMPROVE, and CASTNET monitoring
6 sites for the two domains shown in Figure 1. We used the EPA's Model Attainment Test
7 Software (MATS) (Abt, 2009) which employees the Voronoi Neighbor Averaging (VNA)
8 interpolation technique (Abt, 2008). This technique identifies the set of monitors that are nearest
9 to the center of each grid cell, and then takes an inverse distance squared weighted average of the
10 monitor concentrations. The "fused" spatial fields are calculated by adjusting the interpolated
11 ambient data (in each grid cell) up or down by a multiplicative factor calculated as the ratio of
12 the modeled concentration at the grid cell divided by the modeled concentration at the nearest
13 neighbor monitor locations (weighted by distance).
14 To create the spatial surfaces for use in BenMAP, the 2005 CMAQ-modeled annual
15 average PM2 5 concentrations were "fused" with 2005 total PM2 5 ambient monitoring data from
16 STN, IMPROVE, and CASTNET sites. This was done for both the 36km national domain and
17 the 12km eastern US domain. The spatial surface of annual average PM2.5 air quality
18 concentrations produced by this technique is shown in Figure G-2 for the continental U.S. Where
19 available, the 12km spatial surface was used to supply BenMAP with annual average PM2 5
20 concentrations. In the western part of the U.S., annual average PM2.5 concentrations were
21 supplied from the 36km domain.
22
G-12
-------
Figure G-2: 2005 Predicted Annual Mean PM2.s Levels
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
2005 Fused Surface Baseline Concentrations (ug/m3)
^B I-03 to 4.2
^H 4.3 to 6.5
6.6 to 9.34
9.35 to 12.30
^B l2-31 to 20.57
^H 20.58 to 59.42
Advantages and Limitations
As compared to using monitored data alone, an advantage of using the CMAQ model
output for comparing with health outcomes is that it has the potential to provide more complete
spatial and temporal coverage. In addition, "fusing" the CMAQ data with ambient monitoring
data allows for an improvement over non-fused fields (Timin et al., 2009). Doing so allows for a
combination of the advantages of both sets of data: better spatial coverage and more accurate air
quality estimates. Of course, the more accurate the model estimates of PM2.5, the better the
performance of the "fused" spatial fields. Therefore, it is important to use model outputs that
have adequate PM2.5 performance. As discussed above, we believe that the 2005 CMAQ-
modeled PM2.5 concentration data showed adequate model performance to be used for this
national-scale current conditions analysis.
As with any model estimate of air quality, there are limitations. For example, the
emissions and meteorological data used in CMAQ can each have large uncertainties, in particular
G-13
-------
1 for unusual emission or meteorological events. There are also uncertainties associated with the
2 chemical transformation and fate process algorithms used in air quality models. For these
3 reasons, CMAQ predicts best on longer time scale bases (e.g., synoptic, monthly, and annual
4 scales). These limitations have led us to use modeled air quality estimates in this analysis that
5 are "fused" with measured ambient data and averaged over an annual scale.
6 Air Quality Estimates
7 Figures G-3 through G-6 below illustrate the spatial distribution of air quality impacts.
8 Figure 1 illustrates the modeled 2005 PM2.5 air quality levels across the U.S. Figures 2 and 3
9 display the PM2.5 air quality levels after being adjusted so that the maximum level is no higher
10 than the LML reported in the Krewski et al. (2009) and Laden et al. (2006) studies. Figure G-4
11 displays the PRB by region of the county.
G-14
-------
Figure G-3: 2005 Predicted Annual Mean PM2.s Levels
2005 Fused Surface Baseline Concentrations (ug/m3)
^B I-°3 to 4.2
^H 4.3 co 6.5
6.6 to 9.34
9.35 to 12.30
^H !2.3I to 20.57
20.58 to 59.42
G-15
-------
Figure G-4: 2005 Predicted Annual Mean PM2.s Levels Adjusted for LML of the
Krewski et al. (2009) study
2005 Adjusted Fused Surface Concentrations (ug/m3)
^H 1.03 to 2.74
^B 2.75 to 3.53
3.54 to 4.23
4.24 to 4.85
^B 4-86 to 5-46
5.47 to 5.80
G-16
-------
Figure G-5: 2005 Predicted Annual Mean PM2.s Levels Adjusted for LML of the
Laden et al. (2006) study
2005 Adjusted Fused Surface Concentrations (ug/m3)
^B I-03 to 3.08
^H 3.09 to 4,28
4.29 to 5.58
S.59to 7.16
^B 7.17 to 8.97
^H 8.98 to 10.00
G-17
-------
Figure G-6: PRB by Geographic Area in the U.S.
Policy Relevant Background Level (ug/m3)
0.74
0.84
0.86
1.01
1.72
Figure G-7 displays the distribution of grid cells at different baseline PM2.5 air quality
levels. Figures G-8 through G-10 displays the distribution of grid cells according to the
incremental change in PM2.5 air quality for each of three scenarios: current conditions to 10
|ig/m3, current conditions to 5.8 |ig/m3 and current conditions to PRB.
G-18
-------
Figure G-7: The Number of Grid Cells at Each Level of PMi.s Concentration in
2005 Current Conditions Air Quality Modeling Run
i 2.000
1
,t
1
-
1 ' ,
01234
1 'f
\
,
I 'f
I
L
'
i
i f
f
i I
1
i
i
i
I
t
V,
1 «
J 1
1 r:
1 Ipj
* \
i|
0 11 1
^
r
1 ;
2 1
1
i
[
5 1
s
f
4 1
t
i r
ii
* i »*s 11 (»..., _,..,... ,
5 16 17 IS 19 20 21 22 23 24 ** > ^ 2 z9 30 31 32
Basefine Ambient PfVl; 5 J
Maximum value = 31.3 |igv'm3
Minim'jm value =1,5 ^ig/rn3
G-19
-------
Figure G-8: The Number of CMAQ Grid Cells Experiencing an Incremental
Change in Annual Mean PM2.5 (ug/m3) (Current Conditions - 10
ug/m3)
o 2 500 -<*- -!
a i
JL 2 3 4 5 6 7 3 9 iO 11 12 13 14 15 J.n i~ i ii ^n
Change in Ambient PM; 5 (|ig/m5)
Maximum change =21,3 ftg/m3
Number of cellswith no change: 26..000
G-20
-------
Figure G-9: The Number of CMAQ Grid Cells Experiencing an Incremental
Change in Annual Mean PMi.s (jig/m3) (Current Conditions - 5.8
ug/m3)
3 500 p"
9 10 ii 12 13 14 lc J.i 1~ 1 1 I"
Change in Ambient PM; 5 (|ig/m5)
2* 23 24 25 26
25 30 31 32
Maximum change = 31.3 ftg/'m3
Number of cells with no change: 10,000
G-21
-------
Figure G-10: The Number of CMAQ Grid Cells Experiencing an Incremental
Change in Annual Mean PMi.s (ug/m3) (Current Conditions - Policy
Relevant Background)
1234567
y 10 11 12 13 14 15 L6 L7 13 19 20 21 22
Change in Ambient PM; 5 (ng/m5)
Maximum change = 31 ng/m-^
Number of cellsvvith no change: 0
Figure G-l 1 displays the cumulative distribution of grid cells at each baseline
concentration. Figures G-12 through G-14 display the cumulative distribution of grid cells
experiencing an incremental air quality change.
G-22
-------
Figure G-ll: Cumulative Distribution of Baseline PM2.s Concentrations (ug/m3)
1 2 3 4 5 '5 7 8 9 10 11 12 13 14 15 16 1? IS 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
Baseline Ambient PM3; Leyel (Mg/m5)
G-23
-------
Figure G-12: Cumulative Distribution of PM2.s (ug/m ) Changes (Baseline - 10
ug/m3)
10 11 12 13 14 15 16 17 IS 19 20 21 22 23 24 25 26 27 28 29 30 31 32
Change in Ambient {|ig/m3(
"10 |igv'm3 represents the lowest measured level in the 6-dties cohort
G-24
-------
Figure G-13: Cumulative Distribution of PM2.s (ug/m ) Changes (Baseline - 5.8
ug/m3)
i 2 3 4 5 6
9 10 11 12 13 14 15 16 17 I/ l->
Change in Ambient PMj 5 (
'5.8 ^ig/m3 represents the lowest measured levei in the ACS cohort
G-25
-------
Figure G-14: Cumulative Distribution of PM2.s (ug/m3) (Baseline - Policy Relevant
Background)
1 2 3 4 c P ' 1" 11 12 1 14 IE In i.-1 i: 1' 2n 21 2^ J" -4 _r .1 r ~ 2' -in -1
Change in Ambient PM; 5 <|ig/m5(
Exposure Estimates
Below we provide additional details regarding the estimated exposure changes occurring
as a result of each of the air quality changes assumed in each of the three health impact
assessments: current conditions incremental to 10 |ig/m3, 5.8 |ig/m3 and PRB. Table G-4
summarizes the population-weighted air quality change occurring among populations 30-99 (the
age range considered in the ACS cohort) for each scenario.
Population-weighted air quality change is the average per-person change in PM2 5. It is
estimated by calculating the summation of the population in each grid cell multiplied against the
change in annual mean PM2.5 concentration in that grid cell and then dividing by the total
population.
G-26
-------
Table G-4. Estimated Change in Annual Mean Population-Weighted PM2.s by
Model Scenario
Model scenario
Population-weighted air quality change or
baseline
Current conditions to 10 |ig/m
Current conditions to 5.8 |ig/m3
Current conditions to PRB
Current conditions
2.6 ng/m
6.3 |ig/m3
11 |ig/m3
12 |ig/m3
Health Impact Estimates
Figure G-15 through G-17 illustrate the distribution of total mortality attributable to
PM2.5 exposure for each of three scenarios: current conditions to 10 |ig/m3, 5.8 |ig/m3 and PRB.
G-27
-------
Figure G-15: The Percentage of Total Mortality Attributable to PM2.5 Exposure:
Baseline -10 ug/m3
a
I 4000
l"e J- 5Jc 4% 5% c « T\ 3^ -" 1" , 11\ 1. o I""-*
Percentage o*1 Mortality Attributable to PM2 5 Exposure
Attributable mortality calculated usingKrevvski et al, (2009; risk estimate based on ''99-'00 follow-up period,
Number of grid cells in which the percentage of attributable mortality is equal to 0: 23..000
G-28
-------
Figure G-16: The Percentage of Total Mortality Attributable to PM2.5 Exposure:
Baseline - 5.8 ug/m3
a
I 4.000
1% 2% 3°,, 4°o 5", 6% 7°o 3% 9% 10% 11°,, 12?, 13%
Percentage of Mortality Attributable to PM: , Exposure
'Attributable mortality calculated using Kreivski et al. (2009! risk estimate based on '99-'00 follov-v-up period.
Number of ^rid cells in which the percentage of attributable mortality is equal to 0: 11,000
G-29
-------
Figure G-17: The Percentage of Total Mortality Attributable to PM2.5 Exposure:
Baseline - Policy Relevant Background
I 3 000
Percentage of Mortality Attributable to PM; 5 Exposure
Figures G-18 through G-20 illustrate the cumulative distribution of total mortality
attributable to PM2.5 exposure for each of three scenarios: current conditions to 10 |ig/m3, 5.8
|ig/m3andPRB.
G-30
-------
Figure G-18: The Cumulative Distribution of the Percentage of Total Mortality
Attributable to PM2.s Exposure: Baseline - 10 ug/m3
Percent olf Total Mortality Attributable to PM? =. Exposure
"Attributable mortality calculated using Krewski et al, (2009) risk estimate based on '99-'00 follow-up period.
G-31
-------
Figure G-19: The Cumulative Distribution of the Percentage of Total Mortality
Attributable to PM2.5 Exposure: Baseline - 5.8 ug/m3
1, 5V ', -'., '* f ilf 11' 12 l=i° 1-1 1
Percentage of Total Mortality Attributable to PM; 5 Exposure
ll I"'. 1 , 1-1
'Attributable mortality calculated using Krewski et al. (2009) risk estimate based on '99-'00 follow-up period.
G-32
-------
Figure G-20: The Cumulative Distribution of the Percentage of Total Mortality
Attributable to PMi.s Exposure: Baseline - Policy Relevant Background
n% -% \ ' * 10% 11% 12% 13% 14% 15% 16°,
Percentage of Mortality Attributable to PM2 5
"Attributable mortality calculated usingKre'vVski et al. (2009) risk estimate based on '99-'GO follow-up period.
G-33
-------
APPENDIX H: CONSIDERATION OF RISK ASSOCIATED WITH
EXPOSURE TO THORACIC COARESE PM (PMi0.2 5)
H-l
-------
H.1 OVERVIEW
This appendix discusses the issue of assessing public health risk associated with
exposure to thoracic coarse PM (PMio-2.s). As mentioned in Section 2.6, due to
limitations in available monitoring data characterizing ambient levels of PMio-2.5 in
prospective urban study areas, together with limitations in the epidemiological study data
available for deriving C-R functions for this PM size fraction, EPA staff has concluded
that uncertainties in characterizing risk for PMio-2.5 are potentially significant enough at
this time to limit the utility of those estimates in informing the review of the PM coarse
standard level. Therefore, we have not conducted a PMio-2.5 risk assessment for this
review; instead, we have included a summary of risk estimates for PMio-2.5 generated as
part of the last PM NAAQS review completed in 2005.3
As part of our summarizing PMio-2.5 risk estimates from the last review below in
section H.2, we have included a discussion of the limitations and uncertainties associated
with those risk estimates which resulted in the decision by EPA not to use those risk
estimates in recommending specific standard levels (USEPA, 2006 - Final Rule FR
Notice, p. 61178). This discussion provides the basis for a more detailed discussion (in
Section H.3) of our rationale for not conducting a PMio-2.5 risk assessment as part of the
current review. Specifically, in Section H-3, we consider each of the limitations in the
PMio-2.5 risk assessment from the last review and assess whether data available since the
last review, including more recent ambient monitoring data and epidemiological study
data, address these limitation. Our conclusion is that additional information on PMio-2.5
that has become available since the last review does not substantially reduce overall
uncertainty associated with modeling risk for this PM size fraction, and consequently, we
conclude that conducting a PMio-2.5 risk assessment is not supported at this time.
H.2 SUMMARY OF PMi0-2.5 RISK ESTIMATES GENERATED FOR
THE PREVIOUS REVIEW
This section provides a brief overview of the approach used in completing the
PMio-2.5 risk assessment for the previous review and provides a summary of key
observations resulting from that assessment. Additional details on the risk estimates can
3 We note that inclusion in this appendix of a summary of the PM10_2 5 risk assessment completed
for the previous review should not be construed as implying that overall conclusions regarding limitations
and uncertainties in that risk assessment have changed. Conclusions reached in the last review, that PM10.
2.5 risk estimates should not be used in recommending specific standard levels, still holds. Rather, we have
included a summary of the PM10.2.5 risk assessment completed for the last review in the interest of
completeness.
H-2
-------
be found in the risk assessment report completed for the previous analysis (USEPA,
2005).
The PMio-2.5 risk assessment completed for the previous review is similar in
design to the PM2.5 risk assessment, although the scope is significantly more limited,
reflecting the more limited body of epidemiological evidence and air quality information
available for PMio-2.5- The PMio-2.5 risk assessment assessed risk for populations in three
urban study areas (Detroit, Seattle and St. Louis), with a set of short-term exposure-
related morbidity health endpoints being modeled, including: respiratory hospital
admissions (for Detroit and Seattle), cardiovascular hospital admissions (for Detroit) and
respiratory symptoms (for St. Louis). Selection of these three urban study areas reflected
consideration of the locations included in epidemiological studies providing C-R
functions, as well as availability of co-located PMio and PM2.5 monitoring data used in
deriving estimates of ambient PMio-2.5 levels for urban study areas. EPA staff noted in
the last review that the locations used in the PMio-2.5 risk assessment were not
representative of urban locations in the U.S. that experience the most significant elevated
24-hour PMio-2.5 ambient concentrations. Thus, observations regarding risk reductions
associated with alternative standards in these three urban areas may not be fully relevant
to the areas expected to have the greatest health risks associated with peak daily ambient
PMio-2.5 concentrations. This is a key limitation impacting the PMio-2.5 risk assessment
and remains a primary concern in conducting a PMio-2.5 risk assessment (see below).
In summarizing PMio-2.5 risk estimates from the last review, we focus here on risk
estimates generated for the recent conditions air quality scenario.4 In the risk assessment,
risk estimates are provided for Detroit for several categories of cardiovascular and
respiratory-related hospital admissions and show point estimates ranging from about 2 to
7% of cause-specific admissions being associated with "as is" short-term exposures to
PMio-2.5. The point estimate for asthma hospital admissions associated with short-term
PMio-2.5 exposures for Seattle, an area with lower PMio-2.5 ambient concentrations than
either Detroit or St. Louis, is about 1%. Point estimates for lower respiratory symptoms
and cough in St. Louis are about 12 and 15%, respectively. These estimates use
estimated policy-relevant background as the cutpoint.
The specific set of uncertainties that resulted in EPA staff concluding that the
PMio-2.5 risk estimates should not be used in recommending specific standard levels
include, but are not limited to, the following (see USEPA, 2005, PM SP, Section 5.4.4.2):
4 We have chosen not to discuss risk estimates generated for alternate standard levels here since
uncertainty in those estimates would be even higher than for recent conditions estimates.
H-3
-------
Concerns that the current PMio-2.5 levels measured at ambient monitoring sites
during the study period for the risk assessment may be quite different from the
levels used to characterize exposure in the original epidemiologic studies based
on monitoring sites in different location, thus possibly over- or underestimating
population risk levels;
Greater uncertainty about the reasonableness of the use of proportional rollback to
simulate attainment of alternative PMio-2.5 daily standards in any urban area due
to the limited availability of PMio-2.5 air quality data over time (this uncertainty
only being relevant to risk estimates generated for the alternative standard levels);
Concerns that the locations used in the risk assessment are not representative of
urban areas in the U.S. that experience the most significant 24-hour peak PMio-2.5
concentrations, and thus, observations about relative risk reductions associated
with alternative standards may not be relevant to the areas expected to have the
greatest health risks associated with elevated ambient PMio-2.5 levels; and
Concerns about the much smaller health effects database that supplies the C-R
relationships used in the risk assessment, compared to that available for PM2.5,
which limits our ability to evaluate the robustness of the risk estimates for the
same health endpoints across different locations.
H.3 RATIONALE FOR THE DECISION NOT TO CONDUCT A PMio-2.5
RISK ASSESSMENT AS PART OF THE CURRENT REVIEW
The decision not to conduct a PMio-2.5 risk assessment for the current
review is based on consideration of key uncertainties identified in the last review and an
assessment as to whether newly available information has significantly reduced those
uncertainties. Each of the sources of uncertainty is addressed below:
Concerns that monitoring data that would be used in a PM'10-2.5 risk assessment
(i. e., for the period 2005-200 7) would not match ambient monitoring data used in
the underlying epidemiological studies providing C-R functions: While this is
always a concern in conducting PM-related risk assessments, due to the potential
for greater spatial heterogeneity in PMio-2.5 ambient levels (see final PM ISA,
Sections 2.1.1.2 and 2.2.1, USEPA 2009b), the potential for discrepancies
between the monitoring networks used in epidemiological studies providing C-R
functions and the monitoring network used in the risk assessment introducing
uncertainty is increased relative to PM2.5. That is, the potential for greater spatial
variation in PMio-2.5 levels means that the particular mix of collocated monitors
used in generating an exposure surrogate in epidemiological studies needs to be
more closely matched to the monitoring network used in conducting the risk
assessment if significant uncertainty is to be avoided.
Uncertainty in the prediction of ambient levels under current and alternative
standard levels: This remains a significant factor introducing uncertainty into
PMio-2.5 risk estimates generated for alternative standard levels, and continues to
weigh against the use of these risk estimates in identifying alternative standard
H-4
-------
levels for consideration in this review. Not only is the monitoring network (i.e.,
co-located PMi0 and PM2 5 monitors) available for characterizing PMi0-2.5 levels
in candidate urban study areas limited (see above), given the potential for greater
spatial heterogeneity in PMi0-2.5 levels (relative to PM2.5 levels), generating
representative estimates of ambient air profiles for PMi0-2.5 under alternative
standard levels is substantially more challenging than for PM2 5. In particular, the
use of proportional rollback as a means for conducting rollbacks would be subject
to significant uncertainty given the greater potential for local-scale gradients in
PMio-2.5 levels and the linkage of PMi0-2.5 to local-scale sources.
Concerns that locations used in the risk assessment may not be representative of
areas experiencing the most significant 24-hour peak PMw-2.5 concentrations
(andconsequently, may not capture locations with the highest risk): This concern
still holds since the monitoring network available for characterizing PMi0-2.5
levels in urban areas has not been significantly expanded (final PM ISA, Section
3.5.1.2,). Specifically, the final PM ISA states that: "Given the limited number
of co-located low-volume FRM PMio and FRM PM2.5 monitors, only a very
limited investigation into the intra-urban spatial variability of PMi0-2.5 was
possible using AQS data. Of the 15 cities under investigation, only six (Atlanta,
Boston, Chicago, Denver, New York and Phoenix) contained data sufficient for
calculating PMio-2.5 according to the data completeness and monitor specification
requirements discussed earlier." As noted in the previous risk assessment, these
urban study areas may not capture locations with the highest peak levels of PMio-
2.5 based on consideration of general patterns in PMio and PM2.5 levels.
Concerns about the much smaller health effects database that supplies the C-R
relationships (relative to PM2.5): While a number of epidemiological studies
have been published since completion of the previous PM NAAQS review,
including several large multi-city studies that inform consideration of the effects
of short-term exposure to PMio-2.5, limitations in the available studies still result in
uncertainty in specifying C-R functions for PMio-2.5. For example, while Peng et
al. (2008) and Zanobetti and Schwartz (2009) both provide effect estimates for
short-term exposure-related mortality (with consideration of copollutant
confounding by PM2.s), both have specific limitations that impact their use in risk
assessment. For example, Zanobetti and Schwartz (2009) derives estimates of
PMio-2.5 by subtracting county-level PMio and PM2.s levels, rather than using
collocated monitors. Given the significant spatial gradients associated with PMio-
2.5 relative to PM2.s, the use of this approach for assessing exposure introduces
significant uncertainty (i.e., exposure measurement error). In the case of Peng et
al. (2008), significant uncertainty results from the study not providing regional
and/or seasonally-differentiated effects estimates that control for PM2.s. Given
the potential for regional differences in the composition of PMio-2.5 which could
impact risk estimates, combined with the potential for PM2.5 to vary regionally as
a confounder for the effect of PMio-2.5, EPA staff believes that C-R functions with
control for PM2.s would ideally be available at the regional level. .
When considered together, the limitations outlined above resulted in EPA
staff concluding that a quantitative PMio-2.5 risk assessment would not significantly
H-5
-------
enhance the review of the NAAQS for coarse-fraction PM. Specifically, these limitations
would likely result in sufficient uncertainty in the resulting risk estimates to significantly
limit their utility in informing policy-related questions, including the assessment of
whether the current standard is protective of public health and characterization of the
degree of additional public health protection potentially afforded by alternative standards.
Because of the decision not to conduct a quantitative PMi0-2.5 risk assessment, these
questions will draw more heavily on the results of the evidence-based analysis to be
discussed in the Policy Assessment.
H-t
-------
-------
United States Office of Air Quality Planning and Standards Publication No.
Environmental Protection Health and Environmental Effects Division EPA-452/P-10-001
Agency Research Triangle Park, NC February, 2010
------- |