ro
2
Quantitative Health Risk Assessment for
Particulate Matter
-------
DISCLAIMER
This 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. Questions related to this 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 (Urban Case Study Methods), and Appendices A
(Air Quality Assessment (Summary of Individual and Composite Monitor Data by Urban
Study Area)), C (Epidemiological Study Specific Information for PM Risk Assessment),
E (Risk Analysis - Core Analysis), and F (Sensitivity Analysis Results).
-------
EPA-452/R-10-005
June 2010
Quantitative Health Risk Assessment for
Particulate Matter
US Environmental Protection Agency
Office of Air and Radiation
Office of Air Quality Planning and Standards
Health and Environmental Impacts Division
Research Triangle Park, North Carolina 27711
-------
This page left intentionally blank.
-------
Table of Contents
List of Tables v
List of Figures vi
List of Acronyms/Abbreviations viii
1 Introduction 1-1
1.1 Background 1-3
1.2 Current Risk Assessment: Goals and Planned Approach 1-6
1.3 Organization of Document 1-7
2 Scope 2-1
2.1 Overview of Risk Assessment from Last Review 2-2
2.2 Original Assessment Plan 2-3
2.2.1 Risk Assessment 2-4
2.2.2 Population Exposure Analysis 2-6
2.3 Current Scope and Key Design Elements 2-6
2.4 Alternative Suites of PM2.5 Standards Evaluated 2-10
3 Urban Case Study Analysis Methods 3-1
3.1 General Approach 3-1
3.1.1 Basic Structure of the Risk Assessment 3-1
3.1.2 Calculating PM2.5-Related Health Effects Incidence 3-7
3.1.2.1 Short-term vs. Long-term Exposure 3-9
3.1.2.2 Calculating Annual Incidence 3-10
3.2 Air Quality Inputs 3-12
3.2.1 Characterizing Recent Conditions 3-12
3.2.2 Estimating Policy Relevant Background 3-15
3.2.3 Simulating Air Quality to Just Meet Current and Alternative Standards 3-16
3.2.3.1 Proportional Rollback Method 3-18
3.2.3.2 Hybrid Rollback Method 3-20
3.2.3.3 Locally focused Rollback Method 3-22
3.2.3.4 Presentation of Results for the Three Rollback Methods (with
example calculation) 3-23
3.3 Selection of Model Inputs 3-28
3.3.1 Health Endpoints 3-28
3.3.2 Selection and Delineation of Urban Study Areas 3-29
-------
3.3.3 Selection of Epidemiological Studies and Concentration-response (C-R)
Functions within Those Studies 3-34
3.3.4 Summary of Selected Health Endpoints, Urban Areas, Studies, and C-R
Functions 3-40
3.4 Baseline Health Effects Incidence Data 3-54
3.4.1 Data Sources 3-54
3.4.1.1 Mortality 3-54
3.4.1.2 Hospital Admission and Emergency Department Visits 3-54
3.4.1.3 Populations 3-56
3.4.2 Calculation of Baseline Incidence Rates 3-59
3.5 Addressing Uncertainty and Variability 3-63
3.5.1 Overview 3-63
3.5.2 Treatment of Key Sources of Variability 3-66
3.5.3 Qualitative Assessment of Uncertainty 3-69
3.5.4 Single and Multi-Factor Sensitivity Analyses 3-80
3.5.4.1 Sensitivity Analyses for Long-Term Exposure-Related Mortality 3-80
3.5.4.2 Sensitivity Analyses for Short-Term Exposure-Related Mortality and
Morbidity 3-84
3.5.4.3 Multi-factor Sensitivity Analyses 3-86
3.5.5 Summary of Approach to Addressing Variability and Uncertainty 3-87
4 Urban Case Study Results 4-1
4.1 Assessment of Health Risk Associated with Recent Conditions (core analysis) ... 4-16
4.2 Assessment of Health Risk Associated with Just Meeting the Current and
Alternative Suites of Standards (core analysis) 4-19
4.2.1 Core Risk Estimates for Just Meeting the Current Suite of Standards 4-21
4.2.2 Core Risk Estimates for Just Meeting Alternative Suites of Standards 4-22
4.3 Sensitivity Analysis Results 4-28
4.3.1 Sensitivity Analysis Results to Identify Potentially Important Sources of
Uncertainty and Variability 4-29
4.3.1.1 Single-factor Sensitivity Analysis 4-36
4.3.1.2 Multi-factor Sensitivity Analysis Results 4-43
4.3.2 Additional Set of Reasonable Risk Estimates to Inform Consideration of
Uncertainty in Core Risk Estimates 4-45
4.4 Evaluating the Representativeness of the Urban Study Areas in the National
Context 4-49
11
-------
4.4.1 Analysis Based on Consideration of National Distributions of Risk-
Related Attributes 4-50
4.4.2 Analysis Based on Consideration of National Distribution of PM-Related
Mortality Risk 4-66
4.5 Consideration of Design Values and Patterns of PM2.5 Monitoring Data in
Intrepreting Core Risk Estimates 4-68
4.5.1 Design Values 4-68
4.5.2 Patterns in PM2.5 Monitoring Data 4-77
5 Integrative Discussion of Urban Case Study Analysis of PM2.s-related Risks 5-1
5.1 Overall Confidence in the Risk Assessment 5-1
5.1.1 Use of a Deliberative Process in Designing the Risk Model 5-2
5.1.2 Integration of Key Sources of Variability into the RA Design 5-3
5.1.3 Representativeness of the Urban Study Areas 5-5
5.1.4 Impact of Important Sources of Uncertainty on Core Risk Estimates 5-6
5.1.5 Consideration of Alternative Reasonable Risk Estimates 5-7
5.1.6 Consideration of Composite Monitor Annual Average PM2.5
Concentrations in Relation to the Dataset Used in Deriving C-R Functions
for Long-Term Exposure-Related Mortality 5-8
5.2 Key Observations Related to the Urban Study Area Results 5-9
5.2.1 Nature and Magnitude of Long-Term and Short-Term Exposure-Related
Risk Remaining upon Just Meeting the Current Suite of PM2.5 Standards... 5-10
5.2.2 Nature and Magnitude of Long-term and Short-Term Exposure-Related
Risk Remaining upon Just Meeting the Alternative Suite of PM2.5
Standards 5-12
5.2.3 Nature and Magnitude of Long-Term and Short-Term Exposure-Related
Risk Remaining upon Just Meeting Combinations of Alternative Annual
and 24-Hour PM2.5 Standards 5-14
5.3 Summary of Key Observations 5-16
6 References 6-1
Appendix A Air Quality Assessment (Summary of Individual and Composite Monitor Data by
Urban Study Area)
Appendix B Additional Information Supporting Air Quality Characterization
Appendix C Epidemiological Study Specific Information for PM Risk Assessment
Appendix D Supplement to Representativeness of the 15 Urban Study Areas
Appendix E Risk Estimates (Core Analysis)
in
-------
Appendix F Sensitivity Analysis Results
Appendix G Supplement to the National-Scale Assessment of Long-Term Mortality Related to
PM2 5 Exposure
Appendix H Consideration off Risk Associated with Exposure to Thoracic Coarse PM
(PMiO-2.5)
Appendix I Analysis Comparing Distribution of Short-Term Exposure-Related
Cardiovascular Mortality Incidence to the Distribution of Daily PM2.s Levels tor
Detroit and New York
Appendix J: Provisional Risk Estimates and Additional Results of Simulation Involving the
Alternative Annual Standard of 10 |ig/m3
Appendix K: Maps of the Fifteen Urban Study Areas Evaluated in the Risk Assessment
IV
-------
List of Tables
Table 3-1. Numbers of Monitors in Risk Assessment Locations From Which Composite
Monitor Values Were Calculated 3-12
Table 3-2 Regional Policy-Relevant Background Estimates Used in the Risk Assessment... 3-16
Table 3-3. EPA Design Values for Annual and 24-hour PM2.5 Standards for the Period 2005-
2007 3-19
Table 3-4. Application of the Three Rollback Methods in Simulating Current and Alternative
Standard Levels for the 15 Urban Study Areas 3-25
Table 3-5 Urban Study Areas Selected for the Risk Assessment 3-31
Table 3-6 Locations, Health Endpoints, and Short-Term Exposure Studies Included in the PM2.5
Risk Assessment 3-41
Table 3-7 Locations, Health Endpoints, and Long-Term Exposure Studies Included in the PM2.5
Risk Assessment* 3-42
Table 3-8 Summary of Locations, Health Endpoints, Studies and Concentration-Response
Functions Included in the Core Analysis 3-43
Table 3-9 Summary of Locations, Health Endpoints, Studies and Concentration-Response
Functions Included in Sensitivity Analyses 3-51
Table 3-10 Sources of Hospital Admissions (HA) and Emergency Department (ED)
Visit Data 3-56
Table 3-11 Relevant Population Sizes for PM Risk Assessment Locations 3-57
Table 3-12 Baseline Mortality Rates (Deaths per 100,000 Relevant Population per Year) for
2006 for PM Risk Assessment Locations 3-60
Table 3-13 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 3-62
Table 3-14 Summary of Qualitative Uncertainty Analysis of Key Modeling Elements in the PM
NAAQS Risk Assessment 3-72
Table 4-1 Estimated Annual Incidence of Selected Mortality and Morbidity Endpoints
Associated with Long- and Short-Term Exposure to Ambient PM2.5 Concentrations
that Just Meet the Current Standards, Based on Adjusting 2007 PM2.5 Concentrations
4-6
Table 4-2 Estimated Percent of Total Annual Incidence of Selected Mortality and Morbidity
Endpoints Associated with Long- and Short-Term Exposure to Ambient PM2.5
Concentrations that Just Meet the Current Standards, Based on Adjusting 2007 PM2.5
Concentrations 4-7
Table 4-3 Overview of Sensitivity Analysis Results 4-32
Table 4-4 Derivation of a set of reasonable alternative risk estimates to supplement the core risk
estimates 4-46
Table 4-5 Data Sources for PM NAAQS Risk Assessment Risk Distribution Analysis 4-52
Table 4-6 Summary Statistics for Selected PM Risk Attributes 4-54
Table 4-7 Results of Kolomogrov-Smirnoff Tests for Equality Between National and Urban
Study Area Distributions for Selected National Risk Characteristic Variables 4-58
Table 4-8 Identification of controlling standard (24-hour or annual) for alternative suites of
standard levels 4-74
-------
List of Figures
Figure 3-1 Major components of particulate matter health risk assessment 3-2
Figure 3-2 Flow diagram of risk assessment for short-term exposure studies 3-5
Figure 3-3. Flow diagram of risk assessment for long-term exposure studies 3-6
Figure 3-4 15 urban study areas included in the risk assessment (including seven PM regions
used to guide selection of study areas) 3-32
Figure 4-1 Percent reduction in long-term exposure-related mortality risk (alternative
standards and recent conditions relative to the current suite ofstandards) 4-8
Figure 4-2 Percent reduction in long-term exposure-related mortality risk
(recent conditions relative to the current standards) 4-9
Figure 4-3 Percent reduction in long-term exposure-related mortality risk 4-10
Figure 4-4 Percent reduction in short-term exposure-related mortality and morbidity risk 4-11
Figure 4-5 Percent reduction in short-term exposure-related mortality and morbidity risk
(recent conditions relative to the current standards) 4-12
Figure 4-6 Percent reduction in short-term exposure-related mortality and morbidity risk
(alternative standards relative to the current standards) 4-13
Figure 4-7 Comparison of core risk estimates with reasonable alternative set of risk estimates
for Los Angeles and Philadelphia (IFtD mortality) 4-47
Figure 4-8 Comparison of core risk estimates with reasonable alternative set of risk estimates
for Los Angeles and Philadelphia (all cause mortality) 4-48
Figure 4-9 Comparison of distributions for key elements of the risk equation:
total population 4-59
Figure 4-10 Comparison of distributions for key elements of the risk equation:
98th percentile 24-hour average PM2.5 4-60
Figure 4-11 Comparison of distributions for key elements of the risk equation: all use
mortality rate 4-61
Figure 4-12 Comparison of distributions for key elements of the risk equation:
Mortality risk effect estimate from Zanobetti and Schwartz (2008) 4-62
Figure 4-13 Comparison of distributions for selected variables expected to influence the
relative risk from PM2.s: long term average July temperature 4-63
Figure 4-14 Comparison of distributions for selected variables expected to influence the
relative risk from PMi.s: percent of population 65 and older 4-64
Figure 4-15 Comparison of distributions for selected variables expected to influence the
relative risk from PM2.s: per capita annual personal income 4-65
Figure 4-16 Comparison of distributions for selected variables expected to influence the
relative risk from PMi.s: per capita annual personal income 4-66
Figure 4-17 Cumulative distribution of county-level percentage of total mortality attributable
to PM2.5 for the U.S. with markers identifying where along that distribution the
urban case study area analysis fall 4-67
Figure 4-18 Design values in 15 urban study areas and broader set of U.S. urban areas relative
to the current suite ofstandards (15/35) 4-71
VI
-------
Figure 4-19 Design values in 15 urban study areas and broader set of U.S. urban areas relative
to the 12/35 alternative suite of standards 4-72
Figure 4-20 Design values in 15 urban study areas and broader set of U.S. urban areas relative
to the 12/25 alternative suite of standards 4-73
Figure 4-21 Design values in 15 urban study areas and broader set of U.S. urban areas relative
to the current standard (with regional differentiation) 4-76
Figure 4-22 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) 4-79
Figure 4-23 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) 4-80
vn
-------
List of Acronyms/Abbreviations
ACS
AQS
P
BenMAP
BRFSS
CASAC
CAA
CBS A
CDC
CDF
CFR
CHD
CMAQ
CO
COPD
CPD
C-R
CSA
CV
CVD
df
ED
EPA
FACA
FIPS
GAM
GLM
HA
HCUP
HEI
HS
ICD
IHD
INF
American Cancer Society
EPA's Air Quality System
Slope coefficient
Benefits Mapping Analysis Program
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
Combined Statistical Area
Cardiovascular
Cardiovascular Disease
Degrees of freedom
Emergency Department
United States Environmental Protection Agency
Federal Advisory Committee Act
Federal Information Processing System
Generalized additive model
Generalized linear model
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
Vlll
-------
IRP
ISA
KB
K-S
LML
MCAPS
MSA
NA
NAAQS
NCEA
NEI
NCHS
NMMAPS
NOx
03
OAQPS
PA
PM
PMX
PM
2.5
PMiQ-2.5
PRB
RA
RR
Integrated Review Plan
Integrated Science Assessment Document
Knowledge Base
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 CFR.
Particles with a 50% upper cut-point of 2.5 um aerodynamic
diameter and a penetration curve as specified in the CFR
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
Relative risk
IX
-------
REA
SAB
SEDD
SID
S02
SOX
SES
TRIM
TRIM.Risk
USDA
WHI
WHO
ZCA
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
U.S. Department of Agriculture
Women's Health Initiative
World Health Organization
Zip Code Area
-------
1 INTRODUCTION
The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
the national ambient air quality standards (NAAQS) for particulate matter (PM). Sections 108
and 109 of the Clean Air Act (CAA) govern the establishment and periodic review of the
NAAQS. These standards are established for pollutants that may reasonably be anticipated to
endanger public health and welfare, and whose presence in the ambient air results from
numerous or diverse mobile or stationary sources. The NAAQS are to be based on air quality
criteria, which are to accurately reflect the latest scientific knowledge useful in indicating the
kind and extent of identifiable effects on public health or welfare that may be expected from the
presence of the pollutant in ambient air. The EPA Administrator is to promulgate and
periodically review, at five-year intervals, "primary" (health-based) and "secondary" (welfare-
based) NAAQS for such pollutants. Based on periodic reviews of the air quality criteria and
standards, the Administrator is to make revisions in the criteria and standards, and promulgate
any new standards, as may be appropriate. The Act also requires that an independent scientific
review committee advise the Administrator as part of this NAAQS review process, a function
performed by the Clean Air Scientific Advisory Committee (CASAC).1
The current NAAQS for PM include a suite of standards to provide protection for
exposures to fine and coarse particles using PM2.5 and PMio, as indicators, respectively (71 FR
61144, October 17, 2006). With regard to the primary standards for fine particles, in last PM
NAAQS review completed in 2006, EPA revised the level of the 24-hour PM2 5 standard to 35
ug/m3 (calculated as a 3-year average of the 98th percentile of 24-hour concentrations at each
population-oriented monitor), retained the level of the annual PM2.5 annual standard at 15 ug/m3
(calculated as the 3-year average of the weighted annual mean PM2.5 concentrations from single
or multiple community-oriented monitors), and revised the form of the annual PM2 5 standard by
narrowing the constraints on the optional use of spatial averaging.2 With regard to the primary
1 The Clean Air Scientific Advisory Committee (CASAC) was established under section 109(d)(2) of the
Clean Air Act (CAA) (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.
2
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 to 61167, October 17, 2006).
1-1
-------
standards for coarse particles, in the last review, EPA retained the 24-hour PMio standard at 150
ug/m3 (not to be exceeded more than once per year on average over 3 years) and revoked the
annual standard because available evidence generally did not suggest a link between long-term
exposure to current ambient levels of coarse particles and health or welfare effects (71 FR 61177
to 61203). Decisions related to the primary PM standards were based primarily on a large body
of epidemiological evidence relating ambient PM concentrations to various adverse health
endpoints. In 2006, secondary standards for PM2.5 and PMi0 were revised to be identical to the
primary standards (71 FR 61203 to 61210).
The EPA initiated the current review of the PM NAAQS on June 28, 2007 with a call for
information from the public (72 FR 35462).3 The NAAQS review process includes four key
phases: planning, science assessment, risk/exposure assessment, and policy
assessment/rulemaking.4 A workshop was held on July 11 through 13, 2007 (72 FR 34003) to
discuss policy-relevant scientific and technical information to inform EPA's planning for the PM
NAAQS review. Following the workshop, EPA developed a planning document, the Integrated
Review Plan for the National Ambient Air Quality Standards for Particulate Matter (IRP; US
EPA, 2008a), which outlined the key policy-relevant issues that frame this review, the process
and schedule for the review, and descriptions of the purpose, contents, and approach for
developing the other key documents for this review.5 In December 2009, EPA completed the
process of assessing the latest available policy-relevant scientific information to inform the
review of the PM standards. This assessment the Integrated Science Assessment for P articulate
Matter (ISA; US EPA, 2009a), includes an evaluation of the scientific evidence on the health
effects of PM, including information on exposure, physiological mechanisms by which PM
might adversely impact human health, an evaluation of the toxicological and controlled human
exposure study evidence, and an evaluation of the epidemiological evidence including
information on reported concentration-response (C-R) relationships for PM-related morbidity
and mortality associations, including consideration of effects on susceptible populations.6
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 For more information on the NAAQS review process see http://www.epa.gov/ttn/naaqs/review.html.
5 On November 30, 2007, EPA held a public consultation with the CAS AC PM Panel on the draft IRP. The
final IRP took into consideration comments received from CAS AC and the public on the draft plan as well as input
from senior Agency managers.
6 The ISA also evaluates scientific evidence for the effects of PM on public welfare which EPA will
consider in its review of the suite of secondary PM NAAQS. Building upon the visibility effects evidence presented
in the ISA, OAQPS has also developed a second REA titled Particulate Matter Urban-Focused Visibility
Assessment (US EPA, 201 Ob).
1-2
-------
The EPA's Office of Air Quality Planning and Standards (OAQPS) has developed this
quantitative health risk assessment (RA) describing the quantitative assessments of PM-related
risks to public health to support the review of the primary PM standards. This document is a
concise presentation of the scope, methods, key results, observations, and related uncertainties
associated with the quantitative analyses performed. The RA builds upon the health effects
evidence presented and assessed in the ISA, as well as CASAC advice (Samet, 2009a, b; Samet,
2010) and public comments on a scope and methods planning document for the RA (here after,
"Scope and Methods Plan", US EPA, 2009b) and on the first and second draft RA documents
(US EPA, 2009e; US EPA, 2010a).
The ISA and RA will inform the policy assessment and rulemaking steps that will lead to
final decisions on the primary PM NAAQS. The policy assessment is described in Policy
Assessment for the Review of the P articulate Matter National Ambient Air Quality Standards
(hereafter, "PA"), which include staff analysis of the scientific basis for alternative policy
options for consideration by senior EPA management prior to rulemaking. The PA integrates
and interprets information from the ISA and the RA to frame policy options for consideration by
the Administrator. The PA is intended to link the Agency's scientific and technical assessments,
presented in the ISA and RA, to judgments required of the Administrator in determining whether
it is appropriate to retain or revise the current suite of PM standards. Development of the PA is
also intended to facilitate elicitation of CASAC's advice to the Agency and recommendations on
any new standards or revisions to existing standards as may be appropriate, as provided for in the
Clean Air Act (CAA). The second draft PA is planned for release around the end of June 2010
for review by the CASAC PM Panel and the public during a public teleconference being planned
for late March. Proposed and final rulemaking notices are now scheduled for November 2010
and July 2011, respectively.
1.1 BACKGROUND
As part of the last PM NAAQS review completed in 2006, EPA's OAQPS conducted a
quantitative risk assessment to estimate risks of various health effects associated with exposure
to ambient PM2.5 and PMio-2.s in a number of urban study areas selected to illustrate the public
health impacts of these pollutants (U.S. EPA, 2005, chapter 4; Abt Associates, 2005). The
assessment scope and methodology were developed with considerable input from CASAC and
the public, with CASAC concluding that the general assessment methodology and framework
were appropriate (Hopke, 2002). The final quantitative risk assessment took into consideration
CASAC advice (Hopke, 2004; Henderson, 2005) and public comments on two drafts of the risk
assessment.
1-3
-------
The extensive quantitative assessment conducted for fine particles in the last review
focused on nine urban study areas and included estimated risks of total non-accidental,
cardiovascular-related, and respiratory-related mortality as well as morbidity effects including
hospital admissions for cardiovascular and respiratory causes and respiratory symptoms (not
requiring hospitalization) associated with recent short-term (daily) ambient PM2.5 concentrations.
This assessment also included estimated risks of total, cardiopulmonary, and lung cancer
mortality associated with long-term PM2.5 exposures. The quantitative risk assessment included
estimates of: (1) risks of mortality, morbidity, and symptoms associated with recent ambient
PM2.5 levels; (2) risk reductions and remaining risks associated with just meeting the existing
suite of PM2.sNAAQS (1997 standards); and (3) risk reductions and remaining risks associated
with just meeting various alternative PM2 5 standards.
The quantitative risk assessment conducted in the last review for thoracic coarse particles
was much more limited than the analyses conducted for fine particles. The PMio-2.5 risk
assessment included risk estimates for just three urban areas for two categories of health
endpoints related to short-term PMi0-2.5 exposures: hospital admissions for cardiovascular and
respiratory causes and respiratory symptoms. While one of the goals of the PMio-2.5 risk
assessment was to provide estimates of the risk reductions associated with just meeting
alternative PMio-2.5 standards, OAQPS staff concluded that the nature and magnitude of the
uncertainties and concerns associated with this portion of the quantitative risk assessment
weighed against use of these risk estimates as a basis for recommending specific standard levels
(U.S. EPA, 2005, p. 5-69).
Prior to the issuance of a proposed rulemaking in the last review, CASAC presented
recommendations to the Administrator supporting revisions of the PM2.5 primary standards.
These recommendations placed substantial reliance on the results of the quantitative risk
assessment (Henderson, 2005, pp 6-7). In a letter to the Administrator following the 2006
proposed rule (71 FR 12592, January 17, 2006), CASAC requested reconsideration of the
Agency's proposed decisions and reiterated and elaborated on the scientific bases for its earlier
recommendations which included placing greater weight on the result of the Agency's risk
assessment. With regard to the quantitative risk assessment, CASAC concluded, "While the risk
assessment is subject to uncertainties, most of the PM Panel found EPA's risk assessment to be
of sufficient quality to inform its recommendations." (Henderson, 2006a, p. 3).
In the 2006 final rule, the EPA Administrator recognized that the quantitative risk
assessment for fine particles was based upon a more extensive body of data and was more
comprehensive in scope than the previous assessment conducted for the review completed in
1997. However, as presented in the final rulemaking notice, the Administrator was mindful of
significant uncertainties associated with the risk estimates for fine particles. More specifically,
1-4
-------
Such uncertainties generally related to a lack of clear understanding of a number of
important factors, including, for example, the shape of the concentration-response
functions, particularly when, as here, effect thresholds can neither be discerned nor
determined not to exist; issues related to selection of appropriate statistical models for the
analysis of the epidemiologic data; the role of potentially confounding and modifying
factors in the concentration-response relationships; issues related to simulating how PM2.5
air quality distributions will likely change in any given area upon attaining a particular
standard, since strategies to reduce emissions are not yet defined; and whether there
would be differential reductions in the many components within PM2 5 and, if so, whether
this would result in differential reductions in risk. In the case of fine particles, the
Administrator recognized that for purposes of developing quantitative risk estimates,
such uncertainties are likely to [be] amplified by the complexity in the composition of the
mix of fine particles generally present in the ambient air. (72 FR 61168, October 17,
2006).
As a result, the Administrator viewed that the quantitative risk assessment provided supporting
evidence for the conclusion that there was a need to revise the PM2.5 primary standards, but he
judged that the assessment did not provide an appropriate basis to determine the level of the
standards (72 FR 61168, October 17, 2006).
In a letter to the EPA Administrator following the issuance of the final rule, CASAC
expressed "serious scientific concerns" regarding the final PM standards. In particular, CASAC
was concerned that the Agency "did not accept our finding that the annual PM2.5 standard was
not protective of human health and did not follow our recommendation for a change in that
standard" (Henderson et al, 2006b, p.l). With respect to the use of the risk assessment to inform
EPA's decision on the primary PM2.5 standard, CASAC stated, "While there is uncertainty
associated with the risk assessment for the PM2.5 standard, this very uncertainty suggests a need
for a prudent approach to providing an adequate margin of safety" (Henderson et al., 2006b, p.2)
Several parties filed petitions for review following promulgation of the revised PM
NAAQS in 2006. These petitions for review addressed the following issues with regard to the
primary PM NAAQS: (1) selecting the level of the annual primary PM2.5 standard and (2)
retaining PMio as the indicator of a standard for thoracic coarse particles, retaining the level and
form of the 24-hour PMio standard, and revoking the PMio annual standard. On judicial review,
the D.C. Circuit remanded the annual primary PM25 NAAQS to EPA because the Agency failed
to adequately explain why the standard provided the requisite protection from both short- and
long-term exposures to fine particles including protection for at-risk populations. The court
upheld the Agency's use of the quantitative risk assessment to inform the decision to revise the
PM2.s standards but not to inform the selection of level. The court also upheld the decision to
1-5
-------
retain the 24-hour PMio standard and revoke the annual PMio standard. 7 American Farm
Bureau Federation v. EPA, 559 F. 3d 512, (D.C. Cir. 2009).
1.2 CURRENT RISK ASSESSMENT: GOALS AND PLANNED APPROACH
The goals of the current quantitative health risk assessment remain largely the same as
those articulated in the risk assessment conducted in the last review. These goals include: (a) to
provide estimates of the potential magnitude of premature mortality and/or selected morbidity
effects in the population associated with recent ambient levels of PM and with just meeting the
current and alternative suites of PM standards considered in selected urban study areas,
including, where data are available, consideration of impacts on susceptible populations; (b) to
develop a better understanding of the influence of various inputs and assumptions on the risk
estimates to more clearly differentiate among alternative suites of standards, including potential
impacts on various susceptible populations; and (c) to gain insights into the distribution of risks
and patterns of risk reductions and the variability and uncertainties in those risk estimates. In
addition, this assessment includes nationwide estimates of the potential magnitude of premature
mortality associated with long-term exposure to recent ambient PM2.5 concentrations to more
broadly characterize this risk on a national scale and to support the interpretation of the more
detailed risk estimates generated for selected urban study areas. The overall scope and design of
this quantitative risk assessment, discussed below in chapters 2 and 3, reflect efforts to achieve
these goals.
This current quantitative risk assessment builds on the approach used and lessons learned
in the last PM risk assessment and focuses on improving the characterization of the overall
confidence in the risk estimates, including related uncertainties, by incorporating a number of
enhancements, in terms of both the methods and data used in the analyses. This assessment
considers a variety of health endpoints for which, in staff s judgment, there is adequate
information to develop quantitative risk estimates that can meaningfully inform the review of the
primary PM NAAQS. Evidence of relationships between PM and other health endpoints for
which, in staff s judgment, there currently is insufficient information to develop meaningful
quantitative risk estimates are discussed in the PA as part of the evidence-based considerations
that inform staffs assessment of policy options.
7 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).
1-6
-------
1.3 ORGANIZATION OF DOCUMENT
The remainder of this document is organized as follows. Chapter 2 provides an overview
of the scope of the quantitative risk assessment, including a summary of the previous risk
assessment, the original planned approach and the key design elements reflected in the final
assessment, and the rationale for the alternative standard levels evaluated in this assessment.
Chapter 3 describes the analytical approach, methods, and data used in conducting the risk
assessment, including the approach used to generate risk estimates for the set of urban case
studies included in this analysis and the approaches used in addressing variability and
uncertainty (Appendices A, B, and C provide supplemental information regarding the data and
methods used). Chapter 4 presents selected risk estimates generated for the urban case studies,
including the results of single- and multi-factor sensitivity analyses and a national-scale analysis
of the representativeness of relevant risk-related factors (Appendix D provides supplemental
information on risk-related factors; Appendices E and F provide detailed risk estimates and
sensitivity analysis results, respectively). Chapter 5 provides an integrative discussion of the
various risk estimates generated in the analyses drawing on the results of the urban area case
studies, the uncertainty/variability characterization, the assessment of the representativeness of
the urban study areas in a national context, and the patterns in design values and air quality
monitoring data considered to inform the interpretation of the risk estimates generated in the
urban case study analyses.
1-7
-------
2 SCOPE
This chapter provides an overview of the scope and key design elements of this
quantitative health risk assessment. The design of this assessment began with a review of the
risk assessment completed during the last PM NAAQS review (Abt Associates, 2005; US EPA,
2005, chapter 4), with an emphasis on considering key limitations and sources of uncertainty
recognized in that analysis.
As an initial step in the current PM NAAQS review, EPA invited outside experts,
representing abroad range of expertise (e.g., epidemiology, human and animal toxicology,
statistics, risk/exposure analysis, atmospheric science) to participate in a workshop with EPA
staff to help inform EPA's plan for the review. The participants discussed key policy-relevant
issues that would frame the review and the most relevant new science that would be available to
inform our understanding of these issues. One workshop session focused on planning for
quantitative risk/exposure assessments, taking into consideration what new research and/or
improved methodologies would be available to inform the design of a quantitative health risk
assessment and whether, and if so how, it might be appropriate to conduct a quantitative
exposure assessment. Based in part on the workshop discussions, EPA developed a draft IRP
(US EPA, 2007) outlining the schedule, process, and key policy-relevant questions that would
frame this review. On November 30, 2007, EPA held a consultation with CASAC on the draft
IRP (72 FR 63177, November 8, 2007), which included opportunity for public comment. The
final IRP incorporated comments from CASAC (Henderson, 2008) and the public on the draft
plan as well as input from senior Agency managers. The IRP included initial plans for
quantitative risk and exposure assessments (US EPA, 2008a, chapter 5).
As a next step in the design of these quantitative assessments, OAQPS staff developed a
more detailed planning document, Paniculate Matter National Ambient Air Quality Standards:
Scope and Methods Plan for Health Risk and Exposure Assessment (Scope and Methods Plan;
US EPA, 2009b). This Scope and Methods Plan was the subject of a consultation with CASAC
on April 1-2, 2009 (74 FR 11580, March 18, 2009). Based on consideration of CASAC (Samet,
2009a) and public comments on the Scope and Methods Plan and information in the first draft
ISA, we modified the scope and design of the quantitative risk assessment and completed initial
analyses that were presented in a first draft RA (US EPA, 2009e). The CASAC met on October
5-6, 2009 to review the first draft RA (74 FR 46586, September 10, 2009).8 Based on
8 A public teleconference was held on November 12, 2009, during which CASAC reviewed the draft
comment letter prepared by the CASAC PM Panel.
2-1
-------
consideration of CAS AC (Samet, 2009b) and public comments on the first draft RA, together
with ongoing refinement of elements of the risk assessment approach informed by the second
draft ISA, we prepared a second draft RA (US EPA, 2010a). The CAS AC PM Panel met again
on March 10-11, 2010 to review the second draft RA (75 FR 8062, February 23, 2010). Based on
consideration of CAS AC comments (Samet, 2010) together with public comments on the second
draft RA and ongoing, methods development work, we have prepared this final RA.
In presenting the scope and key design elements of the current risk assessment, this
chapter first provides a brief overview of the quantitative risk assessment completed for the
previous PM NAAQS review in section 2.1, including key limitations and uncertainties
associated with that analysis. Section 2.2 provides a summary of the initial design of the risk
assessment as outlined in the Scope and Methods Plan. Section 2.3 provides an overview of key
design elements reflected in this final risk assessment focusing on those aspects of the final
approach which differ from the originally planned approach reflecting consideration of CAS AC
and public comments and additional EPA analyses. Section 2.4 provides a summary of the
various air quality scenarios evaluated in this assessment, including recent air quality conditions
and simulations of just meeting the current and alternative suites of PM2.5 standards.
2.1 OVERVIEW OF RISK ASSESSMENT FROM LAST REVIEW
The quantitative risk assessment conducted in the last review included a broad
assessment of PM2.s-related risk and a much more limited assessment of PMio-2.s-related risk.
That assessment included estimates of risks of mortality (total non-accidental, cardiovascular,
and respiratory), morbidity (hospital admissions for cardiovascular and respiratory causes), and
respiratory symptoms (not requiring hospitalization) associated with short-term (24-hour) PM2.5
exposures and estimates of risks of total, cardiopulmonary, and lung cancer mortality associated
with long-term PM2.5 exposures in selected urban areas. Nine urban study areas were evaluated
across the U.S.: Boston, MA; Detroit, MI; Los Angeles, CA; Philadelphia, PA; Phoenix, AZ;
Pittsburgh, PA; San Jose, CA; Seattle, WA; and St. Louis, MO.
The EPA recognized that there were many sources of uncertainly and variability inherent
in the inputs to the assessment and that there was a high degree of uncertainty in the resulting
PM2.s risk estimates. Such uncertainties generally related to a number of important factors,
including: (a) the shape of the concentration-response (C-R) function and whether or not a
population threshold exists; (b) the selection of appropriate statistical models for the analysis of
epidemiological data; (c) the role of potentially confounding and modifying factors in the C-R
relationships; (d) the methods for simulating how daily PM2.s ambient concentrations would
likely change in any given area upon meeting a particular suite of standards; and (e) the potential
for differences in the relative toxicity of the components within the mix of ambient PM2.5.
2-2
-------
While some of these uncertainties were addressed quantitatively in the form of estimated
confidence ranges around central risk estimates, other uncertainties and the variability in key
inputs were not reflected in these confidence ranges, but rather were addressed through separate
sensitivity analyses or characterized qualitatively (US EPA, 2005, chapter 4; Abt Associates,
2005). The C-R relationships used in the quantitative risk assessment were based on findings
from epidemiological studies that relied on fixed-site, population oriented, ambient monitors as a
surrogate for actual ambient PM2.5 exposures. The assessment included a series of base case
estimates that, for example, included various cutpoints intended as surrogates for alternative
potential population thresholds. Other uncertainties were addressed in various sensitivity
analyses (e.g., the use of single- versus multi-pollutant models, use of single versus multi-city
models, use of a distributed lag model) and had more moderate and often variable impacts on the
risk estimates in some or all of the selected urban study areas.
These same sources of uncertainty and variability were also applicable to the quantitative
risk assessment conducted for PMi0-2.5 in the last review. However, the scope of the risk
assessment for PMi0-2.5 was much more limited than that for PM2 5 reflecting the much more
limited body of epidemiological evidence and air quality information available for PMio-2.5. The
PMio-2.5 risk assessment included risk estimates for just three urban study areas for two
categories of health endpoints related to short-term exposure to PMio-2.s: hospital admissions for
cardiovascular and respiratory causes and respiratory symptoms. While one of the goals of the
PMio-2.5 risk assessment was to provide estimates of the risk reductions associated with just
meeting alternative PMio-2.5 standards, EPA staff concluded that the nature and magnitude of the
uncertainties and concerns associated with this portion of the risk assessment weighed against
use of these risk estimates as a basis for recommending a range of standard levels for
consideration (US EPA, 2005, see p. 5-69). These uncertainties and concerns were summarized
in the proposed rulemaking notice (FR 71 2662, January 17, 2006) and discussed more fully in
the Staff Paper (US EPA, 2005, chapter 4) and associated technical support document (Abt
Associates Inc., 2005).
2.2 ORIGINAL ASSESSMENT PLAN
The Scope and Methods Plan outlined a planned approach for conducting the current
quantitative PM risk assessment, including broad design issues as well as more detailed aspects
of the analyses. That document also outlined plans for a population exposure analysis based on
micro-environmental exposure modeling. The planned approaches for conducting both analyses
are briefly summarized below.
2-3
-------
2.2.1 Risk Assessment
Key design elements for the quantitative risk assessment, as presented in the Scope and
Methods Plan, included:
PM size fractions: We planned to focus primarily on estimating risk associated with
exposure to PM2.5 with a much more limited assessment of PMio-2.5- Based on
information presented and assessed in the first draft ISA, we concluded that there was
insufficient data to support a quantitative risk assessment for other size fractions (e.g.,
ultrafme particles).
PM components/sources/environments: We considered the extent to which
evidence was available to support a quantitative risk assessment for specific PM
components, sources, and/or environments. Based on review of the evidence
presented and assessed in the first draft ISA, we concluded that there was insufficient
data to support such analyses..
Selection of health effect categories (PMi.s): We planned to focus primarily on
categories for which the evidence supports a judgment that there is at least a likely
causal relationship with PM2.5 exposures. We also planned to consider including
additional categories for which evidence is suggestive of causal relationship with
PM2.5 exposures (e.g., reproductive and developmental outcomes), if sufficient
information was available to develop meaningful risk estimates for these additional
categories.
Selection of health effect categories (PMi0-2.5): We planned to build on the limited
quantitative risk assessment conducted in the last review (US EPA, 2005) with a
focus on health effect categories for which the evidence is suggestive of a causal
relationship with short-term PMio-2.5 exposures, where sufficient information was
available to develop meaningful risk estimates.
Selection of urban study areas: We planned to expand the number of urban study
areas to between 15 and 20, with selection of these study areas being based on
consideration of a number of factors (e.g., availability of location-specific C-R
functions and baseline incidence data, coverage for geographic heterogeneity in PM
risk-related attributes, coverage for areas with more susceptible populations). We
also discussed the possibility of including more refined risk assessments for locations
where more detailed exposure studies had been completed (e.g., Los Angeles, CA,
based on a zip code level analysis of long-term PM2.5-exposure related mortality
presented in Krewski et al., 2009).
Simulation of air quality levels that just meet current or alternative suites of
standards: We planned to consider the use of non-proportional air quality
adjustment methods in addition to the proportional approach that has been used
previously. These non-proportional adjustment methods could be based on (a)
historical patterns of reductions in urban areas, if these supported consideration for
non-proportional reductions across monitors within a specific urban area and/or (b)
model-based (e.g., Community Multiscale Air Quality [CMAQ]) rollback designed to
more realistically reflect patterns of PM reductions across monitors in an urban area.
2-4
-------
Characterization of policy relevant background (PRB): We planned to use the
results of air quality modeling based on a combination of the global-scale circulation
model, GEOS-Chem, with the regional scale air quality model, CMAQ, as presented
in the first draft ISA, rather than empirical data to characterize PRB levels for use in
the risk assessment model.
Selection of epidemiological studies to provide C-R functions: Recognizing
advantages of different study designs, we planned to include C-R functions identified
in both multi- and single-city epidemiological studies using both multi- and single-
pollutant models, where available. We planned to place greater weight on the use of
C-R functions reflecting adjusted single-city estimates obtained from multi-city
studies.
Shape of the functional form of the risk model: We planned to emphasize non-
threshold C-R functions in the risk assessment model, based on the first draft ISA
conclusion that there was little support in the literature for population thresholds for
mortality effects associated with either long-term or short-term PM2.5 ambient
concentrations.9 We also stated that we may consider population thresholds as part of
the sensitivity analysis.
Modeling of risk down to PRB versus lowest measured level (LML): We planned
to model risk down to LML for estimating risk associated with long-term PM2.5
exposures and down to PRB for estimating risks associated with short-term PM2.5
exposures.
Characterization of uncertainty and variability: We planned to include a
discussion in the risk assessment report on the degree to which the risk assessment
covers key sources of variability related to PM risk. For uncertainty, we planned to
include a qualitative discussion of key sources of uncertainty and provide ratings
(low, medium and high) in terms of their potential impact on risk estimates. We also
described the use of sensitivity analysis methods planned both to characterize the
potential impact of sources of uncertainty on risk estimates and to provide an
alternative set of reasonable estimates to supplement the main ("core") set of risk
estimates generated for the urban study areas.
National-scale assessment: We planned to conduct a limited national-scale
assessment of mortality associated with long-term exposure to recent ambient PM2.5
levels.
Representativeness analysis for the urban study areas: We planned to conduct an
analysis to evaluate the representativeness of the selected urban study areas against
national distributions for key PM2 5 risk-related attributes to determine whether they
are nationally representative or more focused on a particular portion of the
distribution for a given attribute.
9 In discussing short-term exposure mortality studies, the first draft ISA (U.S. EPA, 2008a) indicated
support for non-threshold, log-linear models, while acknowledging that the possible influence of exposure error and
heterogeneity of shapes across cities remains to be resolved.
2-5
-------
2.2.2 Population Exposure Analysis
The Scope and Methods Plan also described a population exposure analysis based on
micro-environmental exposure modeling using the Air Pollution Exposure Model (APEX) (US
EPA, 2009b, chapter 4). The planned analysis focused on evaluating PM2.5 exposures in a subset
of the urban study areas included in the quantitative risk assessment to provide insights to inform
the interpretation of the available epidemiological studies.
Following release of the Scope and Methods Plan, we continued development of our
approach for conducting a population exposure analysis, with the goal of completing the analysis
as part of the current PM NAAQS review. However, this additional design work highlighted the
need to more clearly define the intended purpose of the analysis, including specific ways in
which the results would be used to interpret the estimates generated from the risk assessment
(e.g., potentially identifying sources of exposure measurement error associated with the
epidemiological studies from which C-R functions were drawn for the risk assessment and the
magnitude of the impact of those sources of error on risk estimates). Taking CAS AC comments
into consideration, which emphasized the same point regarding the importance of more clearly
defining how the exposure assessment results would be used (Samet, 2009a), as well as the
complexities associated with designing and conducting such an assessment, we decided to
continue methods development work rather than attempt to complete a preliminary population
exposure analysis as part of this review. Development of the population exposure analysis
methodology is ongoing, and we anticipate that such an assessment could be conducted as part of
the next PM NAAQS review.
2.3 CURRENT SCOPE AND KEY DESIGN ELEMENTS
An overview of the scope and key design elements that are the basis for the final RA are
presented below, focusing on those aspects of the risk assessment approach which differ from the
originally planned approach presented in the Scope and Methods Plan and summarized in section
2.2.1.
PM size fractions: This quantitative risk assessment characterizes risk associated
with PM2.s-related exposures only. With regard to PMio-2.5, we conclude that
continued limitations in data available for characterizing PMio-2.5 exposure and risk
would introduce significant uncertainty into a PMio-2.5 quantitative risk assessment
such that the risk estimates generated would be of limited utility for informing
conclusions regarding either the adequacy of the current PMio standard or alternative
standards for consideration. This conclusion was reached by first reviewing the set of
limitations cited in the last PM NAAQS risk assessment for not using the PMio-2.5 risk
estimates in recommending specific standard levels. We then considered whether the
currently available health effects data assessed and presented in the ISA and the
currently available PMio-2.5 air quality monitoring data fundamentally addressed these
2-6
-------
limitations and provided sufficient information to support the development of
meaningful risk estimates. We conclude that significant limitations in both the health
effects data base and the current PMio-2.5 monitoring network continue to exist and
that the currently available information do not support conducting a quantitative risk
assessment for PMio-2.5 at this time (a more in-depth discussion of the rationale
behind the decision not to conduct a quantitative risk assessment for PMio-2.5 is
presented in Appendix H). Furthermore, consistent with the Scope and Methods Plan,
we conclude that the currently available data are too limited to support a quantitative
risk assessment for any specific PM components or for ultrafine particles (UFPs), at
this time. We note, however, that the evidence for health effects associated with
thoracic coarse particles, PM components, and UFPs are presented and assessed in the
ISA and will be discussed as part of the evidence-based considerations presented in
the PA.
Selection of health effects categories (PM2.s): A multi-factor decision framework
was used to select the final set of health effects categories included in the risk
assessment for PM2.5 (section 3.3.1). In evaluating the currently available
epidemiological evidence within the context of the framework, the endpoints focused
on in the quantitative risk assessment focused on total, cardiopulmonary, and lung
cancer mortality associated with long-term PM2.s exposures mortality (total non-
accidental, cardiovascular, and respiratory), morbidity (hospital admissions for
cardiovascular and respiratory causes), and respiratory symptoms (not requiring
hospitalization) associated with short-term (24-hour) PM2.s exposures. The selection
of this set of endpoints is consistent with those endpoints outlined in the Scope and
Methods Plan for PM2.5 and included specific endpoints from health effect categories
classified in the ISA as having a causal or likely causal relationship with PM2.s
exposures. We considered a broader range of endpoints for this quantitative risk
assessment including outcomes within health effect categories classified as having
evidence suggestive of a casual relationship with PM2.5 (e.g., reproductive and
developmental effects). These endpoints were not selected for inclusion in this
analysis for several reasons including limited available information to support the
selection of C-R functions for specific endpoints within these health effect categories
and lack of available baseline incidence data. While the final health endpoints
considered in this quantitative risk assessment are limited to health effect categories
classified as having a causal or likely causal relationship with PM2.s exposures, this
result was a consequence of applying our multi-factor decision framework and not the
sole determining factor. In addition, CASAC members expressed differing views as
to the appropriateness of including health effect categories classified as having
evidence suggestive of a causal relationship.
Selection of urban study areas: We have included 15 urban study areas in the risk
assessment. The selection of these areas is based on a number of criteria including:
(a) consideration of urban study areas evaluated in the last PM risk assessment; (b)
consideration of locations evaluated in key epidemiological studies; (c) preference for
locations with relatively elevated 24-hour and/or annual PM2.s monitored levels so
that the assessment can provide potential insights into the degree of risk reduction
associated with just meeting the current and alternative suites of standards; and (d)
2-7
-------
preference for including locations from different regions across the country, reflecting
potential differences in PM sources, composition, and potentially other factors which
might impact PM-related risk (section 3.3.2). Due in part to time and resource
limitations, we have not included a specialized analysis of risk based on
epidemiology studies using more highly-refined exposure analysis (e.g., the study of
Los Angeles, CA involving zip code-level analysis of long-term PM2 5-exposure
related mortality as presented in Krewski et al., 2009). We have included
consideration of studies with more refined surrogate measures of exposure in our
discussion of uncertainties related to estimating long-term mortality risk, since they
can inform our interpretation of the degree of potential bias associated with the effect
estimates used to model risks (section 3.5.3).
Method used to develop composite monitor values: We revised the methods used
to derive composite monitor values for both the annual and 24-hour air quality
distributions based upon ongoing EPA methods development efforts (section 3.2.1).
The revised methods ensure that ambient measurements from different monitors in a
particular urban study area used to calculate a composite monitor value for that urban
study area are given equal weight. This approach is in contrast to the approach used
in the first draft RA, which effectively weighted ambient measurements from
monitors based on sampling frequency, potentially leading to composite monitor
estimates that were biased high.
Simulation of air quality concentrations that just meet current or alternative
suites of standards: We simulate air quality concentrations using different
approaches. We first use a proportional rollback approach as discussed in the Scope
and Methods Plan and applied in the first draft RA as well as the risk assessment
conducted for the last PM NAAQS review to simulate PM2.5 ambient concentrations
that would "just meet" the current and alternative suites of standards. We also
developed and applied two alternative approaches (hybrid and locally focused) to
improve our understanding of the uncertainty associated with this aspect of the
assessment (section 3.2.3). In addition, we refined our rollback approach for the
Pittsburgh study area, using a dual-zone approach to take into account monitor
locations and the related topography in that area (section 3.2.3).
Characterization of PRB: Consistent with the originally planned approach, we use
regional PRB estimates generated using a combination of GEOS-Chem and CMAQ
modeling as presented and discussed in the ISA (section 3.2.2).
Selection of epidemiological studies to provide C-R functions: In modeling risk
associated with both short-term and long-term PM2.5 exposures, we focus on selecting
C-R functions from larger multi-city studies based on our conclusion that these
studies provide more defensible effect estimates. In modeling short-term exposure-
related mortality and morbidity, we obtained more spatially-refined effect estimates at
the city- and regional-levels, respectively (i.e., effect estimates based on application
of Bayesian methods). We also included C-R functions selected from several single-
city epidemiological studies to provide coverage for additional health effect endpoints
associated with short-term PM2.5 exposures (e.g., emergency department (ED) visits).
Modeling of long-term exposure-related mortality focused on the latest reanalysis of
2-8
-------
the American Cancer Society (ACS) dataset (Krewski et al., 2009). This study
expands upon previous publications presenting evaluations of the ACS long-term
cohort study and, in particular, includes rigorous examination of different model
forms for estimating effect estimates as well as updated and expanded datasets on
incidence and exposure. Our rationale for selecting C-R functions from specific
epidemiological studies to use in the assessment, as well as our rationale for not
selecting C-R functions from alternative epidemiological studies, is discussed below
in section 3.3.3.
Characterization of uncertainty and variability: To characterize uncertainty and
variability, we follow guidance developed by the World Health Organization (WHO)
which presents a four-tiered approach for characterizing uncertainty (and to a lesser
extent variability) in the context of a risk assessment (WHO, 2008). This guidance
includes tiers ranging from qualitative characterization of uncertainty (Tier 1) to use
of full-probabilistic Monte Carlo-based simulation (Tier 3). Sensitivity analysis
methods, which are used in the RA to assess sources of uncertainty and variability,
represent a Tier 2 approach. The application of single- and multi-factor sensitivity
analysis methods in the RA serves two purposes: (a) to characterize the potential
magnitude of impact that a source(s) of uncertainty and/or variability can have on risk
estimates and (b) to provide an additional set of reasonable risk estimates to
supplement the "core" risk estimates10 in characterizing the potential magnitude of
uncertainty in the risk estimates (section 3.5.1 and 3.5.4).
Representativeness of the selected urban study areas: As planned, we conducted
two analyses to evaluate the representativeness of the selected urban study areas for
more broadly characterizing national risks. First, we considered key PM2 5 risk-
related attributes to determine whether the selected urban study areas are nationally
representative or more focused on a particular portion of the distribution for a given
attribute (section 4.4.1). Second, we analyzed estimates of mortality associated with
recent long-term ambient concentrations to assess the extent to which the 31 counties
comprising the 15 urban study areas captured locations within the U.S. likely to
experience the highest PM2.5-related risk (section 4.4.2).11
Consideration of patterns in design values and ambient PMi.s monitoring data
across urban areas12: We examine how 24-hour and annual design values, together
with patterns in PM2 5 monitoring data within an area, can influence the degree of risk
reduction estimated to occur upon simulating just meeting the current or alternative
suites of standards. This analysis improves our understanding of the factors related to
specific patterns of risk reduction. We also compare patterns of design values for the
15 urban study areas with patterns of design values across a broader set of urban areas
10 The "core" risk estimates produced in this assessment refer to those generated using the combination of
modeling elements and input datasets in which we had the highest confidence relative to other modeling choices
11 The National-Scale Mortality analysis planned and discussed in the Scope and Methods Plan and
presented in the first and second draft RA provided the county-level mortality estimates used in this
representativeness analysis (see Appendix G).
12 See section 3.2.3.1 for additional detail on derivation of 24-hour and annual design values.
2-9
-------
in the U.S. for which adequate air quality data are available in order to place core risk
estimates generated for the set of urban study areas in a broader national context.
Integrated discussion of results and key observations: To enhance the utility of the
risk estimates generated for the 15 urban study areas to inform the current review of
the PM2.5NAAQS, we integrate the core risk estimates generated for the 15 urban
study areas with key observations from the sensitivity analyses and the qualitative
analysis of uncertainty, analyses of representativeness, and patterns of design values
across the U.S (chapter 5).
2.4 ALTERNATIVE SUITES OF PM2.5 STANDARDS EVALUATED
The scope of this quantitative risk assessment focuses on consideration of alternative
standard levels only. Simulation of just meeting alternative standard levels is considered in this
assessment in conjunction with the current averaging times (24-hour and annual) and forms of
the existing suite of PM2.5 standards.13 The four basic elements of the NAAQS: indicator14,
averaging time, form, and level, which together serve to define each standard, must be
considered collectively in evaluating the health protection afforded by the primary PM
standards.15
With regard to selecting alternative levels for the annual and 24-hour PM2.5 standards for
evaluation in the quantitative risk assessment, we made initial selections during the development
of the first draft RA based upon information available to us at that time as presented and assessed
in the second draft ISA. In the process of finalizing the risk assessment in consideration of the
final ISA and the ongoing development of the first draft PA, we revisited the selection of
alternative levels and reached the conclusion that it was appropriate to expand the rang of levels
evaluated, as discussed below.
In evaluating the ambient air quality concentrations associated with health effects in
epidemiological studies of long- and short-term exposure to PM2.5 we placed greatest weight on
information from multi-city studies. These studies have a number of advantages compared to
single-city studies including: (1) multi-city studies reflect ambient PM2.5 concentrations and
potential health impacts across a range of diverse locations; (2) multi-city studies "clearly do not
13 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 current 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 current form of the annual PM2 5 standard is an annual arithmetic mean, averaged over 3
years, from single or multiple community-oriented monitors.
14 The "indicator" of a standard defines the chemical species or mixture that is to be measured in
determining whether an area attains the standard.
15 All of the basic elements of the standards are discussed in the Policy Assessment (PA).
2-10
-------
suffer from potential omission of negative analyses due to 'publication bias'" (US EPA, 2004a,
p. 8-30); and (3) multi-city studies generally have higher statistical power.
With regard to selection of alternative levels for the annual PM2 5 standard to be
evaluated in this risk assessment, we first considered long-term average PM2.5 concentrations
associated with health effects observed in long-term epidemiological studies, as summarized in
Figure 2-2 of the second draft ISA. The second draft ISA concluded that the association between
increased risk of mortality and long-term PM2.5 exposure becomes more precise and consistently
positive in locations with mean PM2s concentrations of 13.5 |ig/m3 and above (US EPA, 2009a,
section 2.3.1.2). The second draft ISA also concluded that the strongest evidence for
cardiovascular-related effects related to long-term PM2 5 exposures has been reported in large,
multi-city U.S.-based studies and, specifically, one of these studies, the Women's Health
Initiative (WHI) Study, reports associations between PM2.5 and cardiovascular effects among
post-menopausal women with a mean annual average PM2 5 concentration of 13.5 |ig/m3 (US
EPA, 2009a, section 2.3.1.2). In addition, we evaluated long-term average PM2.5 concentrations
in short-term exposure studies that reported statistically significant effects. More specifically, as
reported in the second draft ISA, both cardiovascular and respiratory morbidity effects (e.g.,
emergency department visits, hospital admissions) have been observed and become more precise
and consistently positive in locations with mean PM2 5 concentrations of 13 |ig/m3 and above
(US EPA, 2009a, section 2.3.1; also see Figure 2-1).16
Based on the available epidemiological evidence indicating effects associated with a
range of annual averaged PM2.s concentrations, as briefly described above, we selected levels of
12 and 13 |ig/m3 as the alternative annual standard levels to be evaluated in the quantitative risk
assessment. Following CAS AC and public review of the first draft RA, we expanded the range
of alternative annual standard levels to include!4 |ig/m3 to provide fuller coverage for the range
of values between the current annual standard level of 15 |ig/m3 and the lowest alternative level
evaluated. Subsequent to the release of the 2nd draft RA, we further expanded the range of
alternative annual standard levels evaluated to include a level of 10 |ig/m3, consistent with
considerations presented in the first draft PA. In so doing, we recognized the increased
uncertainty associated with simulating ambient PM2 5 levels for urban study areas that would just
16 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).
2-11
-------
meet this lower standard level and consequently, the greater uncertainty in the associated risk
estimates, relative to the higher alternative annual standard levels evaluated.
In identifying alternative levels for the 24-hour PM2 5 standard to be evaluated in this risk
assessment, we considered the ambient PM2.5 levels associated with mortality and morbidity
effects as reported in key short-term epidemiological studies. We focused on the 98th percentile
PM2.5 ambient concentrations reported in two multi-city studies that provided C-R functions used
in the core risk assessment, Zanobetti and Schwartz (2009) and Bell et al. (2008). The focus on
the 98th percentile of the 24-hour PM2 5 concentrations observed in the epidemiological studies is
consistent with the approach used in the prior PM NAAQS review and is consistent with the
current form of the 24-hour PM2 5 standard.
The second draft ISA presented 98th percentile 24-hour PM2 5 values for each of the 112
urban areas included in the Zanobetti and Schwartz (2009) short-term mortality study (US EPA,
2009a, Figure 6-22). We evaluated the trend in these county-level 98th percentile 24-hour PM2 5
levels in conjunction with the statistical significance of the associated county-lev el effect
estimates. If we had found an association between the air quality levels and statistically
significant effect estimates (i.e., higher 98th percentile PM2 5 levels were consistently associated
with statistically significant effect estimates), then it would have been reasonable to consider the
lowest 98th percentile PM2 5 level associated with the set of counties for which a statistically
significant effect estimates was observed as the basis for selecting an alternative standard level
for evaluation in this risk assessment. However, no such association was observed. Rather, we
observed mixed results with no apparent correlation between 98th percentile air quality levels and
statistically significant effect estimates. Therefore, we focused on the overall range of 98th
percentile values across the entire set of counties and considered the lower quartile of that
distribution as representative of a reasonably precautionary approach for identifying alternative
levels for consideration in the risk assessment. The overall 98th percentile value across the entire
set of urban areas analyzed was 34.3 |ig/m3 (US EPA, 2009a, Figure 2-1; Zanobetti and
Schwartz, 2009). The 10th and 25th percentiles values were 25.5 and 29.8 |ig/m3, respectively
(Zanobetti, 2009). We also completed a similar analysis of the county-level ambient air quality
data for the 202 counties associated with the Bell et al. (2008) study (Bell, 2009). The overall
98th percentile value across the entire set of counties analyzed in Bell et al. (2008)) was 34.2
|ig/m3 (US EPA, 2009a, Table 6-11; Bell, 2009). The 10th and 25th percentile values were 24.4
and 29.3 |ig/m3, respectively (Bell, 2009). Based on the available epidemiological evidence
indicating effects associated with a range of 98th percentile 24-hour PM2.5 concentrations, as
briefly described above, we selected levels of 25 and 30 |ig/m3 as the alternative 24-hour
standard levels to be evaluated in this quantitative risk assessment.
2-12
-------
Once alternative levels were identified for the annual and 24-hour PM2.5 standards, we
then identified specific combinations of these standard levels to be considered in evaluating
suites of alternative standards in the risk assessment. In selecting the pairing of annual and 24-
hour standard levels, we considered which standard was likely to be controlling across the set of
15 urban study areas. Either the annual or 24-hour standard will be the "controlling standard" at
a given location, depending on the design value associated with that location.17 For this risk
assessment, the goal was to select combinations of annual and 24-hour levels that would result in
a mixture of behavior in terms of which standard would control across the various urban study
areas. For example, with the 12/35 combination (i.e., an annual standard level of 12 |ig/m3 and a
24-hour standard level of 35 |ig/m3), the annual level of 12 |ig/m3 is the controlling standard for
all 15 urban study areas. Alternatively, with the 12/25 combination, the annual standard is the
controlling standard at some locations and the 24-hour standard is the controlling standard at
other locations. Consideration of these factors resulted in a set of five alternative suites of
annual and 24-hour standards being identified for inclusion in the risk assessment.
The air quality scenarios included in the risk assessment for which we felt sufficient
information was available to generate risk estimates with a reasonable degree of confidence,
include the recent conditions air quality scenario and simulations of just meeting the current suite
of standards and five alternative suites of standards as follows:
Recent conditions (risk estimates based on ambient PM2.5 monitoring data for the
analysis period - 2005 to 2007)
Current suite of PM2.5 NAAQS: annual 15 |ig/m3; 24-hour 35 |ig/m3
Alternative suite of PM2.5 standards: annual 14 |ig/m3; 24-hour 35 |ig/m3
Alternative suite of PM2.5 standards: annual 13 |ig/m3; 24-hour 35 |ig/m3
Alternative suite of PM2.5 standards: annual 12 |ig/m3; 24-hour 35 |ig/m3
Alternative suite of PM2.5 standards: annual 13 |ig/m3; 24-hour 30 |ig/m3
Alternative suite of PM2.5 standards: annual 12 |ig/m3; 24-hour 25 |ig/m3
Because of the increased uncertainty associated with estimates of risk generated for the
alternative annual standard level of 10 |ig/m3, we did not incorporate discussion of these
estimates into the discussion of risk estimates presented in Chapter 4; instead, these risk
estimates are presented separately in Appendix J. The two suites of alternative standards based
on simulation of an alternative annual standard level of 10 |ig/m3 include:
17 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.
2-13
-------
Alternative suite of PM2.5 standards: annual 10 |ig/m3; 24-hour 35 |ig/m3
Alternative suite of PM2.5 standards: annual 10 |ig/m3; 24-hour 25 |ig/m3
2-14
-------
3 URBAN CASE STUDY ANALYSIS METHODS
This chapter provides an overview of the methods used in the risk assessment. Section
3.1 discusses the basic structure of the risk assessment, identifying the modeling elements and
related sources of input data needed for the analysis. Section 3.2 discusses air quality
considerations. Section 3.3 discusses the selection of health endpoints, urban study areas and C-
R functions from key epidemiological studies used in modeling those endpoints. Section 3.4
discusses baseline health effects incidence rates. Finally, section 3.5 describes how uncertainty
and variability are addressed in the risk assessment.
3.1 GENERAL APPROACH
3.1.1 Basic Structure of the Risk Assessment
The general approach used in both the prior and the current PM risk assessment relies
upon C-R functions which have been estimated in epidemiological studies. Since these studies
estimate C-R functions using ambient air quality data from fixed-site, population-oriented
monitors, the appropriate application of these functions in a PM risk assessment similarly
requires the use of ambient air quality data at fixed-site, population-oriented monitors.
The general PM health risk model, illustrated in Figure 3-1, combines information about
PM air quality for specific urban areas with C-R functions derived from epidemiological studies,
baseline health incidence data for specific health endpoints, and population estimates to derive
estimates of the annual incidence of specified health effects attributable to ambient PM
concentrations under different air quality scenarios. This assessment was implemented within
Total Risk Integrated Methodology - Risk Assessment component (TRIM.Risk), the component
of EPA's Total Risk Integrated Methodology (TRIM) model that estimates human health risks.18
The analyses conducted for this review focused on estimating risks associated with recent
PM2.5 air quality and estimating changes in these risks associated with air quality simulated to
reflect just meeting the current suite of PM2 5 ambient standards, as well as any additional
reductions in incidence estimated to occur upon just meeting alternative suites of PM2.5
standards.
Consistent with past risk assessments for NAAQS reviews, this risk assessment is
intended to estimate risks attributable to anthropogenic sources and activities only. Therefore, for
all health endpoints associated with short-term exposure to PM2.5, the risk assessment considers
only the incidence of health effects associated with PM2.5 concentrations in excess
18
For more detailed information about TRIM.Risk, see: http://www.epa.gov/ttn/fera/trim_risk.html
5-1
-------
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
Concentratio n-Re s p on s e
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
3-2
-------
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,
5-3
-------
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).
5-4
-------
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
3-5
-------
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
3-6
-------
The risk assessment was carried out using three years of recent air quality data from
2005, 2006, and 2007 (see section 3.2.1). We matched the population data used in the risk
assessment to the year of the air quality data. For example, when we used 2005 air quality data,
we used 2005 population estimates. It was not possible to obtain the necessary data to calculate
baseline incidence rates separately for each of the three years for each of the risk assessment
locations, therefore, we calculated these rates for a single year, under the assumption that these
rates are unlikely to have changed significantly from 2005 to 2007. The calculation of baseline
incidence rates is described in detail in section 3.4.
For this risk assessment, we developed a core (primary) set of risk results based on the
application of modeling element choices (e.g., C-R functions, lag periods) that we believe have
the greatest overall support in the literature (hereafter referred to as the "core" results). While it
is not possible at this time to assign quantitative levels of confidence to these core risk estimates,
we do believe these estimates are generally based on inputs having higher overall levels of
confidence relative to risk estimates that could have been generated using other inputs identified
in the literature.
In addition, as discussed above in section 2.1 and later in section 3.5, we have also used
single-element and multi-element sensitivity analysis techniques to generate a set of reasonable
alternative risk estimates based on the application of alternative modeling element choices that,
while not having as much support in the literature as those used in the core analysis, do still
represent plausible inputs. The results of these sensitivity analyses allow us to gain insights into
which sources of uncertainty and variability may have the greatest impact on risk estimates when
acting alone, or in combination with other sources of uncertainty. The sensitivity analysis-based
risk estimates also provide us with an additional set of reasonable risk results that allow us to
place the results of the core analysis in context with regard to uncertainty. A number of
modeling elements were used in differentiating core analyses from sensitivity analyses (e.g., C-R
function shape, alternative effect estimates, alternative lag structures, different methods used to
rollback air quality to simulate attainment to current or alternative standard levels, application of
PRB versus LML). Specific choices made in relation to individual modeling elements in
differentiating the core analysis from sensitivity analyses are described, as appropriate, in the
sections that follow, which cover specific aspects of the risk assessment design. The potential
utility of the sensitivity analysis-based risk estimates in informing consideration of uncertainty
and variability in the core results is discussed in section 4.3.2.
3.1.2 Calculating PM2.5-Related Health Effects Incidence
The C-R functions used in the risk assessment are empirically estimated relations
between average ambient concentrations of PM2.5 and the health endpoints of interest (e.g.,
5-7
-------
mortality or hospital admissions reported by epidemiological studies for specific locations). This
section describes the basic method used to estimate changes in the incidence of a health endpoint
associated with changes in PM2 5, using a "generic" C-R function of the most common functional
form.
Although some epidemiological studies have estimated linear C-R functions and some
have estimated logistic functions, most of the studies used a method referred to as "Poisson
regression" to estimate exponential (or log-linear) C-R functions in which the natural logarithm
of the health endpoint is a linear function of PM2 5:
(1)
where x is the ambient PM2.5 level, y is the incidence of the health endpoint of interest at
PM2.5 level x, p is the coefficient of ambient PM2.5 concentration, and B is the incidence atx=0,
i.e., when there is no ambient PM2.5. The relationship between a specified ambient PM2.5 level,
XQ, for example, and the incidence of a given health endpoint associated with that level (denoted
asj/o) is then
(2)
Because the log-linear form of a C-R function (equation (1)) is by far the most common
form, we use this form to illustrate the "health impact function" used in the PM2.5 risk
assessment.
If we let x0 denote the baseline (upper) PM2 5 level, and x} denote the lower PM2 5 level,
and.yoand.y7 denote the corresponding incidences of the health effect, we can derive the
following relationship between the change in x, Ax= (XQ- xj), and the corresponding change my,
Ay, from equation (I).19
*y = (y0-y1) = y0V-e-p&x]. (3)
Alternatively, the difference in health effects incidence can be calculated indirectly using
relative risk. Relative risk (RR) is a measure commonly used by epidemiologists to characterize
the comparative health effects associated with a particular air quality comparison. The risk of
19 If Ar < 0 - i.e., if Ar = (xr x0) - then the relationship between Ax and Ay can be shown to be
Ay = (yl - y0) = y0[e13^ - 1]. If Ax < 0, Ay will similarly be negative. However, the magnitude of Ay will be the
same whether Ar>OorAr<0- i.e., the absolute value of Ay does not depend on which equation is used.
-------
mortality at ambient PM2.5 level x0 relative to the risk of mortality at ambient PM2.5 level xj, for
example, may be characterized by the ratio of the two mortality rates: the mortality rate among
individuals when the ambient PM2 5 level is x0 and the mortality rate among (otherwise identical)
individuals when the ambient PM2 5 level is xj. This is the RR for mortality associated with the
difference between the two ambient PM2 5 levels, XQ and xj. Given a C-R function of the form
shown in equation (1) and a particular difference in ambient PM2.5 levels, Ax, the RR associated
with that difference in ambient PM2.5, denoted as RR-Ax, is equal to epAx. The difference in health
effects incidence, Ay, corresponding to a given difference in ambient PM2 5 levels, Ax, can then
be calculated based on this RRAx as:
(4)
Equations (3) and (4) are simply alternative ways of expressing the relationship between
a given difference in ambient PM2 5 levels, Ax > 0, and the corresponding difference in health
effects incidence, Ay. These health impact equations are the key equations that combine air
quality information, C-R function information, and baseline health effects incidence information
to estimate ambient PM2 5 health risk.
3.1.2.1 Short-term vs. Long-term Exposure
Concentration-response (C-R) functions that use as input annual average PM2 5 levels (or
some function of these, such as the average over a period of several years) relate these to the
annual incidence of the health endpoint - i.e., in such studies x in equation (1) above is the
average PM2.5 concentration over a period of one or more years, meant to represent long-term
exposure, and_y is the annual incidence of the health effect associated with that long-term
exposure.
Concentration-response (C-R) functions that use as input 24-hour average PM2.5 levels (or
some function of these, such as the average over one or more days) relate these to the daily
incidence of the health endpoint - i.e., in such studies x in equation (1) above is the average
PM2.5 concentration over a period of one or a few days (short-term exposure), and .y is the daily
incidence of the health effect associated with that short-term exposure.
There are several variants of the short-term (daily) C-R function. Some C-R functions
were estimated by using moving averages of ambient PM2.5 to predict daily health effects
incidence. Such a function might, for example, relate the incidence of the health effect on day t
to the average of PM2.5 concentrations on days t and (M). Some C-R functions consider the
relationship between daily incidence and daily average PM2 5 lagged a certain number of days.
For example, a study might estimate the C-R relationship between mortality on day t and average
5-9
-------
PM2.5 on a prior day (M). A few studies have estimated distributed lag models, in which health
effect incidence is a function of PM2.5 concentrations on several prior days - that is, the incidence
of the health endpoint on day Hs a function of the PM2 5 concentration on day t, day (/-I), day (t-
2), and so forth. Such models can be reconfigured so that the sum of the coefficients of the
different PM2.5 lags in the model can be used to predict the changes in incidence on several days.
For example, corresponding to a change in PM on day tin a distributed lag model with 0-day, 1-
day, and 2- day lags considered, the sum of the coefficients of the 0-day, 1-day, and 2-day lagged
PM2.5 concentrations can be used to predict the sum of incidence changes on days t, (t+l) and
(t+2).
Most daily time-series epidemiological studies estimated C-R functions in which the PM-
related incidence on a given day depends only on same-day PM concentration^.e. lag 0), the
previous-day PM concentration (i.e. lag 1), or some variant of those, such as a two-day average
concentration (e.g. lag 0-1). Such models necessarily assume that the longer pattern of PM
levels preceding the PM concentration on a given day does not affect mortality or morbidity on
that day. To the extent that PM-related mortality on a given day is affected by PM concentrations
over a longer period of time, then these models would be mis-specified, and this mis-
specification would affect the predictions of daily incidence based on the model.
The extent to which time-series studies using single-day PM2.5 concentrations may under
or over-estimate the relationship between short-term PM2 5 exposure and risk of mortality is
unknown. However, there is some evidence, based on analyses of PMio data, that mortality or
morbidity on a given day is influenced by prior PM exposures up to more than a month before
the date of death (Schwartz, 2000). The extent to which short-term exposure studies (including
those that consider distributed lags) may not capture the full impact of long-term exposures to
PM2.5 is similarly not adequately understood, although the current evidence (e.g., Krewski et al.,
2009; Krewski et al., 2000) suggests that there is a substantial impact of long-term exposures on
health effects that is not picked up in the short-term exposure studies.
3.1.2.2 Calculating Annual Incidence
The risk assessment estimated health effects incidence, and changes in incidence, on an
annual basis, for 2005, 2006, and 2007. For mortality, both short-term and long-term exposure
studies have reported estimated C-R functions. As noted above, most short-term exposure C-R
functions estimated by daily time-series epidemiological studies relate daily mortality to same-
day PM2.5 concentration or previous-day PM2.5 concentration (or some variant of those).
To estimate the daily health impacts of 24-hour average ambient PM2 5 levels above PRB,
C-R functions from short-term exposure studies were used together with estimated changes in
24-hour ambient PM2.5 concentrations to calculate the daily changes in the incidence of the
3-10
-------
health endpoint. After daily changes in health effects were calculated, an annual change was
calculated by summing the daily changes.
As part of the PM NAAQS RA completed in 1996, we demonstrated that the majority of
mortality incidence due to short-term exposure (aggregated over a year) was driven by days with
24-hour average PM2.5 levels nearer to the annual average value for the study area rather than
days with relatively higher PM2.5 levels falling in the tail of the annual 24-hour PM2.5
distribution.20 This finding reflects the fact that the number of deaths associated with short-term
exposure to PM2.5 depends both on the number of days at a given concentration and on the
concentration itself. Because the urban areas considered in the 1996 RA had 24-hour PM2.5
distributions that were closer to normal or log-normal in form (i.e., not uniform), overall
incidence of short-term exposure-related mortality was driven by the relatively large number of
days near the center of the distribution, rather than the small number of days out at the tail. This
analysis was updated for the last review completed in 2006.21
As part of the current analysis, we have updated the analysis originally presented in the
1996 PM NAAQS RA to reflect the air quality data and mortality estimates due to short-term
exposure generated as part of the current review. For this updated analysis, we have focused on
Detroit and New York and on cardiovascular mortality as the short-term exposure mortality-
related health effect category.22 The results of this analysis corroborate the findings of the earlier
analysis from 1996 (see Figures 1-1 and 1-2 in Appendix I). Specifically, these figures
demonstrate that for these two urban study areas, a large fraction of short-term exposure-related
cardiovascular mortality incidence is associated with days around the mean of each distribution
(13.9 |ig/m3 for Detroit and 13.8 |ig/m3 for New York - see Appendix A, Table A-5 and A-9,
respectively), rather than days comprising the upper tail of each distribution. In other words,
days with relatively elevated PM2.5 levels contribute a relatively small fraction of short-term
exposure-related cardiovascular mortality incidence at each study area.
The mortality associated with long-term exposure is likely to include mortality related to
short-term exposures as well as mortality related to longer-term exposures. As discussed
previously, estimates of daily mortality based on the time-series studies also are likely influenced
by prior PM exposures. Therefore, the estimated annual incidences of mortality calculated based
on the short- and long-term exposure studies are not likely to be completely independent and
should not be added together. While we can characterize the statistical uncertainty surrounding
20 See Exhibit 7.6 on p. 79 of the 1996 PM NAAQS RA (Abt Associates Inc., 1996).
21 See Figure 4-10 on p. 4-68 of the 2005 PM Staff Paper (US EPA, 2005).
22 As discussed in the introduction to Chapter 4, we have focused on cardiovascular-related endpoints in
summarizing risk estimates for this analysis because this endpoint category (including both cardiovascular-related
mortality and morbidity) has the greatest degree of support in the literature.
3-11
-------
the estimated PM2.5 coefficient in a reported C-R function, there are other sources of uncertainty
associated with the C-R functions used in the risk assessment that are addressed via sensitivity
analyses and/or qualitatively discussed in section 3.5.3.
3.2 AIR QUALITY INPUTS
3.2.1 Characterizing Recent Conditions
Twenty-four hour PM2.5 air quality data for 2005, 2006, and 2007 were obtained for each
of the urban study areas from monitors in EPA's Air Quality System (AQS).23 To characterize
PM2.5 air quality in each risk assessment location as accurately as possible, we used only those
monitors that were located within the county or counties that were analyzed in the
epidemiological studies used to select C-R functions. In a few cases, an urban area was
delineated differently by two or more epidemiological studies used in the risk assessment. For
example, Birmingham, AL was defined as Blount, Jefferson, Shelby, St. Clair, and Walker
Counties in one study and as only Jefferson County in another study. In such cases, we matched
our delineation of the urban study area to that used in each study, resulting in two or more
different delineations of the urban study area and identified them as, for example, Birmingham 1
and Birmingham 2. The counties and the number of air quality monitors included within each
urban area are given in Table 3-1.
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
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
Number of Monitors
8
7
10
8
10
8
6
9
3
6
10
12
7
23 The specific sets of air quality monitoring data for each of the urban study areas are available in the
docket (Docket ID#: EPA-HQ-OAR-2007-0492) and have been posted at: http://www.epa.gov/ttn/analysis/pm.htm.
3-12
-------
Risk Assessment
Location
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO - 1
St. Louis, MO - 2
Tacoma, WA
Counties
Maricopa
Allegheny
Salt Lake
Jefferson, Madison (IL), St.
Madison (IL), St. Louis, St.
Louis, St. Louis City, St. Clair (IL)
Louis City, St. Clair (IL)
Pierce
Number of Monitors
5
12
7
15
14
1
* 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.
**** por 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).
In order to be consistent with the approach generally used in the epidemiological studies
that estimated PM2 5 C-R functions, rather than working directly with individual monitor values
in estimating risk, we first derived composite monitor estimates (i.e., a composite of individual
monitor values for a given study area) and then used those to represent population exposure in
the risk assessment. Two types of composite monitor values were derived including annual
estimates (used in modeling long-term exposure-related mortality) and distributions of 24-hour
average levels (used in modeling short-term exposure-related mortality and morbidity). The
procedure for deriving each of these types of composite monitor estimates is described below.
The approach for creating composite monitors used in this risk assessment reflects the goal of
providing equal weighting of monitors in computing both 24-hour and annual composite monitor
values.24
Composite monitor distributions of 24-hour estimates
To develop composite monitor distributions of 24-hour estimates reflecting equal
weighting of the underlying monitor datasets, we completed an initial step of interpolation to fill
in missing measurements at the individual monitors. We then calculated the average for a
particular day across the contributing monitors to develop a 24-hour PM distribution reflecting
equal weighting of the monitors. The specific step-wise procedure involved:
24 This reflects a change from the approach used in the first draft RA which weighted monitors by sampling
frequency - an approach which could result in bias being introduced into the analysis, in those instances where
monitors with higher measurements were sampled more frequently.
3-13
-------
1. Interpolate missing days at individual monitors: If a monitor had fewer than 11
measurements in a calendar quarter, we left those days "as is" and did not conduct any
interpolation. However, for those quarters with 11 or more measurements, we did interpolate
missing values using the following procedure. For the day(s) (up to a string of seven in a
row) without data, we took the average of the nearest day with measured data above and
below the missing day(s) and assigned that interpolated value to that day(s). Note, however,
that if the string of missing days was longer than seven, we did not interpolate and left those
days blank.25
2. Derive distribution of 24-hour measurements for the composite monitor: For each day in the
simulation year, the composite monitor value is calculated as the average of the daily
estimates across monitors in that study area. If a monitor is missing a measurement on a
particular day (after interpolation described in Step 1), it is not factored into the composite
estimate for that day. If after completing this step, any composite monitor days are missing
values (reflecting the instance where the underlying monitors in that study area were all
missing estimates for a specific day, even after interpolation described in step 1), then a
composite monitor value is interpolated for that day using a 7-day moving average (i.e., the
average of the 3 composite daily estimates before that day, the day on record, and the three
after it) to represent that missing composite 24-hour estimate.
Composite monitor annual average estimates
Composite monitor annual averages were calculated using an approach that did not
require interpolation of missing days at individual monitors (i.e., Step 1 above). Instead, we used
quarterly averages at each monitor within a study area as the basis for deriving composite annual
average estimates. The specific step-wise procedure involved:
1. Calculate quarterly averages at each monitor in the study area: For a given monitor, if a
quarter has less than 11 measurements, we classify that quarter as "missing" and do not use it
in computing the quarterly estimate. For the remaining quarters across the monitors with 11
25 There are a variety of approaches that can be used to interpolate missing data as part of creating
composite monitor 24-hour distributions. While the approach used in this analysis relies on data from the specific
monitor for which interpolation is being conducted, alternative approaches can utilize trend data from across all of
the monitors in a given study area. The presence of a variety of interpolation approaches from which to choose does
represent a source of uncertainty impacting the interpolation process and consequently the risk assessment. To
further examine this source of uncertainty, we have completed a sensitivity analysis for a single location
(Birmingham), in which we apply both the interpolation technique used in the risk assessment, as well as an
alternative approach that utilizes trend data from all of the Birmingham monitors to interpolate missing data at
specific monitors (this sensitivity analysis is fully described in Appendix B, section Bl). The results of this
sensitivity analysis suggest that, in the context of Birmingham, these different approaches for interpolating missing
monitoring data did not produce substantially different results. Furthermore, the small differences that were seen in
the 24-hour PM2 5 distributions generated using these two interpolation approaches, would translate into negligible
differences in short-term exposure-related risk estimates (which utilize the 24-hour PM2 5 distributions involved in
the interpolation). Therefore, we conclude, based on the results of this sensitivity analysis, that uncertainty related
to interpolation of missing data represents a relatively small source of uncertainty in the overall analysis.
3-14
-------
or more measurements, we calculate a quarterly estimate for each quarter/monitor
combination as the average of the existing values (i.e., no interpolation of daily estimates at
individual monitors was completed).
2. Calculate quarterly averages at the composite monitor. For each quarter of the year, we then
calculated the composite monitor quarterly average as the average of the monitor-specific
quarterly averages for that study area.
3. Calculate the annual average for the composite monitor: We then averaged the four
quarterly-average estimates (generated in Step 2) to produce an annual average for the
composite monitor.26
Appendix A summarizes the PM2.5 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 for individual monitors. Appendix A also
provides summary information on the composite monitor annual average estimates and 98th
percentile values generated for each study area.
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, US 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.
26 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).
3-15
-------
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
3.2.3 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 Code of Federal Regulations (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.27
In the last PM risk assessment, just meeting the current or alternative PM2.5 standards was
simulated using a single rollback approach (the proportional) which reflected a uniform regional
pattern of reduction in ambient PM2 5 levels across monitors. For this analysis, we have included
two additional rollback approaches to provide greater coverage for variability associated with
this key aspect of simulating risk for both the current and alternative standard levels. These two
27 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.
3-16
-------
approaches include the hybrid and the locally focused rollback methods. The hybrid rollback
approach involves a combination of an initial step of a more localized reduction in ambient PM2.5
levels at source-oriented monitors followed by a regional pattern of reduction across all monitors
in a study area. The locally focused rollback approach involves a focused reduction of levels
only at those monitors exceeding the daily standard level under consideration. While the
proportional rollback approach is applied to all 15 urban study areas included in the analysis,
both the hybrid and locally focused rollback approaches are applied to a subset of the study areas
meeting specific criteria as outlined below. Each of the three rollback methods is described
below including a description of the step-wise procedure used in implementing each approach.
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,
depending on the type of rollback method used, just meeting the current or alternative PM2.5
standards was simulated by changing 24-hour PM2.5 concentrations at the composite monitor (for
the proportional rollback approach) or by adjusting values at each monitor separately prior to
generating composite-monitor estimates (for the hybrid and locally focused rollback methods).
This change was made because the hybrid and locally focused rollback methods involve non-
uniform degrees of reduction in ambient PM2.5 measurements for the monitors in a given study
area, as contrasted with the proportional approach which involves the same percent reduction
across all monitors in a study area.
The proportional rollback approach was used in generating the core risk estimates in light
of its use in past risk assessments, while the other two rollback approaches (hybrid and locally
focused) have been included as part of the sensitivity analyses to characterize potential
variability in the way urban areas may respond to suites of current or alternative standards. In
considering the three rollback methods collectively, the proportional and locally focused
methods represent approaches more likely to capture "bounding" behavior related to the spatial
pattern of future reductions in ambient PM2.5 levels. By contrast, the hybrid approach can be
interpreted as reflecting a more plausible or representative rollback strategy in principle, since it
(a) reflects consideration for site-specific information regarding larger PM sources and their
potential impact on source-oriented monitors and (b) combines elements of more locally focused
and regionally-focused patterns of reduction.28 However, it is important to note that the hybrid
28 CASAC in providing comments on the 2nd External Review Draft RA placed greater confidence in the
hybrid rollback method relative to the other two methods, identifying both the proportional and locally focused (then
referred to as locally focused) - this edit is wrong- as representing potential bounding approaches and therefore
warranting less focus than the hybrid (REFERENCE)
3-17
-------
approach as implemented is only one of a variety of potential hybrid strategies for considering
site-specific data in simulation spatial patterns of ambient PM2.5 reduction. Because there are a
variety of hybrid rollback strategies that could be considered and because of challenges in
assessing the reasonableness of different strategies in predicting spatial patterns of ambient PM2.5
reductions, we conclude that there is substantial uncertainty associated with predicting rollback
in relation to modeling risk for both the current and alternative standard level. Consequently,
relative importance is placed on the sensitivity analysis examining the issue of rollback, which is
based on application of the three rollback strategies described here. While we describe how the
different rollback methods are implemented in this section, the impact of using different rollback
methods on core risk estimates is discussed as part of the sensitivity analysis presented in section
3.5.4.1.
3.2.3.1 Proportional Rollback Method
The proportional approach involves reducing PM2.5 concentrations by the same
percentage across all monitors in a study area, thereby reflecting a more regional pattern of
ambient PM2.5 reduction. When this approach is used, it does not matter whether (1) PM2.5
concentrations are first rolled back by the same percentage each day at each monitor, and then
the composite monitor values are calculated from these monitor-specific values or (2) first the
composite monitor values are calculated and then these are rolled back by the same percentage
each day - the results will be the same. Therefore, to streamline the analytical process, the
proportional rollback method was applied directly to the composite monitor estimates. The step-
wise procedure used in conducting proportional rollback is described below:
1. Calculate annual and 24-hour design values for the study area under consideration: The
degree of reduction required to simulate attainment of a specific suite of standard levels is
determined by comparing the design values (described here) against the specific standard
level being considered. Therefore, the first step is to calculate 24-hour and annual design
values for the study area.29 The annual design value (in |ig/m3) was calculated as follows:
At each monitor, the annual average PM2.5 concentration was calculated for each of
the years 2005, 2006, and 2007, and these three annual average concentrations were
then averaged.
The maximum of these monitor-specific 3-year averages of annual averages at a
particular study area is the annual design value, denoted dvamuai,
The 24-hour design value (in |ig/m3) was similarly calculated as follows:
29 Note, that as discussed later in section 3.1.3.2, the second phase of the hybrid rollback method involving
proportional reduction also uses the design values for a study area to determine the degree of rollback required.
3-18
-------
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.
-,th
The maximum of these monitor-specific 3-year averages of 98 percentile
concentrations at a particular study area is the 24-hour design value, denoted dvdaily 95
(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
Detroit
Fresno
Houston
Los Angeles
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
St. Louis
Tacoma
Annual
(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
(uS/m3)
35
37
44
26
43
63
31
55
42
38
32
60
55
39
43
3-19
-------
*The calculation of design values is explained in the text above.
2. Calculate the percent reduction required to simulate attainment of the 24-hour and annual
standard levels under consideration: The degree of reduction required to simulate attainment
of a particular standard level is calculated by comparing the standard level under
consideration with its matching design value to determine the percent reduction required to
bring the composite monitor values down to match the standard level. Because pollution
abatement methods are applied largely to anthropogenic sources of PM2.5, rollbacks are
applied only to PM2.5 above estimated PRB levels and consequently, in determining the
percent reduction, we only consider that portion of both the design value and standard level
above PRB. The specific equations used to estimate the percent reduction required to
simulate attainment (with consideration of PRB) are presented as part of a detailed example
of the three rollbacks (as applied to Detroit) in Appendix B, section B3.
3. Determine which standard is controlling for the study area under consideration: The percent
reduction required to simulate attainment of the 24-hour and annual standard levels
(calculated in Step 2) are compared and the larger of the two values is identified as the
controlling standard. Simulated attainment of the controlling standard, by definition, results
in more than sufficient reduction in composite monitor values to produce simulated
attainment of the other (non-controlling) standard level.
4. Apply proportional reduction to the composite monitor values: The percent reduction for the
controlling standard identified in Step 3 above is applied (a) to each 24-hour estimates at the
composite monitor to generate an adjusted 24-hour PM2.5 composite monitor distribution to
be used in modeling short-term exposure-related risk and (b) directly to the annual average at
the composite monitor to generate an adjusted estimate that can be used in estimating long-
term exposure-related mortality.
3.2.3.2 Hybrid Rollback Method
The hybrid rollback approach reflects a combination of a more localized pattern of
rollback focused on source-oriented monitors with relatively elevated ambient PM2.5 levels,
followed by a more generalized regional pattern of rollback across all monitors in the study area
to simulate attainment. The first localized reduction involves reducing levels at the selected
source-oriented monitor(s) such that they match the level of the nearest non-source oriented
monitor(s). This initial localized reduction includes a distance-decay impact on other monitors in
the study area (i.e., monitors further from the source-oriented monitor experience a decreasing
fraction of the reduction experienced by the targeted source-oriented monitor(s)). The second
more generalized regional rollback is implemented using the proportional approach described
above in section 3.2.3.1. However, for the hybrid rollback, the percent reduction is determined
based on consideration of design values calculated after the initial localized reduction phase,
rather than design values based on recent conditions as is the case with proportional rollback
3-20
-------
approach described in 3.2.3.1. Because the hybrid rollback approach requires that specific
criteria be met for a study area to be a candidate (i.e., presence of source-oriented monitors with
clearly identifiable PM sources near-by and elevated PM2 5 levels relative to other monitors in the
study area), this approach was only applied to a subset of the 15 urban study areas that met these
criteria including: Baltimore, Birmingham, Detroit, Los Angeles, New York, and St. Louis.30
The step-wise procedure used to implement the hybrid approach is described below (additional
detail on sample calculations associated with the steps presented below can be found in
Appendix B, section B3 as part of the sample calculations provided for the three rollback
methods).
1. Identify candidate urban study areas: The subset of the 15 urban study areas with high
design values (exceeding the current suite of standards) where those design value monitors
were in close proximity to large sources of PM2.5 were identified as candidate locations.
2. Localized reduction ofPM levels at that source-oriented monitor to meet levels at nearby
non-source oriented monitors: PM levels at the source-oriented monitor identified in Step 1
were reduced such that their design value (after adjustment) matched the design value for the
non-source oriented monitor with the highest design value located close by. This process was
repeated for all source-oriented monitors identified in step 1. This reduction focused on the
design value that was controlling for the source-oriented monitor (i.e., for that specific
monitor, the design value - annual or 24-hour - requiring the greatest percent reduction to
meet the standard). A proportional reduction of all 24-hour estimates at the source-oriented
monitor was conducted to have the design value for this monitor match that for the nearest
non-source oriented monitor.
3. Simulate the impact of the localized reduction ofPM levels at the source-oriented monitor on
levels at other monitors in the study area: While the localized reduction described in Step 2
does primarily impact the source-oriented monitor(s) that is the focus of that reduction, we
do consider the potential for that reduction to impact other monitors in the study area, albeit
with reduced impact the further you move away from the targeted source-oriented monitor.
Specifically, those monitors within one kilometer of the source-oriented monitor were
assigned the same proportional percent reduction as the source-oriented monitor. However,
monitors more than a kilometer away experience a distance-decay effect equal to the percent
rollback multiplied by the inverse of the distance in kilometers between the monitors (e.g., a
30 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 Section 3.2.3.1; 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).
3-21
-------
monitor 10 kilometers away from the source-oriented monitor, would have 1/10th of the
percent-reduction applied to that source-oriented monitor).
4. Conduct proportional rollback to simulate PM levels meeting the suite of standards under
consideration given the initial localized reduction focused on source-oriented monitors: A
regional proportional rollback identical to that described in section 3.1.3.1 is now conducted
except that in this case, we are starting with PM2.5 levels reflecting the initial localized
reduction described above in Steps 1-3 above. Note, that just as with the proportional
rollback approach described in section 3.2.3.1, we begin by first calculating composite
monitor annual average estimates and composite monitor distributions of 24-hour levels
(based on the adjusted PM levels at individual-monitors) and then apply proportional
rollback directly to those composite monitor values.
Additional detail on the hybrid approach, as applied specifically to the Detroit study area, is
presented in Appendix B (sections B2 and B3), including identification of the source-oriented
monitors targeted for focused reduction in the first step of the rollback process.
3.2.3.3 Locally focused Rollback Method
The locally focused rollback approach reflects a local pattern of reduction in ambient
PM2.5 concentrations focused exclusively on those monitors within urban study areas assessed to
exceed the 24-hour standard under consideration. As such, this approach is only considered for
the subset of the 15 urban study areas where the 24-hour standard is controlling and there is no
adjustment to monitors besides those exceeding the 24-hour standard (i.e., no distance-decay
effect as implemented in the hybrid approach). This approach was applied to a subset of the 15
urban study areas meeting the above criteria, including: Baltimore, Detroit, Fresno, Los Angeles,
New York, Philadelphia, Pittsburgh, Salt Lake City, St. Louis and Tacoma. The step-wise
procedure used to implement the hybrid approach is described below (as with the other two
rollback approaches, equations and sample calculations are provided in Appendix B, section B3).
1. Identify candidates for locally focused rollback: Identify the subset of the 15 urban study
areas where the 24-hour standard is controlling. These locations will, by definition, have
monitors with design values exceeding the current standard and consequently are candidates
for locally focused rollback.
2. Determine the degree of reduction required (at each monitor exceeding the 24-hour
standard) to bring that study area into simulated attainment: For each monitor with a 24-
hour design value exceeding the standard under consideration, compare that design value to
the standard under consideration to determine the degree of reduction required to bring that
monitor into simulated attainment (i.e., the percent rollback). As with the proportional
rollback described in section 3.1.3.1, calculation of the percent rollback takes into
consideration PRB and is based on comparing only those portions of design values and
standard levels above PRB.
3-22
-------
3. Rollback individual monitors to meet the 24-hour standard level: Based on the percent
rollback values calculated for individual monitors within each study area in Step 2, adjust the
distribution of 24-hour PM2.5 levels at individual monitors such that their adjusted design
values now meet the 24-hour standard under consideration.
4. Calculate composite monitor 24-hour distributions and annual averages for each study area:
Using the adjusted 24-hour distributions created in Step 3, calculate composite monitor 24-
hour distributions and annual averages for each study area using the approach outlined in
section 3.1.1.31
3.2.3.4 Presentation of Results for the Three Rollback Methods (with example
calculation)
The results of applying the three rollback methods in simulating attainment of the current
and alternative suites of standard levels are presented in Table 3-4 (as noted above, the hybrid
and locally focused methods are only applied to a subset of the study areas). In summarizing the
composite monitor values generated using the three rollback methods, we have included two
types of annual averages: (a) the maximum monitor-specific three-year (2005-2007) annual
average (i.e., "Max. M-S" in both tables) and (b) the composite monitor value for 2007 (i.e.,
"2007 CM" in both tables). The first estimate (Max M-S) allows us to see how the design value
changes in just meeting each suite of standards based on application of the different rollback
methods, while the second estimate (2007 CM) is the surrogate for long-term exposure-related
mortality, as described below in section 3.5.4. As is expected, the Max M-S value is consistently
larger than the CM value for a given combination of urban study area, rollback method and
standard level simulated. In reviewing the results presented in Table 3-4, we see that both the
hybrid and locally focused rollback methods generate larger Max M-S and CM values than the
proportional approach, with the locally focused approach generally resulting in the highest
values of the three rollback methods. This is expected, since both the hybrid and locally focused
rollback methods target a subset of monitors, thereby leaving more of the monitor-signal at a
given study area "unadjusted" compared with the proportional rollback method. The locally
focused rollback method, since it targets only those monitors exceeding the 24-hour standard
with no impact on other non-exceedence monitors, would be expected to have the highest
31 As with the proportional and hybrid rollback methods, rolled back composite monitor values in
Pittsburgh using the locally focused 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.
3-23
-------
remaining annual average composite monitor estimates of the three methods and this is generally
borne out in the values presented in Table 3-4.
To enhance transparency, we have included in Appendix B, section B3, a more detailed
example calculation of the three rollback methods as applied to a single urban study area
(Detroit), showing the step-wise procedure applied to individual monitors as appropriate,
including equations used, input values and sample calculations.
3-24
-------
Table 3-4. Application of the Three Rollback Methods in Simulating Current and Alternative Standard Levels for
the 15 Urban Study Areas (including resulting maximum monitor-specific and composite monitor PM2.s
values)
Risk
Assessment
Location 1
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles,
Rollback
Method
Proportional
Hybrid 3
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Design Value
Annual
16.2
15.6
18.7
12.8
17.2
17.4
15.8
19.6
24-
Hr
35.0
37.0
44.0
26.0
43.0
63.0
31.0
55.0
Recent Air
Quality
(2007)
2007 CM
15.3
13.9
15.7
11.4
13.9
17.4
13.2
14.6
Maximum Monitor-Specific Avg. of 2005, 2006, 2007 Annual Avgs. (Max. M-S) and 2007 Annual
Average at Composite Monitor (2007CM) (in ug/m3)
15/35 2
Max.
M-S
15.0
14.8
14.3
15.2
15.0
15.0
12.8
14.1
13.2
14.1
9.9
10.1
15.0
12.7
2007
CM
14.2
13.1
13.0
13.6
12.7
14.2
11.4
11.4
11.7
12.6
9.9
10.3
12.5
9.5
14/35
Max.
M-S
14.0
14.0
14.0
14.0
14.0
12.8
14.0
13.2
9.9
10.1
14.0
12.7
2007
CM
13.3
12.5
12.7
11.8
13.2
11.4
11.4
11.7
9.9
10.3
11.7
9.5
13/35
Max.
M-S
13.0
13.0
13.0
13.0
13.0
12.8
13.0
13.0
9.9
10.1
13.0
12.7
2007
CM
12.3
11.6
11.8
11.0
12.3
11.4
10.6
11.5
9.9
10.3
10.9
9.5
12/35
Max.
M-S
12.0
12.0
12.0
12.0
12.0
12.0
12.0
12.0
9.9
10.1
12.0
12.0
2007
CM
11.4
10.7
10.9
10.2
11.4
10.7
9.8
10.6
9.9
10.3
10.1
9.0
13/30
Max.
M-S
13.0
12.7
12.3
13.1
13.0
13.0
12.8
12.2
11.4
12.2
8.6
8.8
13.0
10.9
2007
CM
12.3
11.3
11.2
12.0
11.0
12.3
11.4
9.9
10.1
11.0
8.6
8.9
10.9
8.2
12/25
Max.
M-S
11.8
14
10.7
10.3
11.0
11.1
11.3
12.3
12.0
10.2
9.6
10.2
7.3
7.4
12.0
9.2
2007
CM
11.2
11.76
9.5
9.4
10.0
9.4
10.7
11.4
10.7
8.3
8.5
9.2
7.3
7.5
10.1
7.0
5-25
-------
Risk
Assessment
Location 1
CA
New York, NY
Philadelphia,
PA
Phoenix, AZ
Pittsburgh, PA 5
Salt Lake City,
UT
St. Louis, MO
Tacoma, WA
Rollback
Method
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Proportional
Hybrid
Locally focused
Design Value
Annual
15.9
15.0
12.6
19.8
11.6
16.5
10.2
24-
Hr
42.0
38.0
32.0
60.0
55.0
39.0
43.0
Recent Air
Quality
(2007)
2007 CM
13.8
13.4
9.9
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)
15/35 2
Max.
M-S
13.3
13.9
13.3
13.6
14.3
13.9
15.5
12.6
13.3
15.6
7.7
10.8
14.9
15.0
16.5
8.4
8.5
2007
CM
10.5
12.1
11.6
11.8
13.3
12.3
13.0
9.9
11.6
13.2
7.5
9.7
12.9
13.5
14.1
8.0
8.0
14/35
Max.
M-S
13.3
13.9
13.3
13.6
14.3
13.9
15.5
12.6
13.3
15.6
7.7
10.8
14.0
14.0
8.4
8.5
2007
CM
10.5
12.1
11.6
11.8
13.3
12.3
13.0
9.9
11.6
13.2
7.5
9.7
12.1
12.6
8.0
8.0
13/35
Max.
M-S
13.0
13.9
13.0
13.0
13.0
12.6
12.8
15.3
7.7
10.8
13.0
13.0
8.4
8.5
2007
CM
10.3
12.1
11.3
11.3
11.6
9.9
11.2
11.8
7.5
9.7
11.3
11.7
8.0
8.0
12/35
Max.
M-S
12.0
___
12.0
12.0
12.0
12.0
11.8
15.3
7.7
10.8
12.0
12.0
8.4
8.5
2007
CM
9.5
10.4
10.4
10.7
9.4
10.5
11.2
7.5
9.7
10.4
10.8
8.0
8.0
13/30
Max.
M-S
11.5
12.0
11.5
11.7
12.3
11.9
14.1
11.8
12.2
11.5
15.6
6.7
10.8
12.8
13.0
14.2
7.4
7.4
2007
CM
9.1
10.6
10.0
10.2
11.6
10.7
11.3
9.3
9.7
10.0
11.4
6.6
8.8
11.1
11.7
12.4
7.0
7.0
12/25
Max.
M-S
9.6
10.1
9.7
9.8
10.3
10.0
11.8
9.9
10.2
9.7
13.9
5.7
9.1
10.8
11.0
11.9
6.3
6.3
2007
CM
7.7
9.1
8.4
8.5
9.8
9.0
9.5
7.8
9.0
8.4
9.6
5.6
7.7
9.3
9.9
10.4
6.0
6.0
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 PlVb.s (Zanobetti and Schwartz, 2009) is included.
2 The current primary PlVb.s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
5-26
-------
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 locally focused 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 locally
focused method was not applied.
5 The proportional, rollback and locally focused methods were applied to Pittsburgh differently from the way they were applied in the other locations. See text for
details.
6 Percent reduction in composite monitor value with consideration of LML of 5.8 ug/m3 (note: composite monitor value denoted as CMV):
%redUCtion = (CM Vcurrent standard - CMValternative standard)/(CMVCurrernt standard-LML).
5-27
-------
3.3 SELECTION OF MODEL INPUTS
3.3.1 Health Endpoints
The selection of health effect endpoints reflects consideration of a number of factors. The
specific set of factors considered in selecting health effects endpoints to model in this assessment
included:
The overall weight of evidence from the collective body of epidemiological, controlled
human exposure, and toxicological studies and the determination made in the final ISA
regarding the strength of the causal relationship between PM2.5 and the more general
health effect category;
The extent to which particular health effect endpoints within these broader health effect
categories are considered significant from a public health standpoint;
The availability of well-conducted epidemiological studies providing C-R functions for
specific health effect endpoints;
The availability of sufficient air quality monitoring data in areas that were evaluated in
the epidemiological studies;
The availability of baseline incidence data to support population risk (incidence)
modeling; and
The anticipated value of developing quantitative risk estimates for the health effect
endpoint(s) to inform decision-making in the context of the PM NAAQS review.
In selecting the set of health effect endpoint categories (and associated endpoints and
related susceptible populations) to include in the PM2.5 risk assessment, we considered the health
effects evidence presented in the final ISA (US EPA, 2009d), as well as CASAC (Samet, 2009a)
and public comments received on the Scope and Methods Plan and CASAC (Samet, 2009b) and
public comments received on the first draft RA. In reviewing the final ISA in relation to PM2.5,
we focused on the following sections: (a) section 2.3.1.1 (Effects of Short-Term Exposure to
PM2.s), (b) section 2.3.1.2 (Effects of Long-Term Exposure to PM25), (c) section 2.3.2
(Integration of PM2.5 Health Effects), and (d) subsections in Chapter 6 and 7 of the final ISA
providing summaries of causal determination (for both morbidity and mortality endpoints)
related to short-term and long-term exposure, respectively. We also considered information in
the ISA on susceptible populations, which identified the life stages of children and older adults,
people with pre-existing cardiovascular and respiratory diseases, and people with lower
socioeconomic status as populations at increased risk for PM-related health effects.
Based on the evidence presented in the ISA and application of the above criteria, we
identified the following health effects endpoints for inclusion in the risk assessment:
3-28
-------
Health effects associated with short-term PM^exposure:
Mortality (causal relationship)
o non-accidental,
o cardiovascular-related
o respiratory-related,
Cardiovascular effects (causal relationship)
o cardiovascular-related hospital admissions
Respiratory effects (likely causal relationship)
o respiratory-related hospital admissions
o asthma-related emergency department visits
Health effects associated with long-term PM^exposure:
Mortality (causal relationship)
o all-cause
o ischemic heart disease (IHD)-related
o cardiopulmonary-related
o lung cancer
While we selected specific health effect endpoints that were all within broad health effect
categories classified in the ISA as having a "causal" or "likely causal" association with PM2.5
exposure, our selection is a based on applying the multi-factor approach described above.
The evidence available for these selected health effect endpoints generally focused on
the entire population, although some information was available that allowed us to consider
differences in estimated risk for the susceptible populations of older adults and people with pre-
existing cardiovascular and respiratory diseases. While evidence of effects in other important
susceptible populations, including children and people with lower socioeconomic status, was not
judged to be sufficient to support quantitative risk assessment, this evidence will be part of the
evidence-based considerations to be discussed in the PA currently being developed.
3.3.2 Selection and Delineation of Urban Study Areas
This section describes the approach used in selecting the 15 urban study areas included in
this risk assessment (see Table 3-4 for a listing of the urban study areas). This approach builds
upon and expands the approach for selecting urban study areas from the prior risk assessment
(US EPA, 2005, section 3.2, p. 37).
Criteria used in the prior risk assessment and updated in this analysis include:
Availability of sufficient air quality data: Sufficient air quality data was
identified as having at least 11 observations per quarter for a one year period and
3-29
-------
at least 122 observations per year. We assessed prospective study areas by
insuring that there was at least one PM2 5 monitor within the boundaries of the
prospective study area that met these completeness criteria for the period 2005 to
2007 with additional preference given to locations with more than one PM2 5
monitor meeting completeness criteria, since this provided a better
characterization of ambient air levels for that urban location.
Inclusion in epidemiology study: Coverage of the location within one of the key
epidemiology studies included in the risk assessment (at or close to the location
where at least one C-R function for one of the recommended health endpoints has
been estimated by a study satisfying the selection criteria used in the risk
assessment). In this review, because the current risk assessment primarily utilizes
multi-city studies to evaluate risk for short-term and long-term PM2 5 exposures
(whereas the prior risk assessment used city-specific studies in modeling
endpoints associated with short-term exposures), this criterion no longer applies
for most prospective areas.
Availability of city-specific baseline incidence data: Regarding sufficiency of
baseline health effects incidence data, an ongoing effort by EPA to collect county-
level hospital and emergency department admissions data from states to support
this risk assessment (see section 3.4) has resulted in enhanced health effects
baseline incidence data, largely addressing this criterion (i.e., most urban areas in
the U.S. now have coverage with the updated baseline health effects incidence
data).
Two additional factors considered in selecting locations to model in the current
assessment included:
Potential for risk reductions using alternative standard levels: We focused on
those urban areas with PM2 5 monitoring levels suggesting the potential for risk
reduction under the alternative (24-hour or annual) standards being considered
(i.e., urban locations with at least one monitor having an annual average above 12
|ig/m3 and/or a 24-hour value above 25 |ig/m3). Furthermore, locations with
ambient PM2 5 level significantly higher than these levels were favored (with
several urban study areas selected having both annual and 24-hour design values
exceeding the current standards - Table 3-4).
Regional representation: The second criterion we added for study area selection
focused on providing coverage for factors believed to play a role in influencing
risk heterogeneity at the national-level (e.g., PM2 5 source characteristics and
composition, demographics, socio-economic status (SES) status, air conditioner
use). Building on the 7 regions originally identified in the 1996 PM Criteria
Document (US EPA, 1996, section 6.4) (i.e., PM regions), we considered several
urban locations from each of these PM regions with the goal to identify one or
more candidate urban study areas in each region. Ultimately, application of the
criteria described here resulted in one of the PM regions (the Upper Midwest) not
3-30
-------
being covered by an urban study area. However, the remaining six PM regions
each have at least one urban study areas evaluated in the risk assessment. While
the PM regions were originally defined focusing primarily on differences in PM
composition, size and seasonality, by selecting urban study areas from regions
across the continental U.S., we recognize the potential for covering regional
differences in other factors related to risk heterogeneity as well (e.g.,
demographics, SES). The representativeness analysis (section 4.4) specifically
assesses the degree to which the 15 urban study areas provide coverage for
national trends in key risk-related factors such as those listed here.
Based on application of the above criteria, 15 study areas were selected for inclusion in
this risk assessment (see Table 3-4). In addition to identifying the 15 urban study areas, Table 3-
4 also provides additional information including: (a) whether the urban 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 PM regions used to guide the selection of the urban
study areas.
Table 3-5 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
* SE (Southeast), IM (industrial Midwest), SCA (Southern California), NE (Northeast), NW (Northwest), SW
(Southwest) (See S EPA, 1996, section 6.4 for description of these regions).
3-31
-------
Figure 3-4 15 urban study areas included in the risk assessment (including seven PM
regions used to guide selection of study areas).
Once the 15 urban study areas were selected, the next step was to identify the spatial
template to use in defining each study area (i.e., the geographical area associated with each study
area that would be used in identifying which counties and PM2.s monitors were associated with a
particular study area). For 12 of the 15 urban study areas, we either used a combined statistical
area (CSA) as the basis for the spatial template, or if that was not available, we used a core-based
statistical area (CBSA). The three remaining urban study areas were special cases and were
handled as follows:
Baltimore: Used counties in the Baltimore CBSA only and did not consider the larger
Baltimore-DC CSA since we felt it unlikely that the entire larger CSA would behave
similarly with regard to PM2.5 emissions reduction strategies;
Philadelphia: Used the Philadelphia CSA, but excluded Berks County (Reading;:
3-32
-------
Tacotna: Used only Pierce County (since we felt it unlikely that efforts to reduce
emissions at the "elevated" monitor in Pierce County, would significantly impact
monitors in Seattle).
Appendix K provides maps for each of the 15 urban study areas showing: (a) annual and
daily (i.e., 24-hour) design values (DV) for each PM2 5 monitor in each study area with DV
values based on monitoring data from 2005-2007, (b) sources of PM2.5 greater than 50 tons/year
(c) depiction of the highway network within each study area, and (d) counties comprising each
urban study area, together with the CSA or CBSA boundaries depending on location. These
maps allow the reader to visually consider the interplay between both local and more regional
sources of ambient PM2.5 and patterns of long-term (annual) and shorter-term (24-hour) design
values across monitors for a particular study area.
As noted earlier, in a few instances, two or more epidemiological studies used different
geographic boundaries for determining which populations were included in their studies. For
example, in one study conducted in Birmingham, AL populations from Blount, Jefferson,
Shelby, St. Clair, and Walker Counties were included, while another study included the
population residing in only Jefferson County. In such cases, we matched our delineation of the
urban area to that of each study, resulting in two or more different delineations of the urban area.
As we discuss below, two of the studies on which we rely for our core analysis -
Zanobetti and Schwartz (2009) and Bell et al. (2008) - are multi-location studies. Zanobetti and
Schwartz (2009) specified the county or counties included in each of the urban areas they
included in their analysis. Bell et al. (2008), however, did not focus on urban areas, but instead
focused on counties with populations above a specified threshold number. To limit the number
of different "versions" of a risk assessment location, wherever possible we specified the counties
in a risk assessment location for Bell et al. (2008) to match the set specified for Zanobetti and
Schwartz (2009). This was possible in those cases in which Zanobetti and Schwartz (2009)
identified an urban area as a single county, and that county was also included in Bell et al.
(2008). This was the case for several of the risk assessment locations. In some cases, however,
Zanobetti and Schwartz (2009) used a multi-county delineation of an urban area where at least
one of the counties was not among those included in Bell et al. (2008). In those cases, we had to
delineate two definitions of the urban area - one corresponding to Zanobetti and Schwartz (2009)
and the other corresponding to Bell et al. (2008). This was the case for Atlanta, Birmingham,
and St. Louis. In both Atlanta and New York, other delineations by other studies forced
additional delineation of these urban areas, as shown in Table 3-1 above.
Finally, we applied the studies of mortality associated with long-term exposure to PM2.5
to the urban areas as defined by the short-term exposure mortality study, Zanobetti and Schwartz
3-33
-------
(2009), to enable meaningful comparisons between estimates of premature morality associated
with short-term and long-term exposure to PM2.5.
3.3.3 Selection of Epidemiological Studies and Concentration-response (C-R) Functions
within Those Studies
As discussed above, we included in the PM2.s risk assessment only those health effect
endpoint categories (and specific health effects) that met the set of criteria reflected in the multi-
factor approach we developed for selecting health effect endpoints (see section 3.3.1). One of
these factors was the strength of evidence supporting a causal association between PM2.5
exposure and the endpoint of interest. Thus, in cases where the majority of the available studies
did not report a statistically significant relationship, the effect endpoint was not included. Once
it had been determined that a health endpoint would be included in the analysis, however,
inclusion of a study on that health endpoint was not based on statistical significance alone, but
considered other factors (e.g., overall design of the study including degree of control for
confounders, method used to characterize exposure to PM2.5 within the risk assessment).
A significant change since the previous PM risk assessment is the addition to the relevant
epidemiological literature of several multi-city studies. This type of study has several
advantages over single-city studies. First, multi-city studies use the same study design in each of
the cities included in the study, so that city-specific results are readily comparable. Second,
when they are estimating a single C-R function based on several cities, multi-city studies also
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. Moreover, in a multi-city study the statistical power to detect an effect in any given
city can be supplemented by drawing statistical power from data across all the cities included in
the study (or all the cities in the same region) to adjust city-specific estimates towards the mean
across all cities included in the analysis (or in the same region). This is particularly useful in
those instances, where a city has relatively less data resulting in a larger standard error for the
effect estimate. In this situation, the information on the C-R relationship in all the other cities
included in a multi-city study can be used to help inform an assessment of the C-R relationship
in the city in question. Finally, multi-city studies tend to avoid the often-noted problem of
publication bias that single-city studies confront (in which studies with statistically insignificant
or negative results are less likely to get published than those with positive and/or statistically
significant results).
For this risk assessment, we selected what we considered to be the best study to assess
the C-R relationship between PM2.5 and a given health endpoint, and we included other studies
for that health endpoint only if they were judged to contribute something above and beyond what
we could learn from the primary study selected.
3-34
-------
A primary study for a given health endpoint had to satisfy the study selection criteria that
we have used in past PM (and other) risk assessments. In particular:
It had to be a published, peer-reviewed study that has been evaluated in the PM ISA and
judged adequate by EPA staff for purposes of inclusion in this risk assessment based on
that evaluation.
It had to directly measure, rather than estimate, PM2.5 on a reasonable proportion of the
days in the study.
It had to either not rely on Generalized Additive Models (GAMs) using the S-Plus
software to estimate C-R functions or to appropriately have re-estimated these functions using
revised methods.32
Because of the advantages noted above, we selected multi-city studies as our primary
studies for assessing the risks of premature non-accidental, cardiovascular, and respiratory
mortality (Zanobetti and Schwartz, 2009) and cardiovascular and respiratory hospital admissions
(Bell et al., 2008) associated with short-term exposure to PM2 5 in our core analysis. In each of
these studies, the 15 urban areas selected for the PM risk assessment were among the locations
included in their analysis. These two multi-city studies are based on more recent air quality and
health effects incidence data for short-term exposure-related mortality and morbidity and
therefore represent the best studies to use in deriving C-R functions for this risk assessment.
Dominici et al. (2007) was considered as an alternative study in identifying C-R functions for
modeling short-term exposure-related mortality, however its study period and the underlying air
quality data and disease incidence data (1987-2000) are not as current as that of Zanobetti and
Schwartz et al., 2009 (study period of 2001-2005), and therefore, we decided to focus on
Zanobetti and Schwartz et al. (2009) as the source of C-R functions for modeling short-term
exposure-related mortality.
Studies often report more than one estimated C-R function for the same location and
health endpoint. Models can include different sets of co-pollutants, different lag structures, and
different forms to accommodate weather and temporal variables. Once a study has been
selected, the next step is to select one or more C-R functions from among those reported in the
study.
Zanobetti and Schwartz (2009) divided the United States into six regions, based on the
Koppen climate classification (Kottek 2006; Kottek et al. 2006)(http://koeppen-
32 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 (US EPA, 2004a),
PM Staff Paper (US EPA, 2005c), and PM Health Risk Assessment Technical Support Document (Abt Associates,
2005).
3-35
-------
eiger.vuwien.ac.at/).33 They estimated the coefficient of PM2.5 in single-pollutant log-linear
models using Poisson regression for each of 112 cities, as well as in two-pollutant models with
coarse PM. They estimated annual models (which assume that the relationship between
mortality and PM2.5 is the same through the year), as well as four seasonal models per location.
They then used a random effects meta-analysis to combine the city-specific results (Berkey et al.
1998). Pooling of city-specific results was done at the national level as well as at the regional
level, and separately for each season as well as for the annual functions.
With respect to the multi-city study for short-term exposure mortality, at the request of
EPA, the authors produced Empirical Bayes "shrunken" city-specific estimates, adjusted towards
the appropriate regional mean, using the approach described in Le Tertre et al. (2005). This was
done for the annual estimates as well as for each season-specific estimate.34 The annual city-
specific "shrunken" estimates were used in our core analysis.35 The seasonal estimates were
used in a sensitivity analysis. City-specific estimates have the advantage of relying on city-
specific data; however, as noted above, such estimates can have large standard errors (and thus
be unreliable); "shrinking" city-specific estimates towards the regional mean estimate is a more
efficient use of the data.36 Such "shrinking" can be thought of as combining the advantages of a
single-city study (in which the estimation of a city-specific coefficient is not influenced by data
from other locations) with the advantages of a multi-city study (in which there is much greater
statistical power to detect small effects).
In Zanobetti and Schwartz (2009) all PM2.5 models used the same lag structure (i.e., an
average of same-day and the previous day's PM^.s). The study did, however, examine both
single-pollutant and two-pollutant models (with coarse PM). We selected the single-pollutant
33 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, ME, 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).
34 These city-specific "shrunken" estimates were provided to EPA (see Zanobetti, 2009).
35 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.
36 Each "shrunken" city-specific estimate is a weighted average of the regional mean estimate and the city-
specific estimate, where the weight on the city-specific estimate is proportional to the inverse of the standard error,
and the weight assigned to the regional mean estimate is proportional to the inverse of a measure of between-city
variability. If there is a lot of "true" variability between city-specific estimates, the regional mean will receive
relatively less weight in the averaging, compared to a case where there is not a lot of "true" variability. Conversely,
if there is substantial variance in the city-specific estimate, it will receive less weight in the averaging compared to a
case where the city-specific estimate has low variance..
3-36
-------
models, in part to avoid collinearity problems, and in part to be consistent with most of the other
studies used in the risk assessment, which were single-pollutant studies.
Bell et al. (2008) estimated log-linear models relating short-term exposure to PM2 5 and
hospital admissions for cardiovascular and respiratory illnesses among people 65 and older,
using a 2-stage Bayesian hierarchical model, for each of 202 counties in the United States. They
reported both annual and season-specific results, nationally and regionally (for four regions:
Northeast, Southeast, Northwest, and Southwest), but not at the local (city-specific) level. All
cardiovascular hospital admissions models were single-pollutant, 0-day lag models; for
respiratory hospital admissions, both single-pollutant 0-day models and single-pollutant 2-day
models were estimated. We used the regional, annual C-R functions in our core analysis
(identifying the appropriate region for each of our 15 risk assessment locations).37 For
respiratory hospital admissions (for the core analysis), we selected the 2-day lag models, based
on evidence that for respiratory effects the strongest associations with PM exposure may be
associated with longer lag periods (on the order of 2 days or more).38 We used the regional
season-specific functions in a sensitivity analysis.
We identified two studies that estimated C-R relationships between short-term exposure
to PM2.5 and emergency department (ED) visits for cardiovascular and/or respiratory illnesses.
(There were no multi-city studies for this category of health endpoint.) Tolbert et al. (2007)
examined both cardiovascular and respiratory ED visits in Atlanta, GA, using single-pollutant
log-linear models with a 3-day moving average (0-day, 1-day, and 2-day lags) of PM2.5. Ito et al.
(2007) estimated the relationship between short-term exposure to PM2 5 and ED visits for asthma
in New York City. They estimated two single-pollutant models, one for the whole year and one
for the period from April through August; in addition, they estimated several two-pollutant
models for the period from April through August. We selected the single-pollutant model for the
whole year for the core analysis, and we explored the impacts of using the annual versus the
April-through-August model, as well as the single- versus multi-pollutant models in sensitivity
analyses.
For the purpose of conducting a sensitivity analysis to show the impact of different lag
structures, different modeling approaches, and single- versus two-pollutant models on estimates
of the risks of premature mortality and hospital admissions associated with short-term exposure
37 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.
38 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." (U.S. EPA, 2009d, section 2.4.2.2).
3-37
-------
to PM2.5, we selected Moolgavkar (2003). This study reported results for premature non-
accidental, cardiovascular, and respiratory mortality and for cardiovascular and respiratory
hospital admissions associated with short-term exposures to PM2 5 in Los Angeles, using several
different lag structures and several different approaches to modeling the effects of weather and
temporal variables.
In modeling premature mortality associated with long-term exposure to PM2.5 in our core
analysis, we selected Krewski et al. (2009) as our primary study. This study is an extension of
the ACS prospective cohort study (Pope et al., 2002), used in the previous PM risk assessment,.
The Krewski et al., 2009 study (and the underlying ACS dataset) has a number of advantages
which informed our selection of this study as the basis for C-R functions used in the core
analysis, including: (a) extended air quality analysis incorporating data from 1989 to 2000
(extending the period of observation to eighteen years: 1982-2000), which increases the power of
the study and allows the study authors to examine the important issue of exposure time windows,
(b) rigorous examination of a range of model forms and effect estimates, including consideration
of such factors as spatial autocorrelation in specifying response functions, (c) coverage for a
range of ecological variables (social, economic and demographic) which allows for consideration
of whether these confound or modify the relationship between PM2.5 exposure and mortality, (d)
inclusion of a related analysis (focusing on Los Angeles), which allowed for consideration of
spatial gradients in PM2 5 and whether they effect response models (by addressing effect
modification, for example) and (e) large overall dataset with over 1.2 million individuals and 156
MS As. To provide coverage for one of the other larger datasets used in prospective cohort
analyses of long-term mortality (the six-cites dataset), we selected the Krewski et al. (2000)
study to provide C-R functions that were used in the sensitivity analysis completed for this risk
assessment.
A number of other studies were considered as candidates for use in modeling long-term
exposure-related mortality in this analysis. For purposes of transparency, we have included a
brief summary here of our rationale for not selecting a number of the more high-profile studies
for use in the core analysis. The Laden et al. (2006) study (which focused on the six-cities
dataset) was not selected because it used visibility data to estimate ambient PM2 5 levels. The
Goss et al. (2004) study (based on the cystic fibrosis data), while addressing an at-risk population
of concern, was not selected because of a lack of baseline incidence data for this population
which prevents quantitative modeling of mortality incidence. The Miller et al. (2007) study
(focusing on the Women's Health Initiative dataset) while providing coverage for a population of
particular interest, was not used, again due to an absence of baseline incidence data (which is
particularly important for this population which is typically healthier than the general
population). And finally, the Eftim et al. (2008) study (focusing on the Medicare population)
3-38
-------
was not included because this study did not include representative confounder control for
smoking, which introduces uncertainty into C-R functions obtained from the study.
Krewski et al. (2009) (the study selected as the basis for C-R functions used in the core
analysis) considered mortality from all causes, as well as cardiopulmonary mortality, mortality
from ischemic heart disease, and lung cancer mortality. The study presents a variety of C-R
functions, in an effort to show how the results vary with various changes to the method/model
used. It was not readily apparent from review of the Health Effects Institute (HEI) report, that
the authors of the study recommended any one of these as clearly superior to the others.
Therefore, we corresponded with the authors to obtain additional clarification regarding specific
aspects of the study and associated results as presented in the HEI report (Krewski et al., 2009).
In response to the our question of whether the study authors had a preference for a particular
model in the context of using that model and its hazard ratio(s) in risk assessment, the authors
stated that they had "refrained from expressing a preference among the results for their use in
quantitative risk assessment," preferring to "explore several plausible statistical models that we
have fit to the available data." However, the authors went on to state that "...if one had to choose
a model for use in practical applications involved in air quality management, one could argue
that a random effects model (which accounts for apparent spatial autocorrelation in the data)
might be preferable. A model that included ecological covariates, which has the effect of
reducing the residual variation in mortality, might also be of interest. If forced to pick a single
model for risk assessment applications in air quality management, our random effects model with
ecological covariates might be selected" (Krewski, 2009).
In addition to these statements from the study authors regarding the model form to use,
EPA staff also considered the results of an analysis presented in the study examining the
importance of exposure time windows in deriving C-R functions. This analysis suggested that
models developed using both exposure time windows considered in the analysis (1979-1983 and
1999-2000) were equally effective at representing the relationship between PM2.5 exposure and
long-term exposure-related mortality. Therefore, we concluded that C-R functions used in the
core analysis should include functions fitted to both exposure time windows. However, the study
does not provide random effects models with ecological covariates for both exposure time
windows (this form of model is only provided with a fit to the latter exposure window).
Therefore, for the core analysis, we decided to use the Cox proportional hazard model with 44
individual and 7 ecological variables fitted to both exposure time windows.39
39 Note, however, that if the Krewski et al. (2009) study had provided a random effects model with
ecological covariates (for both PM monitoring periods - 1979-1983 and 1999-2000), then we would have used those
models in our core analysis.
3-39
-------
In specifying effect estimates for each set of models, the relative risks for a 10 |ig/m3
change in PM2.5 were back-calculated from Table 33 of Krewski et al. (2009). We selected
several additional C-R functions from Krewski et al. (2009) to use in sensitivity analyses carried
out in two risk assessment locations (Los Angeles and Philadelphia), including the random
effects form (section 3.5.4), as described below. In addition, as mentioned earlier, we used C-R
functions obtained from Krewski et al. (2000) [reanalysis of the Six Cities Study] in the
sensitivity analysis.
3.3.4 Summary of Selected Health Endpoints, Urban Areas, Studies, and C-R Functions
A summary of the selected health endpoints, urban areas, and epidemiological studies used
in the risk assessment is given below in Tables 3-5 and 3-6 for short-term and long-term
exposure studies, respectively. A more detailed overview of the locations, health endpoints,
studies, and C-R functions included in the core analysis is given in Table 3-7. An overview of
the locations, health endpoints, studies, and C-R functions included in sensitivity analyses is
given in Table 3-8.
3-40
-------
Table 3-6 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.
5-41
-------
Table 3-7. 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.
5-42
-------
Table 3-8 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) 1
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 department (ED) visits,
cardiovascular
Short-term exposure emergency department (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
5-43
-------
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
5-44
-------
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
5-45
-------
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
5-46
-------
Risk
Assessment
Location
New York
Philadelphia
Counties
Kings, New York City
(Manhattan), Queens,
Richmond, Bronx
Kings, New York City
(Manhattan), Queens,
Richmond, Bronx
Philadelphia
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)
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 department (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
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
5-47
-------
Risk
Assessment
Location
Phoenix
Pittsburgh
Salt Lake City
Counties
Maricopa
Allegheny
Salt Lake
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)
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)
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 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
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 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
5-48
-------
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)
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)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
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 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
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
NA
0-day lag
2-day lag
*A11 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.
5-49
-------
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
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.
5-50
-------
Table 3-9. 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
locally focused rollback approach as an alternative to the
proportional rollback approach).40
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, N02
,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)
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
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
40 However, as noted in section 3.2.3 and in section 3.5.4, quantitative risk estimates were not generated using the locally focused 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
locally focused rollback approach.
5-51
-------
Sensitivity Analysis
Impact of using season-specific C-R functions (vs. an
annual C-R function)
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
Zanobetti and Schwartz (2009) -
seasonal functions vs. annual
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**
Non-accidental mortality,
cardiovascular mortality, respiratory
mortality associated with short-term
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
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
Zanobetti and Schwartz (2009)
All-cause and ischemic heart disease
mortality associated with long-term
exposure
Non-accidental mortality associated
Los Angeles and
Philadelphia
Baltimore, Birmingham,
5-52
-------
Sensitivity Analysis
functions and proportional vs. hybrid rollbacks to
estimate incidence associated with short-term exposure
to PM2 5 concentrations that just meet the current
standards
Study/C-R Function
Health Endpoint**
with short-term exposure
Risk Assessment
Location(s)
Detroit, Los Angeles, New
York, Pittsburgh, and St.
Louis
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.
5-53
-------
3.4 BASELINE HEALTH EFFECTS INCIDENCE DATA
As noted in section 3.1.2 above, the form of C-R function most commonly used in
epidemiological studies on PM, shown in equation (1), is log-linear. To estimate the change in
incidence of a health endpoint associated with a given change in PM2.5 concentrations using this
form of C-R function requires the baseline incidence (often calculated as the baseline incidence
rate times the population) of the health endpoint, that is, the number of cases per unit time (e.g.,
per year) in the location before a change in PM2.5 air quality (denoted yo in equations 2, 3 and 4).
Incidence rates express the occurrence of a disease or event (e.g., asthma episode, death,
hospital admission) in a specific period of time, usually per year. Rates are expressed either as a
value per population group (e.g., the number of cases in Philadelphia County) or a value per
number of people (e.g., the number of cases per 10,000 residents in Philadelphia County), and
may be age- and sex-specific. Incidence rates vary among geographic areas due to differences in
population characteristics (e.g., age distribution) and factors promoting illness (e.g., smoking, air
pollution levels).
3.4.1 Data Sources
3.4.1.1 Mortality
We obtained individual-level mortality data for 2006 for the whole United States from
the Centers for Disease Control (CDC), National Center for Health Statistics (NCHS). The data
are compressed into a CD-ROM, which contains death information for each decedent, including
residence county Federal Information Processing System (FIPS), age at death, month of death,
and underlying causes (International Classification of Diseases (ICD)-IO codes). The detailed
mortality data allow us to generate cause-specific death counts at the county level for selected
age groups. Below we describe how we generated the county-level death counts.
3.4.1.2 Hospital Admission and Emergency Department Visits
For hospital admissions (HA) and emergency department (ED) visits, there are multiple
data sources:
Healthcare Cost and Utilization Project (HCUP) Central Distributor HCUP is a
family of health care databases developed through a Federal-State-Industry partnership
and sponsored by the Agency for Healthcare Research and Quality (AHRQ). The HCUP
databases are based on the data collection efforts of data organizations in participating
states. We used two HCUP databases: the State Inpatient Database (SID) and the State
Emergency Department Database (SEDD) respectively. SID/SEDD include detailed
HA/ED information for each discharge, including patient county FIPS, age, admission
type (e.g., emergent, urgent), admission/discharge season, and principle diagnosis (ICD-9
codes). The HCUP databases can be purchased from the HCUP Central Distributor,
although not all participant states release the data to the Central Distributor.
3-54
-------
HCUP State Partners. For those HCUP participating states that don't release their data
to the Central Distributor, we contacted the HCUP state partners to obtain the HA and/or
ED data.
Communication with the author(s) of selected epidemiological studies. The ED data
for Atlanta in 2004 were sent to EPA by one of the authors of Tolbert et al. (2007).
Table 3-9 shows the states for which we obtained data from the HCUP Central
Distributor and the HCUP State Partners. The data are at the discharge level if not otherwise
noted, and the data year is 2007 for all the states in the table. The column "PM RA Location"
indicates the selected risk assessment location(s) where the incidence rate is applied.
The necessary baseline incidence data were not available for Atlanta, Birmingham,
Philadelphia, Pittsburgh and St. Louis. Therefore, for each of these five risk assessment
locations EPA instead used the baseline incidence rate for a designated surrogate location.
Surrogate locations were chosen if they were deemed to be sufficiently similar to the urban area
whose baseline incidence data were not available. Surrogate locations are noted in Table 3-9.
3-55
-------
Table 3-10 Sources of Hospital Admissions (HA) and Emergency Department (ED)
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.
*NA denotes "not available, or not
from the HCUP Central Distributor.
available with all variables required for our analysis. If data were not available
we contacted the HCUP State Partner.
3.4.1.3 Populations
To calculate baseline incidence rate, in addition to the health baseline incidence data we
also need the corresponding population. We obtained population data from the U.S. Census
Bureau (http://www.census.gov/popest/counties/asrh/). These data, released on May 14, 2009,
are the population estimates of the resident populations by selected age groups and sex for
counties in each U.S. state from 2000 to 2008. We used 2007 populations for calculating most
incidence rates except for the ED visit rate in Atlanta. Because the ED visit data obtained from
the authors of Tolbert et al. (2007) are for 2004, we used 2004 population estimates for the 20-
county Metropolitan area used in the Tolbert et al. study for the Atlanta area to calculate the ED
incidence rates to be applied when using that study in the risk assessment; we then applied the
2004 rates to the 2007 population, assuming the ED incidence rates in Atlanta did not change
significantly from 2004 to 2007. The sizes of the populations in the assessment locations that are
relevant are shown below in Table 3-10.
3-56
-------
Table 3-11. 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 330
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 3 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
5-57
-------
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 330
2006
1,134,000
437,000
2007
1,134,000
444,000
Ages 3 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.
5-58
-------
3.4.2 Calculation of Baseline Incidence Rates
To calculate a baseline incidence rate to be used with a C-R function from a given study,
we matched the counties, age groupings, and ICD codes used in that study. For example, Bell et
al. (2008) designated Dallas, TX as Dallas County and estimated a C-R function for ICD-9 codes
490-492, 464-466, and 480-487 (respiratory HA) among ages 65 and up; we therefore selected
only those HA records that had corresponding ICD codes for ages 65 and up in Dallas County
and also selected the population for the same age group in the same county. The incidence rate
is simply the ratio of the selected HA count to the population. The same procedure was used to
calculate baseline incidence rates for all of the risk assessment locations.41
If a C-R function was estimated for a specific season, we selected only those HA records
within that season. The season definitions are: winter (December, January, and February), spring
(March, April, and May), summer (June, July, and August) and fall (September, October, and
November). Note that the HA data for some states didn't include information about admission
season but only discharge season or discharge quarter. The admission season was then
approximated using discharge season or discharge quarter.42
Some studies (e.g., Bell et al., 2008) look at the unscheduled hospital admissions (Has)
only, so we excluded scheduled admissions from the analyses to match the study. A HA is
unscheduled if the admission type is emergency or urgent.
The baseline mortality rates are given in Table 3-11. The baseline HA and ED visit rates
are given in Table 3-12.
41 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.
42 Based on communication with the HCUP state partner in Texas, patients are normally admitted and
discharged in the same season.
3-59
-------
Table 3-4. 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
NA
NA
All ages
>30
All ages
>30
NA
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
NA
NA
NA
1,700
NA
1,600
NA
NA
1,020
NA
1,500
NA
1,300
NA
920
NA
1,030
NA
Non-accidental
(AOO-R99)
480
NA
NA
NA
950
NA
920
NA
NA
540
NA
850
NA
620
NA
480
NA
560
NA
630
Cardiovascular
(101-159)
120
NA
NA
NA
270
NA
260
NA
NA
150
NA
300
NA
190
NA
130
NA
190
NA
270
Respiratory
(JOOJ99)
41
NA
NA
NA
85
NA
85
NA
NA
48
NA
67
NA
67
NA
37
NA
57
NA
52
Cardio-
pulmonary
(401-440, 460-
519)
330
NA
NA
NA
690
NA
680
NA
NA
420
NA
700
NA
590
NA
370
NA
510
NA
Ischemic
Heart
Disease
(410-414)
89
NA
NA
NA
300
NA
190
NA
NA
170
NA
360
NA
260
NA
150
NA
250
NA
Lung
Cancer
(162)
51
NA
NA
NA
110
NA
104
NA
NA
66
NA
107
NA
66
NA
57
NA
55
NA
COPD
(490-496)
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
29
NA
NA
5-60
-------
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
NA
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
NA
All ages
>30
All ages
>30
Type of Mortality (ICD-10 or ICD-9 Codes)
All-Cause
1,0800
NA
NA
1,700
NA
1,100
NA
1,800
NA
980
NA
1,500
NA
NA
1,200
810
1,300
Non-accidental
(AOO-R99)
NA
NA
970
NA
600
NA
1,090
NA
480
NA
870
NA
NA
660
NA
750
1,300
Cardiovascular
(101-159)
NA
NA
280
NA
160
NA
330
NA
110
NA
270
NA
NA
190
NA
220
370
Respiratory
(JOO-J99)
NA
NA
83
NA
67
NA
96
NA
45
NA
83
NA
NA
66
NA
76
130
Cardio-
pulmonary
(401-440, 460-
519)
580
NA
NA
720
NA
470
NA
770
NA
350
NA
680
NA
NA
510
340
580
Ischemic
Heart
Disease
(410-414)
380
NA
NA
300
NA
220
NA
350
NA
101
NA
320
NA
NA
240
140
240
Lung
Cancer
(162)
56
NA
NA
120
NA
68
NA
120
NA
37
NA
106
NA
NA
88
53
90
COPD
(490-496)
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
42
71
* Figures in this table are rounded to a two-integer level of precision. NA refers to health endpoint categories that are not relevant for this particular county-
level study area definition (i.e., the epidemiology study and associated effect estimate reflected in this specification county-level of the study area did not include this
particular endpoint category and consequently a baseline incidence value is not shown).
5-61
-------
Table 3-5. 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
NA
>65
All ages
>65
NA
>65
>65
>65
>65
>65
All ages
>65
>65
All ages
>65
>65
>65
>65
NA
>65
>65
Health Endpoints (ICD-9 Codes)
HA, cardio-
vascular (390-
429)
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
5,500
NA
NA
NA
NA
NA
NA
NA
NA
NA
HA (unscheduled),
cardiovascular(426
-429, 430-438,
410-414, 440-449)
NA
5,700
NA
8,600
NA
5,700
5,000
8,800
5,600
5,900
NA
5,500
6,400
NA
8,600
5,020
6,100
3,030
NA
5,600
4,500
HA, COPD
(490-496)
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
223
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
HA (unscheduled),
respiratory (490-492,
464-466, 480-487)
NA
2,020
NA
2,600
NA
2,020
2,000
3,000
2,100
2,200
NA
2,000
2,030
NA
2,600
1,600
1,900
1,200
NA
2,600
1,600
ED visits,
cardiovascular (410-
414, 427, 428, 433-
437, 440, 443-445,
451-453)
NA
NA
690**
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
ED visits, respiratory
(460-465, 466.1, 466.11,
466.19, 477, 480-486, 491-
493, 496, 786.07, 786.09)
NA
NA
2600**
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
ED visits,
asthma
(493)
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
1,100
NA
NA
NA
NA
NA
NA
NA
* Figures in this table are rounded to a two-integer level of precision. NA - see footnote to Table 3-11.
**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.
5-62
-------
3.5 ADDRESSING UNCERTAINTY AND VARIABILITY
3.5.1 Overview
An important component of a population health risk assessment is the characterization of
both uncertainty and variability. Variability refers to the heterogeneity of a variable of interest
within a population or across different populations. For example, populations in different
regions of the country may have different behavior and activity patterns (e.g., air conditioning
use, time spent indoors) that affect their exposure to ambient PM and thus the population health
response. The composition of populations in different regions of the country may vary in ways
that can affect the population response to exposure to PM - e.g., two populations exposed to the
same levels of PM might respond differently if one population is older than the other. In
addition, the composition of the PM to which different populations are exposed may differ, with
different levels of toxicity and thus different population responses. Variability is inherent and
cannot be reduced through further research. Refinements in the design of a population risk
assessment are often focused on more completely characterizing variability in key factors
affecting population risk - e.g., factors affecting population exposure or response - in order to
produce risk estimates whose distribution adequately characterizes the distribution in the
underlying population(s).
Uncertainty refers to the lack of knowledge regarding the actual values of inputs to an
analysis. Models are typically used in analyses, and there is uncertainty about the true values of
the parameters of the model (parameter uncertainty) - e.g., the value of the coefficient for PM2 5
in a C-R function. There is also uncertainty about the extent to which the model is an accurate
representation of the underlying physical systems or relationships being modeled (model
uncertainty) - e.g., the shapes of C-R functions. In addition, there may be some uncertainty
surrounding other inputs to an analysis due to possible measurement errore.g., the values of
daily PM2.5 concentrations in a risk assessment location, or the value of the baseline incidence
rate for a health effect in a population.43 In any risk assessment, uncertainty is, ideally, reduced
to the maximum extent possible through improved measurement of key variables and ongoing
model refinement. However, significant uncertainty often remains, and emphasis is then placed
on characterizing the nature of that uncertainty and its impact on risk estimates. The
characterization of uncertainty can be both qualitative and, if a sufficient knowledgebase is
available, quantitative.
43 It is also important to point out that failure to characterize variability in an input used in modeling can
also introduce uncertainty into the analysis. This reflects the important link between uncertainty and variability with
the effort to accurately characterize variability in key model inputs actually reflecting an effort to reduce uncertainty.
3-63
-------
The selection of urban study areas for the PM2.5 risk assessment was designed to cover
the range of PM2.5-related risk experienced by the U.S. population and, in general, to adequately
reflect the inherent variability in those factors affecting the public health impact of PM2 5
exposure. Sources of variability reflected in the risk assessment design are discussed in section
3.5.2, along with a discussion of those sources of variability which are not fully reflected in the
risk assessment and consequently introduce uncertainty into the analysis.
The characterization of uncertainty associated with risk assessment is often addressed in
the regulatory context using a tiered approach in which progressively more sophisticated
methods are used to evaluate and characterize sources of uncertainty depending on the overall
complexity of the risk assessment (WHO, 2008). Guidance documents developed by EPA for
assessing air toxics-related risk and Superfund Site risks (USEPA, 2004b and 2001, respectively)
as well as recent guidance from the World Health Organization (WHO, 2008) specify multi-
tiered approaches for addressing uncertainty.
The WHO guidance presents a four-tiered approach, where the decision to proceed to the
next tier is based on the outcome of the previous tier's assessment. The four tiers described in the
WHO guidance include:
Tier 0 - recommended for routine screening assessments, uses default uncertainty factors
(rather than developing site-specific uncertainty characterizations);
Tier 1 - the lowest level of site-specific uncertainty characterization, involves qualitative
characterization of sources of uncertainty (e.g., a qualitative assessment of the general
magnitude and direction of the effect on risk results);
Tier 2 - site-specific deterministic quantitative analysis involving sensitivity analysis,
interval-based assessment, and possibly probability bound (high- and low-end)
assessment; and
Tier 3 - uses probabilistic methods to characterize the effects on risk estimates of sources
of uncertainty, individually and combined.
With this four-tiered approach, the WHO framework provides a means for systematically
linking the characterization of uncertainty to the sophistication of the underlying risk assessment.
Ultimately, the decision as to which tier of uncertainty characterization to include in a risk
assessment will depend both on the overall sophistication of the risk assessment and the
availability of information for characterizing the various sources of uncertainly. EPA staff has
used the WHO guidance as a framework for developing the approach used for characterizing
uncertainty in this risk assessment.
The overall analysis in the PM National Ambient Air Quality Standard (NAAQS) risk
assessment is relatively complex, thereby warranting consideration of a full probabilistic (WHO
Tier 3) uncertainty analysis. However, limitations in available information prevent this level of
3-64
-------
analysis from being completed at this time. In particular, the incorporation of uncertainty related
to key elements of C-R functions (e.g., competing lag structures, alternative functional forms,
etc.) into a full probabilistic WHO Tier 3 analysis would require that probabilities be assigned to
each competing specification of a given model element (with each probability reflecting a
subjective assessment of the probability that the given specification is the "correct" description
of reality). However, for many model elements there is insufficient information on which to
base these probabilities. One approach that has been taken in such cases is expert elicitation;
however, this approach is resource- and time-intensive and consequently, it was not feasible to
use this technique in the current PM NAAQS review to support a WHO Tier 3 analysis.44
For most elements of this risk assessment, rather than conducting a full probabilistic
uncertainty analysis, we have included qualitative discussions of the potential impact of
uncertainty on risk results (WHO Tierl) and/or completed sensitivity analyses assessing the
potential impact of sources of uncertainty on risk results (WHO Tier 2). Note, however, that in
conducting sensitivity analyses, we have used both single- and multi-factor approaches (to look
at the individual and combined impacts of sources of uncertainty on risk estimates). Also, as
discussed below in section 3.5.4, in conducting sensitivity analyses, we used only those
alternative specifications for input parameters or modeling approaches that were deemed to have
scientific support in the literature (and so represent alternative reasonable input parameter values
or modeling options). This means that the alternative risk results generated in the sensitivity
analyses represent reasonable risk estimates that can be used to provide a context, with regard to
uncertainty, within which to assess the set of core (base case) risk results (see section 4.5.3).
The sensitivity analysis also includes coverage for potential variability in the pattern of
reductions in ambient PM2.5 concentrations associated with simulations of just meeting the
current and alternative suites of standards. Specifically, as discussed above in section 3.2.3, we
have included three alternative rollback methods (proportional, hybrid and locally focused) to
provide coverage for variability in this potentially important factor influencing risk estimates.
In addition to the qualitative and quantitative treatment of uncertainty and variability
which are described here, we have also completed an analysis to evaluate the representativeness
of the selected urban study areas against national distributions for key PM risk-related attributes
to determine whether they are nationally representative or more focused on a particular portion
of the distribution for a given attribute (section 4.4.1). In addition, we have completed a second
analysis addressing the representativeness issue, which identified where the subset of 31 counties
44 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.
3-65
-------
comprising our 15 urban study areas fall along a distribution of national county-level long-term
exposure-related mortality risk (section 4.4.2). This analysis allowed us to assess the degree of
which the 15 urban study areas capture locations within the U.S. likely to experience elevated
levels of risk related to PM2.5 exposure.
The remainder of this section is organized as follows. Key sources of variability which
are reflected in the design of the risk assessment, along with sources excluded from the design,
are discussed in section 3.5.2. A qualitative discussion of key sources of uncertainty associated
with the risk assessment (including the potential direction, magnitude and degree of confidence
associated with our understanding of the source of uncertainty - the knowledge base) is
presented in section 3.5.3. The methods and results of the single- and multi-factor sensitivity
analyses completed for the risk assessment are presented in section 3.5.4. An overall summary
of the methods used to address uncertainty and variability for the 15 urban study areas (including
the two assessments intended to place the urban study areas in a broader national context) is
presented in section 3.5.5.
3.5.2 Treatment of Key Sources of Variability
The risk assessment was designed to cover the key sources of variability related to
population exposure and exposure response, to the extent supported by available data.45
However, as with all risk assessments, there are sources of variability which have not been fully
reflected in the design of the risk assessment and consequently introduce a degree of uncertainty
into the risk estimates. While different sources of variability were captured in the risk
assessment, it was generally not possible to separate out the impact of each factor on population
risk estimates, since many of the sources of variability are reflected collectively in a specific
aspect of the risk model. For example, inclusion of urban study areas from different PM regions
likely provides some degree of coverage for a variety of factors associated with PM2.5 risk (e.g.,
air conditioner use, PM2.5 composition, differences in population commuting and exercise
patterns, weather). However, the model is not sufficiently precise or disaggregated to allow the
45 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 variability, 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.
3-66
-------
individual impacts of any one of these sources of variability on the risk estimates to be
characterized.
Key sources of potential variability that are likely to affect population risks are discussed
below, including the degree to which they are (or are not) fully captured in the design of the risk
assessment:
PM2.s composition: While information was not available to support modeling risk
associated with different components of PM2.5, the assessment did use effect estimates
(for a number of the short-term exposure-related health endpoints) differentiated by
region of the country, or differentiated for specific urban locations (sections 3.3.3 and
3.3.4). While many factors may contribute to differences in effect estimates (for the
same health endpoint) across different locations, compositional differences in PM2.5
may be partially responsible. Therefore, while the analysis did not explicitly address
compositional differences in generating risk estimates, potential differences in PM2.5
composition may be reflected in those effect estimates that are differentiated by region
and/or urban study area. The effect estimates for mortality associated with long-term
exposure to PM2.5 are not regionally differentiated and instead, a single national-scale
estimate is used. This means that any differences in risks of mortality associated with
long-term exposure to PM2.5 that are linked to differences in PM2.5 composition (or to
any other differences across regions or locations) would not be discernable, since a
single national-scale risk estimate is generated for each mortality category. In addition
to using region- or location-specific effect estimates for health effects associated with
short-term exposures, the selection of urban areas to include in the risk assessment was
designed in part to ensure that areas in different regions of the country, with different
PM2 5 composition, were included.
Intra-urban variability in ambient PM2.s levels: Several recent studies (e.g., Jerrett
et al., 2005) have addressed the issue of heterogeneity of PM concentrations within
urban areas and its potential impact on the estimation of premature mortality associated
with long-term exposure to PM2 5. Most recently, the HEI Reanalysis II (Krewski et
al., 2009), focusing on the ACS dataset, discusses epidemiological analyses completed
for Los Angeles and New York City which included more highly-refined (zip code
level) characterizations of spatial gradients in population exposure within each urban
area based on land-use regression methods and/or kriging. While both analyses
provide insights into the issue of intra-urban heterogeneity in PM2 5 concentrations and
its potential implications for epidemiology-based health assessments, due to the time
and resources necessary to integrate them into the risk assessment, we were not able to
incorporate these studies quantitatively. The implications of these studies for
interpretation of long-term mortality C-R functions and potential exposure error
associated with those functions is discussed below in section 3.5.3.
Variability in the patterns of ambient PMi.s reduction as urban areas: In
simulating just meeting the current or alternative suites of standards, there can be
considerable variability in the patterns of ambient PM2.5 reductions that result from
different simulation approaches (i.e., they can be more localized, more regional, or
some combination thereof). To address this issue in the risk assessment, we have
3-67
-------
included three rollback approaches as part of the sensitivity analysis including:
proportional (reflecting regional patterns of reduction), hybrid (reflecting a
combination of localized and regional patterns of reduction), and locally focused
(reflecting localized patterns of reduction) (see section 3.2.3 for additional detail on
these rollback methods and section 3.5.4 for a description of how this factor is
addressed in the sensitivity analysis).
Copollutant concentrations: Inclusion of copollutant models in short-term exposure-
related time series studies has produced mixed results in terms of the degree of
attenuation of the PM2.5 signal that results from inclusion of other pollutants (see final
PM ISA, sections 6.2.10.9 and 6.3.8.5). The PM ISA (section 6.2.10.9) suggests that
these inconsistent findings associated with controlling for gaseous pollutants are likely
due to differences in the correlation structure among pollutants as well as differing
degrees of exposure measurement error related to the copollutants. Further, the PM
ISA (section 2.1.3) notes that correlations between PM and copollutants (including CO,
Os, SC>2 and NO2) can vary both seasonally and spatially. Therefore, it is possible that
the degree of attenuation of PM2.s-related risk by copollutants may differ across study
areas. However, because the multi-city studies used in the core risk assessment
(Zanobetti and Schwartz., 2009; Bell et al., 2008; and Krewski et al., 2009) provide
single pollutant models, our analysis does not directly address the issue of copollutant
confounding (see section 3.5.3 for additional discussion of uncertainty introduced into
the analysis as a result of not including copollutant models in the core risk assessment).
We did explore the issue of copollutant modeling in the context of modeling long-term
exposure-related mortality as part of the sensitivity analysis (section 3.5.4). In
addition, the potential impact of copollutant confounding on short-term exposure-
related mortality and morbidity was explored in the Moolgavkar et al., 2003 study, as
discussed below in section 4.3.1.1 (although they have limited applicability to the core
risk estimates generated in this RA).
Demographics and socioeconomic-status (SES)-related factors: Variability in
population density particularly in relation to elevated levels of PM2.5 has the potential
to influence population risk. In addition, other aspects of demographics such as age of
housing stock (which can influence rates of air conditioner use thereby impacting rates
of infiltration of PM indoors) can impact exposure and therefore risk (discussed in PM
ISA - sections 2.2.1 and 2.3.2). While risk modeling completed for this analysis is
based on concentrations measured at central-site monitors used as surrogates for
population exposure and does not explicitly consider more detailed patterns of PM
exposure by different populations, potential differences in exposure to PM2.s reflecting
demographic and SES-related factors is covered to some degree by the use of urban
study area-differentiated effects estimates (for short-term exposure-related mortality)
and regionally-differentiated effects estimates (in the case of short-term exposure-
related morbidity). In the case of long-term exposure-related mortality, while the
modeling for this group of endpoints does not utilize location-specific or regionally-
differentiated effects estimates, the national-scale effects estimates that are used do
reflect differences in exposure and health response across urban study areas (which will
reflect, to some extent, differences in demographics and SES-related factors to the
3-68
-------
extent that these factors influence the relationship between PM2.5 exposure and
mortality response, as detected by the underlying cohort studies).
Behavior affecting exposure to PM2.s: We have incorporated, where available,
region- and/or city-specific effect estimates in order to capture behavioral differences
across locations that could affect population exposures to PM2.5 (e.g., time spent
outdoors, air conditioning use). However, while these location-specific effect
estimates may be capturing differences in behavior, they may also be capturing other
differences (e.g., differences in the composition of PM2.5 to which populations are
exposed). As noted above, it was not possible to separate out the impact of these
different factors, which may vary across locations and populations, on effect estimates.
Baseline incidence of disease: We collected baseline health effects incidence data (for
mortality and morbidity endpoints) from a number of different sources (see section
3.4). Often the data were available at the county-level, providing a relatively high
degree of spatial refinement in characterizing baseline incidence given the overall level
of spatial refinement reflected in the risk assessment as a whole. Otherwise, for urban
study areas without county-level data, either (a) a surrogate urban study area (with its
baseline incidence rates) was used, or (b) less refined state-level incidence rate data
were used.
Longer-term temporal variability in ambient PMi.5 levels (reflecting meteorological
trends, as well as future changes in the mix of PM2.5 sources and regulations impacting
PM2.s): Risk estimates for the PM2.5 NAAQS review have been generated using recent
years of air quality data. In other words, efforts have not been made to simulate
potential future changes in either the concentrations or composition of ambient PM2.5
in the risk assessment locations based on possible changes in economic activity,
demographics or meteorology. Actual risk levels potentially experienced in the future
as a result of implementing alternative standard levels may differ from those presented
in this report due, in part, to potential changes in these factors related to ambient PM2.5.
3.5.3 Qualitative Assessment of Uncertainty
As noted in section 3.5.1, we have based the design of the uncertainly analysis carried out
for this risk assessment on the framework outlined in the WHO guidance document (WHO,
2008). That guidance calls for the completion of a Tier 1 qualitative uncertainty analysis,
provided the initial Tier 0 screening analysis suggests there is concern that uncertainty associated
with the analysis is sufficient to significantly impact risk results (i.e., to potentially affect
decision making based on those risk results). Given previous sensitivity analyses completed for
prior PM NAAQS reviews, which have shown various sources of uncertainty to have a
potentially significant impact on risk results, we believe that there is justification for conducting
a Tier 1 analysis. In fact, as argued earlier, given the complexity of the overall risk assessment, a
full Tier 3 uncertainty analysis is warranted for consideration under the WHO guidelines
(although as discussed later, limitations in available data preclude completion of this level of
more-refined uncertainty analysis at this time).
3-69
-------
For the qualitative uncertainty analysis, we have described each key source of uncertainty
and qualitatively assessed its potential impact (including both the magnitude and direction of the
impact) on risk results, as specified in the WHO guidance. 46 As shown in Table 3-13, for each
source of uncertainty, we have (a) provided a description, (b) estimated the direction of influence
(over, under, both, or unknown) and magnitude (low, medium, high) of the potential impact of
each source of uncertainty on the risk estimates, (c) assessed the degree of uncertainty (low,
medium, or high) associated with the knowledge-base (i.e., assessed how well we understand
each source of uncertainty), and (d) provided comments further clarifying the qualitative
assessment presented. Table 3-13 includes all key sources of uncertainty identified for the PM2.5
NAAQS risk assessment. A subset of these sources has been included in the Tier 2 quantitative
assessment discussed in section 3.5.4.
The categories used in describing the potential magnitude of impact for specific sources
of uncertainty on risk estimates (i.e., low, medium, or high) reflect EPA staff consensus on the
degree to which a particular source could produce a sufficient impact on risk estimates to
influence the interpretation of those estimates in the context of the PM NAAQS review.47
Sources classified as having a "low" impact would not be expected to impact the interpretation
of risk estimates in the context of the PM NAAQS review; sources classified as having a
"medium" impact have the potential to change the interpretation; and sources classified as "high"
are likely to influence the interpretation of risk in the context of the PM NAAQS review.
Because this classification of the potential magnitude of impact of sources of uncertainty is
qualitative and not informed directly by any type of analytical results, it is not possible to place a
quantitative level of impact on each of the categories. 48 Therefore, the results of the qualitative
analysis of uncertainty have limited utility in informing consideration of overall confidence in
46 Similar to our discussion of variability in the last section, the term "key sources of uncertainty" 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 (i.e., these sources of uncertainty, if fully addressed could
result in adjustments to the core risk estimates which might impact the interpretation of those risk estimates in the
context of the PM NAAQS review). These key sources of uncertainty have been identified through consideration
for sensitivity analyses conducted for previous PM NAAQS risk assessments, together with information provided in
the final PM ISA and comments provided by CASAC on the analytical plan for the RA, as well as the first draft RA.
47 For example, if a particular source of uncertainty were more fully characterized (or if that source was
resolved, potentially reducing bias in a core risk estimate), could 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?
48 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.
3-70
-------
the core risk estimates and, instead, serve primarily as a means for guiding future research to
reduce uncertainty related to PM2.5 risk assessment.
As with the qualitative discussion of sources of variability included in the last section, the
characterization and relative ranking of sources of uncertainty addressed here is based on
consideration by EPA staff of information provided in previous PM NAAQS risk assessments
(particularly past sensitivity analyses), the results of the sensitivity analyses completed for the
current PM NAAQS risk assessment and information provided in the final PM ISA as well as
earlier PM Criteria Documents. Where appropriate, in Table 3-13, we have included references
to specific sources of information considered in arriving at a ranking and classification for a
particular source of uncertainty.
3-71
-------
Table 3-6. 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
Medium
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
5-72
-------
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
5-73
-------
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. Representa-
tiveness of the
population used
in the
epidemiological
study
If a population was used in the
epidemiological study that is
not representative of the
general (urban) population,
then the effect estimate that
results may not be optimal for
the population being modeled
in the risk assessment (i.e.,
risks may be biased high or
low). The issue of
representativeness would
ideally focus on factors directly
related to PM risk (including
effect modifiers).
Both
Low-
Medium
Low-Medium
KB: often we will have information from the epidemiological
study that allows us to identify potential differences between the
study population and the general (urban) U.S. population related,
for example, to SES factors such as income or education.
However, it can be more difficult to translate these differences
into quantitative estimates of potential bias in effect estimates.
INF: In the case of the ACS dataset underlying our modeling of
long-term exposure-related mortality in the core analysis,
concerns have been raised that the study population has a higher
SES status relative to the U.S. population. With the potential for
this to result in effect estimates that are biased downward
(educational status is one factor that has been cited - see p. 8-118,
U.S. EPA, 2004a).
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 of 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
5-74
-------
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)
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
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 lung-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,
5-75
-------
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)
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
differences for
PM
The composition of PM can
differ across study areas
reflecting underlying
differences in primary and
secondary PM25 sources (both
natural and anthropogenic). If
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.
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-
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
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
5-76
-------
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)
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.
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-
term
exposure
health effect
risk
estimates)
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.
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
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
5-77
-------
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)
typically based on multi-city
prospective cohort studies).
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 (Krewski, 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 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.
* Refers to the degree of uncertainty associated with our understanding of the phenomenon, in the context of assessing and characterizing its uncertainty
(specifically in the context of modeling PM risk)
5-78
-------
The results presented in Table 3-13 consider only the potential impact of each source of
uncertainty when acting in isolation to impact core risk estimates. However, it is likely that a
number of these sources of uncertainty could act in concert to impact risk estimates and
furthermore, that these combined effects could be more than additive in certain circumstances.
EPA staff has identified several combinations of sources of uncertainly addressed in Table 3-13
that should be highlighted due to their potential to produce significant impacts on core risk
estimates when acting in concert. These are briefly described below:
Uncertainty source D (statistical fit of the C-R functions), Source E (shape of the
C-R functions), Source F (addressing copollutants), and Source J (use of single-
city versus multi-city studies in the derivation of C-R functions): Consideration of
uncertainty associated with the shape of C-R functions needs to be considered in light
of overall confidence (uncertainty) associated with a particular model. A number of
factors contribute to an interpretation of confidence in a model including: statistical fit
of the model, degree to which potential confounding by copollutants is considered, and
other aspects of study design including single- versus multi-city study design. While
choice of a particular model (e.g., threshold model, or log-log model) may produce a
significant impact on risk estimates relative to alternative model forms, the overall
scientific support for that particular model form (informed by consideration of the
factors listed above) is an important consideration in assessing overall uncertainty both
from a qualitative and quantitative standpoint.
In addition, there is the potential for sources of uncertainty discussed in Table 3-13 to
interact with sources of variability covered in section 3.5.2 in impacting core risk estimates. One
such interaction is discussed below:
Uncertainty source A (characterizing ambient PM2.s levels for study populations
using the existing ambient monitoring network) and variability related to the
pattern of ambient PM2.s reductions at urban study areas (see section 3.5.2): The
estimation of a composite monitor value to use in modeling risk for a study area under
an alternative suite of standards is dependent both on the specification of the
monitoring network and the approach used in adjusting the concentrations for the
monitors in that network (i.e., the rollback approach used to simulate the pattern of
ambient PM2 5 reductions associated with just meeting the current or alternative suites
of standards). As we have seen in modeling risk for Pittsburgh, refinements in the
approach used to simulate air quality just meeting alternative suites of standards (in the
case of Pittsburgh transit!oning from a single study area to two distinct study areas
each with different design values and separate assessments of rollback) produced
significant differences in composite monitor values for the study area. Therefore, both
of these factors (the definition of the monitoring network and rollback approach) can
work in concert to impact ambient PM2.5 levels and hence risk estimates.
3-79
-------
3.5.4 Single and Multi-Factor Sensitivity Analyses
We quantitatively examined the impact of several inputs to the risk assessment in a series
of single-factor sensitivity analyses summarized above in Table 3-8. A number of these sources
of uncertainty were also examined in-concert to assess their combined impact on core risk
estimates through the multi-factor sensitivity analysis. In addition, the sensitivity analysis
considered variability in the pattern of reductions in ambient PM2 5 associated with just meeting
the current and alternative suites of standards (i.e., consideration of variability in the simulation
of rollback). This section focuses on providing additional detail on the sources of alternative
model specifications and input datasets used in the sensitivity analysis (as alternative to the core
modeling approach).
Rather than present results for each sensitivity analysis for all of the air quality scenarios
considered in the core analysis, we selected a single air quality scenario - PM2.5 concentrations
that just meet the current standards - to use for the sensitivity analyses. The one exception to
this was the sensitivity analyses examining the impact of alternative approaches to simulating
just meeting alternative standards (the hybrid and locally focused rollback methods).49
In discussing the approach used in conducting the sensitivity analysis, we focus first on
methods used in assessing long-term exposure related health endpoints followed by the methods
used in assessing short-term exposure related health endpoints. We then discuss multi-factor
sensitivity analyses completed for both short-term and long-term exposure-related health
endpoints. Note, that the results of the sensitivity analyses (including both single- and multi-
factor analyses) are presented and discussed in section 4.3.
3.5.4.1 Sensitivity Analyses for Long-Term Exposure-Related Mortality
Because Krewski et al. (2009) presented results based on alternative model specifications
only for the later exposure period (1999 - 2000), our sensitivity analyses focusing on the
estimates of health effects incidence associated with long-term exposure to PM2.5 similarly used
the C-R functions based on this later exposure period. Krewski et al. (2009) considered several
alternative modeling approaches to estimate the relationship between mortality (both all cause
and cause-specific) and long-term exposure to PM2.5, providing us the opportunity to examine
the impact of alternative modeling approaches on the estimate of mortality risk associated with
long-term exposure. In particular, we examined the impact of using a random effects log-linear
49 Sensitivity analyses focusing on the hybrid and locally focused 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.
3-80
-------
model and of using a random effects log-log model50 (rather than the standard fixed effects log-
linear model used in the core analysis) to estimate the risks of all cause mortality,
cardiopulmonary mortality, ischemic heart disease mortality, and lung cancer mortality
associated with long-term exposure in Los Angeles and Philadelphia.51 The coefficient of PM2.5
in the random effects log-linear model was back-calculated from the relative risk reported in
Table 9 ("Autocorrelation at MSA and ZCA levels" group - "MSA & DIFF" row) of Krewski et
al. (2009). The coefficient of PM2 5 in the random effects log-log model was back-calculated
from the relative risks reported in Table 11 ("MSA and DIFF" rows) of Krewski et al. (2009).
As noted above, for all health endpoints associated with long-term exposure to PM2.5 we
estimated risk associated with PM2 5 concentrations above 5.8 |ig/m3 (the LML for the later
exposure period used in Krewski et al., 2009). In a sensitivity analysis we examined the impact
of that limitation by comparing those mortality risk estimates to the mortality risk estimates
obtained when we estimated risk associated with PM2 5 concentrations above estimated PRB
levels. This sensitivity analysis was carried out for all cause mortality in all 15 risk assessment
urban areas.
In addition, we compared the impact of using the primary C-R functions used in the risk
assessment, taken from Table 33 of Krewski et al. (2009), versus C-R functions for mortality
associated with long-term exposure reported in another study, Krewski et al. (2000), which was
based on a reanalysis of the Harvard Six Cities Study. The C-R functions estimated in Krewski
et al. (2000) from the Harvard Six Cities cohort were estimated for ages 25 and up, while the C-
R functions estimated in Krewski et al. (2009) from the ACS cohort were for ages 30 and up.
For purposes of consistency in the comparison, however, we applied the C-R functions from
Krewski et al. (2000) to ages 30 and up (and used the baseline incidence rates for that age group
as well).52 This sensitivity analysis was carried out for all cause mortality, cardiopulmonary
mortality, and lung cancer mortality in Los Angeles and Philadelphia.
We also considered the impact of using multi-pollutant models in estimating long-term
exposure-related mortality. Specifically, we obtained 2-pollutant models (considering CO, NO2,
50 i
In the log-log model, the natural logarithm of mortality is a linear function of the natural logarithm of
PM25.
51 As 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.
52 The baseline incidence rates for ages 25 and up and ages 30 and up are likely to be very similar.
3-81
-------
Os or SC>2 together with PM^.s) from Krewski et al., 2000, which is an earlier reanalysis of the
ACS dataset and used them in generating alternative estimates of all-cause mortality to contrast
with the core estimates generated using Krewski et al., 2009.
For all of the sensitivity analyses involving alternative C-R functions, in addition to
calculating the incidence of the health effect when an alternative approach is taken, we
calculated the percent difference in estimates from the core analysis resulting from the change in
analysis input. So for example, when we calculated the incidence of all cause mortality
associated with long-term exposure to PM2.5 using a random effects log-log model (instead of the
fixed effects log-linear model used in the core analysis), we calculated the percent difference in
the result as (incidence estimated using a random effects log-log model - incidence estimated
using a fixed effects log-linear model)/( incidence estimated using a fixed effects log-linear
model).
Finally, we also examined the issue of variability in estimating the pattern of reductions
in ambient PM2.5 levels under the current and alternative standard levels (i.e., conducting
rollback). For the first draft RA, we considered the impact of using a hybrid rollback approach in
addition to the proportional rollback approach which has been more traditionally used in PM
NAAQS risk assessment (this sensitivity analysis was implemented including the generation of
quantitative risk estimates for a full suite of long-term exposure-related mortality categories).
For the final RA, as discussed above in sections 2.3, and 3.2.3.1, we have included consideration
of a locally focused rollback approach in addition to the hybrid as non-proportional methods to
contrast with proportional rollback. As discussed in Section 3.2.3.1, for the final risk
assessment, rather than generating quantitative risk estimates, we have calculated composite
monitor estimates using the different rollback methods (proportional, hybrid and locally
focused). The composite monitor values are surrogates for long-term exposure-related mortality.
Therefore, by comparing composite monitor values generated for the same study area/standard
level combination (using different rollback methods), we can obtain insights into the potential
impact of the rollback method used on long-term exposure-related mortality. Specifically, for
this sensitivity analysis, we compared composite monitor values in two ways:
Potential difference in composite monitor values at the current or alternative standard
level (for the same study area) given application of alternative rollback methods: We
compared the absolute magnitude of composite monitors values produced using different
rollback methods for the same study area/standard level combination to provide insights
into differences in the magnitude of residual risk for a given suite of standards in a study
area using different rollback methods (Appendix F, Table F-50).53 For example, in Table
53
This calculation reflects the fact that we model long-term exposure-related mortality down to LML.
3-82
-------
F-50, for Los Angeles, we see that for the current standard suite of standards, use of
proportional rollback and locally focused rollback methods results in composite monitor
values of 9.5 |ig/m3 and 12.0 |ig/m3, respectively, with the locally focused value being
40% higher than the value derived using proportional rollback. Given that the composite
monitor values are surrogates for long-term exposure-related mortality, we conclude that
for this combination of urban study area and suite of standards, use of the locally focused
rollback method could produce PM2.5-attributable long-term mortality risk estimates that
are approximately 40% higher than use of the proportional rollback method.
Potential difference in the pattern of reduction in composite monitor values across
alternative standards: We compared differences in the percent reduction in composite
monitor values across alternative suites of standards for the same study area using
different rollback methods to provide insights into differences in incremental risk
reduction resulting from the use of different rollback approaches (Appendix F, Table F-
49).54 For example, in Table F-49, for Baltimore, we see that the proportional rollback
and hybrid rollback approaches resulted in composite monitor values for the 13/35
alternative suite of standards of 11.6 |ig/m3 and 11.8 |ig/m3, respectively, with these
translating into a percent reduction (compared with their respective values under the
current suite of standards) of 21% and 16%, respectively. Given that the composite
monitor values are surrogates for long-term exposure-related mortality, we conclude that
use of the two rollback methods (in the case of Baltimore for these two suites of
standards) does not appear to produce notably different patterns of risk reduction (in
terms of percent reduction), although residual risk could differ using the two approaches.
The locally focused and hybrid rollback approaches were not applied to all study areas,
since they are primarily applicable in certain situations.55 The sensitivity analysis results
described above (presented in Appendix F, Tables F-49 and F-50) form the basis for summary
information related to rollback approaches presented in Table 4-3.
In addition to the above insights regarding potential impacts on residual risk and the
degree of risk reduction across standard levels, inclusion of multiple rollback approaches also
allowed us to more fully examine the degree to which alternative 24-hour standards can produce
reductions in annual average PM2.5 concentrations, thereby producing reductions in long-term
exposure-related mortality. As discussed below in section 5.2.3, alternative 24-hour standards,
when controlling, can result in reductions in annual average PM2.5 concentrations, particularly if
proportional rollback is used. In this case, the assumption of more regional patterns of PM2.5
reduction in reducing PM2.5 concentrations to just meet alternative 24-hour standards results in
54 We note that this analysis also reflects calculation of long-term exposure-related mortality down to LML.
55 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).
3-83
-------
an equivalent magnitude of reduction in the annual average. However, in simulating more
localized patterns of PM2.5 reductions to just meet alternative 24-hour standards, the PM2.5
reductions can be more limited to the monitor(s) (and areas) exceeding the 24-hour standard, and
other monitors may not be effected, resulting in a smaller impact on the annual average.
Inclusion of rollback approaches reflecting more localized patterns of ambient PM2.5 reduction
(i.e., the hybrid and particularly the locally focused methods) allows us to assess the degree to
which alternative 24-hour standards (when controlling) produce appreciable reductions in annual
average PM2.5 concentrations and consequently in long-term exposure-related mortality. This
issue is revisited in discussing the results of the sensitivity analysis (section 4.3.1.1) and in the
integrative discussion of the core risk estimates (section 5.2).
3.5.4.2 Sensitivity Analyses for Short-Term Exposure-Related Mortality and
Morbidity
The scope of the sensitivity analysis completed for short-term exposure-related mortality
and morbidity is more limited than that completed for long-term exposure-related mortality.
This reflects, in part, the much greater magnitude of long-term exposure-related mortality. An
additional factor is that while there has been considerable research in the area of short-term
exposure-related mortality and morbidity which sheds light on uncertainty in such factors as C-R
function specification, this information is not directly applicable in a sensitivity analysis. In
order to complete a quantitative sensitivity analysis, we need alternative C-R function
specifications that produce risk estimates that can be directly compared to the core risk estimates.
Ideally, this is done by identifying alternative model forms in the epidemiological study used in
the core risk model. However, in the case of short-term exposure-related mortality, the studies
providing our core risk models (Zanobetti and Schwartz et al., 2009 and Bell et al., 2008), only
provide limited alternative model specifications, as described below. Further, alternative
epidemiological studies, such as Moolgavkar et al., 2003, while providing useful insights into
which factors can impact risk estimates (e.g., lag, multi-pollutant forms), cannot generate
alternative risk estimates that can be readily compared with the core risk estimates given
differences in the underlying study designs and datasets employed.
The primary studies selected to assess mortality risk and risk of hospitalization associated
with short-term exposure to PM2.5 (Zanobetti and Schwartz, 2009, and Bell et al., 2008,
respectively) both provided all-year C-R functions as well as season-specific C-R functions. We
examined the impact of using season-specific functions by applying these functions to each
3-84
-------
season, as defined by the study authors,56 and summing the estimated season-specific incidences
of mortality and hospitalizations. We compared these estimates to the estimates obtained by
applying the corresponding all-year C-R functions to a year of air quality data.57 This sensitivity
analysis was carried out for all 15 of the risk assessment urban areas.
In addition, Ito et al. (2007) estimated an annual C-R function as well as a seasonal
function for April through August for asthma ED visits in New York City. We compared the
results of applying the annual C-R function to a whole year of air quality data to the results of
applying the seasonal function to only those months (April through August) for which it was
estimated.
Moolgavkar (2003) estimated C-R functions for several health endpoints - non-accidental
and cardiovascular mortality; and cardiovascular and respiratory HAs - associated with short-
term exposures to PM2.5 in Los Angeles using different lag structures, different modeling
approaches to incorporating weather and temporal variables, and single-pollutant versus multi-
pollutant models. This study thus provided an opportunity to show the impact of lag structure,
modeling approach, and single- vs. multi-pollutant models, individually, for several health
endpoints associated with short-term exposures, although it is difficult to generalize to other
locations since the study was only conducted in a single urban area. As noted earlier, differences
in study design and the underlying datasets used prevent the results based on application of
models from Moolgavkar et al., 2003 from being compared directly to the core risk estimates.
Finally, as with estimates of long-term exposure-related mortality, we also considered the
impact of variability related to simulating ambient PM2.5 levels under the suite of current
standard levels (i.e., variability in conducting rollback) on estimates of non-accidental mortality
associated with short-term exposures to PM2.5 (using Zanobetti and Schwartz, 2009). However,
in this case, we only considered the hybrid model (consideration of locally focused on the impact
on long-term exposure-related mortality). We note however, that sensitivity analysis findings
based on consideration of locally focused generally will hold for short-term exposure-related
mortality and morbidity since both categories of health endpoints are also driven primary by
annual average PM2.5 levels (see section 6.2). .
56 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.
57 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.
3-85
-------
In all cases except the ED visits sensitivity analysis, in addition to calculating the
incidence of the health effect when an alternative approach is taken, we calculated the percent
difference in estimates from the core analysis resulting from the change in analysis input.58
3.5.4.3 Multi-factor Sensitivity Analyses
Each single-element sensitivity analysis shows how the estimates of PM2.s-related health
effects incidence change as we change a single element of the analysis (such as the form of the
C-R function or the way we simulate just meeting a set of standards). Because each of the
alternative modeling choices is considered to be a reasonable choice, the results of these single-
element sensitivity analyses provide a set of reasonable alternative estimates that may similarly
be considered plausible (see section 4.3). The results of the single-element sensitivity analysis
are presented and discussed in section 4.3.1.1.
The single-element sensitivity analyses provide insight into which sources of uncertainty
may have the greatest impact on risk estimates when acting alone. However, there are several
sources of uncertainty in estimating PM2.5-related health effects. To provide a more complete
picture of the uncertainty surrounding estimates of PM2.s-related health effects incidence - and
to expand the set of reasonable alternative estimates - we next carried out multi-element
sensitivity analyses. The results of the multi-factor sensitivity analysis are presented and
discussed in section 4.3.1.2.
The choice of uncertain analysis elements to include in the multi-element sensitivity
analyses was guided by the single-element sensitivity analyses. In particular, we selected those
modeling choices that had the greatest impacts on the estimates of health effects incidence in the
single-element sensitivity analyses to provide insight into the scope of possible estimates that,
while perhaps not based on our first choice of analysis elements, are nevertheless plausible
alternative estimates.
We identified three analysis elements that substantially affected the estimates of mortality
associated with long-term exposure to PM2.5 the model choice (fixed effects log linear vs.
random effects log-log), whether effects are estimated associated with PM2.5 concentrations
down to the LML in the study (5.8 |ig/m3) or down to PRB, and whether a proportional or a
hybrid rollback is used to simulate PM2 5 concentrations that just meet a given set of standards.
This resulted in 2x2x2 = 8 different estimates of mortality, all of which could be considered
plausible, based on the fact that the underlying model choices are all considered reasonable.
58 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.
3-86
-------
We identified two analysis elements that substantially affected the estimates of mortality
associated with short-term exposure to PM2.5 - whether season-specific or all-year C-R functions
were used and whether a proportional or a hybrid rollback approach was used to simulate just
meeting the current and alternative standards.
3.5.5 Summary of Approach to Addressing Variability and Uncertainty
The characterization of uncertainty and variability associated with the risk assessment
includes a number of elements, which have been discussed in detail above. These include:
Identification of key sources of variability associated with PM2.s-related population
exposure and hazard response and the degree to which they are captured in the risk
assessment (see section 3.5.2). When important sources of variability in exposure
and/or hazard response are not reflected in a risk assessment, significant uncertainty
can be introduced into the risk estimates that are generated. While not explicitly
referenced in the WHO guidance, this assessment (focused on coverage for key sources
of variability) could be considered part of a Tier 1 analysis (i.e., the qualitative
characterization of sources of uncertainty).
Qualitative assessment of uncertainty, including both an assessment of the magnitude
of potential impact of each source on risk estimates (along with the potential direction
of that impact) as well as an assessment of overall confidence associated with our
understanding of that source of uncertainty (see section 3.5.3). This represents a WHO
Tier 1 analysis.
Single-factor sensitivity analysis intended to evaluate the impact of individual sources
of uncertainty and variability on risk estimates (see section 3.5.4). The goal of this
assessment is to evaluate the relative importance of these sources of uncertainty and
variability in impacting core risk estimates. The single-factor sensitivity analysis
represents a WHO Tier 2 analysis. In conducting these assessments, we have used
alternative representations of modeling elements that have support in the literature to
ensure that the risk estimates that are generated represent reasonable alternate estimates
that can supplement the core risk estimates generated in the analysis (see section 4.3).
Multi-factor sensitivity analysis intended to assess the combined impact of multiple
sources of uncertainty and variability on risk estimates (see section 3.5.4). By
considering the combined effect of multiple sources of uncertainty and variability, this
analysis has the potential to identify any non-linearities which can magnify the impact
of uncertainty and variability on risk estimates, especially if several non-linear factors
act in concert. This also represents a WHO Tier 2 analysis. As with the single-factor
sensitivity analysis results, these risk estimates are also generated using modeling
inputs which have support in the literature and consequently, they also represent
reasonable alternate estimates that supplement the core risk estimates (see section
4.3.2).
As noted above, since information was not available to characterize overall levels of
confidence in alternative model inputs, the uncertainty characterization completed for this risk
assessment did not include a full probabilistic assessment of uncertainty and its impact on core
3-87
-------
risk estimates (i.e., a WHO Tier 3 analysis was not completed). Further, the risk estimates
generated using the single- and multi-factor sensitivity analyses do not represent uncertainty
distributions, but rather additional plausible point estimates of risk (i.e., we do not know whether
they represent risk estimates near the upper or lower bounds of a true but undefined uncertainty
distribution and we do not know the actual population percentiles that they represent). The
appropriate use for these reasonable alternate risk estimates in informing consideration of
uncertainty in the core risk estimates is discussed in section 4.3.2.
In addition to the qualitative and quantitative treatment of uncertainty and variability
which are described here, we have also completed an analysis to evaluate the representativeness
of the selected urban study areas against national distributions for key PM risk-related attributes
to determine whether they are nationally representative or more focused on a particular portion
of the distribution for a given attribute (section 4.4.1). In addition, we have completed a second
analysis addressing the representativeness issue, which identified where the subset of 31 counties
comprising our 15 urban study areas fall along a distribution of national county-level long-term
exposure-related mortality risk (section 4.4.2). This analysis allowed us to assess the degree of
which the 15 urban study areas capture locations within the U.S. likely to experience elevated
levels of risk related to PM2.5 exposure.
A third set of analyses that has been added to this final RA focuses on evaluating patterns
in the design values (including both 24-hour and annual) and underlying PM2 5 monitoring data
for the 15 urban study areas (see Section 4.5). The goal of this analysis is to use this information
to enhance our understanding of patterns in risk reduction seen under both the current and
alternative suites of standards across the urban study areas. The interplay of design values and
underlying PM2.5 monitoring data play a key role in determining whether a location will
experience risk reductions when just meeting any given suite of standards is simulated and, if so,
the magnitude of those reduction. As part of this analysis, we contrast patterns in design values
for the 15 urban study areas with patterns seen more broadly across urban areas in the U.S. with
the goal of placing the urban study areas in a national context with regard to this key factor
influencing risk.
-------
4 URBAN CASE STUDY RESULTS
For this risk assessment, we have developed a core set of risk estimates supplemented by
an alternative set of risk results generated using single-factor and multi-factor sensitivity
analysis. The core set of risk estimates was developed using model inputs that staff judge to
have a greater degree of support in the literature relative to inputs used in the sensitivity analyses
(the rationale for selection of specific epidemiological studies and associated C-R functions for
the core analysis is discussed above in section 3.3.3). This chapter presents and discusses the
core set of risk estimates generated for the urban case study area, and also discusses the results of
the sensitivity analyses which serve to augment the core risk estimates. The results of the
sensitivity analyses allow us to evaluate and rank the potential impact of key sources of
uncertainty on the core risk estimates. In addition, because the sensitivity analyses were
conducted using alternative modeling inputs having some degree of support in the literature, the
results of the sensitivity analysis also represent a set of reasonable alternatives to the core set of
risk estimates that can be used to inform characterization of uncertainty in the core results (see
section 4.3.2 below).
As discussed above in section 2.4 and 3.2, this risk assessment includes consideration of
the following air quality scenarios:
Recent conditions: based on PM2.5 concentrations characterized through monitoring for
the period 2005-2007 at each urban case study location;
Current NAAQS: based on rolling back PM2.5 concentrations to just meet the current
suite of standards in each urban study area (annual standard of 15 |ig/m3 and a 24-hour
standard of 35 |ig/m3, denoted 15/35);
Alternative NAAQS: based on rolling back PM2.5 concentrations to just meet alternative
suites of standards in each urban study area:
o annual standard of 14 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted 13/35);
o annual standard of 13 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted 13/35);
o annual standard of 12 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted 12/35);
o annual standard of 13 |ig/m3 and a 24-hour standard of 30 |ig/m3 (denoted 13/30);
o annual standard of 12 |ig/m3 and a 24-hour standard of 25 |ig/m3 (denoted 12/25).
We have also estimated risk for an alternative annual standard level of 10 |ig/m3 (paired
with 24-hour standard levels of 35 |ig/m3 and 25 |ig/m3). However, as discussed in section 2.4,
because of increased uncertainty associated with these risk estimates relative to estimates
generated for the other alternative annual standard levels evaluated in the RA, estimates based on
the alternative annual standard level of 10 |ig/m3 are not discussed in this chapter and instead are
4-1
-------
presented in Appendix J and only briefly addressed as part of the integrative discussion presented
in Chapter 5.
In simulating both current and alternative suites of standards, for the core analysis, we
used a proportional roll-back approach (see section 3.2.3), while a hybrid roll-back approach
reflecting the potential for local source control was used for a subset of urban study areas as part
of the sensitivity analysis conducted for this assessment (see section 3.2.3). In addition, we have
considered the locally focused approach as a further alternative to proportional rollback in
simulating just meeting the current and alternative suites of standards. While we did not generate
risk estimates based on application of the locally focused approach, we did generate composite
monitor-based annual average PM2.5 levels which allow us to assess how long-term exposure-
related risk could vary if this alternative roll-back method was used (see Section 4.3).
As described in section 2.3 and 3.3.2, we assessed risk for 15 urban study areas chosen to
provide coverage for the diversity of urban settings across the U.S. that reflect areas with
elevated annual and/or daily PM2.5 concentrations. At a minimum, all areas selected had recent
air quality levels at or above the lowest annual and/or 24-hour standards analyzed. In addition,
our goal was to select areas reflecting the heterogeneity in PM risk-related attributes such as
sources, composition, demographics, and population behavior.
Risk estimates were generated for the following health effects endpoints: (a) long-term
exposure-related mortality (all-cause, cardiopulmonary disease-related (CPD), ischemic heart
disease-related (IHD) and lung cancer-related), (b) short-term exposure-related mortality (non-
accidental, cardiovascular disease (CVD), respiratory), and (c) short-term exposure-related
morbidity (hospital admissions (HA) for CVD and respiratory illness and emergency department
(ED) visits). Risk estimates are presented separately for each of these 15 study areas, although in
certain circumstances, risk estimates may be restricted to a subset of these locations if, for
example, an endpoint is modeled using a concentration-response (C-R) function derived from an
epidemiological study that was conducted only in a subset of the urban areas. For the core
analysis, long-term exposure mortality risk was modeled down to lowest measured level (LML),
because the LML was higher than estimated PRB and because there is substantial uncertainty as
to the shape of the concentration-response (C-R) function at concentrations below the LML. For
long-term exposure mortality a sensitivity analysis was conducted that estimated risk down to
policy-relevant background (PRB). In contrast, all short-term exposure health effects endpoints
were modeled down to PRB, since this was higher than the LML across all studies and for
purposes of NAAQS decision making, EPA is focused on risks associated with PM2.5 levels that
are due to anthropogenic sources that can be controlled by U.S. regulations (or through
international agreements with neighboring countries).
4-2
-------
In modeling long-term exposure mortality, for the core analysis, we have based estimates
on the latest reanalysis of the American Cancer Society (ACS) dataset, with two sets of risk
estimates being generated; one using a C-R function derived by fitting PM2 5 monitoring data
from 1979-1983 and a second set based on fitting PM2.5 monitoring data from 1999-2000
(Krewski et al., 2009) (see section 3.3.3). In presenting core risk estimates for long-term
mortality, both sets of estimates are given equal weight.
In modeling short-term exposure mortality and morbidity for the core analysis, we have
used the latest multi-city studies (Zanobetti and Schwartz, 2009; Bell et al., 2008) (see section
3.3.3). In the case of short-term exposure mortality, we obtained and used city-specific effects
estimates derived using empirical Bayes methods from the study authors (Zanobetti, 2009).
Multi-city studies were favored for the core analysis, since these studies are not subject to
publication bias and because they reflect a diverse set of locations with regard to the observed
relationship between short-term PM2.5 exposure and health affect response in the population.
Additional detail on the specific C-R functions and related modeling elements such as effects
estimates and lag periods used in the core analysis relative to the sensitivity analysis are
presented above in sections 3.3 and 3.4 and called out where appropriate below as specific risk
estimates are discussed.
The pattern of mortality incidence across the urban study areas is markedly different for
short-term exposure-related mortality compared with long-term exposure-related mortality
reflecting a number of factors including: (a) differences in patterns of daily PM2.5 levels versus
annual average values across the urban study areas and (b) the fact that urban study area-specific
effect estimates are used in modeling short-term exposure-related mortality, while a single effect
estimate is used for all study areas for long-term exposure-related mortality (for a particular
mortality category). Further, effect estimates for short-term exposure-related mortality can be
notably small for some study areas (e.g., the effect estimates for non-accidental mortality for Los
Angeles is significantly smaller than effect estimates for the other study areas, thereby
accounting for the relatively small total incidence estimate for this study area - see Appendix C,
Table C-l).
Because the recent conditions air quality scenario spans three years (2005-2007), risk
estimates are generated for each of these years, reflecting the underlying air quality data for a
particular year. Risk metrics generated for the above health effects endpoints include:
Annual incidence of the endpoint due to PM2.s exposure (annual incidence):
Generated for the population associated with a given urban study area (for a given
simulation year), in most cases, these risk estimates include both a point estimate as well
as a 95th percentile confidence interval, the latter reflecting sampling error as
characterized in the underlying epidemiological study.
4-3
-------
Percent of total annual incidence for the health endpoint due to PM2.s exposure
(percent of total incidence attributable to PM2.s): Again, generated for the population
associated with a given urban study area (and simulation year), this metric characterizes
the fraction of total incidence that is associated with PM2.5 exposure. As with the
underlying PM-related incidence estimates, this risk metric also typically includes a 95th
percentile confidence interval reflecting sampling error associated with the effects
estimate. Compared with the annual incidence metric which reflects underlying
population size for each study area, this risk metric has the advantage of not being
dependent on the size of the underlying population, thereby allowing direct comparison
of the potential impact of PM2.5 for the health effect endpoint of interest across urban
study area locations. For this reason, in discussing risk estimates in this section, the
percent of total incidence attributable to PM2.s risk metric is given somewhat greater
emphasis than the absolute measure of annual incidence attributable to PM2.s.
Percent reduction in PM2.s-related health effect incidence for an alternative set of
standards or the recent conditions scenario, relative to the current standards
(percent change from the current set of standards). Also estimated separately for each
urban study area and simulation year, this metric characterizes the degree of risk
reduction (for alternative standard levels) or increased risk (for the recent conditions
scenario) relative to the current NAAQS. For this metric, a negative value represents an
increase in risk (this is the case for the recent conditions scenario, where risks are higher
than those associated with just meeting the current suite of standards). This metric is
positive, or zero, for alternative suites of standards since they either produce no risk
reduction (if ambient air levels under recent conditions are already at or below that
alternative standard levels), or a positive risk reduction for alternative standards resulting
in a reductions in ambient PM2.5 concentrations. Because this metric is incremental, it
was not possible to generate the 95th percentile confidence intervals included with the
other two "absolute" risk metrics described above. As with the previous risk metric, this
metric is not dependent on the underlying population size and therefore, allows direct
comparison across urban study areas.
In addition to presenting the central-tendency (highest confidence) estimates for each of
these metrics, we also include 95th percentile confidence intervals, reflecting statistical
uncertainty surrounding the estimated coefficients in the reported C-R functions used in deriving
the risk estimates. Note, that these confidence intervals only capture this statistical fit uncertainty
- other sources of uncertainty including shape and form of the function, are addressed separately
as part of the sensitivity analysis (see Section 4.3.1) and the qualitative analysis of uncertainty
(see Section 3.5.3).
Detailed tables presenting estimates for these risk metrics for the complete set of air
quality scenarios (for all 15 urban study areas) are included in Appendix E and referenced as
needed in the discussion of risk estimates presented in the following sections. To support the
discussion of risk estimates presented in this chapter, we have included a subset of tables and
summary figures including:
4-4
-------
Tables summarizing risk for the current standard levels: Two tables are included
which summarize both long-term and short-term exposure-related risk for the 15 urban
study areas associated with just meeting the current suite of standards. Both tables
include a subset of the health endpoints believed to have the greatest support in the
literature including IHD mortality for long-term exposure, cardiovascular mortality and
hospital admissions for short-term exposure. Table 4-1 presents total incidence
attributable to PM2.5 exposure for the endpoints and Table 4-2 presents percent of total
incidence attributable to PM2.5 exposure for these endpoints. Together, these tables
inform consideration of the magnitude of public health impact (related to both long-term
and short-term exposure to PM^.s) associated with just meeting the current suite of
standards in the 15 urban study areas.
Figures illustrating the percent reduction in long-term and short-term exposure-
related risk for the alternative standard levels relative to the current standard (as
well as increases in risk under recent conditions relative to the current standard):
Figures 4-1 and 4-4 provide a snapshot of trends in risk reduction for long-term exposure-
related risk (Figure 4-1) and short-term exposure-related risk (Figure 4-4) across
alternative standard levels relative to the risk under the current standard. These figures
include plots for each of the 15 urban study areas, thereby allowing trends in risk
reduction across standard levels (and urban study areas) to be assessed simultaneously.59
Each of these figures is presented in additional detail by splitting each into (a)
comparison of the recent conditions risk against the current standard level and (b)
comparison of risk under alternative standard level against the current standard, in order
to allow a more detailed look at patterns in risk reduction for individual urban study areas
(splitting Figures 4-1 and 4-4 in this fashion allows greater resolution in tracing the linear
risk plots for each study area). Specifically, Figures 4-2 and 4-3 provide these higher-
resolution plots for long-term exposure-related risk and Figures 4-5 and 4-6 provide
higher-resolution plots for short-term exposure related risk.
Although risk estimates were generated for all three simulation years, in this chapter core risk
estimates primarily from 2007 are presented and discussed for both the recent conditions air
quality scenario and just meeting current and alternative suites of standards. This reflects the
observation that in generally 2007 represents a reasonable central year (in terms of the magnitude
of risk generated for the three simulated years), when considering results for all modeled health
effect endpoints across the 15 study areas. In addition, 2007 is the most recent year of the three
simulated. We note, however, that while we do focus on 2007 in presenting and discussing risk
59 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.
4-5
-------
estimates, we include an assessment of general trends across the three simulation years to gain
perspective on year-to-year variation in PM2.5-related risk estimates as assessed here.
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.ia
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 - 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)
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 - 1 45)
84
(-4 - 1 70)
43
(-9 - 93)
9
(-2 - 20)
106
(24-187)
11
(-6 - 27)
Incidence of
Cardiovascular
Hospitalizations
Associated with Short
term Exposure to
PM2.55
41
(-27-109)
216
(159-273)
16
(-11-43)
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 annua 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 interva s based on statistical
uncertainty surrounding the PM coefficient.
3Estimat.es Based on Krewski et al. (2009), Using Ambient PM2.5 from 1979 - 1983 and from 1999-2000 respectively. Incidence is for 30+
year olds within each urban study area.
4Based 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. Incidence is for all-ages (i.e., all individuals) within each urban study area.
Slncidence estimates were calculated using the appropr ate 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. Incidence is for 65+ year olds within each urban study
area.
4-6
-------
Table 4-2 Estimated Percent of Total 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
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
Percent of Incidence of Ischemic Heart
Disease Mortality Associated with Long-term
Exposure to PM2.53
Exposure Period:
1979-1983
13.2%
(10.9%- 15.5%)
11.7%
(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%- 11%)
10.5%
(8. 6% -12. 3%)
6.7%
(5.5% -7.9%)
9.3%
(7.6%- 11%)
2.9%
(2. 4% -3. 4%)
11.2%
(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%
(11. 3% -16.2%)
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%)
1The current primary PM2. 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/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 Krewski et al. (20CB), Using Ambient PM2.5from 1979- 1983 and from 1999-2000 respectively
4Based on Iccation-spedficsingle 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). Locaticn-specificC-R function estimates were not available from this study.
4-7
-------
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)
"o
°?o
§ E
3 2
o *=
a. c
x o
UJ £
E g
90%
70% -
50%
30%
10%
-10%
-30%
-50%
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%)
SatPLake 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.
4-8
-------
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)
0%
-50%
. m
1"
I =
ra 9>
5 s= -100%
°
X
uj
-150%
8
-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.
4-9
-------
Figure 4-3 Percent reduction in long-term exposure-related mortality risk (alternative standards 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)
60% -
2. B
X o
UJ ^
40% -
20%
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, iui cawi muaii aica, me uiumeiiue estimate vaiiu yj~/o \^i) aiiu 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.
4-10
-------
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)
30%
20%
10%
0%
-10%
-20%
-30%
-40%
-50%
Phoenix, AZ
Baltimore
Houston
New York
Tacoma
WA
Birmingham,
Fresno, Cf
Pittsburgh
PA
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.
4-11
-------
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%
H TJ
(0 ^
| 2 -30%
J2 £
o ° -40%
D g
O £
8j -50%
il
O "D
IT ,g -60%
O "c
(0 O
« -70%
-80%
-90%
Houston, TX
Philadelphia,?/
Atlanta, GA
New York.
Fresno, CA
Pittsburgh, B,
Salt Lake City, UT/
7
. 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%)
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.
4-12
-------
Figure 4-6 Percent reduction in short-term exposure-related mortality and morbidity risk (alternative standards relative to the current standards)
(Note: inset shows PM25 related incidence and percent of total incidence for CV under the current suite of standards)
35%
30%
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%)
-m- Dallas, TX 29 (-19-76); 0.8% (-0.5%-2.2%)
-m- Detroit, Ml 60 (-8-127); 1% (-0.1%-2.2%)
Fresno, CA 12 (-9-33); 0.7% (-0.5%-2%)
-1 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%)
-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%)
Tacoma.WA 11 (-6-27); 0.7% (-0.4%-1.8%)
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.
4-13
-------
As noted above, the risk assessment includes risk estimates for a range of short-term and
long-term exposure-related health effect endpoints. To focus the discussion of these risk
estimates, we have selected a subset of the health endpoints as examples to help illustrate
patterns in the risk estimates that might be of interest from a policy standpoint. Specifically, we
have focused on those endpoints that the ISA identifies as having the greatest support in the
literature (i.e., endpoints related to cardiovascular effects, including both mortality and
morbidity). The subset of health effect endpoints selected as illustrative examples for this
overview include: IHD-related mortality (for long-term exposure) and CV-related mortality and
HA (for short-term exposure). While the discussion does focus on these cardiovascular-related
endpoints, we do address other endpoints modeled in the risk assessment to a limited extent. The
full set of risk estimates generated is presented in the detailed tables in Appendix E.
For a subset of the urban case studies (e.g., Dallas and Phoenix), incremental reductions
across alternative standards are initially very low (or even zero) reflecting the fact that recent
ambient PM2.5 concentrations for these study areas are well below the current annual standard
levels. For these study areas, meaningful reductions in risk may not be seen until relatively
lower alternative standards are assessed (and results in the percent reduction from the current set
of standards tables and figures may be zero for several of the less stringent, alternative sets of
standards). The pattern of risk reductions across alternative standard levels for a given urban
study area is an important factor that is discussed in the integrative discussion in Chapter 6. To
set up that later discussion, in summarizing risk estimates below, we provide observations
regarding trends in risk estimates across alternative suites of standards (for a given urban study
area).
For a number of the urban study areas, confidence intervals (and in some instances, point
estimates) for short-term mortality and morbidity incidence and related risk metrics include
values that fall below zero. Population incidence estimates with negative lower-confidence
bounds (or point estimates) do not imply that additional exposure to PM2.5 has a beneficial effect,
but only that the estimated PM2.5 effect estimate in the C-R function was not statistically
significantly different from zero. In the case of short-term exposure mortality, where study area-
specific effects estimates were used (see section 3.4), several of the urban locations have non-
statistically significant effects estimates; these result in incidence estimates with non-positive
lower bounds and/or best estimates (e.g., Birmingham, Detroit, and Los Angeles for non-
accidental mortality). In the case of short-term morbidity (e.g., HAs), where regional effects
estimates were used, one of the regional coefficients (for the southeast) is not statistically
significant, producing incidence estimates including negative values in the confidence interval
for urban study areas falling within that region (e.g., Atlanta, Dallas, and Houston, for CV-
related HAs). Lack of statistical significance could mean that there is no relationship between
4-14
-------
PM2.5 and the health endpoint or it could mean that there was not sufficient statistical power to
detect a relationship that actually exists. In the case of PM2.5 and both short-term exposure
mortality and morbidity, recognizing that the ISA has concluded that there is either a causal or
likely causal relationship between short-term PM2.5 exposure and these health effects (see section
3.3.1), we believe it is reasonable to assume that instances where effects estimates are not-
statistically significant are likely to reflect insufficient sample size, rather than the absence of an
actual association. We note, however, that (as discussed in section 3.3.3, 3.5.2 and 3.5.3) many
factors can potentially result in variations in the magnitude of effect estimates. In addition to
sample size, these include: source and compositional differences for PM2.5, exposure error
associated with the use of ambient monitors as a surrogate for actual exposure, and differences in
population susceptibility and vulnerability.
An important theme in discussing risk associated with both current and alternative
standard levels is the linkage between the nature and magnitude of risk reductions seen for a
particular study area (for a particular suite of 24-hour and annual standards) and the specific mix
of 24-hour and annual design values associated with that study area. Because design values
determine the degree to which the PM2.5 monitors in a study area are adjusted in simulating
attainment of both current and alternative standard levels, they play a central role in determining
the degree of risk reduction associated with a particular suite of standard levels. Given the
importance of design values in determining risk reduction under both current and alternative
standard levels, we have examined patterns in design values (specifically the relationship
between 24-hour and annual design values) across the 15 urban study areas, as a means for
enhancing our interpretation of patterns in risk reductions for the standard levels modeled. In
addition, we have contrasted the patterns of design values for the 15 urban study areas with
patterns of design values for the broader set of urban areas in the U.S.; this supporting efforts to
place risk estimates for the urban study areas in a broader national context. This exploration of
design values is discussed in section 4.5.
An additional factor to consider in examining patterns in risk estimates is the overall
spread in PM2.5 measurements across monitors at a particular urban study area, including
distributions of both 24-hour and annual averages. This factor works in concert with the patterns
in design values mentioned earlier in determining the degree of risk reduction associated with a
particular suite of standard levels. In addition, the spread in monitor values for a particular urban
study area can also determine the degree to which alternative rollback methods (proportional,
hybrid and locally focused) produce differences in risk estimates for a given study area.
Consequently, in concert with examining patterns in design values (see above) we have also
explored patterns in PM2.5 monitoring data for the 15 urban study areas in an effort to better
4-15
-------
understand how application of different rollback methods results in differing impacts on core risk
estimates. This topic is discussed in section 4.5.
The remainder of this section is organized as follows. Core modeling results for the
recent conditions air quality scenario are presented in section 4.1. Core modeling results for just
meeting the current NAAQS and just meeting alternative NAAQS are presented in section 4.2.
The results of the sensitivity analysis (including single-factor and multi-factor results) are
presented in section 4.3. The results of a representativeness analysis involving comparison of
counties associated with the 15 urban study area locations against the national distribution of
counties with regard to a set of PM-risk related attributes are presented in section 4.4. Section
4.5 discusses the consideration of design values in interpreting risk estimates generated for the
15 urban study areas and helping to place them in a broader national context (section 4.5.1), as
well as consideration of the patterns in ambient PM2.5 data within study areas as a factor
influencing patterns of risk estimates (section 4.5.2). Chapter 5 provides an integrative
discussion of the results of the core risk assessment for the 15 urban study areas informed by
consideration of: (a) the single- and multi-factors sensitivity analysis, (b) the qualitative analysis
of sources of variability and uncertainty, (c) the representativeness analyses, and (d) the role of
design values (and patterns in ambient PM2.5 monitoring data) in influencing overall patterns of
risk estimates across alternative suites of standards.
4.1 ASSESSMENT OF HEALTH RISK ASSOCIATED WITH RECENT CONDITIONS
(CORE ANALYSIS)
This section discusses core risk estimates generated for the recent conditions air quality
scenario, focusing on the 2007 simulation year. Specifically, it provides a set of key observations
regarding core risk estimates generated for the recent conditions air quality scenario. Note, that
while the focus of this section is on identifying key risk-related observations potentially relevant
to the current review of the PM NAAQS, additional review of the risk estimates provided in
Appendix E is likely to result in additional observations that might be relevant to the PM
NAQQS review (EPA staff will continue to review those results as they work on completing the
summary of the RA presented in the PA).
In discussing results for the recent conditions air quality scenario, we have focused on
absolute risk (either above PRB or LML, depending on the health effect endpoint). This reflects
the fact that this air quality scenario represents recent conditions within the urban study areas and
therefore, does not lend itself to an incremental assessment. The section is organized by health
endpoint category, with results discussed in the following order: long-term exposure mortality,
4-16
-------
short-term exposure mortality and short-term exposure morbidity.60 In summarizing estimates
for each endpoint category, we first focus on the central-tendency risk estimates (these are what
is discussed in each of the bullets focusing on a particular endpoint category). A discussion of
the broader risk range reflecting consideration of 95th% confidence interval risk estimates is
presented as a separate bullet towards the end of the discussion. Key observations include:
Long-term exposure-related mortality: Total incidence of PM2.5-related all-cause mortality
ranges from 50-60 (Salt Lake City) to 2,380-3,000 (New York) (Appendix E, Table E-21 and
E-30), with this range reflecting not only differences in baseline incidence across urban study
areas, but also the size of study populations which vary considerably across the study areas.
The percent of total incidence of IHD-related mortality attributable to PM2.5 ranges from 6.3-
8.0% (Tacoma) to 17.7-22.2% (Fresno) (Appendix E, Table E-24 and E-33). Total PM2.5-
attriutable incidence for all-cause mortality and cardiopulmonary mortality is larger than IHD
(for a given study area under recent conditions), while total PM2.5-attributable incidence for
lung-cancer mortality is lower than for IHD. However, the percent of total incidence
attributable to PM2.5 exposure is larger for IHD-related mortality than for any of the other
mortality categories modeled (Appendix E, Tables E-24 and E-33).
Short-term exposure-related mortality: Total incidence of PM2.5-related mortality for
short-term exposure (for all categories modeled) is substantially smaller than estimates for
long-term exposure-related mortality. Estimates for CV mortality for short-term exposure
ranges from 14 (Salt Lake City) to 570 (New York) (Appendix E, Table E-84). The percent
of total non-accidental mortality attributable to PM2 5 ranges from 0.9% (Tacoma) to 2.5%
(New York). (Appendix E, Table E-87). Percent of total incidence attributable to PM2 5
exposure is generally lower for total non-accidental mortality (compared with CV), ranging
from 0.2% (Los Angeles) to 1.8% (Baltimore) (Appendix E, Table E-78). Estimates for
respiratory mortality are usually higher than for CV mortality, ranging from 0.9% (Dallas) to
2.8% (Fresno and New York) (Appendix E, Table E-96). Of the 15 urban study areas
modeled for CV mortality, 12 locations had negative lower bound estimates of incidence
(and two of these head negative point estimates), reflecting use of non-statistically significant
effects estimates (see section 4.0 for additional discussion). The number of study areas
modeled with non-statistically significant effects estimates was lower for the other two short-
term exposure-related mortality endpoints.
Short-term exposure-related morbidity (hospital admissions for respiratory and
cardiovascular illness): Total incidence of PM2.5-related cardiovascular HA range from 15
(Salt Lake City) to 910 (New York City) and are significantly larger than estimates of
respiratory HA attributable to PM2 5 exposure (Appendix E, Tables E102 and E-l 11).
Similarly, the percent of total cardiovascular HA attributable to PM2.5 is larger than estimates
for respiratory HA and ranges from 0.28% (Dallas) to 1.6% (Pittsburgh) (Appendix E, Table
60 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 PM2 5 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.
4-17
-------
E-105). In this case, the pattern of risk across urban study areas reflects both differences in
underlying baseline incidence for these endpoints as well as the use of regionally-
differentiated effect estimates obtained from Bell et al., 2008 (see Appendix C, Table C-l).
Of the 15 urban study areas modeled for cardiovascular-related HAs, five locations had
negative lower bound estimates of incidence, reflecting use of non-statistically significant
effects estimates (see section 4.0 for additional discussion).
Patterns of recent conditions risk across the three simulation years: A comparison of
IHD mortality incidence estimates (based on the C-R function derived using 1979-1982
monitoring data) across the three years (see Appendix E, Tables E-22 through E-24) shows
that, while 2007 does produce incidence estimates that fall between those estimated for 2005
and 2006 for some urban areas (e.g., Tacoma, St. Louis, LA), results for 2007 can be the
highest of the three years (e.g., Fresno) or the lowest (e.g., Baltimore) for some locations.
Generally, results for the same urban study area across the three years are fairly similar
(results for Birmingham vary by less than 7% across the years), although they can vary by as
much as 30% or more in some locations (see results for Tacoma in 2005 and 2006). All of
this temporal variation results from year-to-year variation in the annual average PM2.5 levels
for the study areas (see Appendix A). This is because other candidate input parameters,
which could also involve temporal variability (e.g., demographics and baseline incidence
rates) were not modeled with year-specific values, but rather using one representative year
(see section 3.4.1.3 and 3.4.2 for demographics and baseline health effects incidence rates,
respectively). In terms of short-term exposure-related morbidity and mortality endpoints, the
pattern is similar to that described above for long-term mortality, with risk estimates for 2007
generally falling between those generated for 2005 and 2006 (in terms of magnitude),
although the magnitude of variations across the three simulation years for a given health
endpoint/case study combination was notably lower for the short-term exposure-related
endpoints than for the long-term endpoints. For example, with CV mortality, one of the
urban study areas with the greatest variation across the three years (New York) had a 15%
difference in PM2.5 -related risk across the three years (see Appendix E, Tables E-82 through
E-84). This compares with a spread of 30% for some of the urban study areas modeled for
long-term exposure-related IHD mortality - see above. As with the long-term mortality risk
metrics, all of this temporal variation results from year-to-year variation in the daily PM2 5
levels for the study areas (see Appendix A), given that other candidate input parameters,
which could have temporal variability (e.g., demographics and baseline incidence rates) were
not modeled with year-specific values, but rather using one representative year.
Consideration of the 95th percentile confidence interval risk estimates in assessing
uncertainty related to the statistic fit of effect estimates: As noted above, all of the risk
metrics generated for this analysis include 95th percentiles, reflecting uncertainty in the
statistical fit of the underlying effect estimates in the C-R functions. These results suggest
that this source of uncertainty can be notable. In the case of recent conditions risk estimates,
for long-term mortality, while the central tendency risk estimate for all-cause (long term
exposure-related) mortality incidence in New York range from 2,380-3,000, the 95th
percentile confidence interval for this estimates is 1,960 to 3,500 (Appendix E, Table E-21
and E-30). In this case, this source of uncertainty results in estimates that are -18% lower to
-17% higher than the central tendency estimate range. Using the criteria we applied in
assessing the results of the sensitivity analysis, these would translate as having a "small"
4-18
-------
impact on the core risk estimate (see Section 4.3.1). The impact of statistical fit uncertainty
on the IHD-related long-term exposure-related mortality results (see Appendix E, Tables E-
24 and E-33) are similar in magnitude to those seen for all-cause mortality and also results in
a classification of this uncertainty having a "small" impact on core risk estimates. For short-
term exposure-related mortality and morbidity, the impact of statistical fit (as reflected in the
95th percentile CI risk estimate ranges) is greater than for long-term mortality. For example
with CV-related mortality, the central tendency estimate for New York is 570 cases, while
the 95th percentile CI is 332 to 902 (i.e., -40% lower and -40% higher than the core central-
tendency estimates). This translates into a "moderate" impact by this source of uncertainty
on core risk estimates using the classification scheme developed for the sensitivity analysis.
This suggests that uncertainty related to the statistical fig of effect estimates used in risk
characterization has twice as great an impact on short-term mortality as long-term mortality
risk estimates.
4.2 ASSESSMENT OF HEALTH RISK ASSOCIATED WITH JUST MEETING THE
CURRENT AND ALTERNATIVE SUITES OF STANDARDS (CORE ANALYSIS)
This section discusses core risk estimates generated for just meeting the current suite of
standards and alternative suites of standards, focusing on the 2007 simulation year (although
general trends in observations across the three simulated years are discussed to a limited extent).
In discussing risk estimates for the current and alternative suites of standards, we include
discussion of risk metrics which characterize both incremental reductions in risk (across standard
levels) as well as absolute risk for a particular standard level. In presenting these two categories
of risk metric, we recognize that there is greater uncertainty in estimates of absolute risk relative
to estimates of incremental risk. This reflects the fact that we have greater confidence in the
ability of the risk models to differentiate risk between sets of standards, since this requires the
models to estimate risk for ambient air PM2.5 levels likely near or within the range of ambient air
quality data used in the underlying epidemiology studies. By contrast, estimates of absolute risk
(for a given air quality scenario) require the models to perform at the lower boundary of ambient
air PM2.5 levels reflected in the studies (i.e., down to the LML reflected in the long-term
exposure mortality epidemiology studies or down to PRB levels in the short-term exposure
morbidity and mortality studies). There is greater overall uncertainty in risk estimates generated
based on the contribution to risk of exposures at these lower ambient air PM2.5 levels. While
there is greater uncertainty associated with estimates of absolute risk, these estimates are of
potential use in informing consideration of the magnitude of risk (and therefore public health
impact) for a particular standard level. The overall level of confidence associated with different
risk metrics (and implications for informing their use in the context of the PM NAAQS review)
is discussed in Chapter 5.
This section discusses risk estimates generated for the current standard levels first,
followed by discussion of risk estimates associated the set of alternative standard levels assessed.
4-19
-------
Each of these discussions is further organized by health endpoint category, with results discussed
in the following order: long-term exposure mortality, short-term exposure mortality and short-
term exposure morbidity. Observations presented in the previous section regarding the statistical
significance of effect estimates used in generating risk estimates and their implications for
interpretation of those risk estimates also hold for estimates presented in this section.
Consequently, observations regarding risk results with confidence intervals including negative
estimates are not presented here and the reader is referred back to the earlier discussion in section
4.1.
We note that the lower magnitude of risk reductions (in terms of percent change in PM2.5-
attributable risk) generally seen for short-term exposure-related endpoints relative to long-term
exposure-related endpoints primarily reflects the fact that PM2 5-attributable risk is modeled
down to PRB for short-term, but only down to LML for long-term. This means that an
incremental change (reduction) in long-term risk will be a larger fraction of overall risk
compared with short-term risk and hence, the magnitude of risk reductions for long-term
exposure-related risk is notably larger compared with short-term risk
An important factor to consider in interpreting the risk estimates for both the current set
of standards and sets of alternative standards is whether the annual or 24-hour standard for a
given pairing of standards is controlling for a particular area.61 This factor can have a significant
impact on the pattern of risk reductions predicted for a given location under the simulation of just
meeting a specific set of standards. In addition, the approach used to simulate ambient PM2.5
levels under current and alternative standard levels (i.e., use of proportional, hybrid, or locally
focused) can significantly impact the magnitude risk reduction seen across standard levels
(particularly the degree to which a particular standard produces notable reductions in long-term
exposure-related mortality).62 The potential for different rollback strategies (reflecting
potentially different combinations of local and/or regional controls) to impact patterns of risk
reduction is not discussed here, but rather reserved for discussion as part of the sensitivity
analysis (section 4.3) and the integrative chapter (chapter 5).
61 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.
62 Approaches such as hybrid rollback or locally focused 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.
4-20
-------
An overview of which urban study areas are predicted to have risk reductions under the
current and alternative suites of standards included in the risk assessment is presented below
(Appendix E contains tables presenting the full set of detailed core risk estimates generated for
the current and alternative suites of standards).
4.2.1 Core Risk Estimates for Just Meeting the Current Suite of Standards
This section summarizes risk estimates generated for the 15 urban study areas based on
simulating just meeting the current suite of standards (including the magnitude of risk reductions
relative to recent conditions, where applicable).
Long-term exposure-related mortality: Total incidence of PM25-related IHD mortality
ranges from 15-20 (Salt Lake City) to 1,760-2,220 (New York) (Table 4-1). The percent of
total incidence of IHD mortality attributable to PM2.5 ranges from 3.7-4.7% (Tacoma) to
13.2-16.7% (Atlanta) (Table 4-2). These levels of IHD mortality risk attributable to PM2.5
exposure reflect reductions in risk relative to recent condition ranging from 8.7% (Houston)
to 68.6% (Salt Lake City). Two of the urban study areas (Dallas and Phoenix) do not exhibit
reductions in risk in simulating just meeting the current suite of standards since these two
locations meet the current suite of standards based on recent air quality data. As referenced
above for the recent conditions scenario, total PM2.5-attriutable incidence for all-cause
mortality and cardiopulmonary mortality is larger than IHD (for a given study area under
recent conditions), while total PM2.5-attributable incidence for lung-cancer mortality is lower
than for IHD. However, the percent of total incidence attributable to PM2.5 exposure is larger
for IHD-related mortality than for any of the other mortality categories modeled (Appendix
E, Tables E-24 and E-33).
Short-term exposure-related mortality: As with the recent conditions analysis, total
incidence of PM2.5-related mortality for short-term exposure is substantially smaller than
estimates for long-term exposure-related mortality. Estimates for CV mortality for short-
term exposure ranges from 9 (Salt Lake City) to 470 (New York) (Table 4-1). The percent of
CV mortality attributable to PM2 5 ranges from 0.7% (Fresno) to 2.1% (Philadelphia and New
York). (Table 4-2). The level of risk reduction (comparing risk under the current standard
with risk under recent conditions) is generally lower for short-term exposure-related CV
mortality compared with long-term exposure-related all-cause mortality and ranges from
5.5% (Baltimore) to 36.9% (Los Angeles). As mentioned for long-term exposure-related
risk, both Phoenix and Dallas did not exhibit any risk reduction since these two locations
meet the current suite of standards based on recent air quality data. Percent of total incidence
attributable to PM2 5 exposure is generally lower for total non-accidental mortality (compared
with CV), ranging from 0.1% (Los Angeles) to 1.7% (Baltimore) (Appendix E, Table E-78).
Estimates for respiratory mortality are usually higher than for CV, ranging from 0.9%
(Dallas) to 2.6% (Baltimore) (Appendix E, Table E-96). As noted above, of the 15 urban
study areas modeled for CV mortality, 12 locations had negative lower bound estimates of
incidence (and two of these had negative point estimates), reflecting use of non-statistically
significant effects estimates (see section 4.0 for additional discussion).
Short-term exposure-related morbidity (hospital admissions for respiratory and
cardiovascular illness): Total incidence of PM2.5-related cardiovascular HA range from 9
4-21
-------
(Salt Lake City) to 750 (New York City) and are significantly larger than estimates of
respiratory HA attributable to PM2 5 exposure (Appendix E, Tables E102 and E-l 11).
Similarly, the percent of total cardiovascular HA attributable to PM2.5 is larger than estimates
for respiratory HA and ranges from 0.28% (Dallas) to 1.33% (Baltimore). As noted above,
the pattern of risk across urban study areas reflects both differences in underlying baseline
incidence for these endpoints as well as the use of regionally-differentiated effect estimates
obtained from Bell et al., 2008 (see Appendix C, Table C-l). The level of risk reduction
(comparing risk under the current standard with risk under recent conditions) for both
respiratory and cardiovascular hospital admissions ranges from 5.5% (Baltimore) to 44.8%
(Fresno), again with Phoenix and Dallas not exhibiting any risk reduction since these two
locations meet the current suite of standards based on recent air quality data. As noted above,
of the 15 urban study areas modeled for cardiovascular-related HAs, five locations had
negative lower bound estimates of incidence, reflecting use of non-statistically significant
effects estimates (see section 4.0 for additional discussion).
Patterns of recent conditions risk across the three simulation years: Observations made
earlier regarding patterns of risk across the three simulation years for the recent conditions
simulations generally hold for the current standard level analysis. In other words, (a) 2007
generally represents risks in between the other two years in terms of magnitude, (b) there are
exceptions where 2007 had the highest risks and lowest risk (depending on study area and
endpoint), and (c) generally, long-term exposure-related mortality endpoints showed greater
cross year variation then the short-term exposure-related endpoints (with the magnitude of
this variation similar to what is reported above for the recent conditions simulation).
Consideration of the 95th percentile confidence interval risk estimates in assessing
uncertainty related to the statistic fit of effect estimates: Uncertainty related to the
statistical fit of effect estimates has the same magnitude of effect in modeling risk under the
current standard as it did under recent conditions (i.e., an impact of about +/-18% on the core
risk estimates, translating into a "small" impact based on classification used in the sensitivity
analysis) (see section 4.3.1 for a description of the classification scheme for sources of
uncertainty and Appendix E, Table E-21 and E-30 for risk estimates used to reach this
conclusion). The impact of this source of uncertainty on short-term exposure-related CV
morality was similar (although slightly larger) compared with what was seen with risk
estimates generated for the recent conditions air quality scenarios (i.e., 48% lower to 42%
higher than the core risk estimate - see estimates in Appendix E, Table E-84). This results in
a classification of "moderate" for this source of uncertainty and its impact on short-term
exposure-related mortality, based on the classification scheme developed for the sensitivity
analysis.
4.2.2 Core Risk Estimates for Just Meeting Alternative Suites of Standards
This section summarizes risk estimates generated for the 15 urban study areas when
ambient PM2.5 levels under the alternative standard levels are simulated. As noted in section 4.2,
this discussion focuses on the magnitude of incremental risk reductions for individual standard
levels relative to the current standard, given that overall confidence in incremental risk metrics is
considered higher than estimates of absolute risk for a given standard level. Note, however, that
we do provide limited discussion of absolute risk levels attributable to PM2.5 exposure for
4-22
-------
alternative standard levels, with the provision that these be interpreted in the context of their
greater levels of uncertainty. In discussing risk estimates for the alternative standard levels, we
focus first on patterns of risk reduction across alternative annual levels (i.e., 14/35, 13/35 and
12/35) and then discuss patterns across a combination of alternative 24 hour and annual
standards (i.e., 13/30 and 12/25).
As noted in Section 4.1, although reductions in absolute incidence will differ for health
effect endpoints associated with a particular averaging period across alternative suites of
standards for a given urban study area, the patterns of reduction in terms of percent change in
PM^-attributable risk are very similar for a given urban study area across health endpoints. This
reflects the fact that the C-R functions used in the core analysis are close to linear across the
range of ambient PM2.5 levels considered in this analysis, and consequently the main factor
producing percent reductions in risk across alternative standards is the reduction in the air quality
metric for a given study area (i.e., reductions in annual average PM2.5 concentrations or
reductions in the distribution of 24-hour estimates for a year). Consequently, in discussing
incremental risk reduction in terms of percent change relative to the current suite of standards,
we speak more generally in terms of the category of annual average risk or 2-4hour average
risk, with the assumption that these observations hold for individual health effects endpoints
assessed for each averaging period. These observations regarding patterns of percent risk
reduction for the two averaging periods are reflected in Figures 4-1 through 4-6 which are
referenced in the discussion below.
Alternative annual standard levels (14/35. 13/35. and 12/35) 63
Percent reductions in long-term exposure-related mortality: Reductions in all long-term
exposure-related mortality categories were more limited under the 14/35 alternative standard,
with only 5 of the 15 urban study areas demonstrating notable reductions ranging from 9%
(Baltimore) to 12% (Houston and Birmingham) (see Figure 4-3 and Appendix E, Table E-9).
Reducing the annual standard level to 13 |ig/m3 (i.e., the 13/35 alternative suite of standards)
produced a notable increase in the number of locations (9 of the 15) with risk reductions
relative to the current standard ranging from 5% (New York) to 24% (Houston and
Birmingham). The lowest annual standard evaluated (12 |ig/m3 as reflected in the 12/35
alternative suite of standards) resulted in additional study areas (now 12 of the 15 study
areas) experiencing risk reductions with percentage risk reductions now ranging from 11%
63 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.
4-23
-------
(Phoenix) to 36% (Houston and Baltimore). Note, that even in the 12/35 case, three of the
urban study areas (Tacoma, Fresno and Salt Lake City) did not experience any decreases in
risk, although risk reductions were seen for these three study areas when alternative 24-hour
standards were considered - see below. The specific pattern of risk reduction (including
importantly, the magnitude of risk reduction as well as residual risk associated with a
particular standard level) reflects whether daily or annual standard levels were controlling -
see discussion below regarding patterns of risk reduction.
Percent reduction in short-term exposure-related mortality and morbidity: The pattern
of reductions in the percent of risk attributable to PM2.5 for mortality and morbidity
associated with short-term exposure is similar to that described above for long-term mortality
(see Figures 4-4 through 4-6). Specifically, the same five urban study areas (Atlanta,
Baltimore, Birmingham, Houston and St. Louis) had notable risk reductions under the full set
of alternative annual standards, with the degree of risk reduction for PM2.5-related
cardiovascular mortality for the lowest alternative annual standard level (12/35) compared to
the current standard level, ranging from 20% (St. Louis) to 23% (Birmingham) (see Figure 4-
4 and 4-6 and Appendix E, Table E-90). A number of the other study areas did not exhibit
notable risk reductions until the lowest alternative annual standard was considered (i.e.,
Detroit, Los Angeles, New York, Philadelphia, Pittsburgh), with the degree of reduction in
risk for the lowest alternative suite of standards (12/35) compared with the current standards
ranging from 5% (Phoenix) to 16% (Detroit) (see Figure 4-4 and 4-6 and Appendix E, Table
E-90). As with long-term exposure-related mortality, a number of additional study areas
(Fresno, Salt Lake City, Tacoma) did not exhibit any notable risk reduction under the set of
alternative annual standards considered and only experienced risk reductions when the 24-
hour standard level was reduced. Because the same air quality metric (annual distributions of
24-hour PM2.5 concentrations) is used in generating short-term exposure-related mortality
and morbidity endpoints, patterns of risk reduction are similar for both sets of endpoints (see
Figures 4-4 through 4-6). Specifically, the same groups of urban study areas experience the
same magnitude of risk reductions (in terms of percent changes in PM2.s-related risk relative
to the current standard level) across the alternative standard levels for short-term exposure-
related morbidity (HAs). The specific pattern of risk reduction reflects whether daily or
annual standard levels are controlling - see discussion below regarding patterns of risk
reduction.
Pattern of risk reduction linked to design values: The patterns of risk reduction across the
15 urban study areas for the set of alternative annual standard levels considered here depends
on whether the alternative annual (12, 13 or 14 |ig/m3) or the current 24-hour standard of 35
|ig/m3 is controlling. The approach used to simulate just meeting alternative 24-hour
standards (i.e., proportional, hybrid, or locally focused) can have an impact on the magnitude
of risk reduction, although it does not influence whether the annual or 24-hour design value
was controlling for a given alternative suite of standards (see sensitivity analysis discussion
in 4.3 and the integrative discussion in Chapter 5). The pattern in risk reduction seen across
the 15 urban study areas (given the set of alternative annual standards considered) can be
divided into three categories: (a) all of the alternative annual standard levels are controlling,
resulting in notable risk reductions for all of the annual standard levels considered
(Birmingham, Atlanta, Houston), (b) alternative annual standards only control at lower levels
(i.e., 13/35 and/or 12/35) and consequently notable risk reductions are only seen at the lower
4-24
-------
or lowest annual standard level(s) considered (Dallas, Los Angeles, New York, Philadelphia,
Phoenix, Pittsburgh), and (c) none of the alternative annual standard levels is controlling and
therefore there is no estimated risk reduction for the alternative annual standard levels
considered (Salt Lake City, Tacoma, Fresno).
Absolute levels of PM2.5-attributable risk under alternative annual standards: As
discussed above, we have greater confidence in estimating incremental reductions in risk
between the current and alternative suites of standards, then the estimation of absolute
incidence under a given suite of standards. Nonetheless, we provide a summary of that risk
metric here for long-term and short-term exposure-related mortality and short-term exposure-
related morbidity endpoints:
o Long-term exposure-related mortality: The four study areas displaying the greatest
degree of reduction across the alternative annual standards (Atlanta, Baltimore,
Birmingham and Houston) have PM2 5-related IHD mortality estimates (under the
lowest alternative annual standard of 12/35) ranging from 85-110 (Birmingham) to
220-280 (Houston) (see Appendix E, Table E-21 and E-30). The two urban study
areas with the greatest degree of PM2.5-related risk in absolute terms (Los Angeles
and New York) do not exhibit significant reductions in risk until the lowest annual
standard level of 12/35 is considered, with PM2.5-related IHD mortality estimated at
750-950 and 1,420-1,800, respectively under that alternative standard (see Appendix
E, Table E-21 and E-30).
o Short-term exposure-related mortality: The four study areas displaying the greatest
degree of reduction across the alternative annual standards (Atlanta, Baltimore,
Birmingham and Houston), have PM25-related CV mortality estimates (under the
lowest alternative standard of 12/35) ranging from 25 (Atlanta) to 50 (Baltimore) (see
Appendix E, Table E-84). We note that Birmingham has an incidence estimate of-1,
reflecting application of a non-statistically significant effect estimate in modeling this
endpoint (see section 4.1). The urban study area with the greatest degree of PM2 5-
related risk in absolute terms (New York) does not exhibit significant reductions in
risk until the lowest annual standard level of 12/35 is considered with PM2.5-related
CV mortality estimated at 420 under that alternative standard level (see Appendix E,
Table E-84).
o Short-term exposure-related morbidity: The four study areas displaying the greatest
degree of reduction across the alternative annual standard levels (Atlanta, Baltimore,
Birmingham and Houston), have PM2.s-related cardiovascular HA (under the lowest
alternative standard of 12/35) ranging from 12 (Birmingham) to 170 (Baltimore) (see
Appendix E, Table E-102). The two urban study areas with the greatest degree of
PM2 s-related risk in absolute terms (Los Angeles and New York) do not exhibit
significant reductions in risk until the lowest annual standard level of 12/35 is
considered with PM2.5-related all-cause mortality estimated at 240 and 670,
respectively under that alternative standard level (see Appendix E, Table E-102).
Patterns of recent conditions risk across the three simulation years: Observations made
above regarding patterns of risk across the three simulation years for the recent conditions
and current standards simulations generally hold for the alternative standards analysis. In
other words, (a) 2007 generally represents risks between the other two years in terms of
4-25
-------
magnitude, (b) there are exceptions where 2007 had the highest risks and lowest risk
(depending on study area and endpoint), and (c) generally, long-term exposure-related
mortality endpoints showed greater cross-year variation then the short-term exposure-related
endpoints in terms of both absolute PM2.s risk for a particular alternative suite of standards,
as well as incremental risk reductions relative to the current suite of standards.
Consideration of the 95th percentile confidence interval risk estimates in assessing
uncertainty related to the statistic fit of effect estimates: Continuing the pattern seen with
the current standard level, uncertainty related to the statistical fit of effect estimates has the
same magnitude of effect in modeling risk under alternative standards involving reduction of
the annual level as it did under recent conditions (i.e., an impact of about +/-18% on the core
risk estimates, translating into a "small" impact based on classification used in the sensitivity
analysis) (see Appendix E, Table E-21 and E-30 for risk estimates used to reach this
conclusion). Similarly, the pattern of impact this source of uncertainty on short-term
exposure-related CV morality continues to be similar compared with what was seen for risk
estimates generated for the recent conditions air quality scenarios (i.e., 42% lower to 42%
higher than the core risk estimate - see estimates in Appendix E, Table E-84). This
continues to result in a classification of "moderate" for this source of uncertainty based on
the classification scheme developed for the sensitivity analysis.
Combinations of alternative 24-hour and annual standard levels (13/30, 12/25)
Percent reductions in long-term exposure-related mortality: The combination of suites
of alternative 2-hour and annual standards produced notable reductions in long-term
exposure-related mortality for 14 of the 15 urban study areas, with the lower combination
(12/25) producing a notable reduction in risk relative to the first combination of 13/30. The
only study area that did not exhibit a reduction in risk under the first combination (13/30)
was Dallas, reflecting the fact that its 24-hour and annual design values are below 30 |ig/m3
and 13 |ig/m3, respectively (and consequently, the 13/30 did not produce a reduction in
ambient air PM2.5, or a resulting reduction in risk). Reductions in long-term exposure-related
mortality (across all endpoints) under the 13/30 combination ranged from 14% (Phoenix) to
55% (Salt Lake City), while reductions for the 12/25 combination ranged from 12% (Dallas)
to -100% (Salt Lake City) (see Figure 4-1 and 4-3 and Appendix E, Table E-27). The
reduction for Salt Lake City reflects a very high 24-hour design value which, when reduced
to meet the 24-hour standard of 25 |ig/m3 produced a very large reduction in the annual
design value (given application of proportional adjustment to simulate rollback), such that
the value was very close to 5.8 |ig/m3 (the LML below which long-term exposure-related
mortality is not estimated). The specific pattern of risk reduction reflects whether the 24-hour
or annual standard was controlling - see discussion below regarding patterns of risk
reduction.
Percent reduction in short-term exposure-related mortality and morbidity: The pattern
of reductions in the percent of risk attributable to PM2 5 for mortality and morbidity
associated with short-term exposure is similar to that described above for long-term mortality
in terms of the ordering of sites, however the magnitude of risk reduction (in terms of percent
change in PM2.5-related risk) is lower for short-term exposure-related health endpoints
compared with long-term exposure-related mortality (see Figures 4-4 through 4-6).
Specifically, 14 of the 15 urban study areas (Dallas being the exception), had notable risk
reductions under both the 13/30 and 12/35 alternative suites of standards (Dallas only was
4-26
-------
estimated to have reductions in risk under the lower 12/25 combination - see Figure 4-4 and
4-6 and Appendix E, Table E-108). Reductions in short-term exposure-related mortality and
morbidity (across all endpoints) under the 13/30 combination ranged from 6% (Phoenix) to
15% (Salt Lake City), while reductions for the 12/25 combination ranged from 7% (Dallas)
to 30% (Birmingham).
Pattern of risk reduction linked to design values: As with the set of alternative annual
standards discussed in the previous section, the pattern of risk reduction seen for the two
combinations of alternative 24-hour and annual standards described here depends on which
standard is controlling. In addition, the magnitude of the reduction in risk reflects (a) the
magnitude of the difference between the controlling design value and the standard level
(which determines the degree of reduction in ambient air PM2 5 levels) and (b) the method
used to simulate ambient PM2.5 levels under alternative suites of standards (i.e., proportional,
hybrid or locally focused rollback). For this set of alternative suites of standards, 10 of the 15
study areas had the alternative 24-hour standard controlling under the 13/30 case and that
number was increased to 12 out of the 15 study areas with the 12/25 case (Table 3-5). As
expected, those study areas with the greatest reduction in risk (in terms of percent reduction
compared with the current suite of standards) under the 12/25 case had a controlling 24-hour
standard (e.g., Tacoma, Salt Lake City, Los Angeles and Fresno - see Figure 4-4 and 4-6 and
Appendix E, Table E-90).
Absolute levels of PM2.s-attributable risk under alternative suites of annual and 24-
hour standards: As with the alternative annual standards, below we provide a brief
overview of the magnitude of PM2.5-attributable risk (i.e., absolute risk) associated with the
two alternative suites of annual and 24-hour standards:
o Long-term exposure-related mortality: The four study areas displaying the greatest
degree of reduction across these two alternative suites of standards (Tacoma, St.
Louis, Los Angeles and Fresno), have PM25-related IFID mortality estimates (under
the 12/25 case) ranging from 3-4 (Tacoma) to 290-360 (Los Angeles) (see Appendix
E, Table E-21 and E-30). The other urban study area with the greatest degree of
PM2.s-related risk in absolute terms besides New York (New York) has PM2.s-related
all-cause mortality estimated at 820-1,040 under the 12/25 case.
o Short-term exposure-related mortality: eleven of the 15 study areas had percent
reductions in risk for the 12/25 case (relative to the current standards) of
approximately 29% (the other study areas had lower percent reductions). Of the
locations with -29% reductions in risk, PM2.5-attributable CV mortality for the 12/25
case ranged from 6 (Salt Lake City) to 340 (New York) (see Appendix E, Table E-
84). New York City also represents the study area with the greatest residual risk for
short-term exposure-related mortality under the 12/25 case.
o Short-term exposure-related morbidity Of the 11 urban study areas with -29%
reduction in risk (for the 12/25 case relative to the current standards), the incidence of
PM2.5-attributable cardiovascular HA emissions ranges from 7 (Salt Lake City) to 530
(New York) (see Appendix E, Table E-102). New York City also represents the
study area with the greatest residual risk for short-term exposure-related morbidity
under the 12/25 case.
4-27
-------
Consideration of the 95th percentile confidence interval risk estimates in assessing
uncertainty related to the statistic fit of effect estimates: As with the alternative standards
considering lower annual levels, risk estimates generated for the two standards considering
lower annual and 24-hour levels also suggest that uncertainty related to the statistical fit of
effect estimates will have a greater impact on short-term exposure-related mortality (+/-
-40%) compared with long-term exposure-related mortality (+/- -18%) (see Appendix E,
Tables E-84 and E-21 plus Table E-30, respectively). Again, this results in a classification of
this source of uncertainty as having a "lower" impact for long-term exposure-related
mortality and a "moderate" impact on short-term exposure-related mortality.
4.3 SENSITIVITY ANALYSIS RESULTS
As noted in section 3.5.4 and section 4.0, the sensitivity analysis was conducted in order
to gain insights into which of the identified sources of uncertainty and variability in the risk
assessment model may have significant impacts on risk estimates. A second goal of the
sensitivity analysis was to generate an additional set of reasonable risk estimates to supplement
the core set of risk estimates to inform staffs characterization of uncertainty and variability
associated with those core estimates.
The first goal can be achieved by considering the magnitude of the impact of individual
modeling elements based on results from the sensitivity analysis and identifying those elements
which have the greatest impact on the core risk estimates. Use of the sensitivity analysis results
in this context (i.e., identify those elements that contribute the most to sensitivity in the risk
estimates) is addressed in section 4.3.1. Use of the results of the sensitivity analysis as an
additional set of reasonable risk estimates to augment the core risk estimates in considering the
impact of uncertainty and variability in the core risk model is discussed in section 4.3.2.
In conducting the sensitivity analysis we modeled 2 of the 15 urban study areas
(Philadelphia and Los Angeles - representing east and west coast urban areas, respectively) for
most simulations.64 For some modeling elements (e.g., the hybrid and locally focused alternative
rollback approaches) we included a larger number of urban study areas that were applicable to
the topic being assessed. In conducting the sensitivity analysis, we have also focused on long-
term exposure mortality and to a lesser extent on short-term exposure mortality and morbidity.
Although the sensitivity analysis simulations were completed for all three simulation
years (as reported in Appendix F), we have focused on results for 2007 in this presentation for
comparability with the core results discussed in sections 4.1 and 4.2.
64 These urban study areas were chosen generally to provide coverage for locations with recognized
differences in factors associated with PM-related risk (e.g., meteorology, mix of local and regional PM sources,
demographics), however a rigorous selection framework was not used. It should be noted that for some elements of
the sensitivity analysis (e.g., consideration for alternative rollback methods) a larger number of the urban areas were
included in the sensitivity analysis.
4-28
-------
4.3.1 Sensitivity Analysis Results to Identify Potentially Important Sources of Uncertainty
and Variability
The results of the sensitivity analysis are summarized in Table 4-3 (detailed results tables
are presented in Appendix F). In presenting the results of the sensitivity analysis, we have
compared the risk estimates for the particular simulation to the core set of risk estimates
generated for the same health effect endpoint/urban study area combination. Specifically, we
have calculated a percent difference between the sensitivity analysis result and the associated
core risk estimate to compare the results of the sensitivity analysis across the different modeling
elements that were considered. This metric is used because it is not influenced by location-
specific differences in such factors as population size and baseline incidence rates and therefore,
supports ready comparison of modeling elements (in terms of their impact on core risk estimates)
across the urban locations assessed. These percent difference results are emphasized in Table 4-3
and in the discussion presented below. Note that by contrast, the alternative risk estimates
discussed below in section 4.3.2 focus on absolute risk, since the emphasis with that analysis is
on assessing the potential spread in (absolute) risk that results from considering alternative
modeling element specifications from those used in the core analysis.
In discussing the results of the sensitivity analysis, we have developed four descriptive
categories, based on the general magnitude of the percent difference estimate generated for a
particular modeling element:
Modeling elements estimated to have percent differences of 20% or smaller (i.e., they
produced risk estimates that differed from the core risk estimates by no more than
20%) are classified as having a small contribution to uncertainty in the core risk
estimates.
Modeling elements estimated to have percent difference estimates in the range of 20 to
50% are classified as having a moderate contribution to uncertainty in the core risk
estimates.
Modeling elements estimated to have percent difference estimates in the range of 50 to
100% are classified as having a moderate-large contribution to uncertainty in the core
risk estimates.
Modeling elements estimated to have percent difference results >100% are classified as
having a large contribution to uncertainty in the core risk estimates.
The sensitivity analysis based on Moolgavkar's (2003) study in Los Angeles addressing
model specifications for both short-term mortality and morbidity (e.g., model selection, lag
structure and co-pollutant models) are discussed together as a group. This reflects the fact that
the Moolgavkar-based simulations were based on the same underlying dataset and focused on
Los Angeles. Furthermore, the discussion of the Moolgavkar-based sensitivity analysis results
presented below, as well as the summary of results presented in Table 4-3, focus on the
4-29
-------
difference in the spread of risk results across the Moolgavkar-based model specifications (for a
particular endpoint), rather than the percent difference results based on comparison against the
core result that are emphasized with the other sensitivity analyses.65
The sensitivity analysis examining the impact of alternative rollback approaches for
simulating ambient PM2.5 concentrations in urban study areas under both the current and
alternative suites of standards also deserves additional discussion before presenting the results.
For the first draft RA, we considered the impact of using a hybrid rollback approach in addition
to the proportional rollback approach which has been traditionally used in PM NAAQS risk
assessments. For this second draft, as discussed in sections 2.3, 3.2.3 and 3.5.4, we have
included consideration of a locally focused rollback approach in addition to the hybrid as non-
proportional methods to contrast with proportional rollback.66
As discussed in Section 3.2.3, for the second draft risk assessment, we have calculated
composite monitor estimates based on proportional rollback and hybrid and/or locally focused,
where appropriate. The composite monitor values are surrogates for long-term exposure-related
mortality.67 Therefore, by comparing composite monitor values generated for the same study
area/suite of standards (using different rollback methods), we can obtain insights into the
potential impact of the rollback method used on long-term exposure-related mortality (see
Section 3.5.4 for additional discussion of how the composite monitor values generated using the
different rollback methods are used in the sensitivity analysis). These sensitivity analysis results
based on consideration of composite monitor values generated using the different rollback
methods (which are presented in detail in Appendix F, Tables F-49 and F-50) form the basis for
65 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.
66 The locally focused 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.
67 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).
4-30
-------
summary information related to rollback presented in Table 4-3. Due to the complexity of the
sensitivity analysis conducted examining the issue of rollback, the discussion of results from that
particular analysis presented in section 4.3.1.1 is more detailed than for the other factors
considered as part of the sensitivity analysis.
In discussing the results of the sensitivity analysis, results of the single-factor simulations
are presented first (section 4.3.1.1), followed by the results of the multi-factor simulations
(section 4.3.1.2). Within these categories, results are further organized by health effect endpoint
with results for long-term exposure mortality discussed first and then short-term exposure
mortality, followed by short-term exposure morbidity. An overall conclusion regarding which of
the factors included in the sensitivity analysis represent potentially significant sources of
uncertainty and variability impacting the core risk estimates is presented at the end of each sub-
section.
4-31
-------
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 locally focused 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 locally focused
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
4-32
-------
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 locally focused 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
4-33
-------
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
CV HA: +36%
Table F-33
4-34
-------
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%
C V 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: +26 to +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 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).
4-35
-------
4.3.1.1 Single-factor Sensitivity Analysis
This section presents the results of the single-factor sensitivity analysis, which involved
consideration of alternate model inputs on the core risk estimates, when those alternate inputs are
considered one at a time (consideration of the combined effect of several model inputs being
varied is covered by the multi-factor sensitivity analysis discussed in section 4.3.1.2). The
results of the single-factor sensitivity analysis are characterized qualitatively using the four-
category approach described above (i.e., low, moderate, moderate-large and large, with each of
these representing a defined range of percent difference from the core risk estimates).
Long-term exposure mortality
This section summarizes the results of the sensitivity analysis focused on long-term
exposure-related mortality endpoints (see Table 4-3 for the specific modeling elements
considered in the sensitivity analysis).
Impact of using different model choices for C-R function -fixed effects log-linear (the core
approach) vs. random effects log-linear or random effects log-log models: This simulation
considered two alternative C-R model forms obtained from Krewski et al., 2009 for modeling
all-cause, CPD, IHD and lung cancer mortality, including (a) random effects log-linear
model and (b) a random effects log-log model (note, the core effect estimate was derived
using a fixed effects log-linear model obtained from Krewswki et al., 2009). The simulation
also considered the use of multi-pollutant models that control for CO, NCh, 63 or 862. The
results of the simulation suggest that the use of a random effects log-linear model, rather than
the core fixed effects model, has a relatively small effect on risk estimates, increasing them
by 12 to 23% across the mortality categories and urban study areas modeled (Appendix F,
Table F-3). However, use of a random effects log-log model has a larger impact on risk
estimates, increasing them by 50 to 159% (Appendix F, Table F-3). The greater impact of
the log-log model results from this function having an incrementally steeper slope at lower
PM levels, which quickly increases incidence estimates compared with the core log-linear
model (whose slope has a much more gradual incremental increase in slope at lower PM
levels). The use of multi-pollutant models that control for co-pollutants was shown to have
moderate-large impact on risk estimates, with control for CO, NC>2, or Os resulting in
increased PM2.5-attributable risk estimates, while control for SO2 resulted in a moderate-large
decrease in estimated PM2.5 risk.68
Impact of estimating risks down to PRB rather than down to LML: This simulation
compared long-term exposure mortality incidence associated with modeling risk down to
PRB (which varies by region - see section 3.2.1) with the core approach of modeling down
to LML (5.8 |ig/m3 for long-term mortality - see section 3.1). This simulation involved all
68 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).
4-36
-------
15 urban study areas, given that PRB is stratified by region and therefore, results of the
simulation could differ significantly across the 15 urban study areas, or at least across the six
PM regions represented by those study areas. The results of this simulation suggest that
modeling risk down to PRB could have a moderate to large impact on long-term exposure
mortality incidence, with estimates ranging from 47 to 273% higher than the core estimates
(for matching urban locations) (Appendix F, Table F-6). Note, however, that risk metrics
based on considering the incremental reduction in risk (incidence) between two alternative
suites of standards would not be impacted by this source of uncertainty, since it only affects
estimates of absolute risk.
Impact of C-R function from alternative long-term exposure mortality study: This simulation
considered use of alternative C-R functions (and effect estimates) based on the reanalysis of
the Six Cities study (Krewski et al., 2000). The results suggest that use of the alternative C-R
function could have a moderate to moderate-large effect on CPD mortality (+45 to +74%), a
large effect on all-cause mortality (+123 to +159%), a moderate-large to large effect on IHD
mortality (+80 to +111%) and a moderate-large effect on lung cancer mortality (+67 to
+94%) (Appendix F, Table F-9). The results of this simulation suggest that (at least with
regard to application of C-R functions obtained from the Six Cities study) the potential
impact of functions from alternative studies on long-term exposure mortality depends on the
mortality category being considered. In this analysis, use of the alternative C-R functions
was shown to have a significant impact on all of the long-term mortality categories
considered.
Impact of using alternative rollback approaches (hybrid and locally focused) to simulate just
meeting the current and alternative suites of standards. This sensitivity analysis assessed the
impact of estimating risk for the current and alternative sets of standards using two
alternatives to the proportional rollback strategy: (a) the hybrid rollback approach that
reflects an initial localized pattern of ambient PM2.5 reduction (resulting in non-proportional
rollbacks of monitored PM2.5 concentrations) with a second phase of more regional
reductions in ambient PM2 5 levels (based on proportional adjustments) and (b) locally
focused which represents a primarily local pattern of reductions in ambient PM2 5 (see
Section 3.5.4 for additional discussion of how these alternative rollback methods were
integrated into the sensitivity analysis). We note that the core analysis utilized proportional
rollback exclusively in simulating conditions for the current and alternative sets of standards,
with this approach representing a regional pattern of ambient PM2 5 reduction. A number of
observations can be drawn from this sensitivity analysis including:
o Impact on estimates ofPM2.5-related risk remaining after simulation of just
meeting a given suite of standards: The sensitivity analysis results suggest that
the use of alternative rollback methods can have a notable impact on estimates of
the PM2 5-attributable risk remaining after simulation of a given suite of standards
(see Appendix F, Table F-50 and discussion in section 3.5.4). Generally, use of
the hybrid approach had a small to moderate impact on absolute PM2 5-
attributable risk estimates, compared with the core approach of using proportional
rollback. By contrast, use of the locally focused approach had a moderate to
moderate-large impact on absolute PM2.5-attributable risk estimates. For example,
Los Angeles had composite monitor values for the current suite of standards and
several of the alternative suites of standards that were 40 to 60% greater when the
4-37
-------
locally focused rollback method was used, compared with the proportional
rollback method (see Appendix F, Table F-50). By contrast, composite monitor
values generated using hybrid rollback for Los Angeles, were between 13 and
38% higher than the proportional rollback methods. These results suggest that
more localized spatial patterns of reduction in PM2.5 in response to alternative
standard levels, as reflected in application of the hybrid and even to a greater
extent, the locally focused rollback approach, can result in a larger fraction of risk
remaining after simulating those standards. By contrast, more regional spatial
patterns of PM2.5 reduction, as reflected in the proportional rollback approach
used in the core analysis, results in a greater reduction in risk with less risk
remaining upon simulation of the alternative standard. These results point to the
potentially important role played by the nature of the spatial pattern of PM2.5
reduction (i.e., rollback) in determining the magnitude of public health protection
potentially provided by alternative standard levels.
o Impact on degree of reduction across alternative suites of standards: When the
same rollback methods is used to simulate both the current and any alternative
suite of standards, the pattern of risk reduction across alternative standards is
generally similar regardless of the rollback approaches used (see Table F-49, in
Appendix F). However, if one looks at meeting the current suite of standards with
application of the locally focused approach, followed by application of
proportional rollback to simulate alternative suites of standards, we can see
notable differences in the pattern of risk reduction. This is particularly true for
areas with peaky PM2.5 distributions (i.e., areas with relatively high 24-hour
design values and lower annual average design values). For example, with Los
Angeles, which represents a study area with a relatively peaky PM2 5 distribution,
application of proportional rollback in simulating both the current suite of
standards and the alternative annual standard of 12 |ig/m3 results in a 13%
reduction in long-term exposure-related mortality (see Figure 4-3 and Table E-27
in Appendix E). By contrast, application of the locally focused approach in
simulating the current suite of standard levels followed by proportional reduction
in simulating the same alternative annual standard results in an estimated 48%
reduction in long-term exposure-related mortality.69 These results highlight the
point made in the previous bullet, that the nature of the spatial pattern of PM2 5
reduction in response to an alternative standard level can impact the magnitude of
risk reduction and consequently, the magnitude of risk predicted to remain upon
69 The difference in risk reductions based on application of different rollback methods in simulating the
current suite of standards reflects the fact that locally focused 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 locally focused 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.
4-38
-------
simulation of that standard. These results illustrate further that the type of rollback
used in simulating the current standard level can impact the magnitude of risk
reduction predicted for an alternative (lower) standard level. Particularly in urban
locations with peaky PM2.5 distributions, application of more localized patterns of
PM2.5 reduction (for the current standard), followed by a more regional pattern of
PM2 5 reduction for an alternative standard level can result in a larger estimate of
risk reduction for that alternative standard.
Based on the simulations discussed above covering potential sources of uncertainty and
variability impacting long-term mortality, we conclude that the following factors contribute
potentially large sources of uncertainty to the core risk estimates: (a) use of alternative form of
the C-R function, specifically use of a random-effects log-log model form obtained from the
updated ACS study (Krewski et al., 2009) (b) use of an alternative C-R function with effects
estimates obtained from the reanalysis of the Six Cities study (Krewski et al. 2000), and (c)
estimation of risk down to PRB.70 Other factors considered in the sensitivity analysis had
smaller impacts on core risk estimates.
Short-term exposure mortality
This section summarizes the results of the sensitivity analysis focused on short-term
exposure-related mortality endpoints (see Table 4-3 for the specific modeling elements
considered in the sensitivity analysis).
Impact of using season-specific C-R functions (vs. an annual C-R function): This
simulation considered the impact on short-term exposure mortality risk of using seasonally-
differentiated effects estimates rather than the core approach of using a single C-R function
for the whole year (note, that the seasonal models were based on the same study as the model
used in the core analysis - Zanobetti and Schwartz, 2009). The results of the simulation
suggest that this source of uncertainty can have a wide range of effects across urban study
areas (including not only variation in the magnitude of effect, but also in the direction).
Percent changes compared with the core risk estimate were large, ranging from -116% (Los
Angeles) to +179% (Birmingham) (these results are for non-accidental mortality - see
Appendix F, Table F-15). We note that these two locations also have relatively low overall
incidence estimates, which does raise concerns over the degree of stability in the sensitivity
analysis estimates. Furthermore, for 9 of the 15 urban study areas (for non-accidental
morality), percent changes from the core were small, with absolute values of 12% or less
(Appendix F, Tables F-15). The results for CV and respiratory mortality also demonstrate
considerable variation across locations, but are generally smaller than results cited above for
non-accidental, with one exception. Birmingham is estimated to have short-term CV
mortality that is +500% higher using seasonal effects estimates compared with the core
results (we note, however, that this endpoint category also has very small incidence, again
70 Use of locally focused rollback as an alternative method for simulating ambient PM2 5 concentrations for
alternative standards had a moderate-large impact on risk estimates.
4-39
-------
raising concerns over the stability of the sensitivity analysis results - see Table F-18). The
results for respiratory-related mortality also demonstrate considerable variability with results
that could suggest a moderate to large impact (i.e., -48 to +162% - see Appendix F, Table F-
21). We note, however, that small incidence estimates again raise concerns regarding the
stability of these percent difference results.
Impact of using alternative hybrid rollback approach: This simulation evaluates the
potential impact of using the hybrid approach for simulating just meeting current and
alternative sets of standards, as an alternative to the proportional approach used in the core
analysis.71 The results of this simulation (as contrasted with the impact of using the hybrid
approach on long-term exposure mortality) suggest that use of the hybrid rollback approach
has relatively little effect on short-term mortality risk (e.g., percentage differences relative to
the core risk estimates were in the low single digits for most locations, with one location
having a difference of+17% - see Appendix F, Table F-36).
The sensitivity analysis results discussed above, result in a number of overall
observations regarding sources of uncertainty potentially impacting short-term exposure morality
endpoints. The results of using the seasonally-differentiated effect estimates in modeling short-
term exposure mortality appear to generally have a relatively small impact (e.g., <15%) in most
study areas. For some study areas, the impact does appear to be much larger, with results
including both substantial negative and positive percent differences from the core estimates.
However, in all of these cases, the total incidence estimates involved are very small, raising
concerns over the stability of the risk estimates generated as part of this particular sensitivity
analysis (in many of these instances, the estimates include negative lower bounds, reflecting the
use of non-statistically significant effects estimates). For these reasons, the results of this
sensitivity analysis, while initially appearing to be notable in terms of magnitude in some study
areas, need to be interpreted with care. At this point, we are uncertain as to how important this
source of uncertainty is in the context of short-term exposure mortality estimation. Regarding
the use of the alternative hybrid (non-proportional) approach for simulating conditions under
alternative standard levels, the results suggest that this factor has a modest impact on short-term
exposure mortality (significantly less impact than with the use of the hybrid approach in
estimating long-term exposure mortality). With the exception of factors examined using the
Moolgavkar et al., (2003) study in Los Angeles (see below), it would appear that the factors
examined here do not have a large impact on risk estimates generated for short-term exposure
71 Note, that the locally focused 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 locally
focused 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 locally focused 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.
4-40
-------
mortality. However, we note that the overall scope of the sensitivity analysis completed for
short-term exposure-related mortality and morbidity is far more limited than that completed for
long-term exposure-related mortality.
Short-term exposure morbidity
This section summarizes the results of the sensitivity analysis focused on short-term
exposure-related morbidity endpoints (see Table 4-3 for the specific modeling elements
considered in the sensitivity analysis). The results of individual sensitivity analysis simulations
are presented below, with overall observations presented at the end of the section.
Impact of using season-specific C-R functions (vs. an annual C-R function): This
simulation considered the impact on short-term exposure morbidity (HAs) of using
seasonally-differentiated effects estimates rather than the core approach of using a single C-R
function for the whole year (we note that the seasonal models were obtained from the same
study as the model used in the core analysis - Bell et al, 2008). The results of the simulation
suggest that, as with short-term exposure mortality this source of uncertainty can have a wide
range of impacts on the risk estimates across urban study areas (including not only variation
in the magnitude of risk, but also in the direction) depending on the specific health endpoint
examined. We note, however, that the magnitude of impact appears to be less for short-term
morbidity than for short-term mortality. Percent changes for most of the 15 urban study
areas were small for CV HAs (generally less than a 20% difference in either direction,
although there was a large impact for Tacoma (-105%), see Appendix F, Table F-24). This
source of uncertainty has a moderate to moderate-large impact for respiratory-related HAs
with most locations having greater than a 54% to 74% absolute effect (see Appendix F, Table
F-27).
Impact of using a seasonal function for April through August (applied only to that
period) in modeling asthma-related ED visits in New York, relative to the core
approach of using a single annual effect estimate (and applying that to the whole year):
This sensitivity analysis involved the approach of using a season-specific estimate to model
incidence for the period April through August (obtained from Ito et al., 2007). Because this
sensitivity analysis estimate covers a period shorter than a year, we have not directly
compared it with the annual estimate generated for this endpoint in the core risk assessment
(i.e., we have not generated percent difference estimates as is done with other sensitivity
analysis simulations). However, the results of this sensitivity analysis do suggest that the use
of seasonally-differentiated estimates in modeling this endpoint can impact risk.
Impact of considering models with different lags: To examine the impact of lag on
modeling of short-term exposure-related morbidity, we used a range of effects estimates
obtained from Bell et al., 2008 based on application of different lags, including 0-, 1- and 2-
day lags, (for both respiratory and cardiovascular-related morbidity). Because lag-
differentiated effects estimates were only available as national-averages and were not
regionally-differentiated, we could not directly compare the results using different lag models
to the results generated for the core analysis (i.e., the sensitivity analysis results would have
mixed both the lag effect and the effect of regional differentiation, thereby preventing clear
assessment of the importance of either factor considered in isolation). However,
4-41
-------
consideration of the magnitude of the risk estimates generated using different lag models, for
the same endpoint at the same urban study are, suggests that choice of lag does effect
estimates of short-term exposure-related morbidity (see Appendix F, Table F-48).
Given the results of the set of simulations completed for short-term exposure morbidity,
both of which focused on the use of seasonally-differentiated effects estimates, it would appear
that this factor does not have a substantial impact on risk estimates. The analysis considering
different lag models does suggest that this factor could have a notable impact on risk estimates
and should be carefully considered when specifying C-R functions to use in the risk assessment.
Additional factors potentially impacting short-term exposure morbidity are addressed below in
relation to the sensitivity analysis based on alternative models from Moolgavkar et al. (2003). As
noted earlier, the scope of the sensitivity analysis completed for short-term exposure-related
morbidity is limited.
Short-term exposure-related mortality and morbidity (Moolgavkar et al., 2003 study-based
analysis)
As noted earlier in the introduction to section 4.3, the results of sensitivity analysis based
on Moolgavkar et al., (2003) include percent difference estimates based on considering the range
of risk estimates generated using alternative model specifications from this study for a given
health endpoint and it is these results that are discussed below.
Impact of model selection (e.g., log-linear GAM with 30df, log-linear GAM with lOOdf,
and log-linear GLM with lOOdf) on estimating short-term exposure mortality and
morbidity: Application of models obtained from Moolgavkar et al., (2003) with various
formulations related to model selection (degrees of freedom, GLM vs. GAM) to the Los
Angeles urban case study location results in a range of short-term exposure mortality
estimates (for non-accidental and CV) that differ by 80% and 49%, respectively (see
Appendix F, Table F-33). In the case of short-term exposure morbidity (specifically, CV-
related HAs), incidence estimates differ by 36% (see Appendix F, Table F-33). These results
suggest that these elements of model specification represent a moderate source of uncertainty
in estimating short-term mortality and morbidity.
Impact of lag structure (0-day through 5-day) on estimating short-term exposure
mortality: Consideration of the range of risk estimates for non-accidental mortality
generated using different lag structures (and associated effect estimates) provided in
Moolgavkar et al., (2003), suggest that this factor could have a moderate impact on risk (in
the range of 55% when comparing the lowest and highest positive incidence estimates
generated), (see Appendix F, Table F-33).
Impact of considering multi-pollutant models on estimating short-term exposure
mortality and morbidity: The results of the Moolgavkar-based simulations (when
considering the spread in risk estimates specifically across these simulations) suggest that the
multi-pollutant versus single-pollutant model issue (i.e., including CO in addition to PM2.5),
could have a large impact on the estimation of short-term exposure mortality (106% for all-
cause) and morbidity (140% for CV-related HAs).
4-42
-------
Overall observations regarding key sources of uncertainty impacting short-term exposure
mortality and morbidity risk estimates (based on the Moolgavkar et al., 2003 study) include the
following. The spread in risk estimates generated across the Moolgavkar-based model
specifications suggests that factor related to specifying the C-R model may have a moderate to
large impact. More specifically, variation in the lag structure has a moderate impact on risk and
use of single versus multi-pollutant models could have a potentially large impact on risk. Note,
however, that as discussed earlier, the relevance of these sensitivity analysis results to the
interpretation of core risk estimates is not clear and may be relatively low (see Section 4.3.1).
4.3.1.2 Multi-factor Sensitivity Analysis Results
The results of the multi-factor sensitivity analyses are intended to support both goals of
the sensitivity analysis: (a) identify which factors (now in combination), appear to have a
significant impact on estimation of the core estimates and (b) to derive a set of reasonable
alternative risk estimates for use in considering uncertainty and variability associated with the
core risk estimates. Regarding the latter application, because these multi-factor simulations
combine multiple factors reflecting uncertainty and variability together in generating alternative
risk estimates, they are likely to produce the highest sensitivity analysis results. Therefore, it is
particularly important to consider the reasonableness of the results of these multi-factor
simulations, to insure that only credible estimates are included in the set of reasonable alternative
risk estimates. Consequently, we emphasize consideration of the reasonableness of these multi-
factor simulations in the discussion presented below.
Long-term exposure mortality
This section summarizes the results of the sensitivity analysis focused on long-term
exposure-related mortality endpoints (see Table 4-3 for the specific modeling elements
considered in the sensitivity analysis).
Impact of using log-linear vs. log-log C-R model with fixed or random effects,
estimating incidence down to the LML vs. PRB, and using proportional vs. hybrid
rollback to estimate long-term exposure mortality: This multi-factor sensitivity
analysis focused on a number of model design choices related to modeling long-term
exposure mortality (all-cause and IHD). Modeling elements reflected in the
simulations included: (a) model form (log-linear vs log-log and random vs fixed
effects), (b) modeling risk down to PRB (vs LML), and (c) use of an alternative hybrid
rollback approach (vs proportional rollback) to simulate just meeting the current and
alternative sets of standards. Various permutations of these design elements choices
(relative to the elements selected for the core analysis) were considered. Percent
difference estimates (for all-cause mortality) ranged from 27% (for a model estimating
risk down to PRB and use of the hybrid rollback approach) to 1,089% (for a model
with random effects log-log model, risk estimated down to PRB, and use of the hybrid
rollback approach).
4-43
-------
We believe that application of a log-log model with random effects is a reasonable
alternative to the core model (fixed-effects log-linear model), based on our review of the
discussion in Krewski et al. (2009). Similarly, the use of a hybrid rollback approach involving
non-proportional adjustment where there is the potential for greater use of local control strategies
to address local-sources is a reasonable alternative to solely using a proportional rollback
approach in all study areas. Therefore, we believe that the combinations of modeling elements
including these alternative choices are reasonable. However, there is more concern in predicting
risk down to PRB. This is not because there is evidence for a threshold, but rather because we
do not have data to support characterization of the nature of the C-R function in the vicinity of
PRB. Specifically, there is increasing uncertainty in predicting the nature of the C-R function as
you move below the LML. So, while we believe it is reasonable conceptually to estimate risk
down to PRB, the quantitative process of doing this requires use of a function with very high
uncertainty. Therefore, we concluded that those alternative risk estimates generated using risk
estimated down to PRB should not be used in creating the reasonable alternative set of risk
estimates in considering uncertainty associated with the core risk estimates.
A key limitation of the multi-factor sensitivity analysis is that the approach used did not
allow us to consider the locally focused rollback method in concert with the other modeling
elements described above. This means that the combined impact of locally focused (which has a
greater impact than the hybrid rollback method) with other model specifications was not
characterized. However, as part of the integrative discussion in Chapter 5, we do consider the
results of the single-factor sensitivity analysis examining rollback (with its consideration of
locally focused) along with the multi-factor sensitivity analysis results described here.
Short-term exposure mortality
This section summarizes the results of the sensitivity analysis focused on short-term
exposure-related mortality endpoints (see Table 4-3 for the specific modeling elements
considered in the sensitivity analysis).
Impact of using season-specific vs. annual effect estimates and proportional vs.
hybrid rollback approaches in modeling short -term exposure mortality: This
multi-factor sensitivity analysis focused on a number of model design choices related
to modeling short-term mortality (non-accidental). Modeling elements included in this
sensitivity analysis were use of seasonal vs. annual effects estimates and use of hybrid
vs proportional rollback to simulate just meeting current and alternative standard
levels. Percent difference estimates (for non-accidental mortality) across the 7 urban
study areas included in the simulation ranged from -116% (LA) to +179%
(Birmingham) (see Appendix F, Table F-42). However, we note that the total
incidence estimates associated with these higher-impact locations were relatively low,
again raising the concern for the stability in relative differences with the core estimates.
4-44
-------
Because of the more limited scope of the multi-factor sensitivity analysis completed for
short-term exposure-related mortality, we have concluded that these results should not be used as
an additional set of reasonable risk estimates to inform consideration of uncertainty associated
with this category of risk estimates.
4.3.2 Additional Set of Reasonable Risk Estimates to Inform Consideration of
Uncertainty in Core Risk Estimates
This section discusses the use of the output of the sensitivity analysis completed as part
of this risk assessment as an additional set of reasonable risk estimates to inform consideration of
uncertainty associated with the core risk estimates. Specifically, in the case of long-term
exposure-related mortality endpoints, staff has concluded that the results of the sensitivity
analysis represent a reasonable set of alternate risk estimates that fall within an overall set of
plausible risk estimates surrounding the core estimates.72
While not representing a formal uncertainty distribution, the output of the sensitivity
analysis, when combined with the core risk estimates, represent a set of plausible risk estimates,
which reflect consideration of uncertainty in various elements of the risk assessment model.
Therefore, while the discussion of risk estimates in the context of assessing the degree of risk
reduction associated with suites of alternative standards (see Section 5.2) does focus on the core
risk estimates since these are judged to have the greatest overall confidence, the output of the
sensitivity analysis can be used to provide additional perspective on the potential range of
uncertainty around the core estimates. Note however, that we do not know the confidence
interval captured by this uncertainty set, or the specific percentiles of the risk distribution are
represented by points within that set.
As noted earlier, the quantitative single- and multi-factors sensitivity analyses generated
an additional set of risk estimates for a subset of the urban study areas, air quality scenarios and
health endpoints included in the core risk analysis (i.e., Los Angeles and Philadelphia assessed
for the current standard level). However, the part of the sensitivity analysis focusing on
alternative methods for simulating ambient PM2.5 levels (i.e., rollback), did consider a larger
number of study areas and air quality scenarios. In presenting the alternative sets of reasonable
risk estimates, we focus on Los Angeles and Philadelphia for many of the modeling elements,
72 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 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.
4-45
-------
although we expand the discussion in the context of discussing results related to conducting
rollback.
In using the additional set of reasonable risk results to augment the core risk estimates,
we begin by presenting both the core and alternative sets of estimates for Los Angeles and
Philadelphia in Table 4-4. Then, in Figures 4-7 and 4-8, we present graphical display of the full
uncertainty set comprising the core plus additional reasonable risk estimates for Los Angeles and
Philadelphia, differentiated by mortality category (Figure 4-7 present results for IHD and Figure
4-8 presents results for all cause mortality). This section concludes with a set of observations
resulting from consideration of information depicted in Table 4-4 and Figures 4-7 and 4-8 in the
context of interpreting uncertainty in the core risk estimates.73
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
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
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:
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:
73 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 PM25-attributable risk related to the secondary formation of sulfate. Note, that the risk
estimates for SO2 are presented as open circles in Figure 4-8, to signify that they have lower confidence and are de-
emphasized relative to the other alternative risk estimates presented.
4-46
-------
Core risk estimate
ambient data from
Krewskietal., 2009 -see
section 3.3.3)
Sensitivity analysis
Description of
simulation
function from
alternative long-term
exposure study
(Krewski et al, 2000)
(E) Impact of using
alternative roll-back
approach (hybrid and
locally focused) to
simulate just meeting
alternative standards
Results
(percent difference:
sensitivity analysis versus
core estimate)4
All cause: +121%
Philadelphia:
All cause: +11 9%
Los Angeles:
Both all cause & IHD: +21 to
+40% (hybrid and locally
focused, respectively)
Philadelphia:
Both all cause & IHD: +8%
(locally focused only)
Adjusted set of risk estimate
to supplement core risk
estimates1
All cause: 4.4%
Philadelphia:
All cause: 7.9%
Los Angeles and
Hybrid: IHD: 9.3%, All cause:
2.4%
Locally focused: 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
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
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).
Los Angeles
Philadelphia
les
lia
* *
* *
KEY:
0 10 15 20 25
Percent total incidence attributable to PM2 5 (IHD mortality)
- core risk estimate
- alternative reasonable risk estimate (from
single- and multi-element sensitivity analyses)
4-47
-------
Figure 4-8 Comparison of core risk estimates with reasonable alternative set of
risk estimates for Los Angeles and Philadelphia (all cause mortality).
Los Angeles
Philadelphia
*
KEY:
0.5 1 2 3 45 67 8
Percent total incidence attributable to PM, . (all cause mortality)
25
~J{ - core risk estimate
0 - alternative reasonable risk estimate (from single- and multi-
element sensitivity analyses)
O - alternative reasonable risk estimate for copollutant modeling
including PM2 5 and SO2 (de-emphasized relative to other alternative
risk estimates - see text)
Review of the set of risk estimates presented in Table 4-4 and displayed in Figures 4-7
and 4-8 results in a number of observations regarding uncertainty associated with the core risk
estimates:
Consideration of uncertainty and variability in the core risk estimates results in a notable
spread in risk estimates: Given the factors considered in generating the alternative set of
reasonable risk estimates, there appears to be a factor of 2 to 3 spread in risk estimates if
we consider the lowest (core) estimates generated and the highest alternative risk
estimates generated. This observation holds for both urban study areas considered, as
well as for the two mortality endpoint categories. As noted earlier in this section, we have
de-emphasized risk estimates generated using the copollutant model involving PM2.5 and
SC>2 due to concerns with collinearity between the two pollutants and the potential that
SC>2 represents risk attributable to secondarily formed PM2.5.
Uncertainty set of risk estimates generated to supplement the core risk estimates are
skewed towards higher risk: It appears that, given the factors considered in generating
the alternative set of reasonable risk estimates, consideration of uncertainty could result
in higher (more elevated) risk estimates, compared with the core risk estimates. In other
words, most if not all of the alternative model specifications we considered resulted in
risks that are higher than our core estimates.
4-48
-------
Sensitivity analysis is limited in its scope (potentially important sources of uncertainty
not considered): As noted earlier, the sensitivity analysis did not consider a number of
potentially important sources of uncertainty, some of which were addressed as part of the
qualitative analysis of uncertainty (see Table 3-13). For example, information is not
available to consider compositional differences in PM2 5 and the potential for
differentiation of effects estimates. Further, not considering more refined patterns of
intra-urban exposure to PM2.5 in deriving effects estimates could result in under-
estimation of risk.
It is important to reiterate that this set of alternative realizations presented in Table 4-4
and depicted in Figures 4-7 and 4-8, does not represent an uncertainty distribution. Therefore,
we can not assign percentiles to the individual data points presented and (importantly), we do not
draw any conclusions based on any clustering of the alternative risk estimates seen in Figures 4-7
and 4-8. Further, we do not know whether any of the higher-end estimates generated actually
represent true bounding risk estimates given overall uncertainty associated with the core risk
estimates. Despite these key caveats, having a set of risk estimates reflecting the impact of
modeling element uncertainties does provide information that helps to inform our
characterization of uncertainty related to the core risk estimates.
4.4 EVALUATING THE REPRESENTATIVENESS OF THE URBAN STUDY AREAS
IN THE NATIONAL CONTEXT
The goal in selecting the 15 urban study areas included in this risk assessment was two-
fold: (a) to choose urban locations with relatively elevated ambient PM levels (in order to
evaluate risk for locations likely to experience some degree of risk reduction under alternative
standards) and (b) to include a range of urban areas reflecting heterogeneity in other PM risk-
related attributes across the country. To further support interpretation of risk estimates generated
in this analysis, we included two analyses that assess the representativeness of the 15 urban study
areas in the national context. First, we assessed the degree to which urban study areas represent
the range of key PM2.5 risk-related attributes that spatially vary across the nation. We have
partially addressed this issue by selecting urban study areas that provide coverage for different
PM regions of the country (see section 3.3.2). In addition, we have evaluated how well the
selected urban areas represent the overall U.S. for a set of spatially-distributed PM2 5 risk related
variables (e.g., PM2 5 composition, weather, demographics including SES, baseline health
incidence rates). This analysis, which is discussed in section 4.4.1, helps inform how well the
urban study areas reflect national-level variability in these key PM risk-related variables. The
second representativeness analysis, which is discussed in section 4.4.2, identified where the
subset 31 counties comprising our 15 urban study areas fall along the distribution of national
county-level long-term exposure-related mortality risk. This analysis allowed us to assess the
degree of which the 15 urban study areas capture locations within the U.S. likely to experience
4-49
-------
elevated levels of risk related to PM2.5 exposure. To complete this second representativeness
analysis, we completed a national-scale county-level analysis of long-term exposure-related
mortality risk.
4.4.1 Analysis Based on Consideration of National Distributions of Risk-Related
Attributes
As noted above, the first representativeness analysis evaluated how well the urban study
areas reflect national-level variability in a series of PM risk-related variables.74 That analysis
was conducted as follows. Based on generally available data (e.g. from the 2000 Census, Centers
for Disease Control (CDC), or other sources), distributions for risk-related variables across U.S.
counties and for the specific counties represented in the urban study areas were generated. The
specific values of these variables for the selected urban study areas were then plotted on these
distributions, and an evaluation was conducted of how representative the selected study areas are
with respect to these individual variables, relative to the national distributions.
Estimates of risk (either relative or absolute, e.g. number of cases) within our risk
assessment framework are based on four elements: population, baseline incidence rates, air
quality, and the coefficient relating air quality and the health outcome (i.e., the PM2.5 effect
estimates). Each of these elements can contribute to heterogeneity in risk across urban locations,
and each is variable across locations. In addition, there may be additional identifiable factors
that contribute to the variability of the four elements across locations. In this assessment, we
examine the representativeness of the selected urban area locations for the four main elements,
and also provide additional assessment of factors that have been identified as influential in
determining the magnitude of the C-R function across locations.
The specific choice of variables which may affect the PM2.5 effect estimates for which we
will examine urban study area representativeness is informed by an assessment of the
epidemiology literature. We particularly focused on meta-analyses and multi-city studies which
identified variables that influence heterogeneity in PM2.5 effect estimates, and exposure studies
which explored determinants of differences in personal exposures to ambient PM2.5. While
personal exposure is not incorporated directly into PM epidemiology studies, differences in the
PM2.5 effect estimates between cities clearly is impacted by differing levels of exposure and
differences in exposure are clearly related to a number of exposure determinants. Broadly
74 In selecting these variables, we focused on variables that play a direct role in determining the relative
magnitude of PM-attributable risk, including potential effect modifiers. We did not focus on confounders, as these
were not primary factors we considered in selecting case study areas, and are not expected to impact the
representativeness of our risk estimates. As such, we excluded consideration of variables such as SO2, which are
potential confounders, but have not been identified as effect modifiers in the literature.
4-50
-------
speaking, determinants of the PM2.5 effect estimates used in risk assessment can be grouped into
three areas: demographics, baseline health conditions, and climate and air quality. Based on a
review of these studies, we identified the following variables within each group as potentially
determining the PM2.5 effect estimates:
Demographics: education (see Zeka et al, 2006; Ostro et al, 2006), age and gender (see
Zeka et al, 2006), population density (see Zeka et al, 2005), unemployment rates (see Bell
and Dominici, 2008), race (see Bell and Dominici, 2008), public transportation use (see
Bell and Dominici, 2008),
Baseline health conditions: disease prevalence (diabetes - Bateson and Schwartz, 2004;
Ostro et al, 2006; Zeka et al, 2006; pneumonia - Zeka et al, 2006; stroke - Zeka et al,
2006; heart and lung disease - Bateson and Schwartz, 2004; acute myocardial infarction
- Bateson and Schwartz, 2004).
Climate and air quality: PM2.5 levels (average, 98th percentiles, and numbers of days over
the level of the 24-hour standard, e.g. 35 |ig/m3), co-pollutant levels, PM composition
(see Bell et al, 2009; Dominici et al, 2007; Samet, 2008; Tolbert, 2007), temperatures
(temp) (days above 90 degrees, variance of summer temp, mean summer temp, 98th
percentile temp, mean winter temp see Roberts, 2004; Medina-Ramon et al, 2006; Zeka
et al., 2005), air conditioning prevalence (see Zanobetti and Schwartz, 2009; Franklin et
al, 2007; Medina-Ramon et al, 2006), ventilation (see Sarnat et al, 2006), percent of
primary PM from traffic (see Zeka et al., 2005),
Based on these identified potential risk determinants, we identified possible datasets that
could be used to generate nationally representative distributions for each parameter. We were
not able to identify readily available national datasets for all variables. In these cases, if we were
able to identify a broad enough dataset covering a large enough portion of the U.S., we used that
dataset to generate the parameter distribution. In addition, we were not able to find exact
matches for all of the variables identified through our review of the literature. In cases where an
exact match was not available, we identified proxy variables to serve as surrogates. For each
parameter, we report the source of the dataset, its degree of coverage, and whether it is a direct
measure of the parameter or a proxy measure. The target variables and sources for the data are
provided in Table 4-5. Summary statistics for the most relevant variables are provided in Table
4-6.
4-51
-------
Table 4-5 Data Sources for PM NAAQS Risk Assessment Risk Distribution
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
4-52
-------
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 O O
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
4-53
-------
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
4-54
-------
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
4-55
-------
Formal comparisons of parameter distributions for the set of urban study areas and the
national parameter distributions are conducted using standard statistical tests, e.g. the
Kolmogorov-Smirnov non-parametric test for equality of distributions. In addition, visual
comparisons are made using cumulative distribution functions, and box plots.
The formal Kolmogorov-Smirnov (K-S) test results are provided in Table 4-7. The K-S
tests the hypotheses that two distributions are not significantly different. A high p-value
indicates a failure to reject the null hypotheses that the case-study and national distributions are
the same. We used a rejection criterion of p<0.05, which is a standard rejection criterion. It
should be noted that the K-S test provides a good overall measure of fit, but will not provide a
test of how well specific percentiles of the distributions are matched. As such, the K-S test
results will not be sufficient to determine whether the urban study areas adequately capture the
tails of the distributions of specific risk related variables. Additional visual analyses are used to
assess representativeness for the tails of the distributions. Overall, the K-S test results show that
for many of the important risk variables such as population, air quality, age, and baseline
mortality rates, the urban study areas are not representative of the distributions of these variables
for the U.S. as a whole. However, for some important potential risk determinants, such as
prevalence of underlying hear and lung diseases, the case study areas are representative of the
national distributions. However, for these specific variables, the national distribution is
represented primarily by large urban areas, so it is more accurate in these cases to suggest that
the urban study areas are representative of the overall distribution across urban areas.
Figures 4-9 through 4-12 show for the four critical risk function elements (population, air
quality, baseline incidence, and the PM2 5 effect estimate) the cumulative distribution functions
plotted for the nation, as well as for the urban study areas. These four figures focus on critical
variables representing each type of risk determinant, e.g. we focus on all-cause mortality rates,
but we also have conducted analyses for cardiovascular and respiratory mortality separately. The
complete set of analyses is provided in Appendix D. The vertical black lines in each graph show
the values of the variables for the individual urban study areas. These figures show that the
selected urban study areas represent the upper percentiles of the distributions of population and
air quality, while not representing lower population locations with lower 24-hour PM2.5 levels.
This is consistent with the objectives of our case study selection process, e.g. we are
characterizing risk in areas that are likely to be experiencing excess risk due to PM levels above
alternative standards. The urban case study locations represent the full distribution of PM2.5 risk
coefficients, but do not capture the upper end of the distribution of baseline all-cause mortality.
The interpretation of this is that the case study risk estimates may not capture the additional risk
that may exist in locations that have the highest baseline mortality rates.
4-56
-------
Figures 4-13 through 4-16 shows for several selected potential risk attributes the
cumulative distribution function (CDF) plotted for the nation as well as for the urban study areas.
These potential risk attributes do not directly enter the risk equations, but have been identified in
the literature as potentially affecting the magnitude of the PM2.5 C-R functions reported in the
epidemiological literature. The selected urban study areas do not capture the higher end
percentiles of several risk characteristics, including populations over 65, income, and baseline
cardiovascular disease prevalence. Comparison graphs for other risk attributes are provided in
Appendix D. Summarizing the analyses of the other risk attributes, we conclude that the urban
study areas provide adequate coverage across population, population density, annual and 24-hour
PM2.5 levels, ozone co-pollutant levels, temperature and relative humidity, unemployment rates,
percent non-white population, asthma prevalence, obesity prevalence, stroke prevalence, exercise
prevalence, and less than high school education. We also conclude that while the urban study
areas cover a wide portion of the distributions, they do not provide coverage for the upper end of
the distributions of age (all case study locations are below the 85th percentile), % of population
65 and older (below 85th percentile), percent of primary PM emissions from mobile sources
(below 8oth percentile), prevalence of angina/coronary heart disease (below 85th percentile),
prevalence of diabetes (below 85th percentile), prevalence of heart attack (below 80th percentile),
prevalence of smoking (below 85th percentile), all-cause mortality rates (below 90th %ile),
cardiovascular mortality rates (below 90th percentile) and respiratory mortality rates (below 90th
percentile). In addition, all of the case study locations were above the 25th percentile of the
distribution of personal income.
Based on the above analyses, we can draw several inferences regarding the
representativeness of the urban case studies. First, the case studies represent urban areas that are
among the most populated and most densely population in the U.S. Second, they represent areas
with relatively higher levels of annual mean and 24-hour 98th percentile PM2.5. Third, they
capture well the range of effect estimates represented in the Zanobetti and Schwartz (2009)
study. These three factors would suggest that the urban study areas should capture well overall
risk for the nation, with a potential for better characterization of the high end of the risk
distribution. However, there are several other factors that suggest that the urban study areas may
not be representing areas that may have a high risk per microgram of PM2 5. The analysis
suggests that the urban study areas are not capturing areas with the highest baseline mortality
risks, nor those with the oldest populations. These areas may have higher risks per microgram of
PM2.5, and thus the high end of the risk distribution may not be captured, although the impact on
characterization of overall PM risk may not be as large, for the following reasons.
It should be noted that several of the factors with underrepresented tails, including age
and baseline mortality (R=0.81) are spatially correlated, so that certain counties which have high
4-57
-------
proportions of older adults also have high baseline mortality and high prevalence of underlying
chronic health conditions. Because of this, omission of certain urban areas with higher
percentages of older populations, for example, cities in Florida, may lead to underrepresentation
of high risk populations. However, with the exception of areas in Florida, most locations with
high percentages of older populations have low overall populations, less than 50,000 people in a
county. And even in Florida, the counties with the highest PM2.5 levels do not have a high
percent of older populations. This suggests that while the risk per exposed person per microgram
of PM2.5 may be higher in these locations, the overall risk to the population is likely to be within
the range of risks represented by the urban case study locations.
Table 4-7 Results of Kolomogrov-Smirnoff Tests for Equality Between National
and Urban Study Area Distributions for Selected National Risk
Characteristic Variables
(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
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
0.0001
0.0001
0.0248
0.0003
4-58
-------
Risk Attributes
% Mobile Source PM Emissions
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
Y
N
p-value
0.0133
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/)
.0)
'^
o
o
90% -| Urban case study areas are
all above the 65th Percentile
80% -I of county populations
70%
60%
50%
40%
30%
20%
10%
0%
100
1000
10000 100000
Population
1000000
10000000
All Counties CDF Case Study Counties CDF Case Study Counties
4-59
-------
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%
£ 50%
§ 40% -
ft 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
4-60
-------
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)
lUUVo
90%
80%
£ 70%
1 60%
o
ri 50%
2 40%
o
£ 30%
20%
10%
no/, -
^^^^^^^^^^^^
,_
I 1
Ul
1 1 I 1
I
r,
ii i
/
\ H 1 , 1 1
x
''
1 1 1 1
Ul
Ul
^^
X^
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
4-61
-------
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 (p) to
U.S. Distribution of PM All-cause Mortality Risk
(212 U.S. Urban Areas)
1 UU 70 -
90% -
% 80% -
0
< 70% -
a 60% -
.a
5 50% -
% 40% -
!g 30% -
£ 20% -
10% -
no/, _
*£
^ *
^
I , 1
^
X^
1 , -, 1
A
\ , 1
1
\ , i
,
§
\ i
^
it
i i
^x
1-^ 1
^^
\ \
\ \
r
\
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
4-62
-------
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)
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
4-63
-------
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)
lUUVo
90%
80%
£ 70%
1 60%
0
O
2 40%
0
£ 30%
20%
1 0%
n%
xi
i 1 1
1 1 i
M
W
H 1
1 1
«
l-rl
*^
1 1
Urban case study areas are
all below the 75th percentile
of county % of population 65
and older
i , , , ,
10 15 20 25
% of Population 65 and Older, 2005
30
35
All Counties CDF ^Case Study Counties CDF Case Study Counties
4-64
-------
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)
0%
$10
Urban case study areas are
all above the 25th percentile
of county per capita income
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
4-65
-------
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.
rv~t,--K * of Angina/CHD Prevalence
Distribution fa y s MSA)
100%
Urban case study areas are
all below the 85th percentile
of MSA angina or CHD
prevalence
345678
% Prevalence of Angina or CHD, BRFSS, 2007
All BRFSS MSA CDFCase Study CountyMSA CDEase Study dounty MSA
4.4.2 Analysis Based on Consideration of National Distribution of PM-Related Mortality
Risk
In this section we discuss the second representativeness analysis which identified where
the subset of 31 counties comprising the 15 urban study areas fall along a distribution of
estimated national-scale mortality risk. The national-scale mortality analysis which underpins
this representativeness analysis used 2005 PM2.5 fused air quality estimates from the Community
Model for Air Quality (CMAQ) (Byun and Schere, 2006) in conjunction with the environmental
Benefits Mapping and Analysis Program (BenMAP, Abt Associates Inc, 2008) to estimate long-
term PM2 s-related premature mortality nationwide at the county-level. In relating the 31 counties
comprising the 15 urban study areas to the national county-level distribution of risk, we did not
directly compare the 31 county-level risks 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.s mortality
4-66
-------
risks in the 31 counties modeled in the urban case study areas represented more typical or higher-
end risk relative to the national risk distribution.
The results of this representativeness analysis are presented graphically in Figure 4-17,
which 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.
Figure 4-17 Cumulative distribution of county-level percentage of total
mortality attributable to PMi.s for the U.S. with markers
identifying where along that distribution the urban case study area
analysis fall*
120%
ID
Marker identifies where, and
how many, counties considered
by the Risk and Exposure
Assessme nt fall alongthe
distribution of national baseline
mortality risk.
0% IX 2* 3!i 4% 5% 6% 7% S%
Percentage of mortalfty attributable toPM2 j exposure
'AtblJMiUbhmortjlltypvlod
The results of this analysis, as depicted in Figure 4-17, indicate that most of the 31
counties comprising the 15 urban study areas fall toward the upper end of the national risk
distribution and that 23 of these counties fall within the upper 5th percentile of the risk
distribution suggesting that the PM2.5 mortality risk estimates included in the urban case study
analysis generally represent the upper end of urban area mortality risks within the nation.
Additional details on this second representativeness analysis, together with discussion of the
national-scale mortality assessment underpinning the analysis, are presented in Appendix G.
4-67
-------
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.75 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
urban study areas to help to place risk estimates for the 15 urban study areas in a broader national
context.
This section discusses consideration of patterns of design values (section 4.5.1) and
underlying ambient monitoring PM2.5 data (section 4.5.2) for the 15 urban study areas in the
context of helping to interpret risk estimates. Each of these discussions begins by describing the
methods used in each analysis and concludes with a set of key observations.
4.5.1 Design Values
The set of design values for an urban study area determines whether the 24-hour or
annual standard will be controlling as well as the degree of reduction in ambient PM2 5
concentrations associated with a particular suite of standards. Therefore, by plotting the
relationship between 24-hour and annual design values for each of the 15 urban study areas, we
can obtain a quick visual perspective on (a) which study areas will experience reductions in risk
for a particular suite of standards, (b) whether the 24-hour or annual standard will control, and
(c) the general magnitude of risk reduction. The last observations result from comparing the
controlling standard level with the matching design value, which will determine the fractional
reduction in PM2.5 levels at monitors exceeding the standard level (for locally focused rollback),
or more broadly across all monitors (for proportional rollback).
Figures 4-18 through 4-20 present scatter plots of 24-hour and annual design values for a
combination of the 15 urban study areas (red stars) and the broader set of larger urban areas in
the U.S. (green circles). In addition to depicting the set of design values for these urban areas,
75 See section 3.2.3.1 for additional detail on derivation of 24-hour and annual design values.
4-68
-------
each figure also includes a set of superimposed lines representing the current suite of standards
(Figure 4-18) and three of the alternative suites of standards considered in the risk assessment
(12/35 - Figure 4-19, and 12/25 - Figure 4-20). In each figure, the horizontal line represents the
24-hour standard level, while the vertical line represents the annual standard level. The line that
intercepts the origin (i.e., the "35/15 line" in Figure 4-18) represents the point of demarcation
between those study areas where the 24-hour standard controls (to the left of the intercept line)
and those study area where the annual standard level controls (to the right of the intercept line).
By superimposing these lines related to the current standard level on the scatter plot, we have
created five zones within each figure including:
Zone A: 24-hour design values exceeding the 24-hour standard level, but annual design
values below the annual standard level (i.e., 24-hour standard is controlling). Urban study
areas in this zone are predicted to experience risk reduction with the degree of reduction
reflecting the degree to which the 24-hour design value exceeds the 24-hour standard level.
For example, in Figure 4-18 (depicting the current suite of standards), Tacoma and Salt Lake
City fall in this zone, along with 20-30 additional urban areas in the U.S.
Zone B: 24-hour design values and annual design values exceed 24-hour and annual
standard levels, respectively, and the 24-hour standard is controlling. We have further
transected this zone into Bl and B2, with the former representing those urban areas with
notably high 24-hour design values (Fresno, Los Angeles in Figure 4-18) and B2 those with
lower, although still controlling, 24-hour design values (Pittsburgh, New York, and Detroit in
Figure 4-18). Those urban areas in Bl have exceptionally peaky PM2.5 distributions relative
to urban areas in B2 (i.e., relatively high 24-hour design values and lower annual average
design values).
Zone C: 24-hour design values and annual design values exceed 24-hour and annual
standard levels, respectively, and the annual standard is controlling. Atlanta, Birmingham
and Houston fall into this zone and represent a relatively small number of urban areas in the
U.S..
Zone D: annual design values exceed the annual standard level, but 24-hour design values
are below the 24-hour standard level (i.e., annual standard is controlling). Houston is the only
urban study area falling into this zone for the current standard level, along with a small
number of additional urban areas in the U.S..
Zone E: both the 24-hour and annual design values are below their respective standard levels
(i.e., this is the only zone where urban areas would not be expected to experience risk
reductions under the suite of standards being considered). The majority of urban areas in the
U.S. depicted in these scatter plots fall into Zone E in Figure 4-18.
The five zones presented above are useful in interpreting the risk results generated for the
current suite of standards (for the 15 urban study areas). Specifically, as noted above, they allow
us to (a) quickly identify which of the 15 urban study areas experience risk reductions under the
current standard level, (b) determine whether those reductions are due primarily to a controlling
4-69
-------
24-hour or annual standard and (c) to see how well our set of urban study areas provide coverage
for the broader set of urban areas in the U.S.. We do note that three of the 15 urban study areas
(Baltimore, St. Louis and Birmingham) fall on or near lines demarcating zones B and C depicted
in Figures 4-18 and consequently can not be definitively assigned to either zone. While these
three urban study areas are assessed not to be in attainment of either the current annual or 24-
hour standard level, neither of these standards is definitively controlling for these urban study
areas.
In addition to presenting Figures 4-18 through 4-20 as a means for supporting the
interpretation of risk estimates generated for the 15 urban study areas (based on consideration of
patterns in design values), we have also included Table 4-8 for this purpose. Table 4-8 presents
the annual and 24-hour design values for each urban study area and also identifies which
standard is controlling for a given suite of standards. For example, we see that in Atlanta (which
has design values of 16.2 |ig/m3 and 35 |ig/m3, annual and 24-hour, respectively), the annual
standard controls for the current suite of standards (15/35) as well as the first 4 alternative suites
of standards considered (14/35, 13/35, 12/35 and 13/30). However, the 24-hour standard controls
for the final suite of standards considered (12/25). This matches with information presented in
Figures 4-18 through 4-20 (e.g., Figure 4-18 shows that the Atlanta is just inside of zone C,
suggesting that it meets the 24-hour standard, but not the annual standard).
4-70
-------
Figure 4-18 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 -
65
60 -
55
50 -
45
I
Philadelphia
*»r«Mr ^ Baltimore
Annual Design Value (ug/m3)
10 12 14 16 18 20 22 24 26 28 30
Key:
- urban study area included in risk assessment
-MSA's within the U.S.
4-71
-------
Figure 4-19 Design values in 15 urban study areas and broader set of U.S. urban
areas relative to the 12/35 alternative suite of standards
Phoenix -> ml
" jSQrv5f'' v Houston
t'»mm*am
M*«a><3
e- -3» .
/»»* Dallas
/ ^^^_? '
Annual Design Value (ug/m3)
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Key:
ir- urban study area included in risk assessment
-MSA's within the U.S.
4-72
-------
Figure 4-20 Design values in 15 urban study areas and broader set of U.S. urban
areas relative to the 12/25 alternative suite of standards
Annual Design Value (ug/m3)
Key:
8 10 12 14 16 18 20 22 24 26 28 30
- urban study area included in risk assessment
-MSA's within the U.S.
4-73
-------
Table 4-8 Identification of controlling standard (24-hour or annual) for
alternative suites of standard levels
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 4-18 through 4-20, 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 4-18), 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 24-hour 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 24-hour 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
4-19) Based on 2005-2007 air quality data, as shown in Figure 4-19, reduction in the annual
standard level down to 12 |ig/m3 results in a significant increase in the number of areas that
do not meet the annual standard (zones C and D). And of those areas, roughly similar
4-74
-------
numbers of urban areas do meet the 24hr standard as do not meet the 24hr standard
(comparing numbers of urban areas in B and C to the number in zone D).
Alternative suite of standards involving reductions in both annual and 24-hour levels (see
Figure 4-20): Based on 2005-2007 air quality data, a large fraction of urban areas are
predicted not to meet the 24hr standard (zones A, B and C). Furthermore, the majority of
these have the 24hr controlling (zone A and B). We also note that there are virtually no urban
areas that exceed the annual standard while meeting the 24hr standard (zone C). Of the 15
urban study areas, most do not meet either the 24hr or annual standards, while the 24hr is
controlling in most (zone B).
An additional factor to consider in examining the relationship between annual and 24-
hour design values for the set of larger urban areas in the U.S. is regionality (i.e., do we see
regional differences in the relationship between annual and 24-hour design values across the 15
urban study areas and the broader set of urban areas in the U.S?) To examine this issue, we color
coded the urban areas depicted in Figure 4-18 by PM region to produce a new scatter diagram
that allows us to look for regional patterns in the mix of annual and 24-hour design values across
the urban study areas (see Figure 4-21 - note that as with Figure 4-18, Figure 4-21 provides a
scatter diagram referenced on the current suite of standard levels: 15/35). Visual inspection of
Figure 4-21 suggests that, while urban locations in southern California and the Northwest tend to
dominate locations with relatively elevated 24-hour design values (i.e., urban locations with
higher 24-hour design values in zones A and B), the picture is less clear with regard to regional
patterns in other zones of the scatter diagram. When we look at portions of zones A and B closer
to the current 24-hour standard line, or in zones C, D or E, we do not see patterns of urban study
areas differentiated by PM region.
4-75
-------
Figure 4-21 Design values in 15 urban study areas and broader set of U.S. urban
areas relative to the current standard (with regional differentiation)
2005-2007 24-hour DVs versus Annual DVs in the 15 Study Areas & Additional MSAs, 35 /15
10
65
BO
65
SO
*
a
35
30
25
20
15
10
5
a
Sites in tnis quadrant
violate the 24-hr level only
E
O)
0>
"ro
c
- D)
_
o
*Fi»
Sites in this quadrant
violate both levels
y
Pftt-Utarty
*IA
^35/15 line
0) Phoanac->
Q /}
Annual level
Sitf.s in this quadrant
violate neither level
Sites in this quadrant
violate the annual level only
Annual Design Value (ug/m3)
Stars depict study
areas
Dote denote other
urban areas
Symbol color
inderrifies
geopjaptac report
Northeast
Southeast
Industrial Midwest
Upper Midwest
Northwest
Othe-(AK, HI, VI. PR)
0 2 4 6 8 10 12 14 16 18 ZJ 22 24 36 20 X
4-76
-------
4.5.2 Patterns in PM2.s Monitoring Data
As noted earlier, patterns in PM2 5 monitoring data for each of the 15 urban study areas
can be used (together with consideration of design values as described in section 4.5.1) to
support interpretation of risk estimates generated for current and alternative standard levels. This
is particularly true when considering the impact of using different rollback methods in
supporting risk characterization for current and alternative standard levels, as discussed below.
To facilitate consideration of patterns in PM2.5 monitoring data across the 15 urban study
areas, we have developed Figures 4-22and 4-23. Each of these figures presents 24-hour and
annual design values (blue and green dots, respectively) for each PM2.5 monitor within each
study area. The figures also flag the highest design values for each study area (red and brown
stars for the annual and 24-hour standard levels, respectively).76 Each figure has been scaled to
represent a particular suite of standards, with Figure 4-22 scaled to represent the current suite of
standards (15/35) and Figure 4-23 scaled to represent the 12/25 alternative suite of standards.77
In addition, the figures allow identification of whether a study area had the highest design value
(for the 24-hour and annual averaging periods) occurring at the same or at different monitors.
This factor can influence the degree to which simulation of a controlling 24-hour standard level,
given application of the locally focused rollback approach, results in reduction in annual average
PM2.5 levels for that study area. If an area has both 24-hour and annual design values occurring at
the same monitor, then application of locally focused rollback to reduce the controlling 24-hour
standard will also bring down the annual design value (i.e., the annual average PM2.5 level for
that study area is likely to be reduced to a greater extent). By contrast, if 24-hour and annual
design values are located at different monitors, then locally focused rollback focused on
reduction of the 24-hour design value monitor will potentially not impact the annual design value
(i.e., there will be a smaller impact on the annual average PM2 5 level for that study area).78
76 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.
77 For example, in Figure 4-22, 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).
78 When a star in either Figure 4-22 or 4-23 (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.
4-77
-------
To gain a better understanding of the information provided in Figures 4-22 and 4-23, we
will provide a walkthrough for one of the urban study areas, highlighting key attributes related to
24-hour and annual design values. With Los Angeles (in Figure 4-22) we see that the study area
has a relatively wide spread in 24-hour and annual design values across the monitors (i.e., it has a
relatively peaky PM2.5 distribution), with 24-hour values ranging from -15 to -55 and annual
design values ranging from -7 to -19 (exact values are presented in Appendix A). In addition,
we see that the 24-hour standard is clearly controlling, given how much farther the highest viable
24-hour design value is from the 15/35 line compared with the highest annual design value. In
addition, we can compare these trends in 24-hour and annual design values for Los Angeles to
those for the other urban study area and see that generally, Los Angeles (a) has some of the
widest spreads in both 24-hour and annual design values (i.e., it has one of the more peaky
PM2.5 distributions across monitors) and (b) has one of the highest 24-hour design value of the 15
urban study areas (i.e., it will require more rollback in simulating just meeting the current suite
of standards compared with most of the other study areas). The attributes described above match
well with urban areas falling into zone Bl in Figure 4-18 (i.e., the zone where urban areas do not
meet both the current 24-hour and annual standards, and where the 24-hour standard is
controlling).
4-78
-------
Figure 4-22 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 :
£T 22 ~-
^ 20 ^
O)
s^ -
0
13 :
03 16 :
£= 14:
CT
CO 19 -
0 12
_ 10:
O3
C 8:
^
< 6:
4:
2 :
0 :
u
D
^
O U)
"ra Q
3 .
M
i
N
1 ?
' «5
I
L
+f
l|
^
^^
$ t
n
*i
«t
A
*
*i
i
r
*
*
*
*
i *
* *
A
^
^
i
^
.
!j
*
i|
.*
*
f
|
B *
A
?
i
|
*'
A
A
^
A
A
0
*
^t A
t
*
*
*
1
t
I
!.
A
1
A
'i
!
*
*
Key: stars (*) = area DV for the site (red for annual and brown for 24-hour); dots () = other valid site DVs (green
for annual and blue for 24-hour); triangles (A) = invalid site DVs
ABBDDFHLN
t a i a e r o o e
1 1 r 1 t e u s w
a t m 1 r s s
n i i a o n t A Y
t mnsi oono
a o g t n g r
r n e k
e a 1
m e
s
P P P P S S
h h i i at
i 0 t t 1 .
1 e t t t
a n s - L
d i b L L o
e x u i a u
1 r b k i
p gees
h K r
i t
a y
770
765
760
755
750 ^
: ^
745 °>
^^
740 §
: 03
735 >
730 .2>
: 0
725 Q
^
-20 o
^
i
7 15 CN
7 10
75
1 n
T
a
c
0
m
a
4-79
-------
Figure 4-23 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:
-------
The sensitivity analysis examining uncertainty related to conducting rollback
demonstrated that for some of the study areas (e.g., Los Angeles and Salt Lake City) use of the
locally focused rollback method reflecting application of more localized controls resulted in
composite monitor values that differed notably from values generated when the proportional
rollback approach was used.79 In contrast, many of the other urban study areas displayed little
difference in composite monitor values based on application of proportional or locally focused
rollback methods.
Design value information provided in Figures 4-22 and 4-23 provides explanations for
these sensitivity analysis results. For Los Angeles (which had composite monitor values 40%
higher when using the locally focused rollback method compared with the proportional approach
- see Section 4.3.1.1), the 24-hour standard is controlling. This can be seen by noting that the
maximum 24-hour design value is significantly further away from the 15/35 line in Figure 4-22
compared with the maximum annual design value. In addition, these two maximum design
values do not occur at the same monitor.80 This means that when the proportional rollback
method is used, a relatively large fractional reduction is uniformly applied to all monitors,
resulting in a new (adjusted) composite monitor value that has been reduced to a relatively large
extent. However, if locally focused rollback is used, then only those monitors with 24-hour
design values exceeding the current 24-hour standard level are adjusted and only by the fraction
required to get each 24-hour design value down to the current 24-hour standard level.81 This
means that in an overall sense, there is less adjustment to PM2.5 levels, such that with locally
focused we will see higher composite monitor annual averages than with proportional rollback.
In the case of Salt Lake City (which also has significantly higher composite monitor
annual averages with locally focused than with proportional rollback), while the highest 24-hour
and annual design values occur at the same monitor, which means that even with locally focused,
the monitor with the highest annual averages will be adjusted downward substantially, because
the annual design values for monitors are closer to each other, the impact of locally focused on
the composite annual average is smaller. Specifically, while some of the monitors with 24-hour
design values above the current 24-hour standard level will have their annual averages adjusted
79 Recall that differences in composite monitor estimates represent surrogates for differences in long-term
exposure-related mortality - long-see section 4.3.1.1.
80 In figures 4-22 and 4-23, 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.
81 With the locally focused 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.
4-81
-------
down, there is a fraction of the monitors (with 24-hour design values below the current standard)
that will not be adjusted with application of locally focused rollback.
These two examples illustrate different conditions under which the type of rollback
applied can have a significant impact on the degree of public health protection assessed for a
particular standard level. By contrast, conditions at some of the other urban study areas result in
little difference in risk from application of different rollback methods (i.e., simulation of more
regional versus local control strategies). Specifically, if an urban location has 24-hour and annual
design values at each monitor that display little variation, we expect to see less impact on risk
from varying the type of rollback method used. Examples that fall into this latter category
include Atlanta, Dallas, and St. Louis (see Figure 4-22).
4-82
-------
5 INTEGRATIVE DISCUSSION OF URBAN CASE STUDY ANALYSIS
OF PM2 s-RELATED RISKS
This chapter provides an integrative discussion of the risk-related analyses presented
throughout this final RA, including (a) the core PM2.s-related risk estimates generated for the set
of urban study areas (sections 4.1 and 4.2), (b) the related uncertainty and sensitivity analyses,
including additional set of reasonable risk estimates generated to supplement the core analysis
(sections 3.5.4, and 4.3), (c) assessment of the representativeness of the urban study areas in the
national context (section 4.4), and (d) consideration of patterns in design values and air quality
monitoring data to inform interpretation of risk estimates generated for the urban study areas
(section 4.5). The goal of this integrative discussion is to lay out information in such a way as to
inform consideration of the policy-relevant risk-related questions which are considered in the
PA.
We begin by discussing the overall level of confidence associated with estimates of risk
presented in the RA (section 5.1). We then discuss key observations about the nature and
magnitude of long-term and short-term exposure-related risks estimated for the air quality
scenarios considered in the RA, including: (a) the current suite of PM2.5 standards (section 5.2.1),
(b) alternative annual standard levels paired with the current 24-hour standard level (section
5.2.2) and (c) combinations of alternative annual and 24-hour standard levels (section 5.2.3). As
part of these discussions we consider the role played by 24-hour and annual standards in
influencing the nature and magnitude of estimated risk reductions across the 15 urban study
areas. At the end of this discussion, we summarize our key observations (section 5.3).
This integrative discussion focuses on those health endpoints for which quantitative risk
estimates were generated as part of the RA. However, additional health endpoints for which risks
could not be quantified, but that are of potential concern, are considered in the PA. The PA also
considers the extent to which health endpoints considered in the RA, as well as the additional
endpoints considered in the PA, are of importance from a public health perspective.
5.1 OVERALL CONFIDENCE IN THE RISK ASSESSMENT
This quantitative risk assessment has been designed to generate estimates of risk for a set
of urban study areas likely to represent those urban areas in the U.S. experiencing higher PM2.5-
related risk due to elevated PM2.5 concentrations and/or other attributes related to PM2.5 risk
(e.g., meteorology, baseline health effects incidence rates, differences in PM2.5 emissions sources
and composition). The RA includes design elements intended to increase overall confidence in
the risk estimates generated for the 15 urban study areas including: (a) use of a deliberative
process in specifying components of the risk model that reflects consideration of the latest
5-1
-------
research on PM2.5 exposure and risk, (b) integration of key sources of variability into the design
and interpretation of results from the analysis, (c) assessment of the degree to which the urban
study areas included in the RA are representative of areas in the U.S. experiencing higher PM2 5-
related risk, and (d) identification and assessment of the impact of important sources of
uncertainty on core risk estimates. In addition to these design elements, we also completed two
additional analyses which examine potential bias and overall confidence in the risk estimates.
The first of these analyses explored potential bias in the core risk estimates by considering a set
of alternative reasonable risk estimates generated as part of a sensitivity analysis. The second
analysis compared the annual average PM2.5 concentrations simulated under both current and
alternative standard levels with the air quality distribution used in deriving the C-R functions
applied in modeling mortality risk. Greater confidence is associated with risk estimates based on
annual average PM2.5 concentrations that are within the region of the air quality distribution used
in deriving the C-R functions where the bulk of the data reside (i.e., within one standard
deviation (SD) around the mean). Each of the design elements listed above together with the two
additional analyses is discussed below.
5.1.1 Use of a Deliberative Process in Designing the Risk Model
To increase overall confidence in the RA, a deliberative process has been used in
specifying each of the analytical elements comprising the risk model, including selection of
urban study areas as well as specification of other inputs such as C-R functions. This deliberative
process involved rigorous review of available literature addressing both PM25 exposure and risk
combined with the application of a formal set of criteria to guide development of each of the key
analytical elements in the risk assessment. In addition, the risk assessment design reflects
consideration of CASAC and public comments on the initial risk assessment plan and the first
draft risk assessment. The application of this deliberative process increases overall confidence in
the risk estimates by insuring that the estimates are based on the best available science and data
characterizing PM2.5 exposure and risk, and that they reflect consideration of input from experts
on PM exposure and risk through CASAC and public reviews.
The approach used in specifying several of the key analytical elements used in the risk
assessment is highlighted below for purposes of illustrating the systematic approach used in
developing the model.
Selection of the 15 urban study areas included consideration of (a) whether a city of county
had been included in multi-city epidemiology studies used in specifying C-R functions used
in the core risk estimates, (b) providing coverage for urban areas with relatively high annual
and 24-hour design values, and (c) providing coverage for the seven PM regions which
reflect differences in key PM risk-related attributes (e.g., meteorology, demographic
5-2
-------
attributes, PM sources and composition). See section 3.3.2 for additional detail on selection
of study areas.
Simulation of ambient PMg_5 concentrations under current and alternative standard levels
included the proportional rollback approach used in past risk assessments, which reflects a
regional pattern of ambient PM2.s reductions. To more fully reflect potential variability in
the future pattern of ambient PM2.5 reductions, we added two alternative rollback approaches
(hybrid and locally focused), both of which simulate more localized patterns of ambient
PM2.5 reductions combined with additional regional patterns of reduction (see section 3.2.3).
Selection of health endpoints reflected consideration of the degree of support in the literature
for a causal relationship between PM2.5 exposure and the health effect of interest as assessed
in the ISA, together with consideration of the health significance of the endpoint. In addition,
we considered whether sufficient information existed in the literature to develop C-R
functions and whether we could obtain the baseline incidence data necessary to generate risk
estimates with a reasonable degree of confidence for a particular endpoint (see section 3.3.1).
The selection of epidemiological studies and specification of C-R functions for use in
modeling risk involved a rigorous review of existing literature based on application of
criteria we identified for specifying robust C-R functions. These criteria took into account
both study design as well as the potential scope of the C-R functions that could be drawn
from the studies (e.g., geographic coverage, demographic groups covered and health
endpoints involved). We outlined our rationale for the set of epidemiology studies we
selected and the choices made in specifying C-R functions, and we discussed our rationale
for not including other potential studies and/or forms of C-R functions in the risk assessment
(see section 3.3.3).
The systematic approach described above resulted in a core risk model which included
those model inputs that in our judgment have the greatest degree of support in the literature.
These core risk estimates are emphasized in addressing the policy-related questions considered in
the PA. To provide a more comprehensive assessment of risk for the urban study areas, we have
included an assessment of uncertainty and variability and their impact on the core risk estimates
as part of this analysis, as discussed below.
5.1.2 Integration of Key Sources of Variability into the RA Design
The RA has been designed to provide coverage for key sources of variability which can
impact the nature and magnitude of risks associated with current and alternative standard levels
across the urban study areas. Several of these sources of variability contribute to differences in
risk across urban study areas, but do not directly affect the degree of risk reduction associated
with the simulation of a particular standard levels (i.e., their impact on risk is constant across the
air quality scenarios evaluated for a given study area). These sources of variability include
differences in PM2.5 sources and composition and differences in baseline incidence rates,
demographics and population behavior related to PM2.5 exposure and risk across urban areas.
5-3
-------
Coverage for these sources of variability is provided through inclusion of study areas drawn from
different PM regions in the U.S, since these factors tend to demonstrate regional differences.
In contrast, two additional sources of variability not only introduce variability into risk
estimates across study areas, but also play an important role in determining the magnitude of risk
reductions upon simulation of current and alternative standard levels. These sources of
variability are (a) the degree of "peakiness" in monitored PM2.5 concentrations across urban
study areas 82 and (b) variability in the spatial patterns of ambient PM2.5 reduction resulting from
simulation of current or alternative standard levels. Variability in the peakiness of monitored
PM2.5 concentrations is covered in the RA by including urban study areas from different PM
regions, while variability in the spatial pattern of ambient PM2.5 reduction resulting from
simulation of just meeting current or alternative standard levels is addressed by including
multiple rollback methods in the RA.
As discussed in sections 4.3.1.1 and 4.5, the interplay of 24-hour and annual design
values in a given study area (reflecting the peakiness of ambient PM2.5 concentrations in that
study area) also plays an important role in determining the magnitude of projected risk
reductions under current and alternative standard levels. For example, those study areas with
relative peaky PM2.5 distributions and where the 24-hour standard is controlling can be especially
sensitive to the type of rollback approach used, with the proportional approach resulting in
notably greater risk reduction compared with the locally focused approach. Rigorous
consideration of these factors (i.e., the interplay of the rollback method used with patterns of 24-
hour and annual design values for a given study area), allowed us to obtain a better
understanding of the nature and pattern of risk reductions and risk remaining following
simulation of just meeting both current and alternative standard levels across the urban study
areas. This in turn increased our overall confidence in the RA, since we could better explain
complex patterns of risk reduction seen for some of the study areas.
As discussed in section 4.3.1.1, the nature of the spatial pattern of reductions in ambient
PM2 5 concentrations resulting from simulation of just meeting current or alternative standard
levels, as reflected in application of different rollback methods, can have a substantial impact on
the magnitude of risk reductions estimated for those standard levels. If a more generalized
regional pattern of reduction is assumed (as reflected in the proportional rollback approach), all
monitors in the study area are adjusted by the same proportion and there is a relatively greater
82 The term "peakiness" as used here refers to air quality distributions across urban areas that have high
peak-to-mean ratios relative to distributions in other urban study areas in the U.S. Here, the peak concentration is
represented by the 24-hour design value, while the mean concentration is represented by the annual average design
value for a particular monitor.
5-4
-------
impact on risk. In contrast, if only those monitors exceeding the 24-hour standard are subjected
to an initial stage of reduction to bring the concentrations down to match those at adjacent
monitors (as reflected in the hybrid and to an even greater extent in the locally focused
approaches), there will be a relatively smaller reduction in risk.
5.1.3 Representativeness of the Urban Study Areas
The assessment of the degree to which the 15 urban study areas are representative of
areas within the U.S. likely to experience elevated PM2.5-related risk draws on information
presented in several sections of the RA including: (a) the analysis based on consideration of
national distributions of risk-related attributes (section 4.4.1), (b) analysis based on evaluating
where 31 counties comprising the 15 urban study areas fall along a national distribution of PM-
related mortality risk (section 4.4.2) and (c) consideration of patterns of design values for the 15
urban study areas as contrasted with the broader set of urban areas within the U.S. (section
4.5.1). Key observations from these representativeness analyses are presented below.
Comparison of attributes of the 15 urban study areas (assessed at the county-level) against
national distributions for the same attributes (section 4.4.1) suggests that the 15 urban study
areas represent areas in the U.S. that are among the most densely populated, have relatively
higher levels of annual and 24-hour 98th percentile PM2.5 concentrations, and capture well the
range of effect estimates represented by the Zanobetti and Schwartz (2009) study. Together,
these factors suggest that the urban study areas should likely reflect the distribution of risk
for the nation, with the potential for better characterization of the high end of that
distribution.83
Analysis of where the 15 urban study areas fall along the distribution of U.S. counties
included in the national-scale mortality analysis completed as part of the RA further suggests
that we have captured counties likely to experience elevated PM2.s-related risk (see section
4.4.2). Specifically, this analysis suggests that our urban study areas capture the upper end of
the tail with regard to PM2.5-attributable risk, with 23 of these counties falling within the
upper 5th percentile of the distribution. These findings support the assertion based on the
other analyses described above that the urban study areas are likely to capture risk at urban
areas experiencing relatively elevated levels of PM2.5-attributable mortality.
Consideration of the mix of design values across the 15 urban study areas as contrasted with
design values for the broader set of urban study areas in the U.S. suggests that (a) the 15
urban study areas do a good job of capturing the key groupings of urban areas in the U.S.
83 This representativeness 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 PM25
concentrations (e.g., urban areas in Florida) and consequently failure to include these areas in the set of urban study
areas is unlikely to exclude high PM2 s-risk locations.
5-5
-------
likely to experience elevated risk due to PM (i.e., they provide coverage for the zones
containing urban study areas likely to experience risk reductions under the suites of
alternative standard levels considered - see section 4.5.1) and (b) we have included study
areas likely to experience relatively greater degrees of PM2.5-related risk, based on
consideration of the pattern of design values across urban areas in the U.S.. In addition, for
the current suite of standard levels, the 24-hour standard is controlling for most of our 15
urban study areas, reflecting the pattern seen in the U.S. for urban areas assessed not to be
meeting the current suite of standards (based on 2005-2007 air quality data - see section
4.5.1).
Our overall assessment of the representativeness of the 15 urban study areas in the
national context, based on the three analyses summarized above, is that these areas are
representative of urban areas in the U.S. experiencing elevated levels of risk related to ambient
PM2.5 exposure.
5.1.4 Impact of Important Sources of Uncertainty on Core Risk Estimates
As part of the RA, we completed both a qualitative analysis of uncertainty (section 3.5.3)
as well as a quantitative sensitivity analysis (section 3.5.4 and 4.3), both designed to identify
those sources of uncertainty having a potentially important impact on core risk estimates.84 Key
observations from these analyses are presented below.
The qualitative analysis of uncertainty identified the following sources of uncertainty as
potentially having a medium to high impact on core risk estimates (see Table 3-13 for
additional detail): (a) characterizing intra-urban population exposure in the context of
epidemiology studies linking PM2.5 to specific health effects, (b) statistical fit of the C-R
functions (short-term exposure-related health endpoints), (c) shape of the C-R functions, (d)
specifying lag structure (short-term exposure studies), (e) transferability of C-R functions
from study locations to urban study area locations (long-term exposure-related health
endpoints), (f) use of single-city versus multi-city studies in the derivation of C-R functions,
(g) impact of historical air quality on estimates of health risk from long-term PM2.5 exposures
and (h) potential variation in effects estimates reflecting compositional differences for PM.
The quantitative single-factor sensitivity analysis identified the following sources of
uncertainty as having a moderate to large impact on core risk estimates (see section 4.3.1 for
additional detail):
o Long-term exposure-related mortality: (a) different C-R function model choices (e.g.,
fixed versus random effects, log-linear versus log-log, single- versus multi-pollutant),
(b) modeling risk down to PRB rather than LML, (c) impact of using C-R functions
84 The sensitivity analysis also produced an additional set of reasonable risk estimates that augments the
core risk estimates in considering the impact of uncertainty and variability in the core risk model (this application of
the sensitivity analysis is discussed below in section 5.1.5).
5-6
-------
from different epidemiological studies (e.g., ACS versus six cities), and (d) nature of
the spatial pattern of ambient PM2 5 reductions (i.e., rollback method used).
o Short-term exposure-related mortality and morbidity: (a) use of seasonally-
differentiated versus annual-based C-R functions, and (b) different models, lag
structures and single-versus multi-pollutant model forms (these results based on
Moolgavkar, 2003 study which is not as directly applicable in the context of our RA -
see section 4.3.1.1).
The quantitative multi-factor sensitivity analysis applied both to long-term exposure-related
mortality and short-term exposure-related mortality and morbidity showed that a number of
sources of uncertainty could work in concert to produce notably larger impacts on core risk
estimates (see section 4.3.1.2).
The qualitative analysis of uncertainty and quantitative sensitivity analyses described
above provided us with a comprehensive understanding of which sources of uncertainty were
likely to have a significant impact on the core risk estimates. This information proved useful in
interpreting core risk estimates and increases our overall confidence in the analysis.
5.1.5 Consideration of Alternative Reasonable Risk Estimates
As noted above, the quantitative sensitivity analysis produced an additional set of
reasonable risk estimates that augments the core risk estimates in considering the impact of
uncertainty and variability in the core risk model. Most of the alternative model specifications
supported by available literature produced risk estimates that are higher (by up to a factor of 2 to
3) than the core risk estimates. This is not unexpected, since the epidemiological study used in
obtaining the C-R functions for estimating long-term exposure-related mortality in the core
analysis (Krewski et al., 2009) is based on the ACS dataset which does not provide
representative coverage for lower-SES segments of the general population that are at greater risk
from PM exposure. Because of this, there is the potential that effect estimates, and consequently
risk estimates based on the ACS dataset are biased low, relative to risks estimated for the general
population (which would include the lower SES-population) and especially relative to risks that
might be estimated for the lower-SES population. In contrast, the alternative epidemiological
study considered in the sensitivity analysis for modeling long-term exposure-related mortality
(i.e., Krewski et al., 2000, based on the Six Cities dataset) provides better coverage for lower
SES individuals and has a higher effect estimate and consequently generates higher risk
estimates.
While use of C-R functions from Krewski et al., 2009 does introduce potential for low
bias in the core RA, because of the other strengths associated with this study (e.g., larger number
of cities, inclusion of two time periods which allows us to consider different exposure windows
and analysis of a wide range of C-R function models), the risk assessment team concluded that
5-7
-------
C-R functions obtained from this study had the greatest overall support and should be used in the
core risk model. However, consideration of the alternative set of reasonable risk estimates does
provide several observations relevant to the interpretation of the core risk estimates including: (a)
the core estimates are unlikely to under-estimate risk and (b) the degree of potential bias in the
core risk estimates could range up to at least a factor of 2-3 higher. 85
5.1.6 Consideration of composite monitor annual-average PM2.s concentrations in
relation to the dataset used in deriving C-R functions for long-term exposure-
related mortality
In considering the overall confidence in the core risk estimates, we have compared the
PM2.5 concentrations simulated under both current and alternative standard levels across the
urban study areas to the distribution of PM2 5 concentrations used in deriving the C-R functions
used for long-term exposure-related mortality (as presented in Kreswki et al., 2009). Specifically,
this assessment compares the composite monitor annual average PM2.5 concentrations used in
modeling long-term exposure-related mortality risk in the core analysis to the distribution of
annual-average PM2.5 concentrations from the 1999-2000 ACS exposure period.86 Generally,
when composite monitor annual average values are within one SD of the mean of the ACS
dataset (i.e., in the range of 11 |ig/m3 or above), we have relatively high confidence in those risk
estimates, since they are based on PM2.5 concentrations that roughly match those used in deriving
the C-R functions. However, as composite monitor annual average PM2 5 concentrations extend
below this range, our confidence in the risk estimates decreases, with our confidence being
significantly reduced when composite monitor annual average values reach or extend below the
LML of the ACS dataset (i.e., 5.8 jig/m3).
85 We note that these findings regarding potential bias in the core risk estimates were based on modeling
PM2 s-attributable IHD and all-cause mortality associated with long-term PM2 5 exposure for the current suite of
standards. However, we would expect these observations regarding overall confidence in the core risk estimates to
hold for other long-term exposure-related mortality endpoints modeled in the RA for both the alternative annual and
24-hour standard levels considered. Furthermore, given increased emphasis placed in this analysis on long-term
exposure-related mortality, the uncertainty analyses completed for this health endpoint category are 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.
86 As discussed in sections 3.3.3 and 4.0, each category of long-term exposure-related mortality is
estimated using separate C-R functions derived form the 1979-1983 and 1999-2000 ACS monitoring periods. For
purposes of comparing composite monitor annual-average PM2 5 levels to these ACS datasets used in deriving the C-
R functions, we focus on the later monitoring period (1999-2000), since ambient PM25 levels from this period more
closely match those associated with the study areas in our simulation under recent conditions. The 1999-2000 ACS
monitoring period has a mean PM2 5 level of 14 ug/m3, a SD of 3.0 ug/m3 and an LML of 5.8 ug/m3 (see Table 1 in
Krewskietal.,2009).
-------
5.2 KEY OBSERVATIONS RELATED TO THE URBAN STUDY AREA RESULTS
This section provides key observations from the simulation of risks under current and
alternative suites of standards for the set of 15 urban study areas. These observations are focused
on providing information relevant to addressing the key policy-relevant risk-related questions
that are considered in the PA.
In presenting these observations, we focus on cardiovascular-related endpoints given the
greater overall degree of confidence assigned to this category in the ISA relative to other health
effect categories (e.g., respiratory-related effects). This means that for long-term exposure-
related risk, we focus our discussion on IHD-related mortality and for short-term exposure-
related risk we focus on CV-related mortality and morbidity (the latter in the form of HAs related
to CV symptoms). Although not discussed here, risk estimates were also generated for additional
health effect categories including all-cause, cardiopulmonary, and lung cancer mortality (for
long-term exposure) and non-accidental- and respiratory-related mortality and respiratory
effects-related HAs and asthma-related emergency department visits (for short-term exposure).
It is also important to note that a broader array of health effects beyond those modeled in the RA
has also been associated with PM2.5 exposure, including reproductive and developmental effects.
While information was too limited to consider these effects in this quantitative RA, such effects
are appropriately considered based on the related evidence in the broader characterization of
risks to be discussed in the PA.
The role of annual average ambient PM2.5 concentrations in driving long-term exposure-
related risk is intuitive given that this risk category is modeled using the annual average air
quality metric.87 The fact that changes in the annual average air quality metric can also impact
short-term exposure-related risk is less intuitive, since changes in risk for this category are
modeled using average daily PM2.5 concentrations and not annual averages.88 However, as
discussed in section 3.1.2.2, because the 24-hour PM2 5 distributions tend to be approximately
normal or log-normal in form, overall incidence of short-term exposure-related mortality tends to
be driven by the relatively large number of days near the center of the distribution, rather than
87 As noted in section 3.2.1, estimates of long-term exposure-related mortality are actually based on an
average annual PM2 5 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 Table 3-4 and both are referenced in this
discussion of core risk estimates, as appropriate.
88 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.
5-9
-------
the small number of days out at the tail. Therefore, a shift in the annual average, which will most
directly focus on the bulk of the days near the center of the 24-hour average PM2.5 distribution,
can have a significant effect on short-term exposure-related risk aggregated over a year. This
means that in order to assess the impact of alternative 24-hour standard levels on short-term
exposure-related mortality, it is most appropriate to first consider how those alternative 24-hour
standards impact the composite monitor annual average PM2.5 concentration for a study area,
since this will ultimately determine the magnitude of reductions in both long-term and short-term
exposure-related mortality, as well as short-term exposure-related morbidity.
5.2.1 Nature and Magnitude of Long-Term and Short-Term Exposure-Related Risk
Remaining upon Just Meeting the Current Suite of PMi.s Standards
In considering PM2.s-related risks likely to remain upon just meeting the current PM2.5
annual and 24-hour standards in the 15 areas included in this assessment, we focus on the 13
areas that would not meet the current standards based on recent (2005-2007) air quality. These
13 areas have annual and/or 24-hour design values that are above the levels of the current
standards (Table 4-8).89 Based on the core risk estimates for these areas presented in section
4.2.1, we make the following key observation regarding the magnitude of risk remaining upon
simulation (using proportional rollback) of just meeting the current suite of standards:
Long-term exposure-related mortality risk remaining: The core analysis estimates that the
urban study areas would have IHD-related mortality attributable to long-term PM2.5 exposure
ranging from <100 to approximately 2,000 cases per year, with this variability reflecting to a
great extent differences in the size of study area populations. These estimates represent from
4 to 17% of all IHD-related mortality in a given year for the urban study areas, which is a
measure of risk that takes into account differences in population size and baseline mortality
rates.
Short-term exposure-related mortality and morbidity risk remaining: The core analysis has
short-term exposure-related CV-related mortality across the urban study areas ranging from
<10 to 500 cases per year. These estimates represent from ~1 to 2% of total CV-related
mortality in a given year for the urban study areas. In terms of morbidity risk, CV-related
HA range from -10 to 800 cases per year across the study areas, with this translating into
<1% of total CV-related HA incidence.
Although short-term and long-term exposure-related mortality have similar patterns (in
terms of the subset of urban study areas experiencing risk reductions for a particular standard
level), the magnitude of risk remaining is substantially lower for short-term exposure-related
mortality. These findings are expected, since, as noted in the introduction to section 5.2, changes
89 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.
5-10
-------
in annual average PM2.5 concentrations are expected to drive both short-term and long-term
exposure-related risk, resulting in similar overall patterns in risk reduction for both categories of
endpoint (in terms of the subset of study areas experiencing risk reductions). Note, however, that
variability in the effect estimates used in modeling short-term exposure-related health endpoints
across study areas will introduce additional variation into the pattern of risk reduction across
study areas.
Substantial variability across study areas in the magnitude of risk remaining: Estimated risks
remaining upon just meeting the current suite of standards vary substantially across study
areas, even when considering risks normalized for differences in population size and baseline
incidence rates. This variability in estimated risks is a consequence of the substantial
variability in the annual average PM2.5 concentrations across study areas that result from
simulating just meeting the current standards. This is important because, as noted in the
introduction to this section, annual average concentrations are highly correlated with both
long-term and short-term exposure-related risk. This variability in annual average PM2 5
concentrations occurs especially in those study areas in which the 24-hour standard is the
"controlling" standard. In such areas, the variability across study areas in estimated risks is
largest when regional patterns of reductions in PM2.5 concentrations are simulated (using
proportional rollback, as was done in the core analyses), with less variability when more
localized patterns of PM2.5 reductions are simulated (using locally focused rollback, as was
done in a sensitivity analysis). When simulations are done using locally focused rollback,
estimated risks remaining upon just meeting the current suite of standards can be appreciably
larger than those estimated in the core analysis.
Simulation of risk involves annual average PM2.s concentrations well below the current
standard for some study areas: In simulating just meeting the current suite of standards, the
resulting composite monitor annual average PM25 concentrations range from about 15 |ig/m3
(for those study areas in which the annual standard was controlling) down to as low as about
8 |ig/m3 (for those study areas in which the 24-hour standard was controlling or the annual
average was well below 15 |ig/m3based on recent air quality). As discussed above in section
5.1.6. as the composite monitor annual average PM25 concentrations used in generating risk
estimates extend below 11.0 |ig/m3 (one SD below the mean for the 1999-2000 ACS
monitoring period) we have increasingly less confidence in those risk estimates, with
confidence decreasing significantly as composite monitor concentrations approach the LML
for the ACS dataset (5.8 jig/m3). We make the observation that all four of the urban study
areas with composite monitor annual average PM2 5 concentrations below 11 |ig/m3 under
simulated attainment of the current suite of standards have the 24-hour standard controlling
(see Table 3-4). This illustrates the point that for the 15 urban study areas assessed,
typically, it is the locations where the 24-hour standard is controlling that are simulated to
have the lowest composite monitor annual average PM2 5 concentrations. While these
locations often are estimated to have the greatest risk reductions, there is also reduced
confidence associated with these risk estimates.
5-11
-------
5.2.2 Nature and Magnitude of Long-term and Short-Term Exposure-Related Risk
Remaining upon Just Meeting the Alternative Suite of PM2.s Standards
In characterizing PM2.5-related risks associated with simulation of the alternative annual
standards (14/35, 13/35 and 12/35), we estimate both the magnitude of risk reductions (relative
to risk remaining upon just meeting the current suite of standards) as well as the magnitude of
risk remaining upon just meeting the alternative standards. In discussing these risks, we focus on
the set of urban study areas experiencing risk reductions under each alternative annual standard.
Key policy-relevant observations associated with these risk estimates include:
Reductions in long-term exposure-related mortality risk: Upon simulation of just meeting the
alternative annual standard levels considered (14, 13, and 12 |ig/m3) in conjunction with the
current 24-hour standard (denoted as 14/35, 13/35 and 12/35 suites of standards), the core
analysis estimates reductions in long-term exposure-related mortality for 12 of the 15 urban
study areas, with the degree of risk reduction increasing incrementally across the alternative
standard levels (both in terms of the number of study areas experiencing risk reduction and
the magnitude of those reductions). For the alternative annual standard level of 12 |ig/m3 (in
conjunction with the current 24-hour standard), the core analysis estimates that these study
areas have reductions in risk (relative to risk remaining upon just meeting the current suite of
standards) ranging from about 11 to 35%.
For some of those areas in which the 24-hour standard is controlling, larger risk
reductions would have been estimated in this case (12/35 suite of standards) if locally focused
rollback had been used to simulate just meeting the current suite of standards. This result would
be expected since the magnitude of risk remaining upon just meeting the current suite of
standards would have been higher than that estimated based on the proportional rollback used in
the core analysis. Therefore, while we would have gone down to the same level of risk (under the
12/35 suite of standards), we would have started with a higher level of simulated risk from the
current standard.
Long-term exposure-related mortality risk remaining: For an annual standard level of 14
|ig/m3, the percent of total incidence of long-term exposure-related HID mortality
attributable to PM2.5 (i.e., risk remaining) in the 5 urban study areas experiencing risk
reductions ranges from 9-15%. For an annual standard of 12 |ig/m3, estimated risk remaining
in the 12 urban study areas experiencing risk reductions ranges from 6-11% in terms of
PM2.s-attributable long-term exposure-related mortality. This translates into between 90 and
300 cases per year attributable to long-term PM2.5 exposure for those study areas
experiencing the greatest reductions in risk under the lowest alternative annual standard
simulated.
Simulation of risks for an alternative standard level below 12 jug/m3: Simulation of risks for
an alternative annual standard of 10 |ig/m3 suggests that additional risk reductions could be
expected with alternative annual standards below 12 |ig/m3. However, we recognize that
there is potentially greater uncertainty associated with these risk estimates compared with
estimates generated for the higher alternative annual standards considered in the RA, since
5-12
-------
these estimates require simulation of relatively greater reductions in ambient PM2.5
concentrations. As lower ambient PM25 concentrations are simulated (i.e., ambient
concentrations further from recent conditions), potential variability in such factors as the
spatial pattern of ambient PM2.5 reductions (rollback) increases, thereby introducing greater
uncertainty into the simulation of composite monitor annual average PM2.5 concentrations
and consequently risk estimates.
Short-term exposure-related mortality and morbidity risk: For the alternative annual standard
level of 12 |ig/m3 (in conjunction with the current 24-hour standard), the core analysis
estimates that reductions in both short-term exposure-related CV mortality and morbidity risk
ranged from 5 to 23%.90 In terms of risk remaining upon simulation of 12 |ig/m3 (in
conjunction with the current 24-hour standard), the urban study areas with the greatest
percent reduction have CV-related mortality estimates ranging from 25 to 50 deaths per year.
Substantial variability in magnitude of risk reduction across urban study areas: While there
is a consistent pattern of risk reduction across the alternative annual standards with lower
standard levels resulting in more urban study areas experiencing increasingly larger risk
reductions, there is considerable variability in the magnitude of these reductions across study
areas for a given alternative annual standard level (e.g., as noted above, for the alternative
annual standard level of 12 |ig/m3, risk reduction ranges from 11% to 35% for the study areas
experiencing risk reductions). This variability in risk reflects differing degrees of reduction
in annual average concentrations across the study areas. These differences in annual
averages result in part because the study areas begin with varying annual average PM2 5
concentrations after simulating just meeting the current suite of standards. Therefore, even if
study areas have similar "ending" annual average PM2.5 concentrations after simulation of
just meeting the a given alternative annual standard, because the starting point in the
calculation (the annual average PM2.s concentrations upon just meeting the current suite of
standards) can be variable, the overall reduction in annual average PM2.5 concentrations
across the standards can also be variable. This translates into variation in reductions in long-
term exposure-related risk upon just meeting alternative annual standard levels across the
study areas.
The nature of the spatial pattern in PM2.s reductions (reflected in the rollback method used)
can impact the magnitude of risk reductions: The sensitivity analysis involving application of
locally focused rollback reveals that the pattern of reductions in ambient PM2.5
concentrations upon just meeting the current suite of standards can impact the magnitude of
additional risk reductions estimated for just meeting alternative (lower) annual standard
levels. Specifically, for those study areas with more peaky PM2.5 distributions, application of
locally focused rollback will result in higher annual average PM2 5 concentrations remaining
upon just meeting the current suite of standards. If proportional rollback is then used to
simulate just meeting alternative annual standard levels, a greater degree of reduction in
composite monitor annual average PM2.5 concentrations will result, since the "starting point"
for the calculation (annual average PM2.5 concentrations upon just meeting the current suite
90 Because the same air quality metric (annual distributions of 24-hour PM2 5 concentrations) is used in
generating short-term exposure-related mortality and morbidity endpoints, patterns of risk reduction (as a percent of
risk under the current suite of standards) are similar for both sets of endpoints (see section 4.2.2).
5-13
-------
of standards) will be higher. These findings highlight the important roll played by variability
in the spatial pattern of ambient PM25 concentrations in influencing the magnitude of risk
reductions under alternative annual standard levels.
Based on consideration of the composite monitor annual average PM2.s concentrations
involved in estimating long-term exposure-related mortality, we have varying levels of
confidence in risk estimates generated for the three alternative annual standard levels
considered: With the exception of one study area, those study areas estimated to have risk
reductions under the alternative annual standards of 14 and 13 |ig/m3 have simulated
composite monitor annual average PM2.5 concentrations ranging from just below 10.6 to over
13.3 |ig/m3 (see Table 3-4). In other words, these composite monitor annual average PM2.5
concentrations generally fall well within the range of ambient PM2 5 data used in fitting the
C-R functions used (i.e., within one SD of the mean PM2.s concentration from 1999-2000
ACS dataset). The urban study areas estimated to have risk reductions under the lower
alternative annual standard level of 12 |ig/m3 have lower composite monitor annual average
values ranging from 9.0 to over 11.4 |ig/m3. These values extend to below one SD of the
mean of the ACS dataset and therefore, we have somewhat lower confidence in these risk
estimates, relative to those generated for the two higher alternative annual standards. By
contrast, urban study areas estimated to have risk reductions under the alternative standard
level of 10 |ig/m3 (paired with the current 24-hour standard) have simulated composite
monitor annual estimates ranging from 7.6 to 8.9 |ig/m3 (see Table J-19). These
concentrations are towards the lower end of the range of ACS data used in fitting the C-R
functions (in some cases approaching the LML) and therefore, we conclude that we have
significantly less confidence in these risk estimates, compared with those for the higher
alternative annual standards assessed.
5.2.3 Nature and Magnitude of Long-Term and Short-Term Exposure-Related Risk
Remaining upon Just Meeting Combinations of Alternative Annual and 24-Hour
PM2.5 Standards
In characterizing PM2.5-related risks associated with simulation of the alternative annual
standards combined with alternative 24-hour standards (13/30 and 12/25), we estimate both the
magnitude of risk reductions (relative to risk remaining upon just meeting the current suite of
standards) as well as the magnitude of risk remaining upon just meeting the alternative standards.
In discussing these risks, we focus on the set of urban study areas experiencing risk reductions
under each alternative annual standard.
Additional reduction in long-term exposure-related risk provided by considering alternative
24-hour standards combined with alternative annual standards: In the case of the 12/25 suite
of standards, estimated reductions in long-term exposure-related mortality risk compared
with reductions for the annual standard alone (12 |ig/m3), were roughly twice as large in
many of the study areas, although in a few areas risk reductions were much higher (ranging
up to -100%) and in a few other areas, there was little to no risk reduction.
5-14
-------
These results show that lower 24-hour standards can have an appreciable and highly
variable impact on long-term exposure-related mortality, particularly when just meeting the
lower standards is simulated using a more regional pattern of PM2 5 reductions (i.e., the
proportional rollback used in the core analysis). However, the magnitude of risk reductions
estimated for the lower 24-hour standards was reduced when simulations using a more localized
pattern of PM2.5 reductions (i.e., the locally focused rollback used in the sensitivity analysis).
Based on consideration of the composite monitor annual average PM2.s concentrations
involved in estimating long-term exposure-related mortality, we have lower degrees of
confidence in risk estimates generated for the two alternative 24-hour standard levels
considered (30 and 25 jug/m ): Of the 1 1 urban study areas estimated to have risk reductions
under the alternative 24-hour standard of 30 |ig/m3 (with the 24-hour standard controlling -
see Table 3-4), composite monitor annual average PM2.5 concentrations range from 6.6 to
1 1.3 |ig/m3 with most of the urban study areas having concentrations in the 8 to 10 |ig/m3
range. These concentrations extend into the lower range of PM2 5 data used in the ACS study
to fit the C-R functions and therefore, we have somewhat lower confidence in these
estimates. When we consider composite monitor concentrations for urban study areas
assessed to have risk reductions under the alternative 24-hour standard level of 25 |ig/m3
(again, where the 24-hour standard is controlling), we see composite monitor annual average
PM2.s levels ranging from 5.6 to 1 1.2 |ig/m3 with most study areas have concentrations in the
range of 7 to 9 |ig/m3. Because this range extends well into the lower range of ACS data
used in fitting the C-R functions (in some cases extending below the LML), we have
significantly lower confidence in these risk estimates.
Increased variability associated with simulating composite monitor annual average .
concentrations for the lowest 24-hour standard considered (when it is controlling): We note
that risk estimates generated for the subset of urban study areas where the alternative 24-hour
standards of 30 and 25 |ig/m3 are controlling are subject to additional variability related to
simulating the spatial pattern of ambient PM2 5 concentrations under these alternative levels
(i.e., application of rollback methods). In those scenarios where the alternative 24-hour
standard is controlling, our sensitivity analyses showed the application of alternative rollback
methods (particularly the proportional versus locally focused) produce substantially different
composite monitor annual average PM2 5 concentrations, which translate into differences in
estimated risk. Based on our sensitivity analysis results, this source of variability (rollback)
does not have as great an impact in those instances where alternative annual standard levels
are controlling.
Alternative 24-hour standard levels provide inconsistent degree of risk reduction: The results
of simulating alternative suites of standards including a combination of alternative annual
and 24-hour standard levels suggest that the alternative 24-hour standard can produce
additional estimated risk reduction beyond that provided by the alternative annual standard
alone. However, the degree of estimated risk reduction provided by the alternative 24-hour
standard is highly variable, as illustrated by the considerable spread in the percentage
reductions in long-term exposure-related IHD mortality risk seen across study areas when
comparing risk under the 12/25 suite of alternative standard levels to risk under the current
standard (see section 4.2.2). More consistent reductions in estimated risk and consequently
5-15
-------
degrees of public health protection are estimated to result from simulating just meeting the
alternative annual standard levels considered.
5.3 SUMMARY OF KEY OBSERVATIONS
Key policy-relevant observations drawn from discussions presented in sections 5.1 and
5.2 are summarized below.
The use of a deliberative process in specifying models and inputs used in the RA together
with consideration of key sources of variability and uncertainty in designing the modeling
approach increases our overall confidence in the risk estimates that are generated. In
addition, based on the consideration of both the qualitative and quantitative assessments of
uncertainty completed as part of the analysis, we believe it unlikely that the RA as
implemented has over-stated risk, particularly for long-term PM2.5 exposure-related
mortality. In fact, the core risk estimates for this category of health effect endpoint may well
be biased low based on consideration of alternative model specifications evaluated in the
sensitivity analysis.
Based on the results of several analyses examining the representativeness of the 15 urban
study areas, we believe that the RA provides coverage for urban areas in the U.S. likely to
experience elevated risk due to ambient PM2.5 exposure, consistent with the original goal set
out for the RA.
Simulation of just meeting the current suite of standards (15/3 5) suggests that long-term
exposure-related IHD mortality attributable to PM2.5 exposure in the urban study areas
included in this assessment could range from <100 to 2,000 cases per year across the study
areas, which translates into a range from 4-17% of total IHD-related mortality incidence.
Short-term exposure-related CV mortality risk is lower by up to an order of magnitude.
Simulation of just meeting the alternative annual standard levels evaluated (12, 13, and 14
|ig/m3), combined with the current 24-hour standard (35 |ig/m3), resulted in estimated risk
reductions for most of the urban study areas, with the degree of risk reduction increasing
incrementally across the alternative standard levels (both in terms of the number of study
areas experiencing risk reduction and the magnitude of those reductions). Estimated
reductions in long-term exposure-related IHD mortality (relative to the risk remaining upon
just meeting the current suite of standards) ranged from 11 to 35% across the study areas for
the lowest alternative annual standard considered (12 jig/m3). Additional, but more uncertain
reductions in long-term exposure-related mortality risk were estimated with the simulation of
an alternative annual standard level of 10 |ig/m3.
In general, we have the most confidence in risk estimates based on PM2.5 concentrations near
the mean PM2.5 levels in the underlying epidemiological studies providing the C-R functions.
As PM2.5 concentrations decrease from these mean levels, we have decreasing confidence in
the risk estimates. Risk estimates for the alternative annul standard levels of 14 and 13 |ig/m3
are based on PM2.5 concentrations that are generally within one SD of the mean of the ACS
data from which the C-R functions are derived. Consequently we have a relatively high
degree of confidence in these risk estimates. Risk estimates for the alternative annual
standard level of 12 |ig/m3 are based on PM2.5 concentrations that begin to extend into the
lower range of data used in fitting the C-R functions and consequently our confidence in
5-16
-------
these estimates is somewhat lower. Risk estimates for the alternative standard level of 10
|ig/m3 are based on PM2 5 concentrations that extend well into the lower range of the ACS
data and consequently, we have significantly less confidence in these risk estimates
compared with estimates generated for the other alternative annual standard levels
considered.
Simulation of just meeting combinations of alternative annual and 24-hour standard levels
(13/30 and 12/25) resulted in estimates of additional risk reductions in some study areas
compared with the alternative annual standards alone, particularly for the 12/25 combination,
where estimated reductions were roughly twice as large in many of the study areas, with a
few areas experiencing substantially higher estimated risk reductions (ranging up to -100%
of estimated risk).
Risk estimates for the alternative 24-hour standard level of 30 |ig/m3 are based on composite
monitor annual average PM2.5 concentrations that span a wide range extending from within
one to below two SDs of the mean of the ACS data used in fitting the long-term exposure-
related mortality C-R functions. We have somewhat lower confidence in these estimates
relative to risk estimates based on annual mean concentrations more consistently in the range
of one SD of the mean of the ACS data. In contrast, risk estimates for the alternative 24-hour
standard level of 25 |ig/m3 are based on lower composite monitor annual average PM2.5
concentrations that extend well into the lower range of the ACS data (in some cases
extending down to the LML) and therefore, we have significantly less confidence in these
risk estimates.
Risk estimates generated for the alternative 24-hour standard levels are subject to substantial
variability related to the spatial pattern of ambient PM2.5 concentrations (i.e., rollback)
assumed in simulating these standard levels. Application of more localized patterns of
ambient PM2.5 reduction (locally-focused rollback) versus more regional patterns of
reduction (proportional rollback), can produce significantly different degrees of risk
reduction. This variability in risk reduction associated with application of different rollback
methods was not as pronounced with simulation of the alternative annual standard levels
considered.
While the alternative 24-hour standard levels considered (when controlling) did result in
additional estimated risk reductions beyond those estimated for alternative annual standards
alone, these additional estimated reductions are highly variable, in part due to different
rollback approaches. Conversely, alternative annual standard levels, when controlling,
resulted in more consistent risk reductions across urban study areas, thereby potentially
providing a more consistent degree of public health protection.
5-17
-------
6 REFERENCES
Abt Associates Inc. (1996). A Paniculate Matter Risk Assessment for Philadelphia and Los Angeles. Prepared for
Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle
Park, NC. July 1996. Available electronically on the internet at:
http://www.epa.gov/ttnnaaqs/standards/pm/data/jlv3amd.pdf
Abt Associates Inc. (2003), "Environmental Benefits Mapping and Analysis Program (BenMAP), User's Manual.
Available on the Internet at .
Abt Associates Inc. (2005). Paniculate Matter Health Risk Assessment for Selected Urban Areas. Prepared for
Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle
Park, NC. June 2005. Available electronically on the internet at:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_td.html.
Abt Associates Inc. (2008). Environmental Benefits Mapping and Analysis Program (Version 3.0). Bethesda, MD.
Prepared for Environmental Protection Agency, Office of Air Quality Planning and Standards, Air Benefits
and Costs Group. Research Triangle Park, NC.
Bateson, T.F.; Schwartz, J. (2004). Who is sensitive to the effects of paniculate air pollution on mortality? A case-
crossover analysis of effect modifiers. Epidemiology. 15(2): 143-9.
Bell, M.L.; Ebisu, K.; Peng, R.D.; Walker, J.; Samet, JM; Zeger, S.L.; Dominici, F. (2008). Seasonal and regional
short-term effects of fine particles on hospital admissions in 202 US counties, 1999-2005. Am J
Epidemiol. 168(11):1301-1310.
Bell, M.L.; Dominici, F. (2008). Effect modification by community characteristics on the short-term effects of ozone
exposure and mortality in 98 US communities. Am J Epidemiol. 167(8):986-97.
Bell (2009). Personal communication via e-mail between Dr. Michelle Bell and Beth Hasset-Sipple, June 26th, 2009
(e-mail with attachments placed in docket #EPA-HQ-OAR-2007-0492).
Bell M.L.; Ebisu K.; Peng R.D.; Samet J.M.; Dominici, F. (2009). Hospital admissions and chemical composition of
fine particle air pollution. Am J Respir Crit Care Med. 179(12): 1115-20.
Berkey, C.S.; Hoaglin, D.C.; Antczak-Bouckoms, A.; Mosteller, F.; Colditz, G.A. (1998). Meta-analysis of multiple
outcomes by regression with random effects. Stat Med. 17(22):2537-2550.
Byun, D., and K.L. Schere, (2006). Review of the Governing Equations, Computational Algorithms, and Other
Components of the Models-3 Community Multiscale Air Quality (CMAQ) Modeling System. Applied
Mechanics Reviews, 59, 51-77.
Dominici, F.; Peng, R.D.; Ebisu, K.; Zeger, S.L.; Samet, J.M.; Bell, M.L. (2007). Does the effect of PM10 on
mortality depend on PM nickel and vanadium content? A reanalysis of the NMMAPS data. Environ Health
Perspect. 115(12):1701-3.
Eftim SE, Samet JM, Janes H, McDermott A, Dominici F. (2008). Fine Paniculate Matter and Mortality: A
Comparison of the Six Cities and American Cancer Society Cohorts With a Medicare Cohort.
Epidemiology, 19: 209- 216.
Franklin, M.; Zeka, A.; Schwartz, J. (2007). Association between PM25 and all-cause and specific-cause motrality in
27 US communities. J Expo Sci Environ Epidemiol. 17:279-287.
6-1
-------
Goss CH: Newsom SA, Schildcrout JS, Sheppard L, Kaufman JD. (2004). Effect of ambient air pollution on
pulmonary exacerbations and lung function in cystic fibrosis. Am J Respir Crit Care Med, 169: 816-821.
Henderson, R. (2005). EPA's Review of the National Ambient Air Quality Standards for Paniculate Matter (Second
Draft PM Staff Paper, January 2005): A review by the Paniculate Matter Review Panel of the EPA Clean
Air Scientific Advisory Committee. June 6, 2005. Available electronically on the internet at:
http://www.epa.gov/sab/pdf/casac-05-007.pdf.
Henderson, R. (2006). Clean Air Scientific Advisory Committee Recommendations Concerning the Proposed
National Ambient Air Quality Standards for Paniculate Matter. March 21, 2006. Available electronically
on the internet at: http://www.epa.gov/sab/pdf/casac-ltr-06-002.pdf
Henderson, R; Poirot, R.L.; Cowling, E.; Speizer, F.; Crapo, J.D.; Zielinska, B.; Miller, F.J. (2006). Clean Air
Scientific Advisory Committee Recommendations Concerning the Final National Ambient Air Quality
Standards for Paniculate Matter. September 29, 2006. Available electronically on the internet at:
http://www.epa.gov/sab/pdf/casac-ltr-06-003.pdf.
Henderson, R. (2008). Letter from Dr. Rogene Henderson, Chair, Clean Air Scientific Advisory Committee to the
Honorable Stephen L. Johnson, Administrator, US EPA. Clean Air Scientific Advisory Committee
Paniculate Matter Review Panel's Consultation on EPA's Draft Integrated Review Plan for the National
Ambient Air Quality Standards for Particulate Matter. January 3, 2008. Available:
http://yosemite.epa.gov/sab/sabproduct.nsf/264cbl227d55e02c85257402007446a4/76D069B8191381DA8
52573 C500688E74/$File/EP A-CASAC-08-004-unsigned.pdf
Hopke, P. (2002). Letter from Dr. Phil Hopke, Chair, Clean Air Scientific Advisory Committee (CASAC)
Particulate Matter Review Panel, to Honorable Christine Todd Whitman, Administrator, U.S. EPA. Final
advisory review report by the CASAC Particulate Matter Review Panel on the proposed paniculate matter
risk assessment. EPA-SAB-CASAC-ADV-02- 002. May 23. Available electronically on the internet
at:http://www.epa.gov/sab/pdf/casacadv02002.pdf.
Hopke, P. (2004). Letter from Dr. Phil Hopke, Clean Air Scientific Advisory Committee
(CASAC) Particulate Matter (PM) Review Panel, to Honorable Michael O. Leavitt. CASAC PM Review Panel's
Ongoing Peer Review of the Agency's Fourth External Review Draft of Air Quality Criteria for Particulate
Matter (June 2003); and Peer Review of the Review of the National Ambient Air Quality Standards for
Particulate Matter: Policy Assessment of Scientific and Technical Information (OAQPS Staff Paper - First
Draft) (August 2003) and a Related Draft Technical Report, Particulate Matter Health Risk Assessment for
Selected Urban Areas (Draft Report) (August 2003). EPA-SAB-CASAC-04-004. February 18. Available
electronically on the internet at: http://www.epa.gov/sab/pdf.
Ito, K.; Thurston, G.; Silverman, R.A. (2007). Characterization of PM2 5 gaseous pollutants and meteorological
interactions in the context of time-senes health effects models. J Expo Sci Environ Epidemiol. 17: 45-60.
Jerrett. M.; Burnett, R.T.; Ma, R.; Pope, C.A. Ill; Krewski, D.; Newbold, K.B.; Thurston, G.; Shi, Y.; Finkelstein,
N.; Calle, E.; Thun, M.J. (2005). Spatial analysis of air pollution and mortality in Los Angeles.
Epidemiology. 6(6): 1-10.
Kottek (2006) - use below
Kottek, M.; Grieser, J.; Beck, C.; Rudolf, B.; Rubel, F. (2006). World map of the Koppen-Geiger climate
classification updated. Meteorol Zeit. 15(3): 259-263.
Kottek, M. (2008). World Map of the Koppen-Geiger climate classification updated. Available: http://koeppen-
geiger.vu-wien.ac.at/ [accessed 12/5/2008].
6-2
-------
Krewski (2009). Letter from Dr. Daniel Krewski to HEFs Dr. Kate Adams (dated July July 7, 2009) regarding
"EPA queries regarding HEI Report 140". Dr. Adams then forwarded the letter on July 10, 2009 to EPA's
Beth Hassett-Sipple. (letter placed in docket #EPA-HQ-OAR-2007-0492).
Krewski, D.; Burnett, R.T.; Goldberg, M.S.; Hoover, K.; Siemiatycki, J.; Jerrett, M; Abrahamowicz, M; White,
W.H. (2000). Reanalysis of the Harvard Six Cities study and the American Cancer Society study of
paniculate air matter pollution and mortality. HEI Research Report, Health Effects Institute, Boston, MA.
Krewski, D.; Jerrett, M.; Burnett, R.T.; Ma, R.; Hughes, E.; Shi, Y.; Turner, M.C.; Pope, C.A. Ill; Thurston, G.;
Calle, E.E.; Thun, M.J. (2009). Extended follow-up and spatial analysis of the American Cancer Society
study linking paniculate air pollution and mortality. HEI Research Report, 140, Health Effects Institute,
Boston, MA.
Laden, F., J. Schwartz, F.E. Speizer, and D.W. Dockery. (2006). Reduction in Fine Paniculate Air Pollution and
Mortality. American Journal of Respiratory and Critical Care Medicine 173:667-672
Le Terte, A.; Schwartz, J.; Touloumi, G. (2005). Empirical bayes and adjusted estimates approach to estimating the
relation of mortality to exposure of PM10. Risk Anal. 25(3):711-718.
Medina-Ramon, M.; Zanobetti, A.; Schwartz, J. (2006). The effect of ozone and PM10 on hospital admissions for
pneumonia and chronic obstructive pulmonary disease: a national multicity study. Am J Epidemiol.
163(6):579-88.
Miller KA, Siscovick DS, Sheppard L, Shepherd K, Sullivan JH, Anderson GL, Kaufman JD. (2007). Long-term
exposure to air pollution and incidence of cardiovascular events in women. N Engl J Med, 356: 447-458.
Moolgavkar, S. (2003). Air pollution and daily deaths and hospital admissions in Los Angeles and Cook counties.
pp. 183-198. In: Health Effects Institute Special Report, Revised Analyses of Time-Series Studies of Air
Pollution and Health. Health Effects Institute, Cambridge, MA.
Ostro, B.; Broadwin, R.; Green, S.; Feng, W.Y.; Lipsett, M. (2006). Fine paniculate air pollution and mortality in
nine California counties: results from CALFINE. Environ Health Perspect. 114(l):29-33.
Pope, C.A. Ill; Burnett, R.T.; Thun, M.J. ; Calle, E.E. ; Krewski, D. ; Ito, K. ; Thurston G.D. (2002). Lung cancer,
cardiopulmonary mortality, and long-term exposure to fine paniculate air pollution. JAMA. 287(9): 1132-
1141.
Reid, C.E. ., O'Neill, M.S., Gronlund, C. Brines, S.J., Brown, D.G., Diez-Roux, A.V., Schwartz. J. (2009) Mapping
Community Determinants of Heat Vulnerability. Environmental Health Perspectives, doi:
10.1289/ehp.0900683 (available at http://dx.doi.org/) Online 11 June 2009Roberts, S. (2004). Interactions
between paniculate air pollution and temperature in air pollution mortality time series studies. Environ Res.
96(3):328-37.
Samet, J.M. (2008). Air pollution risk estimates: determinants of heterogeneity. J Toxicol Environ Health A. 71(9-
10):578-82.
Samet, 2009a. Letter from Dr. Jonathan M. Samet, Chair Clean Air Scientific Advisory Committee, to Lisa P.
Jackson, Administrator U.S. Environmental Protection Agency regarding "Consultation on EPA's
Paniculate Matter National Ambient Air Quality Standards: Scope and Methods Plan for Health Risk and
Exposure Assessment." May, 21, 2009. (letter placed in docket #EPA-HQ-OAR-2007-0492).
Samet JM; Rappold A; Graff D; Cascio WE; Berntsen JH; Huang YC; Herbst M; Bassett M; Montilla T; Hazucha
MJ; Bromberg PA; Devlin RB. (2009b). Concentrated ambient ultrafme particle exposure induces cardiac
changes in young healthy volunteers. Am J Respir Crit Care Med, 179: 1034-1042.
6-3
-------
Samet, J. (2010). Letter from Clean Air Scientific Advisory Committee to the Honorable Lisa P. Jackson,
Administrator, US EPA. CASAC Review of Quantitative Health Risk Assessment for Paniculate matter -
Second External Review Draft (February 2010). April 15, 2010. Available:
http://yosemite.epa.gov/sab/sabproduct.nsf/BC4F6E77B6385155852577070002F09F/$File/EPA-CASAC-
10-008-unsigned.pdf
Sarnat, S.E.; Coull, B.A.; Schwartz, J.; Gold, D.R.; Suh, H.H. (2006). Factors affecting the association between
ambient concentrations and personal exposures to particles and gases. Environ Health Perspect.
114(5):649-54.
Schwartz, J. (2000). The distributed lag between air pollution and daily deaths. Epidemiology 11(3):320-326.
Schwartz J, Coull B, Laden F, Ryan L (2008). The effect of dose and timing of dose on the association between
airborne particles and survival. Environ Health Perspect, 116: 64-69.
Tolbert, P.E. (2007). Invited commentary: heterogeneity of paniculate matter health risks. Am J Epidemiol.
166(8):889-91; discussion 892-3.
Tolbert, P.E.; Klein, M; Peel, J.L.; Sarnat, S.E.; Sarnat, J.A. (2007). Multipollutant Modeling Issues in a Study of
Ambient Air Quality and Emergency Department Visits in Atlanta. J Expo Sci Environ Epidemiol. 17:829-
S35.
US Environmental Protection Agency. (2004). Air Quality Critena for Paniculate Matter. National Center for
Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency,
Research Triangle Park, NC 27711; report no. EPA/600/P-99/002aF and EPA/600/P-99/002bF. October
2004. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_cr_cd.html
U.S. Environmental Protection Agency. (2005). Review of the National Ambient Air Quality Standards for
Paniculate Matter: Policy Assessment of Scientific and Technical Information - OAQPS Staff Paper,
Office of Air Quality Planning and Standards, Research Tnangle Park, NC, EPA-452/R-05-005. Available:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm cr sp.html
U.S. Environmental Protection Agency. (2007). Draft Integrated Review Plan for the National Ambient Air Quality
Standards for Paniculate Matter. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA
452/P-08-006.
U.S. Environmental Protection Agency. (2008a). Integrated Review Plan for the National Ambient Air Quality
Standards for Paniculate Matter. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA
452/R-08/004.
U.S. Environmental Protection Agency. (2008b). Integrated Science Assessment for Paniculate Matter: First
External Review Draft. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA/600/R-
08/139 and 139A. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_2007_isa.html.
U.S. Environmental Protection Agency. (2009a). Integrated Science Assessment for Paniculate Matter: Second
External Review Draft. U.S. Environmental Protection Agency, Research Triangle Park, NC, EPA 600/R-
08/139B. Available: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm 2007 isa.html.
U.S. Environmental Protection Agency. (2009b). Paniculate Matter National Ambient Air Quality Standards: Scope
and Methods Plan for Health Risk and Exposure Assessment. U.S. Environmental Protection Agency,
Research Triangle Park, NC, EPA-452/P-09-002.
U.S. Environmental Protection Agency. (2009c). Paniculate Matter Urban-Focused Visibility Assessment - External
Review Draft. Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency,
6-4
-------
Research Triangle Park, NC. EPA-452/P-09/005. September 2009. Available:
http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_index.html.
U.S. Environmental Protection Agency. (2009d). Integrated Science Assessment for Paniculate Matter: Final. U.S.
Environmental Protection Agency, Research Triangle Park, NC, EPA/600/R-08/139F. Available
http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=216546 .
US EPA (2009e). Risk Assessment to Support the Review of the PM Primary National Ambient Air Quality
Standards - External Review Draft. Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, Research Triangle Park, NC. EPA-452/P-09-006. September 2009. Available:
http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_2007_risk.html.
US EPA (2010a). Paniculate Matter Urban-Focused Visibility Assessment - Second External Review Draft. Office
of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park,
NC. EPA-452/P-10-002. January 2010. Available:
http://www.epa.gov/ttn/naaqs/standards/pm/sjm 2007 risk.html.
US EPA (2010b). Quantitative Health Risk Assessment for Paniculate Matter - Second External Review Draft
Report. Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC. EPA-452/P-10-001. February 2010. Available:.
http://www.epa.gov/ttn/naaqs/standards/pm/data/20100209RA2ndExternalReviewDraft.pdf
World Health Organization. (2008). Part 1: Guidance Document on Characterizing and Communicating Uncertainty
in Exposure Assessment, Harmonization Project Document No. 6. Published under joint sponsorship of the
World Health Organization, the International Labour Organization and the United Nations Environment
Programme. WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland
(tel.: +41 22 791 2476).
Zanobetti (2009). Shrunken estimates forPM25 Personal Communication, June 1, 2009. (e-mail with attachments
placed in docket #EPA-HQ-OAR-2007-0492).
Zanobetti, A.; Schwartz, J. (2009). The effect of fine and coarse paniculate air pollution on mortality: A National
Analysis. Environ Health Perspect. 117(6): 898-903.
Zeka, A.; Zanobetti, A.; Schwartz, J. (2005). Short term effects of paniculate matter on cause specific mortality:
effects of lags and modification by city characteristics. Occup Environ Med. 62(10):718-25.
Zeka, A.; Zanobetti, A.; Schwartz, J. (2006). Individual-level modifiers of the effects of paniculate matter on daily
mortality. American Journal of Epidemiology. 163(9): 849-859.
6-5
-------
APPENDIX A: AIR QUALITY ASSESSMENT (SUMMARY OF
INDIVDIUAL AND COMPOSITE MONITOR DATA BY URBAN
STUDY AREA)
A-l
-------
Appendix A. Air Quality Assessment
This Appendix describes the PM data for the 15 urban study areas evaluated in the risk
assessment, including summaries of PM2.5 monitoring data associated with each study area as
well as the composite monitor estimates generated for each study area based on that monitoring
data (see section 3.2 for additional detail regarding selection of monitors and derivation of
composite monitor values).
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
An nu al
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent He
(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
An n ual
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 uarteriy 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 for St Louis - 1
Composite Monitor for 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: ADDITIONAL INFORMATION SUPPORTING AIR
QUALITY CHARACTERIZATION
B-l
-------
Appendix B. Additional Information Supporting Air Quality Characterization
This appendix provides information supporting air quality characterization completed as
part of the risk assessment including both the characterization of recent conditions (as addressed
section 3.2.1) and the simulation of air quality to just meet current and alternative standards (as
addressed in section 3.2.3). Specifically, section Bl considers an alternative approach for
interpolating missing data as part of constructing distributions of 24-hour PM2.5 estimates at
composite monitors, section B2 provides additional detail on the hybrid rollback method (one of
the three rollback methods used in the risk assessment) and section B3 provides example
calculations of the three rollback methods applied to one of the urban study areas (Detroit).
Bl. SENSITIVITY ANALYSIS ADDRESSING THE INTERPOLATION OF MISSING
DATA COMPLETED IN DEVELOPING COMPOSITE MONITOR 24-HOUR
PM2.5 DISTRIBUTIONS
As noted in section 3.2.1, there are a variety of possible approaches for interpolating
missing 24-hour monitoring data as part of generating composite monitor 24-hour PM2.5
distributions for the study areas. For the risk assessment, we have used an approach that relied
on other measurements at the specific monitor where the interpolation was being conducted (i.e.,
the nearest measurements before and after the point of needed interpolation - see section 3.2.1
for additional details on this approach). However, as noted by CASAC, there are other
interpolation approaches available, some of which make use of monitoring trend data across the
entire set of PM2 5 monitors associated with a study area, rather than relying on data only from
the monitor undergoing the interpolation. In addition, these alternative interpolation methods can
address another limitation of the method used in the risk assessment - the restriction of
interpolation to periods shorter than 8 days which excludes interpolation for 1 in 6 day monitors
missing data.
The availability of alternative interpolation methods highlights the potential uncertainty
associated with this component of the risk assessment. To further examine this source of
uncertainty, we have completed a sensitivity analysis based on the application of an alternative
interpolation approach based on centering. The centering approach uses variance-trend data from
all of the monitors in a study area (with emphasis on locations with measurements on the day
being interpolated) as the basis for interpolating missing daily measurements. This sensitivity
analysis has been implemented for Birmingham (specifically the Birmingham 1 grouping of
counties and associated monitors). The alternate approach used the following steps for each of
the simulation years 2005, 2006, and 2007:
B-2
-------
1) The annual mean of the actual data from each individual site was determined, using a
seasonally weighted approach that is also the basis for the calculation of official design
values for the current annual NAAQS. All samples within Jan-March, April-June, July-
September, and October-December were averaged. Then these four quarterly means were
averaged to get the annual mean for the calendar year.
2) The annual averages for the individual sites were averaged, with equal weight, to give the
composite monitor annual average.
3) The annual average for a site was subtracted from each daily value for the site. The residuals
represent the deviation of the site concentration from the site annual average on a given day.
4) For each day, all available site residuals were averaged across sites with equal weight. For
most days, the only available residuals were from the two sites that sampled every day. On
every third day, up to 10 residuals were available. There was only a single day in 2005 for
which no site reported a concentration so no average of residuals could be calculated.
5) On each day, the concentration of the composite monitor was taken to be the sum of the
result from the second step and the result from the fourth step. No composite monitor
concentration was calculated for the single day in 2005 mentioned in step 4.
It was observed that when the weighted annual mean of the resulting estimates of daily
24-hour concentrations for the composite monitor from step 5 were compared to the result of
step 2, slight differences exist. The absolute value of the differences was less than 0.1 |ig/m3 in
2005 and 2006 and less than 0.2 |ig/m3 in 2007.
In comparing same-day estimates of 24-hour concentrations for the composite monitor as
estimated by the method used in the core risk estimates versus the alternative method described
here, differences were larger (as would be expected when comparing 24-hour estimates versus
annual-average estimates) but occurred in both directions. Figure B-l is a scatter plot of the two
sets of daily estimates.
-------
Figure B-l. Comparison of Composite Monitor 24-hour PM2.5 Distributions (2007)
Generated Using the Two Interpolation Methods
Comparison of Two Approaches for Estimating
Birmingham 1 Composite Monitor
10 20 30 40 50
Estimate from Main Analysis Method
60
To investigate how the differences in same-day concentration estimates illustrated in
Figure B-l would affect estimates of risk (specifically short-term exposure-related risk)
aggregated across individual years and across all three years, we used the 24-hour PM2.5
concentration above PRB as a surrogate for risk, since the incidences of (short-term exposure-
related) health endpoints related to 24-hour concentrations are nearly proportional to this metric.
We averaged this across all days in a year, and then across the three years. Averaging rather than
summation was used because the completely missing day in 2005 in the alternate approach
prevents a valid comparison of the sum across all available days. Table B-l shows the results of
this comparison.
B-4
-------
Table B-l. Comparison of Surrogate for Short-Term Exposure-Related Risk (see text)
Generated Using the Two Interpolation Methods
Simulation Year
2005
2006
2007
2005-2007
Interpolation Method
Main method used in RA
14.33
13.61
14.00
13.98
Alternate (centering-based)
method
14.27
13.58
13.80
13.89
The results presented in Table B-l suggest that, while conceptually the two interpolation
methods considered in the sensitivity analysis differ significantly, the impact of switching
between these two methods on short-term exposure-related risk is negligible. While these
findings need to be considered in the context of the sensitivity analysis as conducted (i.e., based
on considering a single alternative interpolation method as applied to one of the 15 urban study
areas), they do reduce concerns that this source of uncertainty significantly impacts short-term
exposure-related risk.
B.2 ADDITIONAL DETAIL ON THE HYBRID ROLLBACK APPROACH
This section provides additional detail on one of the three methods (the hybrid method)
used to simulate ambient PM2.5 levels under both current and alternative standard levels. For
additional detail on the other two methods (proportional and locally focused rollback) used in the
risk assessment to simulate standard levels as well as an overview of how the rollback methods
are applied in the context of assessing risk, see section 3.2.3.
The hybrid approach reflects the combined effects of both local and regional reduction
strategies. As such, in addition to utilizing a traditional proportional rollback to represent the
regional PM reductions, the hybrid approach also includes a distance-weighted rollback was
conducted on a subset of the 15 study areas which contain source-oriented monitors measuring
concentrations higher than those observed at other sites within a particular area.1
Unique sites with high design values exceeding the NAAQS were further investigated to
determine if they were in close proximity to a large source of PM2.5 (Figure B-2). The presence
of possible source-oriented sites in each area was visually determined using satellite photographs
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 approach described here, was considered as part of the
sensitivity analysis along with the locally focused rollback approach described in section 3.2.3.3.
B-5
-------
provided by Google Earth. Areas where source-oriented adjustments were made include Detroit
MI, Pittsburgh PA, St. Louis MO-IL, Baltimore MD, New York NY, Los Angeles CA and
Birmingham AL.
Detroit, Ml (261630033)
' ^^^ANJBK^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-2. 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-
3).
B-6
-------
Drtnt. Ml
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-3. 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-2. For
Detroit, adjustments were based on the difference between the two sites' annual design values.
u « is is
B-7
-------
Table B-2. Comparison of the original and adjusted design values for Detroit, MI
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.
Air quality datasets adjusted using the hybrid rollback approach for the subset of urban
study areas where this approach was used are available in the docket (Docket ID#: EPA-HQ-
OAR-2007-0492) and have been posted at: http://www.epa.gov/ttn/analysis/pm.htm.
-------
B.3 EXAMPLE CALCULATION OF THREE ROLLBACK METHODS AS APPLIED
TO THE DETROIT URBAN STUDY AREA
This section provides a sample calculation of the three rollback methods (hybrid,
proportional, and locally focused) as applied to simulating attainment of the current suite of
standards for Detroit. This sample calculation is intended to illustrate how each rollback method
is applied, including equations (and associated intermediate calculations) as appropriate.
Several details regarding the calculation need to be clarified to ensure that the context for
the sample calculations is well understood before they are reviewed. The current 24-hour
standard level is controlling for this study area and consequently for the sample calculations,
which means that simulated attainment of the 24-hour standard will determine the degree of
reduction in ambient PM2.5 levels needed to simulate attainment of this suite of standards. In
presenting composite monitor values generated using each rollback method, we have focused on
annual-averages rather than the 24-hour PM2.5 distributions. This decision reflects the fact that
composite monitor annual-averages have been shown to drive both short-term and long-term
exposure related risk estimates and consequently are a better metric to consider in comparing and
contrasting the rollback methods than are 24-hour PM2.5 distributions or percentiles summarizing
those distributions.
The calculations using the three rollback methods are presented in Table B-3, which
shows how values at individual monitors are adjusted in applying rollback methods, when
appropriate, together with the composite monitor values that result.2 The table is organized by
rollback method, with the values related to the proportional method presented first followed by
the hybrid and the locally focused method. The first block of columns track monitor-specific
values across the three rollback methods with the last column presenting composite monitor-
related values.
2 As discussed in sections 3.2.3.1 through 3.2.3.3, the rollback methods differ as to whether they involve
direct adjustment at individual monitors (hybrid and locally focused methods) or involve adjustment only at the
composite monitors (proportional).
B-9
-------
Table B-3. Application of Three Rollback Methods (proportional, hybrid and locally
focused) in Simulating Composite Monitor Annual-Average PMi.s Levels for the
Current Suite of Standards (Detroit, 2007)
Row
ID
PI
P2
P3
P4
P5
P6
P7
P8
HI
H2
H3
H4
H5
H6
H7
H8
LF1
LF2
LF3
LF4
LF5
LF6
LF7
LF8
Monitor ID:
261630001
261630015
2616300161
2616300191
261630025J
2616300331
261630036J
261630038J
K>
ON
H^
ON
U)
O
0
L*J
^0
Quarterly
and Annual
Averages at
Composite
Monitor
Recent (2007) Air Quality
Proportional Rollback Method
Quarterly and Annual Averages
2007Q1 Avg.
2007 Q2 Avg.
2007 Q3 Avg.
2007 Q4 Avg.
2007 Ann. Avg.
12.92
10.28
14.00
14.08
12.82
15.15
13.06
15.12
14.82
14.54
13.98
12.12
14.74
14.61
13.86
13.20
11.16
14.36
13.31
13.01
12.23
10.59
13.76
14.42
12.75
18.84
15.20
16.02
17.49
16.89
13.75
11.96
14.60
13.47
13.45
13.63
12.85
15.35
14.23
14.01
13.83
12.98
14.65
13.86
13.83
Composite monitor annual average after proportional rollback
14.17
12.24
14.73
14.48
13.91
11.43
Hybrid Rollback Method
Quarterly and Annual Averages After Initial Adjustment of Recent Air Quality:
2007 Ql Avg.
2007 Q2 Avg.
2007 Q3 Avg.
2007 Q4 Avg.
2007 Ann. Avg.
12.82
10.18
13.90
13.98
12.72
14.87
12.81
14.84
14.56
14.27
13.86
12.00
14.61
14.49
13.74
13.10
11.06
14.26
13.21
12.91
12.13
10.49
13.66
14.32
12.65
16.93
13.64
14.39
15.70
15.16
13.65
11.86
14.50
13.37
13.35
13.44
12.67
15.15
14.04
13.82
13.71
12.85
14.52
13.74
13.70
Quarterly and Annual Averages After Proportional Rollback
13.83
11.95
14.42
14.16
13.59
11.69
Locally-Focused Rollback Method
Monito r-spe cific
Monito r-spe cific
2007 Ql Avg.
2007 Q2 Avg.
2007 Q3 Avg.
2007 Q4 Avg.
2007 Ann. Avg.
36
2.8%
12.57
10.02
13.63
13.71
12.48
40
12.8%
13.32
11.50
13.30
13.04
12.79
41
14.9%
12.02
10.44
12.67
12.56
11.92
40
12.8%
11.63
9.85
12.63
11.72
11.46
34
0.0%
12.23
10.59
13.76
14.42
12.75
43
19.0%
15.43
12.47
13.15
14.33
13.84
36
2.8%
13.38
11.64
14.21
13.11
13.09
40
12.8%
12.00
11.32
13.50
12.52
12.33
37
5.5%
13.12
12.31
13.89
13.14
13.11
12.86
11.13
13.41
13.17
12.64
A step-wise explanation of the values presented in Table B-3 is presented below. Row
identifiers (P# for proportional, H# for hybrid and LF# for locally focused) are included in the
table to facilitate this discussion. In presenting the step-wise explanations we provide only
sufficient explanation of the conceptual approach underpinning each rollback method to insure
that Table B-3 and the step-wise calculations can be understood. The reader is referred back to
relevant sections of the document for a more complete discussion of the conceptual basis for
each rollback method.
Proportional rollback
B-10
-------
The proportional rollback method is applied at the composite monitor levels, which are
themselves based on quarterly-average estimates at individual monitors (see section 3.2.3.1).
Therefore, we begin the step-wise calculation for the proportional method by presenting the
approach used to calculate the composite monitor quarterly- and annual-averages (set of "A"
bullets below). We then present how design values are calculated (set of "B" bullets below). We
then show how the design value is used to implement the proportional rollback at the composite
monitor (set of "C" bullets below).
A. Calculating the composite monitor annual average based on recent air quality
1) Calculate quarterly averages for those monitors with at least 11 observations
in the quarter (middle columns in rows P3-P6).
2) Calculate the quarterly average at the composite monitor as the average of
these monitor-specific quarterly averages (last column in rows P3-P6).
3) Calculate the annual average at the composite monitor as the average of these
4 quarterly averages (last column in row P7). This is the composite monitor
annual-average under recent conditions, prior to proportional rollback.
B. Calculating design values
1) At each monitor, calculate the annual average PM2.5 concentration for each of
the 3 years (2005, 2006, 2007). Average these.
2) The maximum of these monitor-specific 3-year averages of annual averages is
the annual design value. In Detroit this was 17.2 ug/m3, at monitor
261630033 (see Table A-5 in Appendix A).
3) At each monitor, calculate the 98th percentile 24-hour PM2.5 concentration for
each of the 3 years (2005, 2006, 2007). Average these.
4) The maximum of these monitor-specific 3-year averages of 98th percentile
concentrations is the 24-hour design value. In Detroit, this was 43 ug/m3 at
monitor 261630033 (see Table A-5 in Appendix A).
C. Calculating the composite monitor annual average when the 15/3 5 standard is just
met, using the proportional rollback
1) Calculate the percent rollback needed to meet both the daily and the annual
standard:
Inputs for the proportional rollback in Detroit:
o Annual Design Value = 17.2
o Daily Design Value = 43
o Annual Average PRB = 0.86
Percent rollback to just meet the annual standard of 15 is calculated as:
1 - (annual std. - PRB)/(annual design value - PRB) =
1 - (15 - 0.86)7(17.2 - 0.86) = 13.5%
B-ll
-------
Percent rollback to just meet the daily standard of 35 is calculated as:
1 - (daily std. - PRB)/(daily design value - PRB) =
1 - (35 - 0.86)7(43 - 0.86) = 19.0%
Percent rollback to just meet both standards = maximum of 13.5% and
19% = 19% (this determines which standard is controlling)
2) Rolled back annual average at composite monitor = PRB + (annual avg at
composite monitor using recent air quality - PRB)*(1 - 0.19) (see row P8)
Hybrid rollback
As discussed in section 3.2.3.2, the hybrid approach involves a two-phase adjustment
process with the first phase involving targeted reduction of source-oriented monitors to bring
their levels down to that of near-by non-source-oriented monitors and a second proportional
rollback phase to reach simulated attainment of the standard for the study area. In presenting the
step-wise description of the hybrid rollback approach (and the sample calculation in Table B-3),
we illustrate the first phase of adjustment by presenting quarterly values at individual monitors
after the first phase of adjustment has occurred (rows H3-H6). The resulting composite monitor
annual average is shown in the last column of row H7. Then, the second phase of proportional
reduction results in the final adjusted composite monitor annual-average (last column in row
H8). The step-wise procedure is presented in the set of "D" bullets below.
D. Calculating the composite monitor annual average when the 15/35 standard is just
met, using the hybrid rollback
1) Adjust daily concentrations at individual monitors on a first step (see rows
H3-H6);
2) Calculate an annual average at the composite monitor from those adjusted
values, using the same procedure used in Section A above (see last column in
row H7).
3) Roll back this intermediate annual average composite monitor value using the
same procedure used for the proportional rollback, but with design values
based on the adjusted data (see last column row H8).
Inputs for the hybrid rollback in Detroit:
o Annual Design Value =15.4
o Daily Design Value = 41
o Annual Average PRB = 0.86
Percent rollback to just meet the annual standard of 15 = 2.8%; percent
rollback to just meet the daily standard of 35 = 14.9%
Percent rollback to just meet both standards = 14.9%
B-12
-------
4) Rolled back annual average at composite monitor = PRB + (intermediate
annual avg at composite monitor - PRB)*(1 - 0.149) (see last column row
H8).
Locally focused rollback
The locally focused rollback method involves targeted rollback of those monitors
exceeding the 24-hour standard under consideration. As discussed in section 2.3.2.3, this
rollback approach is only applied for study areas where the 24-hour standard is controlling and
therefore, typically, will result in simulated attainment of the annual-standard along with the 24-
hour standard. The illustration of this rollback approach presented in Table B-3 presents the
adjusted quarterly values by monitor (rows LF4-LF7) together with the percent reductions
reflected in those adjusted monitor-specific levels (row LF3). The composite monitor annual-
average that results from this targeted monitor-level reduction is presented in the last column of
row LF8. It is interesting to point out that in the case of Detroit, all of the monitors had some
degree of targeted reduction to achieve simulated attainment of the 24-hour standard, although
the spread in magnitude of percent reduction was substantial (see values in row LF3). The step-
wise procedure used in implementing the hybrid approach is presented in the set of "E" bullets
below.
E. Calculating the composite monitor annual average when the 15/35 standard is just
met, using the locally focused rollback
1) Identify monitor-specific design values (for daily standard this procedure
was applied only when daily standard is controlling) (row LF2)
2) Calculate monitor-specific percent rollbacks (row LF3)
3) Roll back each quarterly average at each monitor, using the monitor-specific
percents for rollback (rows LF4-LF7):
Rolled back quarterly average at monitor = PRB + (quarterly average at
monitor - PRB)*(1 - monitor-specific percent rollback)
4) Calculate the rolled back quarterly average at the composite monitor as the
average of these monitor-specific rolled back quarterly averages (last column
in rows LF4-LF7).
Calculate the rolled back annual average at the composite monitor as the average of these
4 rolled back quarterly averages (last column in row LF8).
B-13
-------
APPENDIX C: EPI STUDY SPECIFIC INFORMATION ON PM25
C-l
-------
Appendix C. Epidemiology Study-Specific Information for PM2.s Risk Assessment
This Appendix provides detailed summary information for the epidemiological studies
used to obtain the concentration-response (C-R) functions used in the risk assessment. For
additional details on selection of epidemiological studies and specification of the C-R functions,
see section 3.3.3.
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
0.0021 1
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
-------
Appendix D. Supplement to the Representativeness Analysis of the 15 Urban Study Areas
(additional graphical comparisons of distributions for key contributors to PMi.s risk )
Following the analysis discussed in Section 4.4.1, this appendix provides graphical
comparisons of the empirical distributions of components of the risk function, and additional
variables that have been identified as potentially influencing the risk associated with PM
exposures.
In each graph, the orange line represents the empirical cumulative distribution function
(CDF) for the complete set of data available for the variable. In some cases, this may encompass
all counties in the U.S., while in others it may be based on a subset of the U.S., usually for large
urban areas. The green line in each graph represents the empirical cumulative distribution
function for the variable based only on the data available for the set of urban case study
locations. The black squares at the bottom of each graph represents the specific value of the
variable for one of the case study locations, with the line showing where that value intersects the
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%
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 15 to U.S. Distribution of % Under
15
(3141 U.S. Counties)
1
0
100%
90%
80% -
70% -
60% -
10
15
20 25
% of Population Under Age 15, 2005
30
35
All Counties CDF ^ Case Study Counties Case Study Counties
D-4
-------
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% -
1 0% -
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
100% -
90% -
80% -
i/)
0)
1 70% -
0 60% -
| 50%-
§ 40%
g
£ 30% -
^ 20% -
10% -
0% -
(
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% -
fl)
1 70% -
0 60% -
1 50%-
§ 40%
g 30% -
^ 20% -
10% -
n%
0
98th %ile PM2.5
(617 U.S. Counties with PM2.5 Monitors)
J
./
^
10 20 30
-
X
X
L "1
»
40 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
| 70% -
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
0
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)
lUUVo T
90% -.
80%
£ 70%
'+-1
1 60%
o
o
50%
CO
2 40%
0
£ 30%
20%
1 0%
n%
U /O T
50
4
A
*S
LI
I 1
II 1
J II 1
,'
1 1 1 1
1 1
1 1 1
LI
1 1
y
i i
X""
i i
^^^
Urban case study areas are
all below the 90th percentile
of county cardiovascular
mortality
i
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)
100%
c
(0
.a
CO
08
N
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 0.002
Bayesian Shrunken PM Cardiovascular Mortality Risk Coefficient (p)
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%
1 0%
0%
(
Comparison of Urban Case Study Area Unemployment Rate to U.S. Distribution of
Unemployment Rates
(3141 U.S. Counties)
)
^^
2
Jj
4
m-t
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% -,
90% -
80% -
£ 70% "
1 60% -
ri 50%
2 40% -
0
5S 30% -
20% -
1 0% -
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
o
O
CO
zi
M
O
100%
90% -
80% -
70% -
60% -
50% -
40%
30% -
20% -
10% -
Comparison of Urban Case Study Area Per Capita Personal Income to U.S.
Distribution of Per Capita Personal Income
(3141 U.S. Counties)
Urban case study areas are
all above the 25th percentile
of county per capita income
i,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
D-20
-------
Figure D-19. Comparison of Distributions for Selected Variables Expected to Influence the
Relative Risk from PMi.s: Air Conditioning Prevalence
w
cc
c
CD
.Q
=>
o
5S
(
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)
/
-~
~
j
^
'
r,
J
0 02 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)
/
-~
~
j
^
/
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%
1 0%
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)
1 UU /O
90%
80%
< 70%
CO
5 60%
CO
£ 50%
a:
2 40%
o
^ 30%
20%
1 0%
<:
nq/, H 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 ^-^^^^^ ^^ ^^ ^ | ^ |
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% -,
90% -
80% -
< 70% -
CO
5 60% -
CO
L£ 50%
or
2 40% -
o
^ 30% -
20% -
1 0% -
0% -
f
(183 U.S. MSA)
^
>. 4
/
/
^
H 1
6
/
f
i i
8
ii
/
/
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)
100%
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%
1 0%
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%
1 0%
^^^^^^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.
Distribution of 4th Highest Daily 8-Hour Ozone
(725 U.S. Counties with Ozone Monitors)
0.04
0.06 0.08
4th Highest Daily 8-Hour Maximum
0.1
0.12
All Counties CDF ^Case Study Counties CDF Case Study Counties
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)
1 0% -]
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%
1 0%
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
iH 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
-------
Appendix E. Risk Estimates (core analysis)
This Appendix provides detailed risk estimates generated for the core analysis for the 15
urban study areas. The tables cover all of the air quality scenarios modeled, including recent
conditions, the current standard, and alternative standard levels. For additional detail on the types
of risk metrics (and figures summarizing key metrics) presented in this Appendix, see section
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 on Krewski et al. (2009), Using Ambient PM2 5 from 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 - 51 1 )
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
(66-138)
826
(536 -1111)
123
(80-166)
15/353
575
(373 - 774)
548
(355 - 738)
297
(192-400)
379
(245-511)
580
(376-782)
89
(57- 120)
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
(245-511)
573
(371 - 772)
89
(57 - 1 20)
491
(318-663)
1045
(676-1413)
1477
(956-1992)
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
(245-511)
502
(325 - 677)
89
(57- 120)
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
(57-120)
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 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
668
(434 - 899)
483
(313-651)
399
(259 - 536)
284
(183-383)
576
(373 - 776)
265
(172-356)
589
(381 - 794)
2054
(1 332 - 2767)
1548
(1 002 - 2087)
471
(305 - 636)
512
(331 -691)
468
(303 - 631 )
84
(54-113)
618
(401 - 834)
83
(54-112)
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 - 1 1 66)
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 - 1 1 66)
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 - 1 1 66)
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 - 1 1 64)
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
(498-1043)
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 - 1 63)
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
(313-650)
418
(271 - 563)
317
(205 - 428)
607
(393-818)
279
(181 -375)
615
(398 - 829)
2134
(1 384 - 2874)
1812
(1174-2443)
466
(302 - 629)
433
(280 - 586)
527
(342-710)
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 - 1 37)
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
(1 66 - 347)
317
(205 - 428)
418
(270 - 564)
101
(65 - 1 37)
494
(320 - 667)
911
(588 - 1 232)
1316
(852-1777)
405
(262 - 546)
433
(280 - 586)
339
(219-457)
38
(24-51)
509
(330 - 687)
50
(32 - 68)
13/35
442
(286 - 596)
347
(224 - 468)
222
(1 44 - 300)
317
(205 - 428)
358
(232 - 484)
101
(65 - 1 37)
427
(276 - 577)
911
(588 - 1 232)
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 - 1 37)
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
(388-813)
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
(1 80 - 375)
192
(1 24 - 260)
36
(23 - 49)
359
(232 - 486)
289
(187-392)
600
(387 - 81 1 )
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%
(11% -11%)
8%
(8% - 9%)
12%
(12% -12%)
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%
(1 1 % - 1 1 %)
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.
-------
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 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
-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%
(11% -11%)
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 PM2 5 Concentrations, Based on Adjusting 2007 PM2 s 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
PM 2.5 Concentrations ina 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%)
-1 0%
(-10% --10%)
-44%
(-43% - -44%)
0%
(0% - 0%)
-43%
(-43% - -44%)
-176%
(-175% --178%)
-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%
(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%)
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%
(46% -46%)
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%
(1 00% - 1 00%)
49%
(49% - 50%)
93%
(93% - 93%)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 in Krewski etal., 2009).
Numbers 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/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
(653-1380)
325
(208 - 439)
780
(498-1058)
2986
(1910-4042)
2560
(1636-3468)
668
(427 - 905)
620
(394 - 843)
759
(485-1026)
132
(84-179)
1056
(676-1429)
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
(317-674)
37
(24-51)
897
(573 - 1 21 5)
103
(66-141)
14/35
657
(420 - 891 )
643
(410-871)
336
(214-457)
486
(310-659)
734
(468 - 996)
114
(72 - 1 55)
630
(401 - 856)
1342
(854 - 1 827)
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 - 1 55)
546
(348 - 743)
1342
(854 - 1 827)
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 - 1 607)
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 - 1 261 )
299
(190-408)
338
(214-460)
244
(155-332)
0
(0-0)
492
(31 3 - 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 PM25 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
(68-146)
792
(506-1075)
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
(704-1509)
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 - 71 3)
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
(848-1813)
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
(77-164)
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 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
821
(525-1112)
618
(394 - 838)
535
(342 - 724)
407
(259 - 553)
778
(496 - 1 054)
357
(228 - 482)
788
(502 - 1 069)
2732
(1746-3702)
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
(82-177)
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 - 1 77)
634
(404 - 861 )
1170
(744-1593)
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 - 1 77)
548
(349 - 745)
1170
(744 - 1 593)
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 - 1 384)
1359
(864 - 1 848)
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
(783-1676)
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%)
-1 85%
(-183% --187%)
-9%
(-9% - -9%)
-1 22%
(-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%
(11% -11%)
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%)
-1 81 %
(-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%
(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%)
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%)
-1 75%
(-173% --177%)
-9%
(-9% - -1 0%)
-1 34%
(-132% --135%)
-37%
(-37% - -38%)
-15%
(-15% --15%)
0%
(0% - 0%)
-55%
(-55% - -56%)
-217%
(-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%
(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 %)
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%
(11% -11%)
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
(286-414)
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
(41 8 - 599)
68
(56 - 81 )
340
(278 - 400)
984
(802-1163)
1959
(1603-2307)
293
(240 - 345)
351
(286 - 41 4)
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
(41 3 - 592)
68
(56 - 81 )
302
(247 - 356)
984
(802-1163)
1959
(1603-2307)
293
(240 - 345)
351
(286 - 41 4)
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 - 41 4)
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 - 1 026)
1610
(1315-1900)
226
(185-267)
316
(258 - 374)
241
(197-285)
12
(9-14)
379
(310-447)
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
(795-1156)
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
(1 200 - 1 736)
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
(1 93 - 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
(173-252)
707
(575 - 837)
1163
(947-1375)
201
(164-238)
335
(273 - 396)
177
(144-209)
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
(169-239)
375
(307-441)
1953
(1 604 - 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
(95-136)
195
(1 59 - 230)
372
(304 - 439)
77
(63 - 92)
304
(249 - 358)
860
(701 -1018)
1755
(1435-2070)
261
(21 4 - 308)
317
(258 - 374)
256
(209 - 302)
15
(12-18)
402
(329 - 474)
38
(31 - 46)
13/35
173
(142-204)
236
(1 93 - 279)
101
(82 - 1 1 9)
195
(1 59 - 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
(61 0 - 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
(140-203)
174
(142-206)
28
(23 - 33)
223
(182-264)
278
(225 - 330)
815
(663 - 966)
130
(106-154)
160
(130-189)
120
(98 - 1 42)
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%)
1 5.9%
(13.1% -18.6%)
11.1%
(9.1% -13.1%)
1 6.4%
(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%)
13.3%
(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% - 1 3.2%)
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%)
1 3.4%
(11% -15.7%)
10.9%
(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% -10.8%)
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%)
1 2.7%
(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%)
1 2.6%
(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%)
1 3.3%
(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%)
-1 74%
(-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%
(11% -11%)
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%
(11% -11%)
25%
(25% - 25%)
0%
(0% - 0%)
34%
(34% - 35%)
12%
(12% -12%)
18%
(18% -18%)
23%
(22% - 23%)
10%
(1 0% - 1 0%)
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%
(430/0 _ 44o/0)
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%)
-1 70%
(-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%
(1 2% - 1 2%)
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%
(1 0% - 1 0%)
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%)
-1 64%
(-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%
(1 1 % - 1 2%)
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 - 1 1 0)
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
(313-448)
471
(384 - 557)
439
(360-516)
42
(34 - 50)
610
(500-716)
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 - 1 229)
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 - 1 229)
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
(31 0 - 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 - 1 20)
195
(158-230)
316
(257 - 375)
92
(75 - 1 08)
270
(220 - 320)
901
(731 - 1 068)
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
(314-456)
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 PM25 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 - 1 296)
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 - 1 296)
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 - 1 28)
218
(178-257)
341
(278 - 404)
98
(80-116)
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
(282-410)
240
(195-284)
9
(7-10)
431
(351 - 509)
27
(21 - 32)
12/25
185
(151 -218)
197
(160-234)
90
(73 - 1 07)
218
(178-257)
222
(180-264)
36
(29 - 43)
284
(231 - 335)
355
(287 - 423)
1040
(843 - 1 234)
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 PM25 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% - 1 3%)
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% -13%)
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% -11%)
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% -13%)
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%)
-1 70%
(-166% --174%)
-9%
(-9% - -9%)
-113%
(-111% --11 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%
(11% -11%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(11% -11%)
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%)
-1 65%
(-162% --169%)
-9%
(-9% - -9%)
-1 28%
(-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 %)
-1 59%
(-156% --163%)
-9%
(-9% - -9%)
-1 24%
(-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%
(11% -11%)
0%
(0% - 0%)
1%
(1%-1%)
0%
(0% - 0%)
11%
(11% -11%)
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%
(11% -11%)
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
PM25 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM25from 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
(1 84 - 295)
499
(380-616)
2357
(1800-2902)
2205
(1683-2717)
439
(335 - 541 )
406
(309 - 503)
529
(405 - 652)
76
(58 - 94)
758
(580 - 933)
110
(84-136)
15/353
455
(347 - 561 )
467
(356 - 575)
258
(196-318)
321
(244 - 396)
547
(41 7 - 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
(1 74 - 282)
321
(244 - 396)
540
(412-667)
85
(65-105)
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 - 1 05)
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
(21 9 - 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
(930-1515)
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
(21 7 - 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 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 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
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
(320-518)
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 - 41 6)
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
(68-112)
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 - 1 067)
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 - 1 36)
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
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.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
507
(387 - 625)
412
(314-508)
361
(276 - 444)
269
(205 - 333)
572
(436 - 705)
263
(201 - 323)
504
(384 - 622)
2160
(1 648 - 2663)
2003
(1 527 - 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 - 1 1 56)
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
(1 70 - 276)
269
(205 - 333)
396
(301 - 490)
97
(74-120)
407
(309 - 503)
933
(708 - 1 1 56)
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
(1 47 - 240)
269
(205 - 333)
340
(259 - 421 )
97
(74-120)
352
(268 - 436)
933
(708 - 1 1 56)
1392
(1059-1722)
304
(231 - 375)
366
(278 - 453)
286
(21 7 - 353)
28
(21 - 35)
410
(31 2 - 507)
45
(34 - 56)
12/35
302
(230 - 374)
253
(192-313)
164
(124-203)
236
(179-292)
284
(216-352)
97
(74-120)
297
(225 - 368)
811
(615-1006)
1179
(895-1459)
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
(812-1324)
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
(103-168)
236
(179-292)
183
(139-227)
35
(26 - 43)
297
(225 - 368)
298
(226 - 370)
671
(508 - 832)
168
(127-208)
183
(139-227)
141
(107-175)
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%)
-1 82%
(-180% --184%)
-9%
(-9% - -9%)
-1 20%
(-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 s 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 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%)
-1 5%
(-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%
(13% -13%)
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 PM25 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%)
-132%
(-130% --133%)
-37%
(-37% - -37%)
-15%
(-15% --15%)
0%
(0% - 0%)
-55%
(-54% - -55%)
-21 5%
(-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%
(1 1 % - 1 1 %)
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.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
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- 1286)
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.
2Numbers rounded to the nearest whole number. Numbers in parentheses are 95% confidence or credible intervals based on
''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).
statistical uncertainty surrounding the PM coefficient.
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 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
744
(586 - 898)
584
(459 - 707)
486
(382 - 587)
344
(268 - 41 7)
770
(604 - 932)
352
(277 - 424)
686
(537-831)
2945
(231 3 - 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
(419-647)
341
(267-414)
344
(268-417)
537
(420 - 652)
129
(100-157)
627
(491 -761)
1263
(985 - 1 538)
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 - 71 9)
486
(381 - 590)
301
(235 - 365)
344
(268 - 41 7)
530
(41 4 - 643)
129
(100-157)
553
(433 - 672)
1263
(985-1538)
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
(332-515)
260
(203-316)
344
(268-417)
453
(354 - 551 )
129
(1 00 - 1 57)
479
(374 - 582)
1263
(985 - 1 538)
1649
(1288-2005)
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 - 31 6)
344
(268 - 41 7)
388
(303 - 472)
87
(67-106)
479
(374 - 582)
825
(642-1007)
1231
(960- 1499)
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
(1 84 - 288)
44
(34 - 53)
404
(315-491)
380
(296 - 465)
713
(555 - 870)
244
(1 90 - 297)
340
(265-415)
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
(1 627 - 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 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
-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%)
-14%
(-140/0 --14%)
0%
(0% - 0%)
-51 %
(-50% - -52%)
-250%
(-248% - -251 %)
-17%
(-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 %)
100%
(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%)
-1 73%
(-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%
(1 00% - 1 00%)
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%)
-1 68%
(-165% --170%)
-9%
(-9% - -9%)
-129%
(-128% --131%)
-36%
(-36% - -37%)
-15%
(-15% --15%)
0%
(0% - 0%)
-54%
(-53% - -54%)
-21 3%
(-211% --21 5%)
-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
(43-178)
26
(10-41)
76
(29-122)
248
(94 - 393)
208
(79-331)
70
(26 - 1 1 1 )
58
(22 - 94)
80
(31 -127)
8
(3-13)
116
(44-184)
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
(26-112)
112
(42-181)
155
(58 - 247)
61
(23 - 98)
58
(22 - 94)
53
(20 - 84)
2
(1 -4)
99
(37 - 1 57)
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
(34-143)
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 - 1 26)
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 - 1 60)
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
(29-125)
116
(43-186)
46
(17-75)
51
(19-82)
40
(15-64)
1
(0-1)
77
(29-123)
8
(3-13)
12/25
45
(17-73)
42
(1 6 - 67)
21
(8 - 34)
44
(17-71)
43
(16-70)
3
(1-5)
46
(17-73)
42
(16-68)
76
(28-123)
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 PM25
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 - 1 1 8)
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 - 1 08)
93
(35 - 1 50)
115
(43 - 1 85)
55
(21 - 88)
62
(23 - 1 00)
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 - 1 50)
115
(43 - 1 85)
55
(21 - 88)
62
(23 - 1 00)
40
(15-64)
1
(0-2)
65
(24 - 1 04)
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 - 1 50)
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 - 1 46)
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
(1 1 - 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
(24-100)
53
(20 - 85)
71
(27 - 1 1 4)
9
(3-15)
96
(36-152)
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 - 1 58)
138
(52 - 221 )
54
(21 - 87)
53
(20 - 85)
46
(17-74)
3
(1-5)
80
(30 - 1 28)
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
(37-158)
138
(52-221)
54
(21 - 87)
53
(20 - 85)
46
(17-74)
3
(1-5)
72
(27 - 1 1 6)
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 - 1 00)
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 - 1 38)
111
(42 - 1 79)
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
(38-163)
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 - 1 02)
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% - 1 1 %)
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 -4.40/0)
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%)
-1 82%
(-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%
(11% -11%)
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%)
-1 77%
(-172% --183%)
-9%
(-9% --10%)
-1 36%
(-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%)
-1 72%
(-166% --177%)
-9%
(-9% --10%)
-1 32%
(-128% --135%)
-37%
(-36% - -38%)
-15%
(-15% --15%)
0%
(0% - 0%)
-55%
(-53% - -56%)
-215%
(-21 2% --21 9%)
-19%
(-19% --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
(45-155)
164
(71 - 253)
224
(99 - 343)
88
(39 - 1 35)
85
(37 -131)
76
(34 - 1 1 7)
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
(39-138)
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
(34-120)
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
(29-102)
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
(34-120)
113
(49-176)
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
(52-180)
39
(18-59)
107
(47- 164)
316
(140-481)
234
(103-359)
91
(40- 140)
90
(39 - 1 39)
92
(41 -140)
9
(4-14)
126
(56-193)
18
(8 - 28)
15/353
101
(45-154)
87
(38-133)
52
(23 - 80)
54
(24 - 84)
82
(36 - 1 27)
14
(6 - 22)
98
(43 - 1 50)
135
(59-210)
167
(73 - 258)
80
(35 - 1 22)
90
(39 - 1 39)
58
(25 - 89)
2
(1-3)
105
(46 - 1 62)
10
(5-16)
14/35
91
(40-138)
79
(35-121)
46
(20-71)
54
(24 - 84)
81
(35 - 1 25)
14
(6 - 22)
87
(38 - 1 33)
135
(59-210)
167
(73 - 258)
80
(35 - 1 22)
90
(39 - 1 39)
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
(33-116)
135
(59-210)
159
(69 - 245)
71
(31 -109)
90
(39-139)
54
(23 - 83)
2
(1 -3)
81
(36-126)
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
(35-125)
46
(20-71)
2
(1 -3)
68
(30-106)
10
(5-16)
13/30
80
(35-123)
66
(29 - 1 02)
40
(18-62)
54
(24 - 84)
59
(26 - 92)
10
(4-15)
75
(33-116)
89
(38-138)
119
(52 - 1 84)
60
(26 - 92)
79
(34-122)
41
(18-64)
0
(0-0)
79
(34-122)
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 - 1 07)
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
(55-189)
41
(18-62)
112
(49-171)
328
(145-499)
273
(121 -417)
91
(40-138)
77
(33 - 1 1 8)
103
(46-157)
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
(33-118)
67
(29 - 1 03)
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 - 1 33)
15
(7 - 24)
90
(40 - 1 39)
143
(62 - 222)
200
(88 - 308)
79
(35 -121)
77
(33-118)
67
(29 - 1 03)
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
(32-115)
15
(7 - 24)
78
(34-121)
143
(62 - 222)
191
(84 - 294)
70
(31 -108)
77
(33-118)
63
(28 - 97)
4
(2-7)
91
(40 - 1 40)
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 - 1 02)
124
(54 - 1 93)
162
(71 - 250)
60
(26 - 92)
68
(30-106)
55
(24 - 84)
4
(2-7)
77
(34 - 1 1 9)
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 - 1 48)
147
(64 - 227)
59
(26 - 91 )
66
(29 - 1 02)
49
(22 - 76)
2
(1-3)
88
(39-136)
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
(1 7 - 60)
38
(17-60)
31
(1 4 - 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%)
1 3.4%
(6% - 20.3%)
8.9%
(3.9% -13.6%)
1 0.4%
(4.6% -15.9%)
8.5%
(3.7% -13%)
9.4%
(4.2% -14.4%)
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% - 1 6.7%)
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%
(1 1 % - 1 1 %)
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%
(1 1 % - 1 1 %)
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%)
-1 73%
(-167% --181%)
-9%
(-9% --10%)
-1 33%
(-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%)
-1 68%
(-161% --175%)
-9%
(-9% - -9%)
-1 29%
(-125% --134%)
-36%
(-35% - -37%)
-15%
(-14% --15%)
0%
(0% - 0%)
-54%
(-52% - -55%)
-213%
(-209% --2 17%)
-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%
(1 1 % - 1 1 %)
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
(552-1323)
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 - 1 1 02)
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 - 1 1 02)
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 - 1 03)
156
(37 - 273)
135
(-24-291)
44
(6-81)
182
(37 - 326)
81
(-117-278)
761
(447-1073)
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
(411 -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 - 1 03)
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
(-171 -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
(-108-257)
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
(-19-234)
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
(-101 -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 - 1 98)
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
(-174-415)
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
(-1 1 0 - 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
(-15-1 86)
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 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% - -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%
(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% -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 fromZanobetti 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 - 1 04)
74
(-5-151)
-1
(-55 - 52)
32
(-21 - 85)
89
(-11-1 88)
20
(-1 4 - 54)
50
(-34-131)
-50
(-223 -121)
605
(353 - 853)
94
(25-163)
84
(-4-170)
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
(-32-124)
-31
(-140-76)
504
(294 -711)
87
(23-150)
84
(-4-170)
47
(-9-103)
8
(-2-18)
122
(27 - 21 5)
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
(-1 40 - 76)
504
(294 - 71 1 )
87
(23 - 1 50)
84
(-4-170)
47
(-9-103)
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
(-27-105)
-31
(-140-76)
491
(286 - 693)
81
(21 -140)
84
(-4 - 1 70)
46
(-9 -101)
8
(-2-18)
106
(24- 187)
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 - 1 72)
12
(-7-31)
13/30
27
(-28-81)
60
(-4-122)
-1
(-36 - 34)
32
(-21 - 85)
62
(-8-131)
10
(-7 - 26)
40
(-27-105)
-27
(-119-65)
431
(251 - 609)
74
(19-129)
79
(-3 - 1 59)
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
(-6-109)
8
(-6-21)
36
(-25 - 96)
-22
(-99 - 54)
358
(208 - 506)
62
(16-107)
65
(-3-132)
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 - 1 05)
65
(-4-132)
-1
(-52 - 50)
27
(-18-71)
72
(-9-152)
22
(-1 5 - 58)
48
(-33-128)
-46
(-205-112)
519
(303 - 733)
89
(23-154)
88
(-4-178)
58
(-12-126)
11
(-3 - 25)
113
(25-199)
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
(-129-70)
432
(252 -611)
82
(21 -142)
88
(-4-178)
41
(-8 - 89)
7
(-2-16)
101
(23-179)
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
(-1 29 - 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- 103)
-29
(-129-70)
421
(246 - 595)
77
(20-133)
88
(-4-178)
40
(-8 - 86)
7
(-2-16)
88
(20-155)
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
(-4-107)
-1
(-35 - 33)
27
(-18-71)
50
(-6-105)
10
(-7 - 27)
39
(-26-103)
-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
(-3-139)
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 - 1 04)
65
(-4-133)
-1
(-54 - 51 )
29
(-19-76)
74
(-9-156)
23
(-1 6 - 59)
49
(-33-130)
-47
(-209-114)
568
(332 - 802)
91
(24-157)
84
(-4-170)
61
(-12-132)
14
(-3-31)
118
(26 - 208)
14
(-7 - 34)
15/353
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 - 1 87)
11
(-6 - 27)
14/35
30
(-30 - 88)
58
(-4-119)
-1
(-39 - 37)
29
(-19-76)
60
(-7 - 1 26)
12
(-9 - 33)
43
(-29 - 1 1 3)
-30
(-1 32 - 72)
473
(276 - 668)
84
(22 - 1 45)
84
(-4-170)
43
(-9 - 93)
9
(-2 - 20)
99
(22 - 1 76)
11
(-6 - 27)
13/35
27
(-28 - 81 )
54
(-4-111)
-1
(-36 - 34)
29
(-1 9 - 76)
55
(-7-116)
12
(-9 - 33)
39
(-27-104)
-30
(-132-72)
461
(269 - 651 )
79
(20-136)
84
(-4 - 1 70)
42
(-8-91)
9
(-2 - 20)
92
(20- 162)
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
(-3-162)
40
(-8 - 87)
9
(-2 - 20)
84
(19-149)
11
(-6 - 27)
13/30
27
(-28-81)
53
(-4-108)
-1
(-36 - 34)
29
(-1 9 - 76)
51
(-6-108)
11
(-7 - 28)
39
(-27-104)
-25
(-112-61)
405
(236 - 572)
72
(19-124)
79
(-3 - 1 59)
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
(-3-133)
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.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 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%)
-12%
(-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%)
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 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
-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%
(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%
(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%)
-12%
(-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%)
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
(-1 0 - 32)
28
(1 - 55)
9
(0-17)
34
(5-61)
57
(6 - 1 08)
106
(37-174)
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
(-10-32)
28
(1 - 54)
9
(0-17)
31
(5 - 57)
57
(6-108)
106
(37 - 1 74)
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
(-1 0 - 32)
26
(1 - 50)
9
(0-17)
29
(5 - 52)
57
(6 - 1 08)
104
(37-169)
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
(5-102)
95
(34 - 1 56)
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
(-1 0 - 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
(8-158)
110
(39-179)
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 - 1 00)
91
(32-149)
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
(5-100)
91
(32 - 1 49)
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 - 1 00)
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 - 1 34)
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 - 1 07)
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 - 1 02)
100
(35-163)
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 - 1 63)
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 - 1 02)
97
(34- 159)
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 - 1 47)
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%
(19% -20%)
22%
(22% - 23%)
7%
(7% - 7%)
16%
(15% -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%)
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.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
-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%
(19% -20%)
22%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 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%
(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.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
-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%
(19% -20%)
22%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 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%
(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 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
43
(-28 - 1 1 5)
262
(192-331)
21
(-1 4 - 56)
31
(-20-81)
345
(253 - 435)
38
(0 - 75)
60
(-39-158)
418
(5 - 827)
952
(700-1204)
233
(171 -294)
108
(1 -213)
222
(163-280)
13
(0 - 25)
231
(170-293)
26
(-65 - 1 1 3)
15/353
40
(-26-105)
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 - 1 1 5)
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 - 1 70)
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
(-29-115)
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 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
44
(-29 - 1 1 7)
227
(167-287)
20
(-1 3 - 53)
26
(-1 7 - 68)
278
(204 - 351 )
40
(0 - 80)
58
(-38-154)
392
(5 - 776)
822
(604-1040)
218
(160-276)
113
(1 - 224)
190
(140-240)
12
(0 - 23)
191
(140-241)
22
(-54 - 95)
15/353
41
(-27-108)
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
(98-169)
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
(98-169)
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 - 1 1 3)
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
(-28-113)
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 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 Ambient
PM2.5
Concentrations
45
(-29 - 1 1 9)
229
(168-289)
21
(-1 4 - 54)
28
(-1 8 - 73)
288
(211 -364)
42
(1 - 83)
60
(-39-158)
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
(-27-109)
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 - 1 1 6)
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
(-29-116)
182
(2 - 362)
534
(392 - 676)
144
(106-183)
84
(1-167)
99
(73 - 1 25)
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%)
Dosure 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%)
Dosure 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%)
Dosure 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 PM2.5 Concentrat
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 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 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%)
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
(14-1 07)
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 - 1 20)
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 - 1 03)
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
(14-1 08)
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
(-26-113)
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%
(0.18% -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%- 1.14%)
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%
(11% -11%)
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%
(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
(-635-1889)
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.s 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):
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 -1730)
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- 1915)
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 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 .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
-------
Appendix F. Sensitivity Analysis Results
This Appendix provides detailed results of the single- and multi-factor sensitivity
analyses completed as part of this risk analysis. For additional detail on the sensitivity analysis
results completed for this analysis, as well as the types of results generated, see section 4.3.
F-2
-------
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.
-------
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 - 1 605)
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-4
-------
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.
F-5
-------
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
(167-355)
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-6
-------
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
(1 943 - 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-7
-------
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
(1 723 - 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-8
-------
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 - 1 827)
1526
(1191 -1856)
164
(71 - 253)
2965
(1 005 - 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-9
-------
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 - 1 509)
1263
(985 - 1 538)
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-10
-------
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 - 1 593)
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-ll
-------
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 - 51 6)
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-12
-------
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
(608-1305)
1108
(704-1510)
16%
1106
(703-1506)
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 - 1 340)
36%
990
(629 - 1 349)
1030
(654 - 1 403)
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
(1 41 - 302)
238
(1 51 - 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-13
-------
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
(1076-2295)
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
(1109-2366)
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
(645-1384)
1171
(744-1594)
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 - 1 053)
1043
(662 - 1 420)
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
(1 57 - 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-14
-------
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 PM2.5 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 - 1 1 7)
279
(1 02 - 453)
93
(20- 165)
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 - 1 66)
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 - 31 9)
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)
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-15
-------
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 PM2S 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-16
-------
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
(7-104)
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
(-6-118)
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
(-9-11 2)
-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-17
-------
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-18
-------
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-19
-------
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 - 1 27)
12
(-9 - 33)
46
(-31 - 122)
-30
(-1 32 - 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-20
-------
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 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-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-22
-------
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-23
-------
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 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-24
-------
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-25
-------
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-26
-------
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-27
-------
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
(-18-43)
1
(-7 - 8)
1
(-1 - 3)
2
(-13- 16)
-1
(-20- 15)
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-28
-------
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-29
-------
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 PM25
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). 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-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 PM2.5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m):
Recent PM25
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
(1707-3654)
1649
(1071 -2212)
1Based on Ito et al. (2007). 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-31
-------
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.s
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m):
Recent PM25
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
(1 1 40 - 2354)
Based on Ito et al. (2007). 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-32
-------
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 PM2S - 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 PM2S - Impactof 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 PNI25 - 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-lhear, 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 PM2S - 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
(-1 1 7 - 278)
-436%
-309%
-309%
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 PM2S Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM25Concentrations1
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 PM 25 - Impact of Changing the Type of Model, with a 0-Day Las
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%
'/W 25 - Impact of Changing the Type of Model, with a 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
Percentdiff. =
48%
-31
(-140-76)
120%
72%
48%
Cardiovascular Mortality Associated with Short-Term Exposure to PMzs - Impact of 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
Percentdiff. =
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 PM 2S - Impact of Changing the Type of Model, with a 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
Percentdiff. =
340%
F-34
-------
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 PM2S Concentrations that Just Meet the Current Standards, Based on Adjusting 2005 PM25Concentrations1
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
Percentdiff. =
36%
35
(-60-130)
880%
621%
683%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changjng 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
Percentdiff. =
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-linear, GAM (stringent), 100df
log-lhear, GLM, 100df
log-linear, 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 PM2.5 - Im
HA, respiratory (COPD+)
HA, respiratory (COPD+)
HA, respiratory (COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100df
log-linear, GLM, 100 df
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
Percentdiff. =
140%
35
(-60-130)
143%
262%
51%
69%
fpe of Model, with a 0-Day Lag
Max. positive est. =
336
Min. positive est. =
278
Percentdiff. =
21%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.5 - 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), 100 df
log-linear, GLM, 100 df
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
Percentdiff. =
58%
F-35
-------
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 PM2SConcentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Respiratory Hospital Admissions Associated with Short-Term Exposire 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 Changing 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
Percentdiff. =
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
Percentdiff. =
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 ana lysis estimates for non-accidental mortality and cardiovascuter mortality associated with short-term exposure to PM2.5 a re from Zanobetti and Schwartz (2009). The core ana lysis 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-36
-------
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 PM2S - 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 PM2S - Impactof 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 PNI25 - 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-lhear, 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 PM2S - 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 -118)
-156
(-497-183)
-156
(-555 - 240)
75
(-108- 257)
-435%
-308%
-308%
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 PM2S Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 PM25Concentrations1
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
Percentdiff. =
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
Percentdiff. =
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
Percentdiff. =
320%
F-38
-------
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 PM25Concentrations1
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
Percentdiff. =
36%
248
(3-491)
893%
631%
693%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM 2.5 - Impact of Changjng 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
Percentdiff. =
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-linear, 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, 100 df
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
Percentdiff. =
141%
248
(3-491)
147%
267%
52%
71%
fpe of Model, with a 0-Day Lag
Max. positive est. =
310
Min. positive est. =
256
Percentdiff. =
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), 100 df
log-linear, GLM, 100 df
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
Percentdiff. =
58%
F-39
-------
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 PM25 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 Chang ng 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
Pencentdiff. =
79%
oact of Changing the Lag Structure, with a Copdlutant 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-40
-------
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 PM2S - 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 PM2S - Impactof 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 PNI25 - 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-lhear, 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 PM2S - 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-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 PM25 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 PM 25 - Impact of Changing the Type of Model, with a 0-Day Las
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
Pencentdiff. =
2%
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
1
-30
(-132 -72)
109%
105%
105%
'/W 25 - Impact of Changing the Type of Model, with a 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
Pencentdiff. =
49%
-30
(-132 -72)
118%
69%
47%
Cardiovascular Mortality Associated with Short-Term Exposure to PMzs - Impact of 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, G LM, 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
289
(123-453)
305
(109-498)
148
(-21 -316)
148
(-57-351)
Max. positive est. =
305
Min. positive est. =
148
Pencentdiff. =
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 PM 2S - Impact of Changing the Type of Model, with a 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), 100df
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-42
-------
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, 100df
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-43
-------
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 PM2S 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 Changing 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
Percentdiff. =
79%
Respiratory Hospital Admissions Associated with Short-Term Exposire to PM 2.5 - Impact of Changing the Lag Structure, with a Copdlutant 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
Percentdiff. =
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 nearestwhole number. Numbers in parentheses are 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
cardiovascular 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-44
-------
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 - 1 97)
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-45
-------
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
(-108-257)
84
(-120-287)
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
(54-319)
192
(55 - 328)
3%
12/35
180
(73 - 286)
184
(75 - 293)
2%
26
(-40 - 90)
29
(-45-102)
12%
99
(-1 7-215)
109
(-1 9 - 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
(-1 8-218)
103
(-1 8 - 223)
2%
64
(-92 - 21 9)
72
(-1 02 - 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
(-15-185)
2%
53
(-76 - 1 82)
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-46
-------
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-47
-------
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-48
-------
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-49
-------
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 PM2.5 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-50
-------
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-51
-------
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-52
-------
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 - 1 37)
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-53
-------
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 - 1199)
708
(412- 1012)
129
(-227 - 526)
0%
45%
73%
45%
-74%
1The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 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 copolutant).
F-54
-------
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 5 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-55
-------
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 are 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-56
-------
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
(1 1 8 - 200)
1-Day Lag
35
(-60-130)
14
(-24 - 52)
2-Day Lag
30
(-58- 118)
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-57
-------
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 available from this study.
F-58
-------
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-59
-------
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
Locally Focused4
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
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
14.1
9.9
10.1
15.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
12.5
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
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
10.1
14.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
11.7
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
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
10.1
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
10.3
10.9
9.5
10.3
12.1
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
10.1
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
10.3
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
13.1
13.0
13.0
12.8
12.2
11.4
12.2
8.6
8.8
13.0
10.9
11.5
12.0
11.5
11.7
12.3
11.9
14.1
11.8
12.2
2007
CM
12.3
11.3
11.2
12.0
11.0
12.3
11.4
9.9
10.1
11.0
8.6
8.9
10.9
8.2
9.1
10.6
10.0
10.2
11.6
10.7
11.3
9.3
9.7
12/25
Max. M
S
11.8
14
10.7
10.3
11.0
11.1
11.3
12.3
12.0
10.2
9.6
10.2
7.3
7.4
12.0
9.2
9.6
10.1
9.7
9.8
10.3
10.0
11.8
9.9
10.2
2007
CM
11.2
11.76
9.5
9.4
10.0
9.4
10.7
11.4
10.7
8.3
8.5
9.2
7.3
7.5
10.1
7.0
7.7
9.1
8.4
8.5
9.8
9.0
9.5
7.8
9.0
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%
0%
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%
21%
24%
22%
0%
27%
27%
25%
32%
31%
24%
34%
30%
24%
27%
27%
22%
26%
23%
14%
12/25
35%
49%
50%
46%
47%
42%
13%
55%
54%
50%
64%
62%
36%
68%
60%
48%
55%
54%
47%
52%
48%
50%
F-60
-------
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, PA 5
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Rollback
Method
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
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
8.5
2007
CM
11.6
13.2
7.5
9.7
12.9
13.5
14.1
8.0
8.0
14/35
Max. M
S
13.3
15.6
7.7
10.8
14.0
14.0
8.4
8.5
2007
CM
11.6
13.2
7.5
9.7
12.1
12.6
8.0
8.0
13/35
Max. M
S
12.8
15.3
7.7
10.8
13.0
13.0
8.4
8.5
2007
CM
11.2
11.8
7.5
9.7
11.3
11.7
8.0
8.0
12/35
Max. M
S
11.8
15.3
7.7
10.8
12.0
12.0
8.4
8.5
2007
CM
10.5
11.2
7.5
9.7
10.4
10.8
8.0
8.0
13/30
Max. M
S
11.5
15.6
6.7
10.8
12.8
13.0
14.2
7.4
7.4
2007
CM
10.0
11.4
6.6
8.8
11.1
11.7
12.4
7.0
7.0
12/25
Max. M
S
9.7
13.9
5.7
9.1
10.8
11.0
11.9
6.3
6.3
2007
CM
8.4
9.6
5.6
7.7
9.3
9.9
10.4
6.0
6.0
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%
18%
0%
0%
23%
23%
0%
0%
12/35
19%
27%
0%
0%
35%
35%
0%
0%
13/30
27%
24%
55%
21%
25%
23%
21%
46%
46%
12/25
54%
48%
110%
51%
50%
47%
44%
93%
93%
For 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.
^The current primary PM2.s standards include an annual standard set at 15 ug/m° and a daily standard set at 35 ug/m°.
0 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 locally focused 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 locally focused method was not applied.
° The proportional rollback and locally focused methods were applied to Pittsburgh differently from the way they were applied in the other locations. See text for details.
B Percent reduction in composite monitor value with consideration for LML of 5.8 ug/m3 (note: composite monitor value denoted as CMV): %reduction = (CMVcurrentstandard - CMVa|ternaft/estandard)/(CMVcurrernt
standard-LML). Note, greyed cells identify instances where percent change differs by >10% across alternative rollback methods (for a given alternative standard level/study area combination).
F-61
-------
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
i
Rollback
Method
Proportional
Hybrid 3
Locally Focused4
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
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
14.1
9.9
10.1
15.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
12.5
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
9.9
11.6
13.2
14/35
Max. M
S
14.0
14.0
14.0
14.0
14.0
12.8
14.0
13.2
9.9
10.1
14.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
11.7
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
9.9
___
11.6
13.2
13/35
Max. M
S
13.0
13.0
13.0
13.0
13.0
12.8
13.0
13.0
9.9
10.1
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
10.3
10.9
9.5
10.3
12.1
11.3
11.3
11.6
9.9
___
11.2
11.8
12/35
Max. M
S
12.0
12.0
12.0
12.0
12.0
12.0
12.0
12.0
9.9
10.1
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
10.3
10.1
9.0
9.5
10.4
10.4
10.7
9.4
___
10.5
11.2
13/30
Max. M
S
13.0
12.7
12.3
13.1
13.0
13.0
12.8
12.2
11.4
12.2
8.6
8.8
13.0
10.9
11.5
12.0
11.5
11.7
12.3
11.9
14.1
11.8
12.2
11.5
15.6
2007
CM
12.3
11.3
11.2
12.0
11.0
12.3
11.4
9.9
10.1
11.0
8.6
8.9
10.9
8.2
9.1
10.6
10.0
10.2
11.6
10.7
11.3
9.3
9.7
10.0
11.4
12/25
Max. M
S
11.8
14
10.7
10.3
11.0
11.1
11.3
12.3
12.0
10.2
9.6
10.2
7.3
7.4
12.0
9.2
9.6
10.1
9.7
9.8
10.3
10.0
11.8
9.9
10.2
9.7
13.9
2007
CM
11.2
11.76
9.5
9.4
10.0
9.4
10.7
11.4
10.7
8.3
8.5
9.2
7.3
7.5
10.1
7.0
7.7
9.1
8.4
8.5
9.8
9.0
9.5
7.8
9.0
8.4
9.6
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%
18%
8%
21%
41%
3%
23%
9%
22%
14/35
cells use
cells use
4%
cells use
19%
cells use
cells use
6%
cells use
8%
cells use
cells use
21%
41%
cells use
3%
23%
cells use
9%
cells use
cells use
22%
13/35
;d as ba
;d as ba
4%
;d as ba
20%
;d as ba
;d as ba
16%
;d as ba
8%
;d as ba
;d as ba
17%
41%
;d as ba
0%
;d as ba
;d as ba
-
;d as ba
11%
12/35
sis for ca
sis for ca
4%
sis for ca
21%
sis for ca
sis for ca
18%
sis for ca
8%
sis for ca
sis for ca
13%
sis for ca
0%
sis for ca
sis for ca
sis for ca
13%
13/30
Iculation
Iculation
-2%
10%
Iculation
20%
Iculation
___
Iculation
5%
21%
Iculation
10%
Iculation
Iculation
26%
49%
Iculation
4%
28%
Iculation
12%
Iculation
10%
Iculation
25%
12/25
9%
-3%
12%
26%
36%
___
7%
26%
14%
...
38%
64%
5%
34%
15%
36%
31%
F-62
-------
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
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
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
8.5
2007
CM
7.5
9.7
12.9
13.5
14.1
8.0
8.0
14/35
Max. M
S
7.7
10.8
14.0
14.0
8.4
8.5
2007
CM
7.5
9.7
12.1
12.6
8.0
8.0
13/35
Max. M
S
7.7
10.8
13.0
13.0
8.4
8.5
2007
CM
7.5
9.7
11.3
11.7
8.0
8.0
12/35
Max. M
S
7.7
10.8
12.0
12.0
8.4
8.5
2007
CM
7.5
9.7
10.4
10.8
8.0
8.0
13/30
Max. M
S
6.7
10.8
12.8
13.0
14.2
7.4
7.4
2007
CM
6.6
8.8
11.1
11.7
12.4
7.0
7.0
12/25
Max. M
S
5.7
9.1
10.8
11.0
11.9
6.3
6.3
2007
CM
5.6
7.7
9.3
9.9
10.4
6.0
6.0
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
55%
8%
15%
0%
14/35
cells use
55%
cells use
6%
cells use
0%
13/35
d as ba
55%
d as ba
7%
d as ba
0%
12/35
sis for ca
55%
sis for ca
7%
sis for ca
0%
13/30
Iculation
74%
Iculation
10%
19%
Iculation
0%
12/25
1 09%
13%
23%
0%
'For 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.
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 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 locally focused 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 locally focused method was not applied.
5 The proportional rollback and locally focused 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 LMLof 5.8 ug/m3. Percent reduction = (CM V|0ca^ fbcused or hybrid - CM Vproportionai)/(CMVIocally focused orhybrid-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-63
-------
APPENDIX G: SUPPLEMEMNT TO THE NATIONAL-SCALE
ASSESSMENT OF LONG-TERM MORTALITY RELATED TO PM2 5
EXPOSURE
G-l
-------
Appendix G. National-Scale Assessment of Long-Term Mortality Related to PM2.s
Exposure (additional discussion of the analysis and technical detail regarding
methods and inputs used)
As mentioned in section 4.4, as part of assessing the representativeness of the 15 urban
study areas in the national-context, we completed a national-scale assessment of long-term
exposure-related mortality. This allowed us to assess where along the national-scale distribution
the 31 counties comprising our 15 urban study areas fell and specifically, whether they provide
coverage for areas in the country likely to experience relative elevated PM2.5-related long-term
exposure-related mortality risk. This appendix provides additional information related to the
national-scale assessment of long-term mortality that was completed as part of the risk analysis.
It begins with an overview of the national-scale analysis (section G.I), including presentation of
the national-scale mortality estimates that were generated as part of the analysis and a brief
description of the technical approach used in conducting the analysis. Then, additional technical
detail on air quality (CMAQ) modeling is presented in section G.2. Additional detail on the
outputs of the analysis, including air quality, exposure and risk estimates (focusing on graphical
presentation of results), is presented in section G.3.
G.I OVERVIEW OF NATIONAL-SCALE MORTALITY ANALYSIS
In this section we present the estimated nationwide premature mortality resulting from
recent exposures to ambient PM2 5 To perform this assessment we use 2005 PM2 5 fused air
quality estimates from the Community Model for Air Quality (CMAQ) (Byun and Schere, 2006)
in conjunction with the environmental Benefits Mapping and Analysis Program (BenMAP, Abt
Associates Inc, 2008) to estimate long-term PM2 5-related premature mortality nationwide. We
estimate excess PM2.5-related long-term mortality by applying two estimates of all-cause
mortality risk found in the Krewski et al. (2009) PM2 5 mortality extended analysis of the
American Cancer Society (ACS) cohort, and an estimate of all-cause mortality risk found in the
Laden et al. (2006) PM2.5 mortality extended analysis of the Six-Cities cohort. We estimate that
total PM2.5-related premature mortality ranges from 63,000 (39,00087,000) (95th percentile
confidence interval) and 88,000 (49,000130,000), respectively; in each case we estimated
deaths per year down to the lowest measured levels (LMLs) in each epidemiological study.
We had considered expanding the national-scale mortality to include additional health
endpoints (related to short-term PM2.5 exposure) or additional air quality scenarios that simulate
just meeting the current and alternative suites of standards. However, we have concluded that
any expansion of this assessment is beyond the scope of what is needed or can reasonably be
G-2
-------
done within the time and resources available for this review. The rationale for our decision not to
expand the scope of the national-scale analysis is as follows.
The goal of the national-scale analysis is two-fold: to provide perspective on the
magnitude of PM2.5 health impacts on a national-scale and to help to place the risk estimates
generated for the urban study areas in a national context. We note, however that the second goal
(assessing the representativeness of the 15 urban study areas in the national context) plays a
greater role in the context of the risk assessment completed as part of the PM NAAQS review.
The analysis as currently implemented achieves the first goal by providing estimates of long-
term exposure-related all-cause mortality under recent conditions. While simulation of risk for
the current and alternative standard levels would provide additional perspective on the magnitude
of national-scale risk, that assessment would be resource-intensive and subject to considerable
uncertainty if it were conducted using air quality simulation methods similar to those used in the
urban study area analysis (i.e., application of a combination of rollback methods that reflects
both local and regional patterns in ambient PM2.5 reductions implemented at the monitor-level).
A particular area of uncertainty (and technical complexity) related to air quality simulation
would be addressing the interplay between regional-scale reductions in ambient PM2.5 in adjacent
urbanized areas. In the urban study area analysis, each location is treated independently with
regard to simulating ambient PM2.5 under alternative suites of standards. However, if we were to
expand the national analysis to include alternative standards, then simulation of rollbacks in
ambient PM2.5 levels would necessarily have to address this contiguity issue between adjacent
urban areas and even between suburban areas and adjacent urbanized areas in the context of
simulating monitor rollback.
In addition, because long-term exposure-related mortality dominates PM2.5 in terms of
total incidence, providing coverage for this endpoint category ensures that the majority of PM2 5-
related mortality incidence is reflected in the analysis, without including short-term exposure-
related mortality.
The national-scale mortality analysis, as currently implemented, also achieves its second
goal: to help place risk estimates for the urban study areas in a national context. Because the
national-scale analysis focuses on the long-term exposure-related mortality, which is the primary
driver for PM2 5-related health impacts, the analysis allows us to assess how the urban study
areas "fall" across a national distribution of risk for this key health endpoint category (see section
4.4.2 of the RA). This then allows us to characterize the degree to which the set of urban study
areas provides coverage for areas of the country likely to experience relatively elevated levels of
PM2.5-related health impacts.
This assessment combines information regarding estimated PM2 5 air quality levels,
population projections, baseline mortality rates, and mortality risk coefficients to estimate PM2.5-
G-3
-------
related premature mortality. Figure G-l below provides a conceptual diagram, detailing each of
the key steps involved in performing this BenMAP-based health impact assessment.
Figure G-l 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
PM2 ^-Related
Premature Mortality
Population
Projections
Air Quality
Modeling
Incidence and
Prevalence Rates
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).
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
employed a data fusion approach, which joined 2005 monitored PM2.s concentrations with 2005
CMAQ-modeled air quality levels using the Voronoi Neighbor Averaging (VNA) technique
(Abt, 2003). CMAQ was run at a horizontal grid resolution of 12km for the east and 36km in the
G-4
-------
west using 2005 estimated emission levels and meteorology. Figure G-2 shows the geographic
distribution of baseline annual mean PM2.5 concentrations across the continental U.S. The
maximum predicted value within the U.S. is 31 |ig/m3, the mean PM2 5 value is 8.7 |ig/m3,
-th
median is 8.8 |ig/m and the 95 percentile value is about 14 |ig/m
Figure G-2 2005 fused surface baseline PM2.s concentrations
2005 Fused Surface Baseline Concentrations (ug/m3)
^B 1.03 to 4.2
4.3 to 6.5
6,6 to 9.34
9.35 to 12.30
^B I2-3I to 20.57
^B 20.58 to 59.42
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.
Premature Mortality Estimates
In this assessment of PM2.5-related premature mortality we considered risk estimates
drawn from studies based on two prospective cohorts. The first study is the recently published
G-5
-------
Krewski et al. (2009) extended reanalysis of the ACS cohort. To remain consistent with the
urban study areas analysis, we applied the two log-linear all-cause mortality risk coefficients
based on the 1979-1983 and the 1999-2000 time periods that control for 44 individual and 7
ecologic covariates. We also applied a log-linear all-cause mortality risk coefficient drawn from
the extended analysis of the Six Cities cohort as reported by Laden et al. (2006). When
estimating premature mortality using these functions we considered air quality levels down to the
lowest measured levels (LML) in each study; for the Krewski et al. (2009) study this is 5.8 |ig/m3
and for the Laden et al. (2006) study this is 10 |ig/m3. In general, we place a higher degree of
confidence in health impacts estimated at air quality levels at or above the LML because the
portion of the concentration-response curve below this point is extrapolated beyond the observed
data. We also estimated health impacts down to Policy Relevant Background (PRB) levels
(EPA, 2008). The final ISA presents estimates of annual mean PRB for each of 7 Health Effects
Institute PM regions; this value ranges from 0.62 |ig/m3 in the southwest to 1.72 |ig/m3 in the
southeast.
BenMAP contains baseline age-, cause- and county-specific mortality rates drawn from
the CDC-WONDER. Current baseline mortality estimates are an average of a three year period
from 1996-1998. EPA is in the process of updating these rates with 2006-2008 data; a sensitivity
analysis suggests that the results reported here are largely insensitive to the use of more current
mortality rates.
Results
Table G-l and figure G-3 below summarize the results of the national-scale mortality
analysis. Table G-l summarizes the total PM2.5-related premature mortality associated with
modeled 2005 PM2.5 levels.
Table G-l
Estimated PMi.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
th
or policy relevant background (90 percentile confidence interval)
Air Quality
Level
10 ng/m3 (LML
for Laden et al.,
2006)
Estimates Based on Krewski et al. (2009)
'79- '83 estimate
(90th percentile
confidence interval)
26,000
(16,00036,000)
'99-' 00 estimate
(90th percentile
confidence interval)
33,000
(22,000^4,000)
Estimates Based on
Laden et al. (2006)
(90th percentile
confidence interval)
88,000
(49,000130,000)
G-6
-------
5.8 ug/m3 (LML
for Krewski et
al., 2009)
Policy-Relevant
Background
Estimates Based on Krewski et al. (2009)
63,000
(39,00087,000)
110,000
(68,000150,000)
80,000
(54,000110,000)
140,000
(94,000180,000)
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.3 Figure G-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.
G-7
-------
Figure G-3 Percentage of premature mortality attributable to PM2.s exposure at various
2005 annual average PM2.s levels*
n=12
n=15,082
n=13.14S
5.8 to 10 us/ri3
10 to IS ii^/rnS 15to 20 ug/m3
2005 Average Ambient Baseline PM25 Levels
20ug,m3 and above
'Attributable mortality calculated usingKreaski et al. (2009) risk estimate based on '99-'00 follow-up period.
n= number of 12km or 36km grid cells at each baseline air quality level
Figure G-3 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.
We note that the results of the national-scale analysis used in assessing the
representativeness of the 15 urban study areas in the national context have already been
discussed in section 4.4 and are not discussed further here.
G.2 ADDITIONAL TECHNICAL DETAIL ON AIR QUALITY MODELING
The Community Model for Air Quality (CMAQ) model was used to estimate annual
PM2.5 concentrations for the year 2005 for the continental US. These data were then combined
with ambient monitored PM2.5 measurements to create "fused" spatial surfaces supplied to
BenMAP.
CMAQ Model Application and Evaluation
G-8
-------
CMAQ is a non-proprietary computer model that simulates the formation and fate of
photochemical oxidants, including PM2.5 and ozone, for given input sets of meteorological
conditions and emissions. This analysis employed a version of CMAQ based on the latest
publicly released version (i.e. CMAQ version 4.74).
Model Domain and Grid Resolution
The CMAQ modeling analyses were performed for two domains covering the continental
United States, as shown in Figure G-4. These domains consist of a horizontal grid of 36 km
covering the entire continental US and a finer-scale 12-km grid covering the Eastern U.S. The
model extends vertically from the surface to 100 millibars (approximately 15 km) using a sigma-
pressure coordinate system. The 36-km grid was used to establish the incoming air quality
concentrations along the boundaries of the 12-km grids. Table G-2 provides some basic
geographic information regarding the CMAQ domains. The 36-km and both 12-km CMAQ
modeling domains were modeled for the entire year of 2005. All 365 model days were used in
the annual average levels of PM2.5.
Table G-2. 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
4CMAQ 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-9
-------
36km Domain
Specs:
,_, _ . x,y:-1(X)6000,-1620000
12km Domain coi.row: 279,240
Figure G-4. 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).
CMAO Model Inputs
Emissions
The 2005 emissions inputs to CMAQ included five source sectors: a) Electric Generating
Units (EGUs); b) Other Stationary Sources (Point and Nonpoint); c) Onroad and Nonroad
Mobile Sources; d) Biogenic Emissions; and e) Fires. The fires portion of the inventory included
emissions from wildfires and prescribed burning computed as hour-specific point sources.
Electric Generating Units (EGUs)
G-10
-------
Annual emissions estimates for EGUs for all National Emissions Inventory (NEI) air
pollutants for 2005 were developed using data reported to the USEPA's Clean Air Marketing
Division's (CAMD) Acid Rain database. The Acid Rain database contains hourly emissions for
SO2 and NOx emissions plus hourly heat input amounts. These three values are reported to the
database by the largest electric generating facilities, usually based upon Continuous Emissions
Monitors (CEMs). For all pollutants except the directly monitored SO2 and NOx, the ratio of the
Acid Rain heat input for 2005 to the Acid Rain heat input for 2002 was used as the adjusting
ratio to estimate the 2005 emissions.
Other Stationary Sources (Point andNonpoint)
Emission estimates for other stationary sources including both point and nonpoint
stationary sources were held constant at the level in Version 3 of the 2002 NEI. The only
exception to this was that some information on plants that closed after 2002 was incorporated
into the emissions modeled. Emissions for plants that closed were set to zero. U.S. EPA, 2008c
provides complete documentation on the development of the 2002 NEI.
Onroad and NonroadMobile Sources
Emission estimates for all pollutants were developed using EPA's National Mobile
Inventory Model (NMIM), which uses MOBILE6 to calculate onroad emission factors. A full
VMT database at the county, roadway type, and vehicle type level of detail was developed from
Federal Highway Administration (FHWA) information. However, state and local agencies had
the opportunity to provide model inputs (vehicle populations, fuel characteristics, VMT, etc) for
2002 and 2005. If the state or local area submitted 2005 VMT estimates, these data were used.
However, if the state or local area only provided 2002 VMT estimates that were incorporated in
the 2002 NEI, the 2002 NEI VMT data were grown to 2005 using growth factors developed from
the FHWA data, and these grown VMT data replaced the baseline FHWA-based VMT data.
Otherwise, the FHWA-based VMT data were used.
Emission estimates for NONROAD model engines were developed using EPA's National
Mobile Inventory Model (NMIM), which incorporates NONROAD2005. Where states provided
alternate nonroad inputs, these data replaced EPA default inputs, as described above. For more
G-ll
-------
information on how NMIM is run, refer to the 2005 NEI documentation posted at
ftp://ftp.epa.gov/EmisInventory/2005 nei/mobile/2005 mobile nei version 2 report.pdf.
Fires
Fires in the 2005 emissions inventory were modeled with the same methodology as used
for the 2002 NEI (U.S. EPA, 2008). However, as described in Raffuse et al., 2008, the wildland
fire emission inventories for 2005 were produced using the BlueSky framework for the
conterminous United States, which used the Satellite Mapping Automatic Reanalysis Tool for
Fire Incident Reconciliation (SMARTFIRE) as the fire information source. SMARTFIRE is an
algorithm and database system designed to reconcile these disparate fire information sources to
produce daily fire location and size information (Sullivan et al., 2008).
Biogenic Emissions
Biogenic emissions were computed for CMAQ based on 2005 meteorology data using the
BEIS3.13 model (Schwede, et. al, 2005) from the Sparse Matrix Operator Kernel Emissions
(SMOKE). The BEIS3.13 model creates gridded, hourly, model-species emissions from
vegetation and soils. It estimates CO, VOC, and NOX emissions for the U.S., Mexico, and
Canada. The inputs to BEIS include:
temperature data at 10 meters which were obtained from the CMAQ
meteorological input files, and
land-use data from the Biogenic Emissions Landuse Database, version 3
(BELD3), which provides data on the 230 vegetation classes at 1 km resolution over most
of North America.
Meteorological Input Data:
The gridded meteorological input data for the entire year of 2005 were derived from
simulations of the Pennsylvania State University / National Center for Atmospheric Research
Mesoscale Model. This model, commonly referred to as MM5, is a limited-area, nonhydrostatic,
terrain-following system that solves for the full set of physical and thermodynamic equations
which govern atmospheric motions (Grell et al., 1994). Meteorological model input fields were
G-12
-------
prepared separately for both of the domains shown in Figure G-4 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-3
and do not vary by horizontal grid resolution.
Table G-3. 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
12
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
0.9
Approximate
Height (m)
0
38
77
115
154
232
310
389
469
550
631
712
794
Approximate
Pressure (mb)
1000
995
991
987
982
973
964
955
946
937
928
919
910
G-13
-------
CMAQ
Layers
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
MM5
Layers
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Sigma P
0.88
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)
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)
892
874
856
838
820
793
766
730
685
640
595
550
505
460
415
370
325
280
235
190
145
100
The meteorological outputs from the MM5 sets were processed to create model-ready
inputs for CMAQ using the Meteorology-Chemistry Interface Processor (MCIP), version 3.4, to
derive the specific inputs to CMAQ.
Before initiating the air quality simulations, it was important to identify the biases and
errors associated with the meteorological modeling inputs. The 2005 MM5 model performance
evaluations used an approach which included a combination of qualitative and quantitative
analyses to assess the adequacy of the MM5 simulated fields. The qualitative aspects involved
comparisons of the model-estimated synoptic patterns against observed patterns from historical
weather chart archives. Additionally, the evaluations compared spatial patterns of monthly
average rainfall and monthly maximum planetary boundary layer (PEL) heights. Qualitatively,
the model fields closely matched the observed synoptic patterns, which is not unexpected given
the use of nudging. The operational evaluation included statistical comparisons of
model/observed pairs (e.g., mean normalized bias, mean normalized error, index of agreement,
root mean square errors, etc.) for multiple meteorological parameters, including temperature,
humidity, shortwave downward radiation, wind speed, and wind direction (Baker and Dolwick,
G-14
-------
2009a, Baker and Dolwick, 2009b). It was ultimately determined that the bias and error values
associated with the 2005 meteorological data were generally within the range of past
meteorological modeling results that have been used for air quality applications.
Initial and Boundary Conditions
The lateral boundary and initial species concentrations are provided by a three-
dimensional global atmospheric chemistry model, the GEOS-CHEM model (Yantosca, 2004).
The global GEOS-CHEM model simulates atmospheric chemical and physical processes driven
by assimilated meteorological observations from the NASA's Goddard Earth Observing System
(GEOS). This model was run for 2002 with a grid resolution of 2.0 degrees x 2.5 degrees
(latitude-longitude) and 24 vertical layers. The 2005 CMAQ 36km simulation used non-year
specific GEOS-CHEM data, which was created by taking the median value for each month in
each individual grid cell of the 2002 GEOS-CHEM data described above. The predictions were
used to provide one-way dynamic boundary conditions and an initial concentration field for the
CMAQ simulations. More information is available about the GEOS-CHEM model and other
applications using this tool at: http://www-as.harvard.edu/chemistry/trop/geos.
CMAQ Model Performance Evaluation
An operational model performance evaluation for PM2.5 and its related speciated
components was conducted for 2005 using state/local monitoring sites data in order to estimate
the ability of the CMAQ modeling system to replicate the concentrations for the 12-km Eastern
domain and 36-km domain in the west. The principal evaluation statistics used to evaluate
CMAQ performance included two bias metrics, normalized mean bias and fractional bias; and
two error metrics, normalized mean error and fractional error. For the 12-km Eastern domain,
performance evaluation statistics were computed for the entire domain as well as its subregions.
For the 36-km domain, evaluation focuses on the parts of the US not covered by the 12-km
Eastern domain by computing performance evaluation statistics for the states included in the
Western Regional Air Partnership (WRAP).
The PM2.5 evaluation focuses on PM2.5 total mass and its components, including sulfate
(SO4), nitrate (NO3), total nitrate (TNO3 = NO3 + HNO3), ammonium (NH4), elemental carbon
(EC), and organic carbon (OC). PM2.5 ambient measurements for 2005 were obtained from the
following networks for model evaluation: Speciation Trends Network (STN), Interagency
Monitoring of PROtected Visual Environments (IMPROVE), and Clean Air Status and Trends
Network (CASTNET). For PM2.5 species that are measured by more than one network, we
G-15
-------
calculated separate sets of statistics for each network. Table G-4 provides annual model
performance statistics for PM2.5 and its component species. Based on the bias and error values
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-4. 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
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
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
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
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
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
G-16
-------
CMAQ 2005 Annual
Nitrate
Total Nitrate
(NO3+HNO3)
Ammonium
CASTNet
STN
IMPROVE
CASTNet
STN
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
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
No. of
Obs.
9693
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
NMB (%)
-5.2
-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
NME (%)
45.2
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
FB (%)
9.6
-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
FE (%)
47.6
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
G-17
-------
CMAQ 2005 Annual
Elemental
Carbon
Organic Carbon
CASTNet
STN
IMPROVE
STN
IMPROVE
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
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
No. of
Obs.
16,680
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
NMB (%)
8.1
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
NME (%)
47.2
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
FB (%)
12.8
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
FE (%)
48.9
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
G-18
-------
CMAQ 2005 Annual
36-km
West
WRAP
No. of
Obs.
17,295
NMB (%)
-22.5
NME (%)
57.5
FB (%)
-40.8
FE (%)
67.6
"Fused" Spatial Surfaces
Spatial surfaces of the 2005 data were created by fusing CMAQ-modeled annual average
PM2.5 concentrations with total PM2.5 data from STN, IMPROVE, and CASTNET monitoring
sites for the two domains shown in Figure G-4. We used the EPA's Model Attainment Test
Software (MATS) (Abt, 2009) which employees the Voronoi Neighbor Averaging (VNA)
interpolation technique (Abt, 2008). This technique identifies the set of monitors that are nearest
to the center of each grid cell, and then takes an inverse distance squared weighted average of the
monitor concentrations. The "fused" spatial fields are calculated by adjusting the interpolated
ambient data (in each grid cell) up or down by a multiplicative factor calculated as the ratio of
the modeled concentration at the grid cell divided by the modeled concentration at the nearest
neighbor monitor locations (weighted by distance).
To create the spatial surfaces for use in BenMAP, the 2005 CMAQ-modeled annual
average PM2.5 concentrations were "fused" with 2005 total PM2.5 ambient monitoring data from
STN, IMPROVE, and CASTNET sites. This was done for both the 36km national domain and
the 12km eastern US domain. The spatial surface of annual average PM2.5 air quality
concentrations produced by this technique is shown in Figure G-5 for the continental U.S. Where
available, the 12km spatial surface was used to supply BenMAP with annual average PM2 5
concentrations. In the western part of the U.S., annual average PM2.5 concentrations were
supplied from the 36km domain.
G-19
-------
Figure G-5: 2005 Predicted Annual Mean PM2.s Levels
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-20
-------
for unusual emission or meteorological events. There are also uncertainties associated with the
chemical transformation and fate process algorithms used in air quality models. For these
reasons, CMAQ predicts best on longer time scale bases (e.g., synoptic, monthly, and annual
scales). These limitations have led us to use modeled air quality estimates in this analysis that
are "fused" with measured ambient data and averaged over an annual scale.
G.3 ADDITIONAL DETAIL ON ESTIMATES (AIR QUALITY, EXPOSURE AND
RISK)
Air Quality Estimates
Figures G-6 through G-9 below illustrate the spatial distribution of air quality impacts.
Figure G-6 illustrates the modeled 2005 PM2.5 air quality levels across the U.S. Figures G-7 and
G-8 display the PM2.5 air quality levels after being adjusted so that the maximum level is no
higher than the LML reported in the Krewski et al. (2009) and Laden et al. (2006) studies. Figure
G-9 displays the PRB by region of the county.
G-21
-------
Figure G-6: 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 12.31 to 20.57
20.58 to 59.42
G-22
-------
Figure G-7: 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-23
-------
Figure G-8: 2005 Predicted Annual Mean PM2.5 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-24
-------
Figure G-9: 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-10 displays the distribution of grid cells at different baseline PM2.5 air quality
levels. Figures G-l 1 and G-13 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-25
-------
Figure G-10: The Number of Grid Cells at Each Level of PM2.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 ng/rn3
G-26
-------
Figure G-ll: 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-27
-------
Figure G-12: The Number of CMAQ Grid Cells Experiencing an Incremental
Change in Annual Mean PMi.s (jig/m3) (Current Conditions - 5.8
ug/m3)
9 10 ii 12 13 14 15 IS 17 18 15 20 21 22
Change in Ambient PM; 5 (|ig/m5)
25 30 31 32
Maximum change = 31.3 ftg/'m3
Number of cellswith no change: 10,000
G-28
-------
Figure G-13: The Number of CMAQ Grid Cells Experiencing an Incremental
Change in Annual Mean PMi.s (ug/m3) (Current Conditions - Policy
Relevant Background)
a
O
;o
Cf
<~>
!
i
,.
1 2 3 4 5 '5 7
Maximum change = 31 ng/m3
Number of cellswith no change: 0
9 10 11 12 13 14 15 16 17 IS 19 20 21 22 23 24 25 26 27 23 29 30 31 32
Change in Arrbient PM; 5 ||ig/m5)
Figure G-14 displays the cumulative distribution of grid cells at each baseline
concentration. Figures G-15 through G-17 display the cumulative distribution of grid cells
experiencing an incremental air quality change.
G-29
-------
Figure G-14: Cumulative Distribution of Baseline PM2.s Concentrations (ug/m3)
1 2 3 4 5 6 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-30
-------
Figure G-15: 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-31
-------
Figure G-16: 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-32
-------
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 PMj 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-5
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-33
-------
Table G-5. 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-14 through G-16 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-34
-------
Figure G-14: The Percentage of Total Mortality Attributable to PM2.5 Exposure:
Baseline -10 ug/m3
a
I 4000
Percentage of 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-35
-------
Figure G-15: The Percentage of Total Mortality Attributable to PM2.5 Exposure:
Baseline - 5.8 ug/m3
a
I 4.000
1% 2% 3°,, 4% 5% 6% 7% 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-36
-------
Figure G-16: 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-17 through G-19 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-37
-------
Figure G-17: 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-38
-------
Figure G-18: 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-39
-------
Figure G-19: 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-40
-------
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 2.3, 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 2005.5
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 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.s risk assessment, although the scope is significantly more limited, reflecting the more
5 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 PM10.25
risk assessment completed for the last review in the interest of completeness.
H-2
-------
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 PMi0 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.6 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):
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
6 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
-------
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 PMi0-2.5 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 PM10-2.s risk assessment (i.e., for
the period 2005-2007) 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 levels for consideration in this
review. Not only is the monitoring network (i.e., co-located PMio and PM2.5 monitors)
available for characterizing PMio-2.5 levels in candidate urban study areas limited (see
above), given the potential for greater spatial heterogeneity in PMio-2.5 levels (relative to
PM2.5 levels), generating representative estimates of ambient air profiles for PMio-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 PMio-2.5 to local-scale sources.
H-4
-------
Concerns that locations used in the risk assessment may not be representative of areas
experiencing the most significant 24-hour peak PM'10-2.5 concentrations (and
consequently, may not capture locations with the highest risk): This concern still holds
since the monitoring network available for characterizing PMio-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 PMio-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 affect 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 PM2.5 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.s to vary regionally as a confounder for the effect of PMio-2.5,
EPA staff believes that C-R functions with control for PM2.5 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 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 PMio-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-5
-------
APPENDIX I: ANALYSIS COMPARING DISTRIBUTION OF SHORT-
TERM EXPOSURE-RELATED CARDIOVASCULAR MORTALITY
INCIDENCE TO THE DISTIRBUTION OF DAILY PM25 LEVELS FOR
DETROIT AND NEW YORK
1-1
-------
1.1 OVERVIEW
As discussed in section 3.1.2.2, we have updated an earlier analysis from the 1996 PM
NAAQS Risk Assessment which showed that short-term exposure-related mortality incidence
was driven more by days with PM2.5 levels near the typical annual level, than by the days with
higher PM2.5 levels that comprise the tail of the annual daily PM2.5 distribution. This analysis
was completed using short-term exposure-related incidence estimates for cardiovascular
mortality for the simulation year 2007 (recent conditions) for Detroit (Figure 1-1) and New York
(Figure 1-2). The daily PM2.5 distributions are the composite monitor 24-hour annual distributions
for 2007 for each urban study area. Methods used in developing these composite monitor
distributions are discussed in section 3.2.1, while the methods used in generating the short-term
mortality estimates are presented in section 3.1.2.
As discussed in section 3.1.2.2, these figures clearly demonstrate that short-term
mortality cardiovascular incidence for these two urban study areas is driven more by daily PM2.5
levels nearer to the annual average levels, than by higher-end concentrations.
6/30/2010 1-2
-------
Figure 1-1: Comparison of Short-Term Exposure-Related Cardiovascular Mortality
Against 24-hour PMi.s Distribution (Detroit - recent conditions for
2007 - total PM-attributable incidence: 74)
120
100 -
80 -
o
o
ui
S
40 -
20 -
II Frequency of Daily Average PM2.5
Concentrations
-Associated Cardiovascular Mortality
-r 60
-- 50
-- 40
-- 30
20 :g
8
O
10 'I
W
3
-- 0
-10
12 16 20 24 28 32 36
Upper Bound of 4 ug/m3 PM2.5 "Bins"
40
44
48
52
1-3
-------
Figure 1-2: Comparison of Short-Term Exposure-Related Cardiovascular Mortality
Against 24-hour PMi.s Distribution (New York - recent conditions for
2007 - total PM-attributable incidence: 568)
Frequency of Daily Average PM2.5
Concentrations
Associated Cardiovascular Mortality
-T 250
-- 200
12
16
20 24 28 32 36
Upper Bound of 4 ug/m3 PM2.5 "Bins"
40
44
48
52
1-4
-------
APPENDIX J: PROVISIONAL RISK ESTIMATES AND ADDITIONAL
RESULTS OF SIMULATION INVOLVING THE ALTERNATIVE
ANNUAL STANDARD OF 10 jig/m3
J-l
-------
This appendix provides risk estimates generated for two sets of alternative standard levels
involving an alternative annual standard of 10 |ig/m3, including: (a) a pairing with the current 24-
hour standard (10/35) and (b) a pairing with the alternative 24-hour standard of 25 |ig/m3
(10/25). As noted in section 2.4, these risk estimates are potentially subject to greater uncertainty
than risk estimates generated for the other alternative annual standard levels considered in the
RA due to the need to simulate ambient PM2.5 levels for urban study areas assuming attainment
of this lower annual standard level.
To facilitate comparison of estimates for these two additional sets of alternative standards
against the other standards assessed as part of the formal risk assessment, we have repeated risk
estimates for the additional sets of alternative standard levels, as well as recent conditions and
the current set of standard levels (as appropriate). This does mean that the risk estimates
presented in these tables repeat many of the same estimates presented in tables in Appendix E.
The organization of the tables in this appendix mirror those presented in Appendix E, although
the tables presented here focus on long-term exposure-related health endpoints and do not
present estimates for any of the short-term exposure-related health endpoints modeled in the
formal risk assessment. The final tables presented in this appendix (Table J-19 and J-20) provide
the same type of information related to application of the different rollback methods as is
presented in Tables F-49 and F-50.
J-2
-------
Table J-l. 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
NewYork, 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 PMZ5 Concentrations in a RecentYear 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
(1746-2489)
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- 1163)
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- 1163)
1959
(1603-2307)
293
(240 - 345)
351
(286-414)
291
(238 - 343)
12
(9-14)
496
(406 - 583)
61
(50-72)
13/35
176
(1 44 - 207)
298
(244-351)
103
(84 -121)
231
(189-272)
444
(363 - 523)
68
(56-81)
263
(215-310)
984
(802 -1163)
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
(212 -306)
87
(71 - 103)
206
(169-243)
383
(313-452)
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
(310-447)
61
(50 - 72)
10/35
105
(85-124)
180
(147-212)
56
(46 - 66)
142
(116- 168)
258
(211 -306)
68
(56-81)
143
(116- 169)
515
(419-611)
1071
(873-1268)
152
(124- 180)
198
(161 -235)
162
(132- 191)
12
(9-14)
259
(21 1 - 307)
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- 1156)
153
(125-181)
194
(158-230)
146
(119-172)
0
(0-0)
305
(249-361)
20
(16-24)
10/25
105
(85 - 1 24)
180
(147-212)
56
(46 - 66)
142
(116- 168)
258
(21 1 - 306)
22
(18-26)
143
(116- 169)
373
(303 - 443)
976
(795-1156)
152
(124- 180)
194
(158-230)
146
(119- 172)
0
(0-0)
259
(211 -307)
20
(16-24)
'Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 in Knewski etal., 2009).
2Numbers roundedto the nearest whote number. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-J
-------
Table J-2. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PMis 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
NewYork, 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 PMZ5
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
(211 -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 - 41 0)
33
(27-39)
484
(397 - 569)
63
(51 -75)
1 5/353
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- 129)
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
(1 80 - 260)
6
(5-7)
363
(297 - 428)
36
(29- 43)
13/35
181
(149- 214)
237
(194- 279)
95
(77-112)
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
(129 -185)
202
(1 65 - 239)
80
(65 - 95)
153
(125-181)
249
(203 - 294)
72
(59 - 85)
213
(173-252)
707
(575 - 837)
1163
(947- 1375)
201
(164-238)
335
(273 - 396)
177
(144-209)
6
(5-7)
263
(215-311)
36
(29 - 43)
10/35
108
(88 - 1 28)
131
(107- 155)
50
(41 - 59)
97
(79-115)
145
(118- 172)
72
(59 - 85)
133
(108- 157)
379
(308 - 450)
689
(560-817)
131
(107- 155)
212
(172- 251)
107
(87 -127)
6
(5-7)
161
(131 - 191)
36
(29 - 43)
13/30
181
(149- 214)
228
(186- 268)
95
(77 - 1 12)
175
(143-207)
257
(209 - 304)
49
(40 - 58)
252
(206 - 298)
534
(434 - 633)
1043
(850- 1235)
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)
10/25
108
(88 -128)
131
(107- 155)
50
(41 - 59)
97
(79-115)
145
(118- 172)
25
(20 - 29)
133
(108- 157)
247
(200 - 293)
606
(492-719)
131
(107- 155)
207
(1 68 - 245)
93
(76-110)
0
(0-0)
161
(131 - 191)
2
(1-2)
Based on follow-upthrough2000, using modete
2Numbers roundedto the nearest whote number.
3The current primary PM2 5 standards include an
/In 44 individual and 7 ecological covariates (see Tabte 33 in Krewski etal., 2009).
Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficent.
annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-4
-------
Table J-3. 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
NewYork, 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 PM2.5
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
(159-230)
532
(436 - 625)
204
(169-239)
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)
1 5/353
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- 136)
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)
10/35
101
(82 -119)
131
(106- 155)
54
(44 - 64)
112
(91 -133)
162
(132- 192)
77
(63 - 92)
140
(114- 166)
413
(335 - 490)
906
(737-1073)
129
(105- 153)
164
(133- 195)
135
(109- 159)
15
(12- 18)
189
(154-224)
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
(1 54 - 223)
7
(6-8)
340
(278-401)
21
(17-25)
12/25
146
(119 -172)
155
(126-184)
71
(58 - 84)
172
(140-203)
174
(142 -206)
28
(23 - 33)
223
(182 -264)
278
(225 - 330)
815
(663 - 966)
130
(106-154)
160
(130 -189)
120
(98 - 1 42)
0
(0-0)
231
(188-273)
3
(2-3)
10/25
101
(82 -119)
131
(106- 155)
54
(44 - 64)
112
(91 -133)
162
(132- 192)
28
(23 - 33)
140
(114- 166)
278
(225 - 330)
815
(663 - 966)
129
(105- 153)
160
(130- 189)
120
(98-142)
0
(0-0)
189
(154-224)
3
(2-3)
'Based on follow-upthrough2000, using modete win 44 individual and 7 ecological covariates (see Tabte 33 inKnewski etal., 2009).
2Numbers rounded to the nearest whote number. Numbers in parentheses a re 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-5
-------
Table J-4. Estimated Percent of Total Annual Incidence oflschemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM25 Concentrations in a Recent Year (2005) and PM25 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 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 of Total Incidence of Ischemic Heart Disease Mortality Associated with Long- Term Exposure to PMzs Concentrations in a RecentYear and
PMZ5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
15.8%
(13%- 18.6%)
15.6%
(12.8% -18.2%)
15.9%
(13.1% -18.6%)
1 1 .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%)
13.3%
(10.9% -15.6%)
8%
(6.5% -9.4%)
1 5.7%
(12.9% -18.4%)
8.2%
(6.7% -9.7%)
1 6.1 %
(13.3% -18.9%)
9.2%
(7.5%- 10.8%)
1 5/353
14.1%
(11. 6% -16. 6%)
14.4%
(11. 8% -16. 9%)
11.3%
(9.3%- 13.4%)
11.1%
(9.1%- 13.1%)
12.2%
(10% -14. 3%)
6.1 %
(5% -7.2%)
11.2%
(9.2%- 13.2%)
7%
(5.7% -8.3%)
10.6%
(8.6%- 12.4%)
11.7%
(9.6%- 13.8%)
8%
(6.5% -9.4%)
10.5%
(8.6%- 12.3%)
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%- 11.9%)
1 1 .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
11.2%
(9.2% -13.2%)
11.7%
(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% -11.9%)
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% -11. 5%)
10.2%
(8.3%- 12%)
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%- 11.4%)
6.1%
(4.9% -7.2%)
10/35
6.7%
(5. 4% -7.9%)
7.1%
(5. 8% -8.4%)
4.8%
(3.9% - 5.7%)
6.8%
(5.5% - 8%)
6.2%
(5% - 7.3%)
6.1%
(5% - 7.2%)
4.7%
(3.8%- 5.6%)
3.7%
(3% - 4.4%)
5.8%
(4.7% - 6.8%)
6.1%
(5% - 7.2%)
4.5%
(3.7% - 5.3%)
5.8%
(4.7%- 6.9%)
2.4%
(1.9% -2.8%)
6.6%
(5.4% - 7.8%)
6.1%
(4.9% - 7.2%)
13/30
11.2%
(9.2% - 13.2%)
11.3%
(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%- 11.2%)
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%)
10/25
6.7%
(5.4% - 7.9%)
7.1%
(5.8% - 8.4%)
4.8%
(3.9% - 5.7%)
6.8%
(5.5% - 8%)
6.2%
(5% - 7.3%)
2%
(1.6% -2.4%)
4.7%
(3.8% - 5.6%)
2.7%
(2. 2% -3.2%)
5.3%
(4.3%- 6.2%)
6.1%
(5% -7.2%)
4.4%
(3.6% - 5.2%)
5.2%
(4.3%- 6.2%)
0%
(0% - 0%)
6.6%
(5.4%- 7.8%)
2%
(1.6%- 2.4%)
Based on follow-upthrough2000, using modete win 44 individual and 7 ecological covariates (see Table 33 inKtewski etal., 2009).
2Petcents rounded to the nearest tenth. Numbers in parentheses ate 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.
J-6
-------
Table J-5. Estimated Percent of Total Annual Incidence oflschemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM25 Concentrations in a Recent Year (2006) and PM25 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 PMzs Concentrations in a Recent Year and
PMZ5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
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%)
10.9%
(9% -12.9%)
12.1%
(9.9%- 14.3%)
8.1%
(6.6% -9.6%)
12.6%
(10.4% -14.8%)
6.5%
(5.3% -7.7%)
12.2%
(10%- 14.4%)
6.1%
(5% - 7.3%)
1 5/353
14.1%
(11. 6% -16. 6%)
11.7%
(9.6%- 13.7%)
10.5%
(8.6%- 12.3%)
8.2%
(6.7% -9. 7%)
8.5%
(6.9% -10%)
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%)
1 0.6%
(8.7%- 12.5%)
8.1%
(6.6% - 9.6%)
8%
(6.5% - 9.4%)
1.2%
(1 % - 1 .5%)
9.2%
(7.5%- 10.8%)
3.5%
(2.9% - 4.2%)
13/35
11.2%
(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% -11.5%)
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%)
10/35
6.7%
(5.4% - 7.9%)
5.1%
(4. 2% -6.1%)
4.2%
(3.4% - 5%)
4.6%
(3.7% -5.4%)
3.5%
(2. 8% -4.1%)
6.4%
(5.2% - 7.5%)
4.2%
(3.4% -5%)
2.7%
(2. 2% -3.2%)
3.7%
(3%- 4.4%)
5.3%
(4.3% - 6.2%)
4.6%
(3.8% - 5.5%)
3.9%
(3.2%- 4.6%)
1.2%
(1 % - 1 .5%)
4. 1 %
(3.3%- 4.8%)
3.5%
(2.9%- 4.2%)
13/30
11.2%
(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%- 11.2%)
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%)
10/25
6.7%
(5. 4% -7.9%)
5.1%
(4. 2% -6.1%)
4.2%
(3.4%- 5%)
4.6%
(3. 7% -5.4%)
3.5%
(2. 8% -4.1%)
2.2%
(1.8%- 2.6%)
4.2%
(3.4% - 5%)
1 .8%
(1.4% -2.1%)
3.2%
(2.6%- 3.8%)
5.3%
(4.3%- 6.2%)
4.5%
(3.7%- 5.4%)
3.4%
(2.7%- 4%)
0%
(0% - 0%)
4.1 %
(3.3%- 4.8%)
0.1 %
(0.1%- 0.2%)
Based on follow-up thro ugh 2000, using modete win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Petcents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistical uncertainty surrounding the PM coefficent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-7
-------
Table J-6. 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 PM25 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 PM25 from 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
Percent of Total Incidence of Ischemic Heart Disease Mortality Associated with Long- Term Exposure to PMzs Concentrations in a Recent Year and
PMZ5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.s
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%)
1 7.7%
(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%)
14.2%
(11.7% -16.7%)
9%
(7.4%- 10.6%)
13.3%
(10.9% -15.7%)
6.3%
(5.2% - 7.5%)
1 5/353
13.2%
(10.9% -15. 5%)
11.7%
(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%)
11.2%
(9.2%- 13.2%)
3.7%
(3% -4. 4%)
14/35
1 1 .8%
(9.7%- 13.9%)
10.6%
(8.7%- 12.5%)
9.7%
(7.9%- 11.4%)
9%
(7.3%- 10.6%)
9%
(7.3%- 10.6%)
6.7%
(5.5% - 8%)
9.5%
(7.8%- 11.2%)
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%- 11.9%)
3.7%
(3% - 4.4%)
13/35
10.4%
(8.5% -12. 3%)
9.3%
(7.6% - 1 1 %)
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% - 1 1 %)
6.7%
(5.5% -7.9%)
8.7%
(7.1% -10.3%)
2.9%
(2.4%- 3.4%)
8.8%
(7 .2% -10. 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%)
10/35
6.1%
(4.9% -7.2%)
5. 1 %
(4.2%- 6.1%)
4.5%
(3.7% -5.4%)
5.2%
(4.2%- 6.1%)
3.9%
(3.2%- 4.6%)
6.7%
(5.5% - 8%)
4.4%
(3.6%- 5.2%)
2.9%
(2.4%- 3.5%)
4.8%
(3.9%- 5.7%)
5.2%
(4.2%- 6.1%)
3.5%
(2. 8% -4.1%)
4.9%
(4% - 5.8%)
2.9%
(2. 4% -3. 4%)
4.8%
(3.9% -5.7%)
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%)
10/25
6.1 %
(4.9% -7.2%)
5.1 %
(4.2%- 6.1%)
4.5%
(3.7% - 5.4%)
5.2%
(4.2%- 6.1%)
3.9%
(3.2% - 4.6%)
2.4%
(2% - 2.9%)
4.4%
(3.6% - 5.2%)
2%
(1 .6% - 2.3%)
4.3%
(3.5%- 5.1%)
5.2%
(4.2%- 6.1%)
3.4%
(2. 8% -4%)
4.4%
(3.5%- 5.2%)
0%
(0% - 0%)
4.8%
(3. 9% -5.7%)
0.3%
(0.2% - 0.3%)
Based on follow-upthrough2000, using modete with 44 individual and 7 ecological covariates (see Table 33 in Knewski etal., 2009).
2Petcents rounded to the nearest tenth. Numbers in parentheses a re 95% confidence or credible intervals based on statist bal uncertainty surround ing the PM coefficent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-8
-------
Table J-7. 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 CurrentStandards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMZ5
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%
(-8% - -8%)
-40%
(-39% - -41 %)
0%
(0% - 0%)
-35%
(-35% - -36%)
-174%
(-171% --177%)
-9%
(-9% - -9%)
-116%
(-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%)
1 5/353
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%- 11%)
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%
(1 0% - 1 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
20%
(1 9% - 20%)
0%
(0% - 0%)
12/35
31%
(31% -31%)
29%
(29% - 29%)
34%
(34% - 34%)
11%
(11% -11%)
25%
(25% - 25%)
0%
(0% - 0%)
34%
(34% - 35%)
12%
(12% -12%)
18%
(1 8% - 1 8%)
23%
(22% - 23%)
10%
(1 0% - 1 0%)
17%
(1 7% - 1 7%)
0%
(0% - 0%)
30%
(30% -31%)
0%
(0% - 0%)
10/35
53%
(52% - 53%)
51%
(50% -51%)
58%
(57% - 58%)
39%
(38% - 39%)
49%
(49% - 50%)
0%
(0% - 0%)
58%
(58% - 58%)
48%
(47% - 48%)
45%
(45% - 46%)
48%
(48% - 48%)
44%
(43% - 44%)
44%
(44% - 45%)
0%
(0% - 0%)
52%
(52% - 53%)
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%)
100%
(100% -100%)
44%
(44% - 44%)
67%
(67% - 67%)
10/25
53%
(52% -53%)
51%
(50% -51%)
58%
(57% - 58%)
39%
(38% - 39%)
49%
(49% - 50%)
68%
(67% - 68%)
58%
(58% - 58%)
62%
(62% -62%)
50%
(50% - 50%)
48%
(48% - 48%)
45%
(45% - 45%)
50%
(50% - 50%)
100%
(100%- 100%)
52%
(52%- 53%)
67%
(67% - 67%)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coefficent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-9
-------
Table J-8. 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 CurrentStandards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMZ5
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%)
-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%)
1 5/353
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%
(1 1 % - 1 1 %)
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%
(1 0% - 1 0%)
20%
(20% - 20%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
10/35
53%
(52% - 53%)
56%
(56% - 56%)
60%
(60% - 60%)
45%
(44% - 45%)
59%
(59% - 59%)
0%
(0% - 0%)
60%
(59% - 60%)
54%
(53% - 54%)
53%
(53% - 53%)
50%
(50% - 51 %)
43%
(43% - 43%)
51%
(51% -51%)
0%
(0% - 0%)
60%
(60% -61%)
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%)
100%
(100% -100%)
51%
(50% - 51 %)
96%
(96% - 96%)
10/25
53%
(52% -53%)
56%
(56% - 56%)
60%
(60% - 60%)
45%
(44% - 45%)
59%
(59% - 59%)
66%
(66% - 66%)
60%
(59% - 60%)
70%
(70% - 70%)
59%
(59% - 59%)
50%
(50% - 51 %)
44%
(44% - 45%)
58%
(58% - 58%)
100%
(100%- 100%)
60%
(60% - 61 %)
96%
(96% - 96%)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coefficent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-10
-------
Table J-9. 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 CurrentStandards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMZ5
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% --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% --212%)
-19%
(-18% --19%)
-72%
(-71 % - -72%)
1 5/353
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%
(1 1 % - 1 1 %)
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%)
10/35
54%
(54% - 55%)
56%
(56% - 56%)
59%
(58% - 59%)
43%
(42% - 43%)
57%
(57% - 57%)
0%
(0% - 0%)
59%
(59% - 59%)
52%
(52% - 52%)
48%
(48% - 49%)
51%
(50% - 51 %)
48%
(48% - 48%)
47%
(47% - 48%)
0%
(0% - 0%)
58%
(57% - 58%)
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%)
100%
(100% -100%)
48%
(48% - 49%)
93%
(93% - 93%)
10/25
54%
(54% - 55%)
56%
(56% - 56%)
59%
(58% - 59%)
43%
(42% -43%)
57%
(57% - 57%)
64%
(64% - 64%)
59%
(59% - 59%)
68%
(68% - 68%)
54%
(53% - 54%)
51%
(50% -51%)
50%
(49% - 50%)
53%
(53% - 53%)
1 00%
(1 00% - 1 00%)
58%
(57% - 58%)
93%
(93% - 93%)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surround ing the PM coeffbent.
3The current primary PM2 s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-ll
-------
Table J-10. 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 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 Ischemic Heart Disease Mortality Associated with Long- Term Exposure to PM2.5 Concentrations in a Recent Yearand PMZ5
Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM25
Concentrations
312
(257 - 364)
497
(409 -581)
233
(192 -273)
292
(239 - 344)
862
(711-1 007)
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- 138)
1 5/353
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
(2024-2914)
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 - 1 03)
382
(31 2 - 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
(213 -307)
485
(396 - 572)
87
(71 - 1 03)
284
(231 -335)
1103
(897- 1306)
2040
(1665-2408)
287
(234 - 338)
401
(327 - 475)
305
(249-361)
15
(12 -18)
480
(392 - 566)
78
(63 - 92)
10/35
133
(108- 158)
228
(186- 270)
72
(58 - 85)
180
(147- 213)
329
(267 - 389)
87
(71 -103)
182
(148- 216)
658
(533- 781)
1363
(1108- 1615)
194
(158-229)
252
(205 - 300)
206
(167-244)
15
(12- 18)
330
(268 - 390)
78
(63 - 92)
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- 1030)
1871
(1525-2210)
284
(232 - 335)
389
(317-461)
278
(227 - 329)
5
(4-6)
539
(441 - 635)
52
(42 - 62)
12/25
189
(154-223)
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)
10/25
133
(108- 158)
228
(1 86 - 270)
72
(58 - 85)
180
(147- 213)
329
(267 - 389)
28
(23 - 34)
182
(148- 216)
477
(386 - 567)
1243
(1010- 1474)
194
(158-229)
247
(200 - 293)
185
(151 -220)
0
(0-0)
330
(268 - 390)
26
(21 - 31)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest whote number. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coeffbent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-12
-------
Table J-ll. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2.s
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 Ischemic Heart Disease Mortality Associated with Long- Term Exposure to PM2.5 Concentrations in a Recent Yearand PMZ5
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
(313-448)
471
(384 - 557)
439
(360-516)
42
(34 - 50)
610
(500-716)
80
(65 - 95)
1 5/353
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 - 1 08)
368
(301 - 434)
1038
(843 - 1 229)
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
(1 58 - 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)
10/35
137
(112- 162)
167
(135- 198)
64
(52 - 75)
124
(101 - 147)
185
(150- 220)
92
(75-108)
169
(137- 201)
484
(392 - 576)
880
(713-1045)
167
(136- 198)
270
(219-320)
137
(111 - 163)
8
(6-10)
205
(167-244)
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)
10/25
137
(112- 162)
167
(135- 198)
64
(52 - 75)
124
(101 - 147)
185
(150- 220)
32
(26 - 38)
169
(137- 201)
316
(255 - 376)
774
(627 - 920)
167
(136- 198)
264
(214-313)
119
(96 -141)
0
(0-0)
205
(167-244)
2
(2-2)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest whote number. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coeffbent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-13
-------
Table J-12. Estimated Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient PM2.s
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 Ischemic Heart Disease Mortality Associated with Long- Term Exposure to PM2.5 Concentrations in a Recent Yearand PMZ5
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
(211 -298)
473
(387 - 556)
2456
(2017 -2879)
3003
(2462 - 3525)
378
(309 - 444)
402
(327 - 476)
490
(403 - 574)
59
(48 - 70)
665
(546 - 780)
84
(68 - 99)
1 5/353
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)
14/35
248
(203-291)
342
(280 - 402)
146
(120- 173)
247
(202- 291)
471
(385 - 556)
98
(80- 116)
385
(31 4 - 453)
1094
(890 - 1 296)
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
(1 78 - 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)
10/35
128
(104- 152)
167
(135- 197)
69
(56 - 82)
143
(116- 169)
207
(168- 246)
98
(80-116)
179
(145- 212)
528
(427 - 627)
1155
(937-1369)
164
(133- 195)
209
(170-249)
171
(139-203)
19
(16- 23)
241
(196-286)
49
(40 - 58)
13/30
219
(179-258)
288
(235 - 339)
128
(104- 151)
247
(202- 291)
352
(286- 416)
68
(55 - 80)
335
(273 - 395)
730
(592 - 866)
1641
(1336- 1941)
249
(203 - 295)
347
(282- 410)
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)
10/25
128
(104- 152)
167
(135- 197)
69
(56 - 82)
143
(116- 169)
207
(168- 246)
36
(29 - 43)
179
(145- 212)
355
(287 - 423)
1040
(843-1234)
164
(133- 195)
204
(165-242)
153
(124- 181)
0
(0-0)
241
(196-286)
4
(3-4)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest whote number. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coeffbent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-14
-------
Table J-13. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM25 Concentration s in a Recent Year (2005) and PM25 Concentration s that Just Meet the Current and Alternative Standards,
Based on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM25 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 Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMzs Concentrations in a Recent Year and
PMZ5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
19.9%
(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%- 11.9%)
19.7%
(16.2% -23%)
10.4%
(8.5%- 12.3%)
20.2%
(16.6% -23.6%)
1 1 .6%
(9.5%- 13.7%)
1 5/353
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%- 11.9%)
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%- 11.9%)
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% -11.9%)
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%- 11.2%)
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%- 13%)
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%)
10/35
8.5%
(6.9%- 10%)
9%
(7 .3% -10. 6%)
6.1%
(5% - 7.2%)
8.6%
(7% -10.2%)
7.8%
(6.4% -9.3%)
7.8%
(6.3% -9.2%)
6%
(4.9% -7.1%)
4.7%
(3.8% -5.6%)
7.3%
(6% - 8.7%)
7.7%
(6.3% -9.2%)
5.7%
(4.6% -6.8%)
7.4%
(6% - 8.8%)
3%
(2. 4% -3.6%)
8.4%
(6.8%- 9.9%)
7.7%
(6.3% -9.1%)
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%)
11.9%
(9 .7% -14%)
5.2%
(4.2% -6.2%)
11%
(9%- 13%)
6.2%
(5% -7. 4%)
10.1%
(8.2%- 11.9%)
11.4%
(9.3%- 13.4%)
8.8%
(7.2% - 10.5%)
10%
(8.2%- 11.8%)
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% -11.6%)
2.6%
(2.1% -3.1%)
10/25
8.5%
(6.9% -10%)
9%
(7 .3% -10. 6%)
6.1 %
(5% -7.2%)
8.6%
(7%- 10.2%)
7.8%
(6.4% -9.3%)
2.5%
(2.1% -3%)
6%
(4. 9% -7.1%)
3.4%
(2. 8% -4.1%)
6.7%
(5.4% -7.9%)
7.7%
(6.3% - 9.2%)
5.6%
(4.5%- 6.6%)
6.7%
(5.4% -7.9%)
0%
(0% - 0%)
8.4%
(6.8% - 9.9%)
2.6%
(2.1%- 3.1%)
'Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Petcents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistbal 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.
J-15
-------
Table J-14. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM25 Concentration s in a Recent Year (2006) and PM25 Concentration s that Just Meet the Current and Alternative Standards,
Based on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM25 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 Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMzs Concentrations in a Recent Year and
PMZ5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
Concentrations
19.8%
(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%)
1 5/353
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%)
1 0. 1 %
(8.2%- 11.9%)
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%- 11.9%)
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%)
11.8%
(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% -11.2%)
11.9%
(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% -11.9%)
4.5%
(3.6% - 5.3%)
12/35
12.3%
(10%- 14.5%)
10.1%
(8.2%- 11.9%)
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%- 11%)
8.1%
(6.6% -9. 6%)
1.6%
(1.3% -1.9%)
8.5%
(6.9%- 10%)
4.5%
(3.6% -5.3%)
10/35
8.5%
(6.9%- 10%)
6.5%
(5.3% -7.8%)
5.4%
(4.4% - 6.4%)
5.8%
(4.7% -6.9%)
4.4%
(3.6% -5.3%)
8.1%
(6.6% -9.6%)
5.4%
(4.4% -6.4%)
3.4%
(2. 8% -4.1%)
4.7%
(3.8% -5.6%)
6.7%
(5.4% -7.9%)
5.9%
(4.8% - 7%)
5%
(4% - 5.9%)
1 .6%
(1.3%- 1.9%)
5.2%
(4.2%- 6.2%)
4.5%
(3.6% -5.3%)
13/30
14.2%
(11.6%- 16.6%)
11.3%
(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% - 11.5%)
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%)
10/25
8.5%
(6.9% -10%)
6.5%
(5.3% -7.8%)
5.4%
(4.4%- 6.4%)
5.8%
(4.7% -6.9%)
4.4%
(3.6% - 5.3%)
2.8%
(2. 3% -3.3%)
5.4%
(4.4% - 6.4%)
2.2%
(1 .8% - 2.7%)
4.1%
(3.3% - 4.9%)
6.7%
(5.4% - 7.9%)
5.8%
(4.7% - 6.8%)
4.3%
(3. 5% -5.1%)
0%
(0% - 0%)
5.2%
(4.2% - 6.2%)
0.2%
(0.2%- 0.2%)
'Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Petcents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistbal 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.
J-16
-------
Table J-15. Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to Ambient
PM25 Concentration s in a Recent Year (2007) and PM25 Concentration s that Just Meet the Current and Alternative Standards,
Based on Adjusting 2007 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM25 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 Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMzs Concentrations in a Recent Year and
PMZ5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m)2:
Recent PM2.5
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%)
1 6.8%
(13.8% -19.7%)
8%
(6.5% -9.5%)
1 5/353
16.7%
(13.7% -19.5%)
14.7%
(12.1% -17.3%)
13.8%
(11.3% -16.2%)
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%)
14/35
14.9%
(12.2% - 17.6%)
13.4%
(11%- 15.8%)
12.2%
(1 0% - 1 4.4%)
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%)
11.8%
(9.6% -13. 9%)
10.7%
(8.7% -12.6%)
11.4%
(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%)
11.2%
(9 .2% -13. 2%)
11.8%
(9.6% -13.9%)
8.5%
(6.9% -10.1%)
11%
(9% -13%)
3.7%
(3% -4. 4%)
11.2%
(9.1% -13.2%)
4.7%
(3.8% - 5.6%)
12/35
1 1 .4%
(9.3%- 13.4%)
10.1%
(8.2%- 11.9%)
9.1%
(7.4%- 10.7%)
10%
(8.2%- 11.9%)
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%- 11.3%)
10.1%
(8.2%- 11.9%)
7.6%
(6.2% - 9%)
9.6%
(7.8%- 11.4%)
3.7%
(3% - 4.4%)
9.5%
(7.7%- 11.2%)
4.7%
(3.8% -5.6%)
10/35
7.7%
(6.3% -9.1%)
6.5%
(5.3% -7.8%)
5.8%
(4.7%- 6.8%)
6.6%
(5.3% -7.8%)
5%
(4% - 5.9%)
8.5%
(7% -10.1%)
5.6%
(4.5% -6.6%)
3.7%
(3% - 4.4%)
6.1%
(5% - 7.3%)
6.6%
(5.3% -7.8%)
4.4%
(3.6%- 5.3%)
6.2%
(5.1% -7.4%)
3.7%
(3% - 4.4%)
6.1%
(4.9%- 7.2%)
4.7%
(3.8% -5.6%)
13/30
13.2%
(10.8%- 15.5%)
11.3%
(9.2%- 13.3%)
10.7%
(8.7% - 12.6%)
11.4%
(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
1 1 .1 %
(9.1%- 13.1%)
7.8%
(6.3% -9. 2%)
7.5%
(6.1% -8.9%)
10%
(8.2%- 11.9%)
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%)
10/25
7.7%
(6.3%- 9.1%)
6.5%
(5.3% -7.8%)
5.8%
(4.7%- 6.8%)
6.6%
(5.3% -7.8%)
5%
(4% -5. 9%)
3.1%
(2. 5% -3.7%)
5.6%
(4.5% - 6.6%)
2.5%
(2% - 3%)
5.5%
(4. 5% -6.5%)
6.6%
(5.3% - 7.8%)
4.3%
(3.5%- 5.1%)
5.6%
(4. 5% -6.6%)
0%
(0% - 0%)
6.1%
(4.9% - 7.2%)
0.3%
(0.3%- 0.4%)
'Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Petcents rounded to the nearest tenth. Numbers in parentheses are 95% confidence or credible intervals based on statistbal 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.
J-17
-------
Table J-16. 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 CurrentStandards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMZ5
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 2%
(-11% --12%)
-8%
(-8% - -8%)
-39%
(-38% - -40%)
0%
(0% - 0%)
-34%
(-34% - -35%)
-170%
(-166% --174%)
-9%
(-9% - -9%)
-113%
(-111% --11 6%)
-33%
(-32% - -33%)
-13%
(-13% --14%)
0%
(0% - 0%)
-49%
(-48% - -50%)
-244%
(-242% --247%)
-16%
(-16% --17%)
-51 %
(-50% --51%)
1 5/353
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% - 11%)
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%
(1 0% - 1 0%)
0%
(0% - 0%)
6%
(6% - 6%)
0%
(0% - 0%)
19%
(1 9% - 1 9%)
0%
(0% - 0%)
12/35
31%
(30% -31%)
29%
(28% - 29%)
34%
(33% - 34%)
11%
(11% -11%)
24%
(24% - 25%)
0%
(0% - 0%)
34%
(34% - 34%)
12%
(12% -12%)
18%
(1 7% - 1 8%)
22%
(22% - 23%)
10%
(1 0% - 1 0%)
17%
(1 7% - 1 7%)
0%
(0% - 0%)
30%
(29% - 30%)
0%
(0% - 0%)
10/35
52%
(59% - 42%)
50%
(58% - 40%)
57%
(64% - 48%)
38%
(48% - 25%)
49%
(56% - 38%)
0%
(16% --23%)
58%
(64% - 48%)
47%
(55% - 35%)
45%
(53% - 33%)
48%
(55% - 36%)
43%
(52% - 30%)
44%
(52%- 32%)
0%
(1 6% - -24%)
52%
(59% - 41 %)
0%
(16% --23%)
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%
(11% -11%)
46%
(46% - 46%)
67%
(67% - 67%)
34%
(34% - 34%)
62%
(62% - 62%)
50%
(49% - 50%)
47%
(47% - 48%)
45%
(44% - 45%)
50%
(49% - 50%)
100%
(100% -100%)
43%
(43% - 44%)
67%
(66% - 67%)
10/25
52%
(52%- 53%)
50%
(50% - 51 %)
57%
(57% - 58%)
38%
(38% - 39%)
49%
(48% - 49%)
67%
(67% - 67%)
58%
(57% - 58%)
62%
(62% -62%)
50%
(49% - 50%)
48%
(47% - 48%)
45%
(44% - 45%)
50%
(49% - 50%)
100%
(100%- 100%)
52%
(51 % - 52%)
67%
(66% - 67%)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coeffbent.
3The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-18
-------
Table J-17. 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 CurrentStandards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMZ5
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 2%
(-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%)
1 5/353
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%
(1 0% - 1 0%)
20%
(1 9% - 20%)
0%
(0% - 0%)
35%
(34% - 35%)
0%
(0% - 0%)
10/35
52%
(52% - 53%)
56%
(55% - 56%)
59%
(59% - 60%)
44%
(44% - 44%)
59%
(59% - 59%)
0%
(0% - 0%)
59%
(59% - 60%)
53%
(53% - 54%)
53%
(53% - 53%)
50%
(50% - 50%)
43%
(43% - 43%)
51%
(51 % - 51 %)
0%
(0% - 0%)
60%
(60% - 60%)
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%)
100%
(100% -100%)
50%
(50% - 51 %)
96%
(96% - 96%)
10/25
52%
(52%- 53%)
56%
(55% - 56%)
59%
(59% - 60%)
44%
(44% - 44%)
59%
(59% - 59%)
66%
(65% - 66%)
59%
(59% - 60%)
70%
(69% - 70%)
58%
(58% - 59%)
50%
(50% - 50%)
44%
(44% - 44%)
57%
(57% - 58%)
100%
(100%- 100%)
60%
(60% - 60%)
96%
(96% - 96%)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coeffbent.
3The current primary PM2 s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-19
-------
Table J-18. 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 CurrentStandards: Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PMZ5
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 2%
(-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%)
1 5/352
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%
(11% - 11%)
0%
(0% - 0%)
1%
(1%- 1%)
0%
(0% - 0%)
11%
(11% - 11%)
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%
(1 9% - 1 9%)
24%
(23% - 24%)
11%
(1 1 % - 1 1 %)
18%
(1 8% - 1 8%)
0%
(0% - 0%)
33%
(33% - 33%)
0%
(0% - 0%)
10/35
54%
(53% - 54%)
56%
(55% - 56%)
58%
(58% - 59%)
42%
(42% - 43%)
57%
(56% - 57%)
0%
(0% - 0%)
59%
(58% - 59%)
52%
(52% -52%)
48%
(48% - 48%)
50%
(50% - 51 %)
48%
(48% - 48%)
47%
(47% - 47%)
0%
(0% - 0%)
57%
(57% - 57%)
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%)
100%
(100% -100%)
48%
(47% - 48%)
93%
(93% - 93%)
10/25
54%
(53% - 54%)
56%
(55% - 56%)
58%
(58% - 59%)
42%
(42% - 43%)
57%
(56% - 57%)
63%
(63% - 64%)
59%
(58% - 59%)
68%
(67% - 68%)
53%
(53% - 54%)
50%
(50% - 51 %)
49%
(49% - 50%)
53%
(52%- 53%)
100%
(100%- 100%)
57%
(57% - 57%)
93%
(93% - 93%)
Based on follow-upthrough2000, using models win 44 individual and 7 ecological covariates (see Table 33 inKnewski etal., 2009).
2Numbers rounded to the nearest percent. Numbers in parentheses a re 95% confidence or credibte intervals based on statistical uncertainty surrounding the PM coeffbent.
3The current primary PM2 s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
J-20
-------
Table J-19. 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
Locally Focused4
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
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
14.1
9.9
10.1
15.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
12.5
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
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
10.1
14.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
11.7
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
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
10.1
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
10.3
10.9
9.5
10.3
12.1
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
10.1
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
10.3
10.1
9.0
9.5
10.4
10.4
10.7
9.4
...
10/35
Max. M
S
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
9.9
10.1
10.0
10.0
10.0
10.0
10.0
10.0
10.0
...
2007
CM
9.5
8.9
9.1
8.5
9.5
8.9
8.2
8.9
9.9
10.3
8.5
7.6
8.0
8.7
8.7
8.9
7.9
...
13/30
Max. M
S
13.0
12.7
12.3
13.1
13.0
13.0
12.8
12.2
11.4
12.2
8.6
8.8
13.0
10.9
11.5
12.0
11.5
11.7
12.3
11.9
14.1
11.8
12.2
2007
CM
12.3
11.3
11.2
12.0
11.0
12.3
11.4
9.9
10.1
11.0
8.6
8.9
10.9
8.2
9.1
10.6
10.0
10.2
11.6
10.7
11.3
9.3
9.7
12/25
Max. M
S
11.8
14
10.7
10.3
11.0
11.1
11.3
12.3
12.0
10.2
9.6
10.2
7.3
7.4
12.0
9.2
9.6
10.1
9.7
9.8
10.3
10.0
11.8
9.9
10.2
2007
CM
11.2
11.76
9.5
9.4
10.0
9.4
10.7
11.4
10.7
8.3
8.5
9.2
7.3
7.5
10.1
7.0
7.7
9.1
8.4
8.5
9.8
9.0
9.5
7.8
9.0
10/25
Max. M
S
10.0
10.0
10.0
10.0
10.0
10.0
10.0
9.6
7.3
7.4
10.0
9.2
9.6
10.1
9.7
9.8
10.3
10.0
9.9
10.2
2007
CM
9.5
8.9
9.1
8.5
9.5
8.9
8.2
8.5
7.3
7.5
8.5
7.0
7.7
9.1
8.4
8.5
9.8
8.9
7.8
9.0
Percent reduction in a surrogate for long-term
exposure-related mortality (alternatve standard
compared w th current standard)6
14/35
11%
...
9%
4%
12%
11%
0%
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%
10/35
56%
...
57%
54%
60%
56%
44%
58%
48%
0%
0%
60%
53%
54%
50%
51%
52%
...
49%
13/30
22%
...
25%
25%
21%
24%
22%
0%
27%
27%
25%
32%
31%
24%
34%
30%
24%
27%
27%
22%
26%
23%
14%
12/25
35%
...
49%
50%
46%
47%
42%
13%
55%
54%
50%
64%
62%
36%
68%
60%
48%
55%
54%
47%
52%
48%
50%
10/25
56%
...
57%
54%
60%
56%
44%
58%
54%
64%
62%
60%
68%
60%
48%
55%
54%
47%
52%
...
50%
J-21
-------
Table J-19 (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
Locally Focusei
Proportional
Hybrid
Locally Focusei
Proportional
Hybrid
Locally Focusei
Proportional
Hybrid
Locally Focusei
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
8.5
2007
CM
11.6
13.2
7.5
9.7
12.9
13.5
14.1
8.0
8.0
14/35
Max. M
S
13.3
15.6
7.7
10.8
14.0
14.0
8.4
8.5
2007
CM
11.6
13.2
7.5
9.7
12.1
12.6
8.0
8.0
13/35
Max. M
S
12.8
15.3
7.7
10.8
13.0
13.0
8.4
8.5
2007
CM
11.2
11.8
7.5
9.7
11.3
11.7
8.0
8.0
12/35
Max. M
S
11.8
15.3
7.7
10.8
12.0
12.0
8.4
8.5
For some locations (e.g., Atlanta) more than one "version" (group of counties) was used in the risk assessment. In this
2007
CM
10.5
11.2
7.5
9.7
10.4
10.8
8.0
8.0
10/35
Max. M
S
10.0
7.7
10.8
10.0
10.0
8.4
8.5
2007
CM
8.8
7.5
9.7
8.7
9.0
8.0
8.0
13/30
Max. M
S
11.5
15.6
6.7
10.8
12.8
13.0
14.2
7.4
7.4
2007
CM
10.0
11.4
6.6
8.8
11.1
11.7
12.4
7.0
7.0
12/25
Max. M
S
9.7
13.9
5.7
9.1
10.8
11.0
11.9
6.3
6.3
2007
CM
8.4
9.6
5.6
7.7
9.3
9.9
10.4
6.0
6.0
10/25
Max. M
S
9.7
5.7
9.1
10.0
10.0
6.3
6.3
2007
CM
8.4
5.6
7.7
8.7
9.0
6.0
6.0
Percent reduction
exposure-related m
compared w
n a surrogate for long-term
ortality (alternative standard
th current standard)6
14/35
0%
0%
0%
0%
10%
12%
0%
0%
13/35
7%
18%
0%
0%
23%
23%
0%
0%
12/35
19%
27%
0%
0%
35%
35%
0%
0%
10/35
49%
0%
0%
59%
58%
0%
0%
13/30
27%
24%
55%
21%
25%
23%
21%
46%
46%
12/25
54%
48%
1 1 0%
51%
50%
47%
44%
93%
93%
able only the version that was used for mortality associated with short-term exposure to PM2 5 (Zanobetti and Schwartz, 2009)
10/25
54%
1 1 0%
51%
59%
58%
93%
93%
s
included.
2The current primary PM2s 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 locally focused 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
location, the annual standard was controlling and so the locally focused method was not applied.
5 The proportional rollback and locally focused methods were applied to Pittsburgh differently from the way they were applied in the other locations. See text for details.
b Percent reduction in composite monitor value with consideration for LML of 5.8 ug/m3 (note: composite monitor value denoted as CMV): %reduction = (CM Vcurrent standard- CMVaitemativestandardy(CMVcurrerntstandard-LMI
where percent change differs by >10% across alternative rollback methods (fora given alternative standard level/study area combination).
annual and daily standards in that
L). Note, greyed cells identify instances
J-22
-------
Table J-20. 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
Rollback Method
Proportional
Hybrid 3
Locally Focused4
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
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
14.1
9.9
10.1
15.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
12.5
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
9.9
11.6
13.2
14/35
Max. M
S
14.0
14.0
14.0
14.0
14.0
12.8
14.0
13.2
9.9
10.1
14.0
12.7
13.3
13.9
13.3
13.6
14.3
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
10.3
11.7
9.5
10.5
12.1
11.6
11.8
13.3
12.3
13.0
9.9
11.6
13.2
13/35
Max. M
S
13.0
13.0
13.0
13.0
13.0
12.8
13.0
13.0
9.9
10.1
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
10.3
10.9
9.5
10.3
12.1
11.3
11.3
11.6
9.9
11.2
11.8
12/35
Max. M
S
12.0
12.0
12.0
12.0
12.0
12.0
12.0
12.0
9.9
10.1
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
10.3
10.1
9.0
9.5
10.4
10.4
10.7
9.4
10.5
11.2
10/35
Max. M
S
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
9.9
10.1
10.0
10.0
10.0
10.0
10.0
10.0
10.0
10.0
2007
CM
9.5
8.9
9.1
8.5
9.5
8.9
8.2
8.9
9.9
10.3
8.5
7.6
8.0
8.7
8.7
8.9
7.9
8.8
13/30
Max. M
S
13.0
12.7
12.3
13.1
13.0
13.0
12.8
12.2
11.4
12.2
8.6
8.8
13.0
10.9
11.5
12.0
11.5
11.7
12.3
11.9
14.1
11.8
12.2
11.5
15.6
2007
CM
12.3
11.3
11.2
12.0
11.0
12.3
11.4
9.9
10.1
11.0
8.6
8.9
10.9
8.2
9.1
10.6
10.0
10.2
11.6
10.7
11.3
9.3
9.7
10.0
11.4
12/25
Max. M
S
11.8
14
10.7
10.3
11.0
11.1
11.3
12.3
12.0
10.2
9.6
10.2
7.3
7.4
12.0
9.2
9.6
10.1
9.7
9.8
10.3
10.0
11.8
9.9
10.2
9.7
13.9
2007
CM
11.2
11.76
9.5
9.4
10.0
9.4
10.7
11.4
10.7
8.3
8.5
9.2
7.3
7.5
10.1
7.0
7.7
9.1
8.4
8.5
9.8
9.0
9.5
7.8
9.0
8.4
9.6
10/25
Max. M
S
10.0
10.0
10.0
10.0
10.0
10.0
10.0
9.6
7.3
7.4
10.0
9.2
9.6
10.1
9.7
9.8
10.3
10.0
9.9
10.2
9.7
2007
CM
9.5
8.9
9.1
8.5
9.5
8.9
8.2
8.5
7.3
7.5
8.5
7.0
7.7
9.1
8.4
8.5
9.8
8.9
7.8
9.0
8.4
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%
18%
8%
21%
41%
3%
23%
9%
-
22%
14/35
4%
19%
6%
8%
21%
41%
3%
23%
9%
...
22%
13/35
cells use
cells use
4%
cells use
20%
cells use
cells use
16%
cells use
8%
cells use
cells use
17%
41%
cells use
0%
cells use
cells use
cells use
11%
12/35
d as ba
d as ba
4%
d as ba
21%
d as ba
d as ba
18%
d as ba
8%
d as ba
d as ba
13%
d as ba
0%
d as ba
d as ba
d as ba
13%
10/35
sis for ca
sis for ca
6%
sis for ca
26%
sis for ca
sis for ca
23%
sis for ca
8%
sis for ca
sis for ca
19%
sis for ca
0%
sis for ca
sis for ca
sis for ca
13/30
culation
culation
-2%
10%
culation
20%
culation
culation
5%
21%
culation
10%
culation
culation
26%
49%
culation
4%
28%
culation
12%
culation
10%
culation
25%
12/25
9%
-3%
12%
26%
36%
-
7%
26%
14%
-
38%
64%
5%
34%
15%
36%
31%
10/25
-
6%
26%
-
13%
14%
-
38%
64%
5%
34%
-
36%
-
J-23
-------
Table J-20 (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
Locally Focused
Proportional
Hybrid
Locally Focused
Proportional
Hybrid
Locally Focused
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
8.5
2007
CM
7.5
9.7
12.9
13.5
14.1
8.0
8.0
14/35
Max. M
S
7.7
10.8
14.0
14.0
8.4
8.5
2007
CM
7.5
9.7
12.1
12.6
8.0
8.0
13/35
Max. M
S
7.7
10.8
13.0
13.0
8.4
8.5
2007
CM
7.5
9.7
11.3
11.7
8.0
8.0
12/35
Max. M
S
7.7
10.8
12.0
12.0
8.4
8.5
2007
CM
7.5
9.7
10.4
10.8
8.0
8.0
10/35
Max. M
S
7.7
10.8
10.0
10.0
8.4
8.5
2007
CM
7.5
9.7
8.7
9.0
8.0
8.0
13/30
Max. M
S
6.7
10.8
12.8
13.0
14.2
7.4
7.4
2007
CM
6.6
8.8
11.1
11.7
12.4
7.0
7.0
12/25
Max. M
S
5.7
9.1
10.8
11.0
11.9
6.3
6.3
2007
CM
5.6
7.7
9.3
9.9
10.4
6.0
6.0
10/25
Max. M
S
5.7
9.1
10.0
10.0
6.3
6.3
2007
CM
5.6
7.7
8.7
9.0
6.0
6.0
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
55%
8%
15%
0%
14/35
55%
6%
0%
13/35
cells use
55%
cells use
7%
cells use
0%
12/35
d as ba
55%
d as ba
7%
d as ba
0%
10/35
sis for ca
55%
sis for ca
9%
sis for ca
0%
13/30
culation
74%
culation
10%
19%
culation
0%
12/25
109%
13%
23%
0%
10/25
109%
9%
0%
1 For 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 PM25 (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 locally focused 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 locally focused method was not applied.
5 The proportional rollback and locally focused 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 LMLof 5.8 ug/m3. Percent reduction = (CMVIocally focused orhybnd - CMVprQpDrtlDnai)/(CMVIocally focused Drhybnci-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).
J-24
-------
APPENDIX K: MAPS OF THE FIFTEEN URBAN STUDY AREAS
EVALUATED IN THE RISK ASSESSMENT
K-l
-------
This appendix provides maps for each of the 15 urban study areas included in the
risk assessment. The spatial templates used in defining 13 of the 15 study areas were
based either on the combined statistical area (CSA) or the core-based statistical area
(CBSA), if the CSA was not available.7 The three remaining urban study areas
(Baltimore, Philadelphia and Tacoma) are special cases and were handled as described in
section 3.3.2. The maps presented in this appendix provide several types of information
including: (a) annual and daily (i.e., 24-hour) design values (DV) for each PM2.5 monitor
in each study area with DV values based on monitoring data from 2005-2007, (b) sources
of PM2.5 greater than 50 tons/year (c) major highways within each study area, and (d)
counties comprising each urban study area, together with the CSA or CBSA boundary as
relevant.
7 The spatial template defines the geographical area associated with each study area that would be
used in identifying which counties and PM2 5 monitors were associated with a particular study area.
K-2
-------
Figure K-l: Atlanta Urban Study Area
Chemical Products Corporation
Primary PM25 Annual Emission
84.9 tpy
AQSitel 30670003
Annual DV16ug/m3
Daily DV35ug/m3
Pickens Dawson
~7
Cherokee
Georgia Power Company, Bowen Steam-Electric Generating Pfant
Primary PM25 Annual Emission
9027.5 tpy
AQSitel 30670004
Annual DV15.3ug/m3
Daily DV32ug/m3
AQSitel 32230003
Annual DV14.3ug/m3
Daily DV33ug/m3
Georgia Power Company. McDonough Steam-E
Primary PM25 Annual Emission
1375.4 tpy
Owens Corning - Fairburn Plant
Primary PM25 Annual Emission
272 4 tpy
Georgia Power Company, Yates Steam-Electric Generati
Primary PM25 Annual Emission
2209 8 tpy
AQSite131390003
Annual DV13,9ug/m3
Daily DV30ug/m3
AQSite131350002
Annual DV15.7ug/m3
Daily DV31 ug/m3
Johns Manville
Primary PM25 Annual Emission
54 3 tpy
AQSitel 30892001
Annual DV15.2ug/m3
Daily DV34ug/m3
AQSitel 31210032
Annual DV15.6ug/m3
Daily DV34ug/m3
AQSite131210039
Annual DV17.7ug/m3
Daily DV32ug/m3
AQSitel 31210048
Annual DV15.3ug/m3
Daily DV34ug/m3
AQSitel 30890002
Annual DV15.2ug/m3
Daily DV33ug/m3
AQSitel 30630091
Annual DV16.2ug/m3
Daily DV34ug/m3
Georgia-Pacific Corp - Monticello Plywood Pit
Primary PM25 Annual Emission
54.3 tpy
Georgia Power Company, Wansley Steam-Electric Generating Plant
Primary PM25 Annual Emission
3632 6 tpy
The Sherwin-Williams Company
Primary PM25 Annual Emission
57 2 tpy
Georgia-Pacific Corp. - Monticello Panelboard
Primary PM25 Annual Emission
236.9 tpy
Georgia Pacific Corp - Warm Springs, GA Plywood
Primary PM25 Annual Emission
75.1 tpy
K-2
-------
Figure K-2: Baltimore Urban Study Area
AQSite 245100006
AQSite 240051007
AQSite 245100007 Annual DV 13.6 ug/m
Annual DV 13.9 ug/m3Daily DV 33jjgiro
Daily DV 34 ug/m3
AQSite 2451 °0008
Annual DV 13.8 ug/mAnnual DV 15-6 u9/m3
Daily DV 33 ug/m3 DailV DV37 u9/m3
AQSite 240053001
Annual DV14.5 ug/m3
Daily DV 35 ug/m3
AQSite 240032002
Annual DV15.7 ug/m3
Daily DV 36 ug/m3
AQSite 240030014
Annual DV 12 ug/m
Daily DV 31 ug/m3
AQSite 245100040
Annual DV15.2 ug/m3
Daily DV 37 ug/m3
AQSite 240251001
Annual DV 12.4 ug/m3
Daily DV 31 ug/m3
CONSTELLATION POWER SOURCE GENERATION - BRANDON SHO
Primary PM25 Annual Emission
1968 tpy
TATE & LYLE NORTH AMERICAN~SUGARS|
Primary PM25 Annual Emission
88.4 tpy'A
RED STAR YEAST$K^ TSCflL
Primary PM25 Annual Emission^
56.6 tpy f^ \\ \\ ^UNITED STATES GYPSUM COMPANYl
~^\\\.^r Primary PM25 Annual Emission
FEElfef
AQSite 245100049
Annual DV 16.1 ug/m
Daily DV 37 ug/m3
AQSite 245100035
Annual DV 15.1 ug/m
Daily DV 36 ug/m3
AQSite 240031003
Annual DV 14.3 ug/m
Daily DV 34 ug/m3
BETHLEHEM STEEL
Primary PM25 Annual Emission
1326.2 tpy
GRACE - DAVISON CHEMICAL
Primary PM25 Annual Emission
I7.2 tpy
^^^
|CONSTE!LATION POWER SOURCE GENERATION -WAGNER
[Primary PM25 Annual Emission
39.3 tpy
ONSTELLATION POWER SOURCE GENERATION - BRANDON SHORES
Primary PM25 Annual Emission
1660 tpy
K-4
-------
Figure K-3: Birmingham Urban Study Area
AQSite 011270002 AQSite 010735003
Annual DV 14.1 ug/m3 Annual DV 14.6 ug/m
Daily DV 33 ug/m3 Daily DV 36 ug/m3
ALABAMA POWER COMPANY
Primary PM25 Annual Emission
1666.5 tpy
Alabama Power Company
Primary PM25 Annual Emission
2738.8 tpy
AQSite 010732003
Annual DV 17.5 ug/m
Daily DV 41 ug/m3
AQSite 010731009
Annual DV 13.7 ug/m:
Daily DV 35 ug/m3
OAK GROVE RESOURCES, LLC
Primary PM25 Annual Emission
169.1 tpy
U. S. PIPE & FOUNDRY COMPANYJNC.(BESSEMER PLAN
Primary PM25 Annual Emission
295.4 tpy
AQSite 010731005
Annual DV15.6 ug/i
Daily DV 33 ug/m
AQSite 010730023
Annual DV18.7 ug/m3
Daily DV 44 ug/m3
AQSite 010735002
Annual DV14.6 ug/m3
Daily DV 35 ug/m3
National Cement Co of Alabama
Primary PM25 Annual Emission
129.6 tpy
RUMMOND COMPANY, INC.
Primary PM25 Annual Emission
1222.2 tpy
TCH CEMENT COMPANY
Primary PM25 Annual Emission
215.2 tpy
U. S. PIPE & FOUNDRY COMPANY INC.(NO. B'HAM PLANT)
Primary PM25 Annual Emission
356.8 tpy
AQSite 010731010
Annual DV 15.9 ug/m
Daily DV 34 ug/m3
R POWER PLAN
AME
Primary
436.9 tpy
UNITED STATES STEE
Primary PM25 Annual Emi
187.2 tpy
AQSite 010732006
Annual DV 15.1 ug/m
Daily DV 32 ug/m3 /LaFarge BiluldtBg-Mateiicfls
Primary PM25 Annual Emission
Southern Lime 344 tpy
Primary PM25 Annual Emission Alabama Power Company
51 8 tpy Primary PM25 Annual Emission
6804.7 tpy
DUSTRIES CORPORATION - COKE/UTILITIES/BTF
lual Emission
478.2 tp ^
SLbsS.INDUSTRItS-CORPORATIO^hHVUJ\IERAL WOOL PLANT
SMI STEEL, INC.
Primary PM25 Annual Emission
i34tpy
NUCOR STEEL BIRMINGHAM,INC.
CAST IRON PIPE^SQMPANY Primary PM25 Annual Emission
Annual Emission ^\^ 85.6 tpy
MCWANE CAST IRON PIPE CO.
Primary PM25 Annual Emission
308.3 tpy
ORPORATION-FAIRFIELD PIPE MIL
n
UNITED STATES STEEL CORPORATION - FAIRFIELD WORKS
Primary PM25 Annual Emission
635.1 toy
K-5
-------
Figure K-4: Dallas Urban Study Area
AM ELECTRIC STATION^8"6 481130069
AQSite 481130050 Fannin
Annual DV 12.8 ug/m3
Daily DV 26 ug/m3
Annual DV 11.5 ug/m3
AQSite 484391002
laily DV 25 ug/m
Primary PM25 Annual^&jiission
Annual DV 11.2
AQSite 481130
-^-t-^ J . T 11
13.6 ug/m
Daily DV 27 ug/m3
1
HANDLEY STEAM ELEC. S
Primary PM25 Annual Emiss
105.1
Site 48439101
iualDV11.9ua/m3 "
KL tr1-
QSite 481130057
CERAMIC TILE GLAZED MFG.
Primary PM25 Annua
58.2 tpy
nual DV 14.2 ug/
DV 29 ug/m3
AQSite484393006
Annual DV 12.5 ug/m3
Daily DV 26 ug/m3 Parker
AQSite 48113300
^Ann;ua,:DV12.4ug/
ilyDV|26rug7m3
NORTH TEXAS CEMENT CO.
Primary PM25 Annual Emission
HOLCIM ([TEXAS) LP
Primarw pkyl?^ Annim
AQSite 481390016
rimary PM25 Annual E
aily-DV-24 ug/m3
RRAL STEEL MIDLOTHIAK
Primary PM25 Annual Emission
Hood
DEROSA PINE ENERGY
Primary PM25v\nnual Emission
QSite 481130087
al DV 10.8 |jg/m
\ Daily>DV 22 ug/m3
Primary PM25 Annutel Emission
WAXAHACHIE PLA
Primary PM25 Annual^mission
419.8tpv
Annual DV 12.5vug/m3
AQSite 481390015
G S FFING PRODUCTS
IDLOTHIAMPNT
Primary PM25 Annual Emission Annual DV 12.<2 \iglm
ission Primary F1M25 Annual En|j , inn ~
Daily DV 27 Md/m=
K-6
-------
Figure K-5: Detroit Urban Study Area
AQSite 26099000
Afinual DV 12.5 ug/m
Daily DV 35 ug/
QS te 260490021
nualDV 11.6 ug/m3
ily DV 29 ua/m
AQSite 261470005
Annual DV 13.2 ug/m
Daily DV 41 ug/m3
ST.
AQSite 261250001
nnual DV 13.6 ug/m
"
AQSite 261630019 AQSite 261630016
Annual DV 14.1 ug/m3 Annual DV 14.3 ug/m
Daily DV 40 ug/m3 , Daily DV 41 ug/m"
T
261610005
ual DV 13.2 {jg/m3
ily DV 3
AQSite 261630039
Annual DV 14.4 ug/m3
Daily DV 37 ug/m3
AQSite 261610008
nnual DV 13.7 ug/
aily DV 39 ug/m3
UGE STEEL CO
Prim a
~9tpy
NATIONAL STEEL CORP
AQSite 26163
Annual DV 14 ug/m3
ily
GUARDIAN INDUSTRIES
Primary PM25 Annual Emission
124.9tpy
HOLCIM (US) INC.
Primary PM25 Annual Emission
656.5 tpy
J.R. WHITING CO
Primary PM25 Annual Emission
1029.2 tpy
DETROIT EDISON GREENWOOD ENERGY CENTER
Primary PM25 Annual Emission
87.6 tpy
AQSite 261150005
Annual DV 13.8 ug/m3
Daily DV 38 ug/m3
CARGILLSALT
Primary PM25 Annual Emission
80.5 tp
/ BELLE RIVER POWER PLANT
mary PM25 Annual Emission
1593.1 tpy
SIX S INC. PT4
Primary PM25 Annual Emission
64.9 tpy
AQSite 261630038
Annual DV 14.3 ug/m3
Daily DV 40 ug/m3
AQSite 261630015
Annual DV 15.5 ug/m3
Daily DV 40 ug/m3
AQSite 261630033
_Annual DV 17.2 ug/m3
Daily DV 43 ug/m3
MARATHON ASHLAND PETROLEUM
Primary PM25 Annual Emission
144.6 tpy
DETROIT EDISON RIVER ROUGE
Primary PM25 Annual Emission
378 tpy
AQSite 261630036
Annual DV 14.3 ug/m3
Daily DV 36 ug/m3
DETROIT EDISON TRENTON CHANh
Primary PM25 Annual Emission
1127.9 tpy
ROCKWOOD QUARRY
Primary PM25 Annual Emission
52.5 tpy
DETROIT EDISON/MONROE
rimaryPM25 Annual Emission if rimary F
2814.8 tpy ^2108.7 tp
K-7
-------
Figure K-6: Fresno Urban Study Area
-GOBAIN CONTAINERS,
Primary PM25 Annual E
51.4tpy
ADERA POWER, LLC
Primary PM25 Annual Emission
77.8 tpy
^77=^
AES MENDOTA, L.P.
rimarv PM25 Annual Emissio
^-sP"tP>t«i-r<' .^
QSite 060195001
Alikual-DVJl'6.
Daily^DV58|jg/m
AQSite 060195025
nnual DV 17.1 |jg/m3
Dat
K-8
-------
Figure K-7: Houston Urban Study Area
LEWIS CREEK^LANT AQSite 483390078 '
Primary PM25 4inual Emission *&,££&£ -, , ^ -, .._,_3 ^N
Primary PMJ25 AnnyJI Emi:
67.9 tpy
CHEWfafl CHEMICAL
Primary PM25 Aiual Emission
Daily-DV25|jg/m3
AQSite 48201 0024
ual DV13.1 ug/m3
Da ly DV 25 ug/m
iary^*M25 Annual Efnissior
29.9ti
Daily
CHANNELVIEW PL
Primary PM25 Ann
'
AQSite 482011034
MontgomeryJKfSnnual DV 13.9 ug/m3
ally DV28 |jg/m3
QUALITY" ELECTRIC ST
Primafv PM25\Annual
WN,OLEFINS PLANT
.25 Annual Emission
62.9 tpy
ON
Primary PW25 Annual Emissic
915.2 tpy
CEPAR'B'AYOU
afimary PM25 Annual Emisgj
164.2 tpy
J/ftBgBfe^t L*
Prima«)IPM25 Anniffl Emja#ior
.7 tpy
DEER PARK PLANT
_Primary PM25 Annual Emi
194.3 ta~
SAN JACINTO STEAM ELE STA
Primary PM25
ONGHORN GLASS
rimary PM25fAnnual Er
72.2 tpy
PASADENA PLAN
rimary PM25; Annual
COMPLEX
Annual Emissi
AQSite 48201
Annual DV15.8 ug
Daily DV 31 ug/m3
Harris
MAXWELL HOIISE DIVISION
Primary PM25 Ajnual Emission
73.1 tpy
DEEgJ*fiRK PLANT
ary PM25 Annual Emission
233.1 tpy
X PETROCHEMICALS LP
STATION Primary PM25 Annual Emission
M25 Annual Emission119-1 'PY
TEXAS CITY REFINERY
Primary PM25 Annual Emission
912.1 tpy
ON REFINERY
Primary PM25 Ann/al Emission1-YONgH_L-CJ
NNEL-ENERGY ENERGY
majyiJM25 Annual Emission
tpy
TEXAS CITY REFINERY
Primary PM25 Annual Emission
4.4 tpy
TEXAS CITY REFINERY
Primary PM25 Annual Emission
178.7 tpy
P AMOCO CHEMICAL TS(AS C VINYL ACETATE FACILITY 5
Prfrnary PM25 Annual Emissio
W A PARISH STATION
Primary PM25 Annual Emission
86.7 tpy
NYxCOMPLE
Primary PM55, Annual Emis
218.2 tpy
Primary PM25 Annual Emission
tpy
\
AQSite 481670014
imary PM25 Annual Emission
171.7 tpy
FREEPORTSITE D^^^'^T'"^'"''" Annual DV 10.5 |jg/m3
Primary PM25 Annual Emissiorgj1^ 25 Annual Em'SS'°n Daily DV 21 M9/m3
M CLEAR LAKE COGENERATION
25 Annual Emission
CHOCOLATE BAYOU PLNT
K-9
-------
Figure K-8: Los Angeles Urban Study Area
CONS. INC/ LI
PM25 Annual Emission
-ORMOND BEACH GEN STATION
nary PM25 Annual Emissh
77.7 tpy
/Ventura
AQSite 061113001
Annual DV 10.3 ug/m
Daily DV 25 ug/nf
AQSite^p61110007
AnnualDXjp.S^ug/m3,
f» A ^-.
Daily DV 24|jgfm3
VALLEY AGGREGATE N
Primary PM25 Annual E
RSIDE CEMENT COMPANY
ry PM25 Annual Emission
tpy
257.2 tpy
ite 06037903
-fAnnual DV 8.1 ug/m
Daily DV 17 ug/m3
AQSite 06037
Annual DV 14
EMEX - BLACK MOUNTAIN QUARR'
Primary PM25 Annual Emission
AQSitefoei
\
Annual DV 9.6 ug/m
Daily DV 21 ua/m
QSite 0607103
DX
DV1
HANSON AGGREG
JPrimary PM25
75.2 tpy
TAMCO
Primary PM25
AQSite 060370002
Annual DV 16 ug/m
Daily DV 47 ug/m3
aily DV 48 ug/n
AQSite 060371
TATION DISTRICT
Annual Emission
San Bernardino
Site 06071
Ijrf
Annual DV 18.
rimary PM25 Annual
260.2 tpy
AQSite 060718001
Annual DV 11.2 ug
Annual DV 18.4 ug/
Daily DV 52 ug/m
AQSite 060371602
DV 16.7 ug/m
Daily DV 49 ug/m
Da y DV 38
. i \
AQSite,0607l
Annual DV 17.
Daily DV 54 ug/m3
ALIFORNIA PORTLAND CEMENT
Primary PM25 Annual Emission
AQSite 060658001
Annual DV 19.6 ug/m3
r-=>
ily DV 55 ug/m3
AQSite 06065800
nnual DV 20.9 ug/m
DV 56 ug/m
UL
Primary FM
140.5tby
PRISES, LLC
AnnSaf Emission
C (NSR USE O
Annual Emissio
AN CLEMENTE ISLAND RANG
rimary PM25 Annual Emission
AQSite 0603740
Annual DV 14.9 u
AQSite 060651003
Annua
ily DV 49sijg/
V 17.7 ug/m3
Daily D,V 39 ug/m
AES ALAMITO
Primary PM25 A
243.4 tpy
AQSite 0605920
Annual DV 10.9 ug
D'^ tsssass mwx
JOHNS MANVILLE CORP MARCH AiR FORCE BAS
Primary PM25 Annual Emission Primary PM25 Annual Emis
221.1 tpy
WEST COAS^egpD
Primary PM25 Ann
72.4 tpy
AQSite 061112002 ^QSite.06l
Annual DV 11.1 ug/m3*"3'
Daily DV 29 ug/m3 Da.ly DV 34 ug/m
LA CITY, DWP SCATTERGOOD GENERATING S
Primary PM25 Annual Emission
50.8 tpy
EL SEGUNDO POWER, LLC
Primary PM25 Annual Emission
108.7 tpy
CHEVRON U.S.A. INC.
Primary PM25 Annual Emission
223 tpy
AES REDONDO BEACH, LLC
Primary PM25 Annual Emissi
MOBIL OIL CORP
Primary PM25 Annual Emission
119 tpy
EQUILOI
Primary PJ
89 tpy
TOSCO REFINING COMPANY
Primary PM25 Annual Emission
140.5 toy
AQSite 060374004
Annual DV 14.3 ug/m3
Daily DV 36 ua/m3
\f sx
AQSite 060590007
Aniiihu DV 14.3 ug/m3
Daily DV 42 ug/m3
K-10
-------
Figure K-9: New York Urban Study Area
AQSite361191002
Annual DV 13 ug/m3 Annual DV 11.7 |jg/m3
Daily DV 35 ug/m3 DajL DV 33 ug/m
AQSite 340310005
Annual DV12.9ug/
/ i'ff\ 1 P. v
Daily DV 37 ug/m'
QSite 340030003
Annual DV 13.2 ug/m3
Daily DV 38 ug/m3 AQSite 360050080
' Annual DV 15.5 |jg/m3
AQSite 340130015
nnual DV 13.3 ug/m3
Daily DV 38 ug/m3
IAGENERATI
Primary PM25 Annual Emission
254.9 t
ite 360050110
Annual DV 13,ug/m3
Daily DV 36,j!jg/m3
Daily DV38 ug/m
QSite 360610079
ug/rn
AQSite 361030001
nnual DV12 |jg/m3
ily DV34 ug/m3
Annual DV 1
Daily DV 36 ug/
Annual DV 15.4 |jg/m3
Daily V 41 Mg/m3
AQSite 36061005
Annual DV 15.9 ug/m
9/uMm
AQSite 361030002
Annual DV 10.6 ug/m
Annu'aUDVil 3r3Tug/n)'
Daily DV 38 ug/m
Daily DV 30 ug/m3
AQSite 360810124
Annual DV11.8ugf "
Hudson Generatm
imary PM25 Ann
1846.1 tpy
AQSite,360590008
AnnUialDV11.4 ug/m3
Daily DV 32 ug/m3
e3402
nual DV 11.3 ug/m
V32 ugVm3
AQSite 340390004
Annual DV 14.4 |jg/m
Daily~DV42 |jg/
Annual DV 14\|jg/m3
QSite"360470122
Anrtual DV
Daily DV-36 ug/m
Daily DV 38TJgVm
Bayway KSTinery
mary PM25 A(inual Emission
147.9 tpy
A,
>V Annual DV 14 8 ug/m3 Pnma\/PM25 Annual Emlbsjon
f Hnnuai uv it.o ng;m \ "*?
Dail>DV 37 |jg/m3 80.1 tpy^
AQSite 360610062
DDM
Primary PM25 Ann
70 8 tpy
Site 340392003
nnual DV 13 ug/m
Daily DV 36
EF BARRETT POWER STATION
Primary PM25 Annual Emission
CON ED-74TH STREET STA
Primary PM25 Annual Emission
RAVENSWOOD GENERATING STATION
ry PM25 Annual Emission
269.7~
CON ED-WATERSIDE STATION
ary PM25 Annual Emission
3BUHOUUOO AQSite 3'6B850067 AQSiteX360610134 AQSite 360610128 \
I DV 13.2 ug/m3 Annua, DV^5 Mg/m3 Annua| DV 13.3 ug/m3Annua, DV 15.3 ^.^^B^^ GB^RAT,NG STAT
Daily DV 34,ug/m3^ Dal|y DV 31 Mg/fi Dai|y py 37 Ug/m3 Daily DV 38 Mg/m3 96 8 ^
K-ll
-------
Figure K-10: Philadelphia Urban Study Area
EXELON GENERATION CO/CROMBY GENERATION STATION
AQSite 340210008
^Annual DV 12.5
Daily DV 34 ug/m
AL CHEM CORP/POTT
imary PM25 Annual Emission
.2 toy
cer GeneFattncfStatio
Primary PM25 Annual Emissi
Primary PM25 Annual Emissi
.6 tpy
AQSite 420170012/
Annual DV^3.2 Mg/m3 AQSite 421010024
Daily DV 35 uglm
Annual DV 12.5 Mg/
Annual DV 12.7 ug/m3
*Sy\J~^
Daily DV '
Daily DV 33 Mg/
e421010004
Annual DV 13.8 Mg/m3 AQSite 340071007
aily DV 37 Mg/m3 V^nnual DV 13.5
^vC
SUNOCO CHEMICALS'(FORMER ALLI ED SIG
PrimaryfJMSS'Annual Emission
P
QSite 421010020
nnual DV 15.
AQSite 421010057 AQSite^21010047
Annual DV 12 Mg/m3 Annual DV
>>rxV\^r^^r>_:i.. r>i i ^l
QSite 42101013
AQSite 340155001
EXELON &ENERATIO
Annual DV 12.2 M9/m3 Primary
Daily DV 29 Mg/m3 400.4 tr^
Delaware
UNOCO INC. (R
imary PM25 Ann
212.1 tpy
efining Co.- N.J.
Site 420450002 /^SS^^
nnual DV 15 Mg/m3 > CO^TER
Daily DV 36 Mg/m' <miuy PM25^-al Emlss^n
ER
rimary PM
7.8 tpy
CONOCOP
Primary PM
SUNOCO INC (R&M)/MARCUS
imary PM25 Annual Emission
Site 42029010
Annual DV,14.2 M9/m3
''
100031003
Annual DV 13.4 Mg/m3
Dailv DV 33 ""'m3
QSite 100032004
Annual DV 14.7 M9/m3
NERATING PLANT
Deepwater Generating Static
imary PM25 Ah^Tal Emissio
K-12
-------
Figure K-ll: Phoenix Urban Study Area
AQSite 040139997
QSite 040137020
Annual DV7.9 ug/m3
NvDV15ug
AQSite 0401300119
Daily DV 19 ug/m
AQSite 040134003
Annua
Asarco Ray Mi
Primary PM25 Annu
99-K^PYjL
RCO - RAY COMfLEX HAYC
ry PM25 Annual/Emission
y. /(
QSite 0402130,13
K-13
-------
Figure K-12: Pittsburgh Urban Study Area
,QSite 420031008
ALLEGHENY LUDLUM CORP - BRACKENRIDGE
Primary PM25 Annual Emission J^XW
ite 420070014
Annual DV 16.5 ug
Daily DV 43 ug/m3
QSite 420030093 Annual DV 15 ug/m3
nual DV 13 ug/m3 Daily DV 40 ug/m3
JtsJ A~>c
Daily DV 40 ug/m3
RBILIANT ENERG
mary PM25 An
t
AES BEAVER VALLEY LLC/BEAVER VALLY COGE
Primary fiM25 Annual Emission
417.8 tpy
ite 420030116
Annual DV 16.1 ug/m
Daily DV 39 ug/m3
RION POWER MIDWBST, CHESWICK STATI
INC.
mary PM25 Annual Emission
GLASS COMPANY, INC.
25 Annual Emission
AQSite 4200300
Annual DV
[QSite 420030008
QSite'421255001 Daily DV 36 ug/m
3
Daily DV40.Mg/m3
AQSite 42003002
Annual DV 12.9 ug/m
W. H. SAMMIS PLANT \ Daily DV 35 ug/ /
rimary PM25 Arnual Emission
QSite 42003130
AQSite 421290008
nnual DV 15.
L CORPORATI
JPrimary PM|5 Annual Emission
36101
US STEEL CORPORATION - IRVIN PLANT
rimary PM25 Annual Emi
63:
nnualDV19.8Mg/m3
aily DV 60 |jg/m3
Brooke
WHEELING1PITTSBUR
imary PM^S An
23.7tpy
NAQSite420033007;
USS^etAlETONWORKS
Primary PM25 AnriDal-Etnissio
394.3 tpy
TjrKxr/"^
Annual DV 15.3 |jg/m3
AQSite 420030133
Annual DV 14 ug/m3
Daily-DV 28 ug/m3
NG COMPANY)
I I-
I EmissionAQSjte 421250200
Annual DV 14.6 Mg/m
aily DV 35 u
K-14
-------
Figure K-13 Salt Lake City Urban Study Area
^°°^Sf^
-Annual DV8.2j|5g/m3
Production
Primary PM
65.2 tpy
AQSite 490571003
Annual DV9.1 ug/m
AQSite 490570002
\nnual
Daily DV 36 ug/m
AQSite 49057000
Annual DV 8.9 u
Daily DV 31 ug/m
QSite 490351001
Smelter & Refinery
Primary PM25 Annua
,,,,
AQSite 490110004'
I
Annual DV 10 ug/m"
Annual DV 8.9 ug/m
Daily DV 32 ug/m
Power Pit/ Lab/Tailings Impound
M25 Annual Emissi
Annual DV 14.4 ug/m3
Daily DV 55 ug/m3 Salt Lake City Refi
* Primary PM25 Ann
QSite 490450003
Annual DV/7.6 ug/
Daily DV'sYjjgYrn?
AQSite 490353010 '0.7 tpy
Annual DV 10.5 ug/m3
Daily DV 30 ug/m
AQSite 4^90353006
Annual E)V 10.7 ug/m
Daily DV 48 ua/m
Mine & Copperton Concentrator-
Primary PM25 Annual Emi
AQSite 490353007
Annual DV 11.6 ug/m3
Salt Lake Valley Landfill ^Transfer Station
rimary PM25 Annual Emission
U.S. Army-Dugway Proving Gro
Primary PM25 Annual Emission
174.7tpyA.
Site 490353008
nnual DV 7.8 ug/m
Western Fiberglass - Salt Lake City Pla
Primary PM25 Annual Emission
Daily DV 47 ug/m
K-15
-------
Figure K-14 St. Louis Urban Study Area
DYNO HO BEL INC
Primary PM25
83 .3 tpy
ADUALON DIV OF HERCULES INC-MISSOURI CHEMICAL WORKS
Priman/ PM25 Annual Emission
,109.5 tpy JTJtn-
AMEREN ENERGY GENERATING CO
Priman7>M2 5 Annual Em ission_£19 ^ PM25 Ann ljal ^'^^
1 175.2 ti* : ' TT53TO tPV
AM EREHUE-S10 UX P LANT
HOLCM(US)INC-
Primary PM25 Annual Emission
101.1 toy
AQSite 295100085
Annu^ DV14.5Mg/m=
Daily BV34ug/m
YNE6YMIDWEST GENERATION INC
ry PM2S Annual Emission
72.8 tpy
'TOSCOPETRO CfoRP
Primary PM25 Annual Emission
1353.1 tpy |
AQSite 235100086
I
-vf *rSf-*v>£
AQSite 295100093
Annua DV 15.5 LQ/m3
Daily DV 35 ug/m
ASF-KEYSTONE IN
Primary PM25
5 3 tpy
CAf-J COMMERCIAL TERMINAL
n ual D V 14 .b ug.1 m
Daily DV35
AMgBUSen C OMM ERCIAL TE R\fl
nman/Protifi Annual Emission
AQSite 291892003
.Annual DV 13.4
NATIBNAL STEEL CORP -GRANITE CITiYDIV
Primaru' PM26 Annual Emission
WAS H IN GT 0 N U NMME Dl C.AL S CH 0 0 L- B 01LER P LAN!
Primary PM25 AnnuaJEmissio
98tPV < »
Daily DV 34 Mgi-.:
UMAX FOILS INC-.4LCOAFOILPRODUCTS/AUJMAX
Primary PM25 Annual Emission
AMERENUEMER.AV1EC '
Prima ry PfulSS Arm ual Emission LL-tt CONTINENTALT
36S.8tpy I SAJNT.COBAJN CONTAINERS LLC-PEVELY tfrimary Plw?5Ani
Primary PM25 Annual Emission | 546 tpy
71.1tpy DYNBGYMIDWEST GENERATION INC
nrnary PM25 Annual Emission
1705.4 tpy
flERENUE-LARA ME PLAf-JT
man/ PM25 Anm 3\ Emission
6.9 tpy
Site 291890004
Hpnual DV12.6u^/m5 Franklin
Daily DV31 vgtm]
AQSite29099iH1S
OMPANYINC-RTvER CEM
AMERENUE-RUSH laLAfJD PLANT
rimary PM25 Annual Emission
riman/PM25 Annual Emissio
465.7 tpy
K-16
-------
Figure K-15 Tacoma Urban Study Area
SIMPSON TIMBER CO
Primary PM25 Annual Emission
137 tpy
Jefferson
Samt-Gobam Containers me
Primary PM25 Annual Emission
R CO
irrjary PM25 Annual Emissio
57.1
Graymont Western US Inc
Primary PM25 Annual Emis/fon
53.5 tpy
WE^rERHAEUSER CO.
PrimaVv PM25 Annual Emission
113.
GRAYS HARBOR PAPER LP
Primary PM25 Annual Emission
174.6 r. ._,..,
Grays Harbo
SIMPSON TACOMA KR
Primary PM25 Annual Emiss
79.3 tpy^'
~
AQSite 530530029
Annual DV 10.2 ug/m3
Daily DV 43 ug/m3
ransAlta Centralia Mining, LLC TransAlta Centralia Generation
Primary PM25 Annual Emission Primary PM25 Annual Emission
Primary PM25 Annual E
K-17
-------
-------
United States Office of Air Quality Planning and Standards Publication No.
Environmental Protection Health and Environmental Effects Division EPA-452/R-10-005
Agency Research Triangle Park, NC June, 2010
-------
-------
United States Office of Air Quality Planning and Standards Publication No.
Environmental Protection Health and Environmental Effects Division EPA-452/R-10-005
Agency Research Triangle Park, NC June, 2010
------- |