Risk Assessment to Support the Review of
the PM Primary National Ambient Air
Quality Standards

External Review Draft
September 2009

-------
                                    DISCLAIMER
       This draft document has been prepared by staff from the Ambient Standards
Group, Office of Air Quality Planning and Standards, U.S. Environmental Protection
Agency. Any opinions, findings, conclusions, or recommendations are those of the
authors and do not necessarily reflect the views of the EPA. This document is being
circulated to obtain review and comment from the Clean Air Scientific Advisory
Committee (CASAC) and the general public. Comments on this draft document should
be addressed to Zachary Pekar, U.S. Environmental Protection Agency, Office of Air
Quality Planning and Standards, C504-06, Research Triangle Park, North Carolina 27711
(email: pekar.zachary@epa.gov).
       Elements of this report (see Acknowledgements below) have been provided to the
U.S. Environmental Protection Agency (EPA) by Abt Associates, Inc. in partial
fulfillment of Contract No. EP-D-08-100, Work Assignment 0-02. Any opinions,
findings, conclusions, or recommendations are those of the authors and do not necessarily
reflect the views of the EPA or Abt Associates, Inc. Any analyses, interpretations, or
conclusions presented in this report based on emergency department, hospitalization and
mortality baseline incidence data obtained from outside sources, are credited to the
authors and not the institutions providing the raw data.
                              ACKNOWLEDGEMENTS

       In addition to EPA staff, personnel from Abt Associates, Inc. contributed to the
writing of this document.  Specific chapters and appendices where Abt Associates, Inc.
made contributions include: chapter 3 (Overview of Methods Used in Estimating Risk for
the Urban Case Studies), and appendices A (Air Quality Assessment), C
(Epidemiological - Epi - Study Specific Information On PM2.5), E (Risk Analysis - Core
Analysis), and F (Sensitivity Analysis Results).

-------
                                                       EPA 450/P-09-006
                                                         September 2009
Risk Assessment to Support the Review of the PM Primary National Ambient Air
             Quality Standards - External Review Draft
                   US Environmental Protection Agency
                       Office of Air and Radiation
               Office of Air Quality Planning and Standards
                Health and Environmental Impacts Division
                        Ambient Standards Group
               Research Triangle Park, North Carolina 27711
                               11

-------
                                Table of Contents

List of Acronyms/Abbreviations	ix
1   INTRODUCTION	1
    1.1     Background	3
    1.2     Current Health Risk Assessment: Goals and Planned Approach	6
    1.3     Organization of Document	7
2   SCOPE	8
    2.1     Overview of the PM NAAQS risk assessment from the last review	9
    2.2     Original assessment plan	11
      2.2.1     Risk assessment	11
      2.2.2     Population exposure analysis	13
    2.3     CASAC comments provided on the Scope and Methods Plan	13
    2.4     Current scope and key design elements	15
      2.4.1     Risk Assessment	15
      2.4.2     Population exposure analysis	17
    2.5     Alternative Standard Levels Included in the Risk Assessment	18
3   METHODS USED IN URBAN CASE STUDY ANALYSIS	22
    3.1     General Approach	22
      3.1.1     Basic Structure of the Risk Assessment	22
      3.1.2     Calculating PM-Related Health Effects Incidence	29
         3.1.2.1     General approach	29
         3.1.2.2     Short-term vs. long-term exposure	30
         3.1.2.3     Calculating annual incidence	31
    3.2     Air Quality Inputs	32
      3.2.1     Characterizing recent conditions	32
      3.2.2     Estimating policy relevant background	34
      3.2.3     Simulating air quality to just meet current and alternative standards.. 34
    3.3     Selection of model inputs	38
      3.3.1     Health endpoints	38
      3.3.2     Selection and delineation of urban study areas	41
      3.3.3     Selection of epidemiological studies and concentration-response
                (C-R) functions within those studies	45
                                       in

-------
       3.3.4     A summary of selected health endpoints, urban areas, studies, and
                 concentration-response (C-R) functions used in the risk assessment.. 50
    3.4    Baseline Health Effects Incidence Data	63
       3.4.1      Data sources	63
          3.4.1.1     Mortality	63
          3.4.1.2     Hospital admission and emergency department visits	63
          3.4.1.3     Populations	65
       3.4.2     Calculation of baseline incidence rates	68
    3.5    Addressing Uncertainty and Variability	72
       3.5.1      Overview	72
       3.5.2     Key sources of variability	75
       3.5.3      Qualitative assessment of uncertainty	77
       3.5.4     Single and multi-factor sensitivity analyses	85
       3.5.5      Summary of approach to addressing variability and uncertainty	89
4   RESULTS	91
    4.1    Assessment of health risk associated with recent conditions (core
           analysis)	96
       4.1.1      Long-term mortality	96
       4.1.2     Short-term mortality	100
       4.1.3      Short-term morbidity	102
    4.2    Assessment of health risk associated with Just meeting the current and
           alternative standards (core analysis)	103
       4.2.1      Long-term mortality	105
       4.2.2     Short-term mortality	117
       4.2.3      Short-term morbidity	123
    4.3    Sensitivity analysis results	127
       4.3.1      Single-factor sensitivity analysis results	132
          4.3.1.1     Sensitivity analysis results associated with long-term exposure
                     mortality	132
          4.3.1.2     Sensitivity analysis results associated with short-term
                     mortality	134
          4.3.1.3     Sensitivity analysis results associated with short-term
                     morbidity	136
          4.3.1.4     Single-factor sensitivity analysis addressing model selection,
                     lags, and co-pollutant models	137
       4.3.2     Multi-Factor Sensitivity Analysis	138
                                         IV

-------
         4.3.2.1     Multi-factor sensitivity analyses - long-term exposure
                    mortality	139
         4.3.2.2     Multi-factor sensitivity analyses - short-term exposure
                    mortality	140
    4.4    Evaluating the representativeness of the urban study areas in the national
           context	141
    4.5    Overall Summary andKey Observations	158
       4.5.1     Core risk results from the recent conditions, current NAAQS, and
                alternative NAAQS analyses	159
       4.5.2     Use of the reasonable set of alternative risk estimates generated
                through the sensitivity analysis	162
       4.5.3     Representativeness of the urban study areas in the national context
                based on consideration for coverage of PM2.5 risk-related
                parameters	166
       4.5.4     Use of the results of the national-scale long-term mortality analysis
                to evaluate degree of coverage of the 15 urban study areas for
                national-distribution of risk	166
5   NATIONAL-SCALE ASSESSMENT OF LONG-TERM  MORTALITY
    RELATED TO PM2.5 Exposure	168
    5.1    Overview	168
    5.2    Methods	169
       5.2.1     Population Estimates	169
       5.2.2     Population Exposure	169
       5.2.3     Premature Mortality Estimates	171
    5.3    Results	171
REFERENCES	175
Appendix A. Air Quality Assessment
Appendix B. Hybrid (Non-Proportional) Rollback Approach
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)
Appendix F. Sensitivity Analysis Results
Appendix G. Supplement to the National-Scale Assessment of Long-Term
             Mortality Related to PM2.s  Exposure

-------
                                  List of Tables
Table 3-

Table 3-

Table 3-

Table 3-
Table 3-

Table 3

Table 3

Table 3

Table 3

Table 3-
Table 3-

Table 3-


Table3-

Table 4-
Table 4-
       •1. Numbers of Monitors and Numbers of Missing Days at Composite Monitors
          in Risk Assessment Locations from 2005 Through 2007.                 33
       •2  Regional Policy-Relevant Background Estimates Used in the Risk
          Assessment.                                                        34
       •3. EPA Design Values for Annual and \24-hour PM2 5 Standards for the Period
          2005-2007.                                                         36
       •4. Urban Study Areas Selected for the Risk Assessment.                    43
       •5. Locations, Health Endpoints, and Short-Term Exposure Studies Included in
          the PM2.5 Risk Assessment                                          51
       -6. Locations, Health Endpoints, and Long-Term Exposure Studies Included in
          the PM2.5 Risk Assessment                                          52
       -7. Summary of Locations, Health Endpoints, Studies and Concentration-
          Response Functions Included in the Core Analysis.                      53
          Summary of Locations, Health Endpoints, Studies and Concentration-
          Response Functions Included in   Sensitivity Analyses.                  61
       -9. Sources of Hospital Admissions (HA) and Emergency Department (ED)
          Visit Data.                                                          64
        10.Relevant Population Sizes for PM Risk Assessment Locations.            66
        11.Baseline Mortality Rates (Deaths per 100,000 Relevant Population per Year)
          for 2006 for PM Risk Assessment Locations.                           69
        12.Baseline Hospital Admission (HA) and Emergency Department (ED) Rates
          (Admissions/Visits per 100,000 Relevant Population per Year) for 2007 for
          PM Risk Assessment Locations.                                       71
        13. Summary of Qualitative Uncertainty Analysis of Key Modeling Elements in
          the PM NAAQS Risk  Assessment.                                  79
        1. Overview of Sensitivity Analysis Results.                             130
        2. Data Sources for PM NAAQS Risk Assessment Risk Distribution Analysis.
                                                                            143
Table 4-3. Summary Statistics for Selected PM Risk Attributes.                    146
Table 4-4  Results of Kolomogrov-Smirnoff Tests for Equality Between National  and
          Urban Study Area Distributions for Selected National Risk   Characteristic
          Variables                                                         150
Table 4-5  Derivation of a set of reasonable alternative risk estimates to supplement the
          core risk estimates (Los Angeles, current standards, for long-term all cause
          mortality).                                                         164
Table 5-1  Estimated PM2.s-related premature mortality associated with   incremental
          air quality differences between 2005 ambient mean  pm2.5 levels and lowest
          measured level from the epidemiology studies  or policy relevant
          background (90th percentile confidence interval)                        172
                                       VI

-------
                                  List of Figures
Figure 3-1. Major components of particulate matter health risk assessment.            24
Figure 3-2. Flow diagram of risk assessment for short-term exposure studies.         27
Figure 3-3. Flow diagram of risk assessment for long-term exposure studies.          28
Figure 3-4 15 urban study areas included in the risk assessment (including seven PM
          regions used to guide selection of study areas).                          44
Figure 4-1. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in All Cause Mortality Associated with Long- Term Exposure
          to PM2.5 (Exposure Period: 1979 - 1983):  Based on 2007 Air Quality Data.
                                                                             109
Figure 4-2. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in All Cause Mortality Associated with Long-Term Exposure to
          PM2.5 (Exposure Period: 1999 -    2000): Based on 2007 Air Quality Data.
                                                                             110
Figure 4-3. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in Ischemic Heart Disease Mortality Associated with Long-Term
          Exposure to PM2.5 (Exposure Period: 1979 - 1983): Based on 2007 Air
          Quality Data.                                                       111
Figure 4-4. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in Ischemic Heart Disease Mortality Associated with Long-Term
          Exposure to PM2.5 (Exposure Period: 1999 - 2000):  Based on 2007 Air
          Quality Data.                                                       112
Figure 4-5. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in Cardiopulmonary Disease Mortality Associated with Long-
          Term Exposure to PM2.5 (Exposure Period: 1979 - 1983): Based  on 2007 Air
          Quality Data.                                                       113
Figure 4-6. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in Cardiopulmonary Disease Mortality Associated with Long-
          Term Exposure to PM2.5 (Exposure Period: 1999 - 2000): Based  on 2007 Air
          Quality Data.                                                       114
Figure 4-7. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in Lung Cancer Mortality Associated with Long-Term Exposure
          to PM2.5 (Exposure Period: 1979 - 1983): Based on 2007 Air Quality Data.
                                                                             115
Figure 4-8. Estimated Percent Reductions From the Current Standards to Alternative Set
          of Standards in Lung Cancer Mortality Associated with Long-Term Exposure
          to PM2.5 (Exposure Period: 1999 - 2000): Based on 2007 Air Quality Data.
                                                                             116
Figure 4-9. Estimated Percent Reductions From the Current Standards to Alternative Sets
          of Standards in Non-Accidental Mortality Associated with Short-Term
          Exposure to PM2.5:  Based on 2007 Air Quality Data.                   120
Figure 4-10.Estimated Percent Reductions From the Current Standards to Alternative
          Sets of Standards in Cardiovascular Mortality Associated with Short-Term
          Exposure to PM2.5: Based on 2007 Air Quality Data.                    121
                                       vn

-------
Figure 4-11. Estimated Percent Reductions From the Current Standards to Alternative
           Sets of Standards in Respiratory Mortality Associated with Short-Term
           Exposure to PM2.5: Based on 2007 Air Quality Data.                    122
Figure 4-12. Estimated Percent Reductions From the Current Standards to Alternative
           Sets of Standards in Cardiovascular Hospital Admissions Associated with
           Short-Term Exposure to PM2.5: Based on 2007 Air Quality Data.         125
Figure 4-13. Estimated Percent Reductions From the Current Standards to Alternative
           Sets of Standards in Respiratory Hospital Admissions Associated with Short-
           Term Exposure to PM2 5: Based on 2007 Air Quality Data.               126
Figure 4-14. Comparison of distributions for key elements of the risk equation: total
           population.                                                          151
Figure 4-15. Comparison of distributions for key elements of the risk equation: 98th
           percentile 24-hour average PM2.5                                       152
Figure 4-16. Comparison of distributions for key elements of the risk equation: all cause
           mortality rate.                                                        153
Figure 4-17. Comparison of distributions for key elements of the risk equation:
           mortality risk effect estimate from Zanobetti and Schwartz (2008).        154
Figure 4-18. Comparison of distributions for selected variables expected to influence the
           relative risk from PM2.s: long term average July  emperature.            155
Figure 4-19. Comparison of distributions for selected variables expected to influence the
           relative risk from PM2.5: percent of population 65 and older.              156
Figure 4-20. Comparison of distributions for selected variables expected to influence the
           relative risk from PM2 5: per capita annual personal income.              157
Figure 4-21. Comparison of distributions for selected variables expected to influence the
           relative risk from PM2.5: cardiovascular disease prevalence days.         158
Figure 5-1  Conceptual diagram of data inputs and outputs for national long-term
           mortality risk assessment                                              169
Figure 5-2  2005 fused surface baseline PM2 5 concentrations                       170
Figure 5-3  Percentage of premature mortality attributable to PM2.5 exposure at various
           2005 annual average PM2 5 levels                                      173
Figure 5-4  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                       174
                                        Vlll

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
A/C
ACS
Act
AMI
AQS
P
BenMAP
BMI
BRFSS
CASAC
CAA
CBS A
CDC
CDF
CFR
CHD
CMAQ
CO
COPD
CPD
C-R
 CSA
CV
CVD
df
ED
ER
EPA
FACA
FIPS
GAM
GEOS-CHEM
GLMs
HA
                     LIST OF ACRONYMS/ABBREVIATIONS

                         Air conditioning
                         American Cancer Society
                         Clean Air Act
                         Acute Myocardial Infarction
                         EPA's Air Quality System
                         Slope coefficient
                         Benefits Mapping Analysis Program
                         Body Mass Index
                         Behavioral Risk Factor Surveillance System
                         Clean Air Scientific Advisory Committee
                         Clean Air Act
                         Core-based Statistical Area
                         Centers for Disease Control
                         Cumulative Distribution Function
                         Code of Federal Regulations
                         Coronary Heart Disease
                         Community Multiscale Air Quality
                         Carbon Monoxide
                         Chronic Obstructive Pulmonary Disease
                         Cardio-pulmonary Disease
                         Concentration-response
                         Consolidated Statistical Area
                         Cardiovascular
                         Cardiovascular Disease
                         Degrees of freedom
                         Emergency Department
                         Emergency Room
                         United States Environmental Protection Agency
                         Federal Advisory Committee Act
                         Federal Information Processing System
                         Generalized additive model
                         Goddard Earth Observing  System-Chemical Model
                         Generalized linear model
                         Hospital Admissions
September 2009
                                             IX
                                            Draft - Do Not Quote or Cite

-------
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
HCUP
HEI
HS
ICD
IHD
INF
IRP
ISA
KB
km
K-S
LML
MCAPS
MSA
NA
NAAQS
NCEA
NEI
NCHS
NMMAPS
NOx
03
OAQPS
PA
PM
PMX









PM10



                           Healthcare Cost and Utilization Project
                           Health Effects Institute
                           High School
                           International Classification of Diseases
                           ischemic heart disease
                           Influence of uncertainty on risk estimates
                           Integrated Review Plan
                           Integrated Science Assessment Document
                           Knowledge Base
                           Kilometer
                           Kolmogorov-Smirnov
                           Lowest Measured Level
                           Medicare Air Pollution Study
                           Metropolitan  Statistical Area
                           Not Applicable
                           National Ambient Air Quality Standards
                           National Center for Environmental Assessment
                           National Emissions Inventory
                           National Center for Health Statistics
                           National Morbidity, Mortality, and Air Pollution Study
                           Nitrogen oxides
                           Ozone
                           Office of Air  Quality Planning and Standards
                           Policy Assessment Document
                           Particulate Matter
                           The legal definition for PMX, as defined in the Code of Federal
                           Regulations, includes both a 50% cut-point and a penetration
                           curve.  A 50% cut-point of X um diameter means that 50% of
                           particles with aerodynamic diameter of X are removed by the inlet
                           and 50% pass through the inlet and are collected on the filter.
                           Depending on the specific penetration curve specified, particles
                           larger than X um aerodynamic diameter are collected with an
                           efficiently than decreases rapidly for particles larger than X while
                           the collection efficiency for particles smaller than X increases
                           rapidly with decreasing size until 100 % efficiency is reached.
                           Particles with a 50% upper cut-point of 10± 0.5 um aerodynamic
                           diameter and  a penetration curve as specified in the Code of
                           Federal Regulations.
September 2009
Draft - Do Not Quote or Cite

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
PM2.5


PM10.2.5


PRB
RA
RR
REA
SAB
SEDD
SID
S02
sox
SES
TRIM
TRIM.Risk
UFP
USDA
VNA
WHI
WHO
ZCA
                          Particles with a 50% upper cut-point of 2.5 um aerodynamic
                          diameter and a penetration curve as specified in the Code of
                          Federal Regulations.
                          Particles with a 50% upper cut-point of 10 um aerodynamic
                          diameter and a lower 50% cut-point of 2.5 um aerodynamic
                          diameter.
                          Policy-Relevant Background
                          Risk Assessment Document
                          Relative risk
                          Risk and Exposure Assessment
                          Science Advisory Board
                          State Emergency Department Databases
                          State Inpatient Database
                          Sulfur Dioxide
                          Sulfur Oxides
                          Socio-economic Status
                          Total Risk Integrated Methodology
                          Total Risk Integrated Methodology - Risk Assessment component
                          Ultrafine Particles
                          U.S. Department of Agriculture
                          Voronoi Neighbor Averaging
                          Women's Health Initiative
                          World Health Organization
                          Zip Code Area
September 2009
XI
Draft - Do Not Quote or Cite

-------
 1
 2                                       1   INTRODUCTION

 3           The U.S. Environmental Protection Agency (EPA) is presently conducting a review of
 4    the national ambient air quality standards (NAAQS) for particulate matter (PM). Sections 108
 5    and 109 of the Clean Air Act (Act) govern the establishment and periodic review of the NAAQS.
 6    These standards are established for pollutants that may reasonably be anticipated to endanger
 7    public health and welfare, and whose presence in the ambient air results from numerous or
 8    diverse mobile or stationary sources. The NAAQS are to be based on air quality criteria, which
 9    are to accurately reflect the latest scientific knowledge useful in indicating the kind and extent of
10    identifiable effects on public health or welfare that may be expected from the presence of the
11    pollutant in ambient air.  The EPA Administrator is to promulgate and periodically review, at
12    five-year intervals, "primary" (health-based) and "secondary" (welfare-based) NAAQS for such
13    pollutants. Based on periodic reviews of the air quality criteria and standards, the Administrator
14    is to make revisions in the criteria and standards, and promulgate any new standards, as may be
15    appropriate.  The Act also requires that an independent scientific review committee advise the
16    Administrator as part of this NAAQS review process, a function performed by the Clean Air
17    Scientific Advisory Committee (CASAC).l
18           The current NAAQS for PM include a suite of standards to provide protection for
19    exposures to fine and coarse particles using PM2.5 and PMio, as indicators, respectively (71 FR
20    61144, October 17, 2006).  With regard to the primary and secondary standards for fine particles,
21    in 2006 EPA revised the level of the 24-hour PM2.5 standard to 35 ug/m3 (calculated as a 3-year
22    average of the 98th percentile of 24-hour concentrations at each population-oriented monitor),
23    retained the level of the annual PM2.5 annual standard at 15  ug/m3 (calculated as the 3-year
24    average of the weighted annual mean PM2 5 concentrations from single or multiple  community -
25    oriented monitors), and revised the form of the annual PM2.5 standard by narrowing the
26    constraints on the optional use of spatial averaging.2  With regard to the primary and secondary
      1 The Clean Air Scientific Advisory Committee (CAS AC) was established under section 109(d)(2) of the Clean Air
      Act (CAA or Act) (42 U.S.C. 7409) as an independent scientific advisory committee. CASAC provides advice,
      information and recommendations on the scientific and technical aspects of air quality criteria and NAAQS under
      sections 108 and 109 of the CAA. The CASAC is a Federal advisory committee chartered under the Federal
      Advisory Committee Act (FACA).  See
      http://yosemite.epa.gov/sab/sabpeople.nsfAVebComniitteesSubcomniittees/CASAC%20Particulate%20Matter%20R
      eview%20Panel for a list of the CASAC PM Panel members and current advisory activities.
       In the revisions to the PM NAAQS finalized in 2006, EPA tightened the constraints on the spatial averaging
      criteria by further limiting the conditions under which some areas may average measurements from multiple
      community-oriented monitors to determine compliance (see 71 FR 61165-61167, October 17, 2006).
      September 2009                              1          Draft - Do Not Quote or Cite

-------
 1    standards for PMio, EPA retained the 24-hour PMio standard at 150 ug/m3 (not to be exceeded
 2    more than once per year on average over 3 years) and revoked the annual standard because
 3    available evidence generally did not suggest a link between long-term exposure to current
 4    ambient levels of coarse particles and health or welfare effects. These standards were based
 5    primarily on a large body of epidemiological evidence relating ambient PM concentrations to
 6    various adverse health endpoints. Secondary standards for PM2.5 and PMio were revised to be
 7    identical to the primary standards.
 8          The next periodic review of the PM NAAQS is now underway.3  The EPA outlined the
 9    science-policy questions that will frame this  review, outlined the process and schedule that the
10    review will follow, and provided more  complete descriptions of the purpose, contents, and
11    approach for developing the key documents for this review in the Integrated Review Plan for the
12    National Ambient Air Quality Standards for Particulate Matter, henceforth referred as the IRP
13    (EPA, 2008a).4 The EPA is currently completing the process of assessing the latest available
14    policy-relevant scientific information to inform the review of the PM standards. The  latest draft
15    of this assessment is contained in the Integrated Science Assessment for P articulate Matter:
16    Second External Review Draft, henceforth referred to as the draft ISA (EPA, 2009a) which was
17    released in July 2009 for review by  the CAS AC and for public comment.  The draft ISA includes
18    an evaluation of the scientific evidence on the health effects of PM, including information on
19    exposure, physiological mechanisms by which PM might damage human health, and an
20    evaluation of the epidemiological evidence including information on reported concentration-
21    response (C-R) relationships and lag structures for PM-related morbidity and mortality
22    associations, including consideration of effects in at-risk populations..
23           Drawing from the health effects evidence presented in the draft ISA as well as CASAC
24    advice (Samet, 2009) and public comments on &Particulate Matter National Ambient Air
25    Quality Standards: Scope and Methods Plan for Health Risk and Exposure Assessment,
26    henceforth referred to as the Scope and Methods Plan (EPA, 2009b), EPA's Office of Air
27    Quality Planning and Standards (OAQPS) has developed this draft Risk Assessment (RA)
28    describing the quantitative assessments being conducted by the Agency to support the review of
29    the primary PM standards.  This draft document is a concise presentation of the scope, methods,
30    key results, observations, and related uncertainties associated with the quantitative analyses
      3 See http://www.epa.gov/ttn/naaqs/standards/pm/sjm index.html for more information on the current and
      previous PM NAAQS reviews.
      4 On November 30, 2007, EPA held a consultation with the CASAC on the draft IRP (72 FR 63177, November 8,
      2007). Public comments were also requested on the draft plan and presented at that CASAC teleconference.  The
      final IRP incorporated comments received from CASAC and the general public on the draft plan as well as input
      from senior Agency managers.

      September 2009                             2          Draft - Do Not Quote or Cite

-------
 1    performed.  Revisions to this draft RA will draw upon the final ISA and will reflect
 2    consideration of CASAC and public comments on this draft.
 3          The final ISA and final RA will inform the policy assessment and rulemaking steps that
 4    will lead to final decisions on the primary PM NAAQS. A Policy Assessment (PA) is now being
 5    prepared by OAQPS staff to provide a transparent staff analysis of the scientific basis for
 6    alternative policy options for consideration by senior EPA management prior to rulemaking.  The
 7    PA is intended to help "bridge the gap" between the Agency's scientific assessments, presented
 8    in the ISA and RA, and the judgments required of the Administrator in determining whether it is
 9    appropriate to retain or revise the standards.  The PA will integrate and interpret information
10    from the ISA and the RA to frame policy options and to facilitate CASAC's advice to the
11    Agency and recommendations on any new standards or revisions to existing standards as may be
12    appropriate, as provided for in the Clean Air Act. A preliminary draft PA is planned for release
13    in September 2009 to facilitate discussion on the overall structure, areas of focus, and level of
14    detail to be included in an external review draft of the document, which EPA plans to release for
15    CASAC  review and public comment later this year. A discussion of the preliminary draft PA
16    with CASAC will be held in conjunction with CASAC review and public comment of the draft
17    ISA, this draft RA, and a draft assessment document that will inform the review of the secondary
18    PM standards - Paniculate Matter Urban-Focused Visibility Assessment - External Review
19    Draft (EPA, 2009c).

20         1.1  BACKGROUND
21          As part of the last PM NAAQS review completed in 2006, EPA's OAQPS conducted a
22    quantitative risk assessment to estimate risks of various health effects associated with exposure
23    to ambient PM2.5 and PMio-2.5 in a number of urban study areas selected to illustrate the public
24    health impacts of these pollutants (U.S. EPA, 2005, Chapter 4; Abt Associates, 2005). The
25    assessment scope and methodology were developed with considerable input from the CASAC
26    Review Panel and the public, with CASAC concluding that the general assessment methodology
27    and framework were appropriate (Hopke, 2002).  In addition, the final risk assessment took into
28    consideration CASAC advice (Hopke, 2004; Henderson, 2005) and public comments on two
29    drafts of the risk assessment.
30          The extensive assessment conducted for fine particles in the last review included
31    estimates of risks of mortality (total non-accidental, cardiovascular, and respiratory), morbidity
32    (hospital admissions for cardiovascular and respiratory causes), and respiratory symptoms (not
33    requiring hospitalization) associated with recent short-term (daily) ambient PM2 5 levels and risks
34    of total, cardiopulmonary, and lung cancer mortality associated with long-term exposure to PM2.5
35    in nine urban study areas. The risk assessment included estimates of: (1) risks of mortality,


      September 2009                            3          Draft - Do Not Quote or Cite

-------
 1    morbidity, and symptoms associated with recent ambient PM2.5 levels; (2) risk reductions and
 2    remaining risks associated with just meeting the existing suite of PM2.5 NAAQS (1997
 3    standards); and (3) risk reductions and remaining risks associated with just meeting various
 4    alternative PM2.5 standards.
 5           The quantitative risk assessment conducted in the last review for thoracic coarse particles
 6    was much more limited than the analyses conducted for fine particles. Estimates of hospital
 7    admissions attributable to short-term exposure to PMi0-2.5 were developed for Detroit
 8    (cardiovascular and respiratory admissions) and Seattle (respiratory admissions), and estimates
 9    of respiratory symptoms were developed for St. Louis.  While one of the goals of the PMio-2.5
10    risk assessment was to provide estimates of the risk reductions associated with just meeting
11    alternative PMi0-2.5 standards, EPA staff concluded that the nature and magnitude of the
12    uncertainties and concerns associated with this portion of the risk assessment weighed against
13    use of these risk estimates as a basis for recommending specific standard levels (U.S. EPA, 2005,
14    p. 5-69).
15           Prior to the issuance of a proposed rulemaking in the last review, CASAC presented
16    recommendations to the Administrator supporting revisions of the PM2.5 primary standards.
17    These recommendations placed substantial reliance on the results of the  quantitative risk
18    assessment (Henderson, 2005, pp 6-7). In a letter to the Administrator following the 2006
19    proposed rule (71 FR 12592, January 17, 2006), CASAC requested reconsideration of the
20    Agency's proposed decisions and reiterated and elaborated on the scientific bases for its earlier
21    recommendations which included placing greater weight on the result of the Agency's risk
22    assessment.  With regard to the quantitative risk assessment, CASAC concluded, "While the risk
23    assessment is subject to uncertainties, most of the PM Panel found EPA's risk assessment to be
24    of sufficient  quality to inform  its recommendations." (Henderson, 2006a, p. 3).
25           In the 2006 final rule, the EPA Administrator recognized that the quantitative risk
26    assessment for fine particles was based upon a more extensive body of data and was more
27    comprehensive in scope than the previous assessment conducted for the  review completed in
28    1997. However, as presented in the final rulemaking notice, the Administrator was mindful of
29    significant uncertainties associated with the risk estimates for fine particles. More specifically,
30
31           Such uncertainties generally related to a lack of clear understanding of a number of
32           important factors, including, for example, the shape of the concentration-response
33           functions, particularly when, as here, effect thresholds can neither be discerned nor
34           determined not to exist; issues related to selection of appropriate statistical models for the
35           analysis of the epidemiologic data; the role of potentially confounding and modifying
36           factors in the concentration-response relationships; issues related to simulating how PM2.5
37           air quality distributions will  likely change in any given area upon attaining a particular
38           standard, since strategies to reduce emissions are not yet defined; and whether there

      September 2009                              4          Draft - Do Not Quote or Cite

-------
 1           would be differential reductions in the many components within PM2.5 and, if so, whether
 2           this would result in differential reductions in risk. In the case of fine particles, the
 3           Administrator recognized that for purposes of developing quantitative risk estimates,
 4           such uncertainties are likely to [be] amplified by the complexity in the composition of the
 5           mix of fine particles generally present in the ambient air. (72 FR 61168, October 17,
 6           2006).
 7
 8    As a result, the Administrator viewed that the quantitative risk assessment provided supporting
 9    evidence for the conclusion that there was a need to revise the PM2.5 primary standards, but he
10    judged that the assessment did not provide an appropriate basis to determine the level of the
11    standards (72 FR 61168, October 17, 2006).
12           In a letter to the EPA Administrator following the issuance of the final rule, CASAC
13    expressed "serious scientific concerns" regarding the final PM standards.  In particular, CASAC
14    was concerned that the Agency "did not accept our finding that the annual PM2.5 standard was
15    not protective of human health and did not follow our recommendation for a change in that
16    standard" (Henderson  et al, 2006b, p. 1). With respect to the use of the risk assessment to inform
17    EPA's decision on the primary PM2.5 standard, CASAC stated, "While there is uncertainty
18    associated with the risk assessment for the PM2.5 standard, this very uncertainty suggests a need
19    for a prudent approach to providing an adequate margin of safety" (Henderson et al., 2006b, p.2)
20           Several parties filed petitions for review following promulgation of the revised PM
21    NAAQS in 2006.  These petitions for review addressed the following issues with regard to the
22    primary PM NAAQS:  (1) selecting the level of the annual primary PM2.5 standard, (2) retaining
23    PMio as the indicator for coarse particles and retaining  the level and form of the 24-hour PMi0
24    standard, and (3) revoking the PMio annual standard. On judicial review, the D.C. Circuit
25    remanded the annual primary PM2.5 NAAQS  to EPA because the Agency failed to adequately
26    explain why the standard  provided the requisite protection from both short- and long-term
27    exposures to fine particles including protection for at-risk populations. The court upheld the
28    Agency's use of the quantitative risk assessment to inform the decision to revise the PM2.5
29    standards but not to inform the selection of level.5  The court also upheld the decision to retain
30    the 24-hour PMio standard and revoke the annual PMio standard. American Farm Bureau
31    Federation v.  EPA, 559 F. 3d 512, (D.C. Cir. 2009).
32
33
      5 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).
      September 2009                              5          Draft - Do Not Quote or Cite

-------
 1         1.2  CURRENT HEALTH RISK ASSESSMENT: GOALS AND PLANNED
 2              APPROACH
 3          The goals of the current risk assessment remain largely the same as those articulated in
 4    the risk assessment conducted as part of the last review. These goals include: (a) to provide
 5    estimates of the potential magnitude of premature mortality and/or selected morbidity effects in
 6    the population associated with recent ambient levels of PM and with just meeting the current
 7    suite of PM standards and any alternative standards that might be considered in selected urban
 8    study areas, including, where data are available, consideration of impacts on at-risk populations;
 9    (b) to develop a better understanding of the influence of various inputs and assumptions on the
10    risk estimates to more clearly differentiate alternative standards that might be considered
11    including potential impacts on various at-risk populations; and (c) to gain insights into the
12    distribution of risks and patterns  of risk reduction and uncertainties in those risk estimates. In
13    addition, EPA is conducting an assessment to provide nationwide estimates of the potential
14    magnitude of premature mortality associated with long-term exposure to ambient PM2.5 to more
15    broadly characterize this risk on a national scale and to support the interpretation of the more
16    detailed risk estimates generated for selected urban study areas. The overall scope and design of
17    the risk assessment reflect efforts to achieve these goals.
18          The current risk assessment builds on the approach used and lessons learned in the last
19    PM NAAQS risk assessment and attempts to reduce overall uncertainty associated with the
20    analysis through incorporation of a number of enhancements, in terms of both the methods and
21    data used in the analyses. In preparing the Scope and Methods Plan for the health risk/exposure
22    assessment, EPA considered the  scientific evidence presented in the first draft ISA (EPA, 2008b)
23    and the key science policy issues raised in the IRP (EPA, 2008a). The EPA held a consultation
24    with CASAC to solicit comments on the Scope and Methods Plan during an April 2009 CASAC
25    meeting.  Public comments were also requested (74 FR 11580, March 18, 2009). CASAC
26    (Samet, 2009) and public comments were considered in advance of the conduct of the analyses
27    and results presented in this draft REA. The design of the current risk assessment builds upon
28    information presented in the draft ISA (EPA, 2009b) with particular emphasis on conclusions
29    regarding causality determinations for specific PM-related health effect categories and discussion
30    of the scientific strengths and weaknesses underlying key epidemiological studies addressing
31    specific health effect endpoints of interest.
32          The risk assessment described in this draft document covers a variety of health endpoints
33    for which there is adequate information to develop quantitative risk estimates. Evidence of
34    relationships between PM and other health endpoints for which there currently is insufficient
35    information to develop quantitative risk estimates will be discussed in the OAQPS staff Policy
36    Assessment.


      September 2009                             6          Draft - Do Not Quote or Cite

-------
 1         1.3  ORGANIZATION OF DOCUMENT
 2          The remainder of this document is organized as follows.  Chapter 2 provides an overview
 3   of the scope of the risk assessment, including a summary of the previous risk assessment (section
 4   2.1), the planned approach as presented in the Scope and Methods Plan (section 2.2), a summary
 5   of CASAC comments on the Scope and Methods Plan (section 2.3) and how these comments as
 6   well as public comments were addressed in the design of the analyses (section 2.4), and
 7   summary of the  alternative levels evaluated in the risk assessment including the rationale for
 8   their selection (section 2.5).  Chapter 3 describes the analytical approach, methods, and data used
 9   in conducting the risk assessment. This includes a description of the approach used to generate
10   risk estimates for the set of urban case studies included in this analysis, as well as the approaches
11   used in addressing variability and uncertainty as part of the analysis (Appendices A, B, and C
12   provide supplemental information regarding the data and methods used in the analysis). Chapter
13   4 presents the risk estimates generated for the urban case studies, including key observations
14   resulting from review and interpretation of the results (Appendices E and F provide detailed risk
15   estimates and sensitivity analysis results, respectively).  Chapter  5 presents the approach used
16   and results of a national-scale assessment of PM2.s-related long-term mortality risks (Appendix G
17   provides supplemental information to the national-scale mortality analysis). In addition,
18   Appendix D provides supplemental information to a representativeness analysis completed for
19   the 15 urban  study areas (see Section 2.4.1).
      September 2009                             7          Draft - Do Not Quote or Cite

-------
 1                                          2   SCOPE

 2          This chapter provides an overview of the scope and key design elements of the PM
 3    risk assessment being conducted for this review, including the process that has been followed
 4    to design the analyses. Following initiation of the current PM NAAQS review, we began the
 5    design of this risk assessment by reviewing the risk assessment completed during the previous
 6    PM NAAQS review (Abt Associates, 2005; EPA, 2005, chapter 4) with an emphasis on
 7    considering key limitations and sources of uncertainty recognized in that analysis.
 8    Furthermore, as an initial step in the overall PM NAAQS review, EPA invited a wide range of
 9    external experts as well as EPA staff, representing a variety of areas of expertise (e.g.,
10    epidemiology, human and animal toxicology, statistics, risk/exposure analysis, atmospheric
11    science) to participate in a workshop titled, "Workshop to Discuss Policy-Relevant Science to
12    Inform EPA's  Integrated Plan for the Review of the Primary PM NAAQS" (72 FR 34003,
13    June 20, 2007). This workshop provided an opportunity for the participants to broadly
14    discuss the key policy-relevant issues around which EPA would structure the PM NAAQS
15    review and to discuss the most meaningful new science that would be available to inform our
16    understanding of these issues. One session of this workshop was centered around planning
17    for the quantitative risk/exposure assessments.  Specifically,  the discussions focused on the
18    extent to which new research and/or improved methodologies were available to inform how
19    EPA designed a quantitative risk assessment and whether it was appropriate to conduct a
20    quantitative exposure assessment, and, if so, how that assessment might be structured.
21          Based in part on these workshop discussions, EPA developed a draft IRP outlining
22    the schedule, the process, and the key policy-relevant science issues that would guide the
23    evaluation of the air quality criteria for PM and the  review of the primary and secondary PM
24    NAAQS including initial thoughts for conducting quantitative assessments (EPA,  2007,
25    chapter 5). On November 30, 2007, CAS AC held a teleconference with EPA to provide its
26    comments on the draft IRP (72 FR 63177, November 8, 2007). Public comments were also
27    presented at that teleconference. A final IRP incorporating comments received from CASAC
28    and the general public on the draft plan was issued in March 2008 (EPA, 2008a).
29          As a next step in the design of the quantitative assessments, EPA developed a  planning
30    document outlining the initial design for the PM NAAQS risk assessment - Paniculate Matter
31    National Ambient Air Quality Standards: Scope and Methods Plan for Health Risk and
32    Exposure Assessment, henceforth Scope and Methods Plan (EPA, 2009b). This planning
33    document was released for CASAC consultation and public review in February 2009. Based
34    on consideration of CASAC and public comments on that Scope and Methods Plan, along

    September 2009                            8          Draft - Do Not Quote or Cite

-------
 1    with ongoing review of the latest PM-related literature, we made modifications to the scope
 2    and design of the risk assessment and completed our initial analyses.
 3          In presenting the scope and key design elements of the current risk assessment, this
 4    chapter first provides a brief overview of the risk assessment completed for the previous PM
 5    NAAQS review in section 2.1, including key limitations and uncertainties associated with that
 6    analysis.  Section 2.2 provides a summary of the initial design of the risk assessment as
 7    outlined in the Scope and Methods Plan. Next, section 2.3 summarizes comments received
 8    during the CASAC consultation on the Scope and Methods Plan.  Key design elements for the
 9    current risk assessment, including modifications made to the overall scope of the assessments
10    relative to the initial Scope and Methods Plan and explanations to support these changes are
11    outlined in section 2.4. Finally, section 2.5 provides a summary of the alternative air quality
12    scenarios modeled in the current assessment, including the rationale behind selection of
13    specific alternative levels.

14         2.1   OVERVIEW OF THE PM NAAQS RISK ASSESSMENT FROM THE
15              LAST RE VIEW
16          The PM NAAQS risk assessment from the last review completed  in 2006 included a
17    broad assessment of PM2.5-related risk and a more limited treatment of PMi0-2.5-related risk.
18    The assessment conducted for the review completed in 2006 included estimates of risks of
19    mortality  (total non-accidental,  cardiovascular, and respiratory), morbidity (hospital
20    admissions for cardiovascular and respiratory causes), and respiratory symptoms (not
21    requiring  hospitalization) associated with recent short-term (daily) ambient PM2 5 levels and
22    risks of total, cardiopulmonary, and lung cancer mortality associated with long-term exposure
23    to PM2.5 in selected urban areas. Nine urban areas were included in this assessment to provide
24    some sense of the variability in the PM2 5-related risk estimates across the U.S.  The areas
25    evaluated were: Boston, MA; Detroit, MI; Los Angeles, CA; Philadelphia, PA; Phoenix, AZ;
26    Pittsburgh, PA; San Jose, CA; Seattle, WA; and St. Louis, MO.
27          The EPA recognized that there were many sources of uncertainty  and variability
28    inherent in the inputs to this assessment and that there was a high degree of uncertainty in the
29    resulting PM2.5 risk estimates. Such uncertainties generally  related to a lack of clear
30    understanding of a number of important factors, including, for example: (a) the shape of the
31    concentration-response (C-R) function (and whether or not a population threshold exists); (b)
32    issues related to the selection of appropriate statistical models for the analysis of
33    epidemiological data; (c) the role of potentially confounding and modifying factors in the C-R
34    relationships; (d) the method for simulating how daily PM2.5 ambient concentrations would
35    likely change in any given area upon meeting a particular suite of standards, since strategies
36    to reduce  emissions had not yet been defined; and (e) the issue of whether there would be


      September 2009                            9            Draft - Do Not Quote or Cite

-------
 1    differential reductions in the many components within PM2.5 and, if so, whether this would
 2    result in differential reductions in risk.
 3          While some of these uncertainties were addressed quantitatively in the form of
 4    estimated confidence ranges around central risk estimates, other uncertainties and the
 5    variability in key inputs were not reflected in these confidence ranges, but rather were
 6    addressed through separate sensitivity analyses or characterized qualitatively (EPA, 2005,
 7    chapter 4; Abt Associates, 2005). The C-R relationships used in the quantitative risk
 8    assessment were based on findings from human epidemiological studies that relied on fixed-
 9    site, population oriented, ambient monitors as a surrogate for actual ambient PM2.5 exposures.
10    The assessment included a series of base case estimates that, for example, included various
11    cutpoints intended as surrogates for alternative potential population thresholds.  Other
12    uncertainties were addressed in various sensitivity analyses  (e.g., the use of single- versus
13    multi-pollutant models, use of single versus multi-city models, use of a distributed lag model)
14    and had a more moderate and often variable impact on the risk estimates in some or all of the
15    cities.
16          The general overview and discussion of key components of the quantitative risk
17    assessment used to develop risk estimates for PM2.5 presented above is also applicable to the
18    risk assessment conducted for PMio-2.5 as part of the last review. However, the scope of the
19    risk assessment for PMi0-2.5 was much more limited than that for PM2 5 reflecting the much
20    more limited body of epidemiological evidence and air quality information available for
21    PMio-2.5. As discussed in section 4.5 of the Staff Paper (EPA, 2005), the PMi0-2.5 risk
22    assessment included risk estimates for just three urban areas for two categories of health
23    endpoints related to short-term exposure to PMio-2.5: hospital admissions for cardiovascular
24    and respiratory causes and respiratory symptoms (see also Abt, 2005, chapter 9).
25          Estimates of hospital admissions attributable to short-term exposure to PMio-2.5 were
26    developed for Detroit, MI (cardiovascular and respiratory admissions) and Seattle, WA
27    (respiratory admissions), and estimates of respiratory symptoms were developed for St. Louis,
28    MO. While one of the goals of the PMio-2.5 risk assessment was to provide estimates of the
29    risk reductions associated with just meeting alternative PMio-2.5 standards, EPA staff
30    concluded that the nature and magnitude of the uncertainties and concerns associated with this
31    portion of the risk assessment weighed against use of these risk estimates as a basis for
32    recommending specific standard levels (EPA, 2005, see p. 5-69). These uncertainties and
33    concerns were summarized in the proposal notice (see FR 71 2662, January 17, 2006) and
34    discussed more fully in the Staff Paper (EPA, 2005, chapter 4) and associated technical
35    support document (Abt Associates, 2005).
36
      September 2009                            10           Draft - Do Not Quote or Cite

-------
 1         2.2   ORIGINAL ASSESSMENT PLAN
 2           As noted earlier, the Scope and Methods Plan reflected consideration of the design of
 3    the risk assessment completed for the last review (specifically the uncertainties and
 4    limitations associated with that assessment) as well as more recent PM-related research
 5    published since completion of the last assessment. The Scope and Methods Plan outlined a
 6    planned approach for conducting the current PM risk assessment, including broader design
 7    issues as well as more detailed aspects of the analyses. The Scope and Methods Plan also
 8    outlined plans for a population exposure analysis based on micro-environmental exposure
 9    modeling. The planned approaches for conducting both analyses are briefly summarized
10    below.

11         2.2.1  Risk assessment
12           Key design elements for the risk assessment, as presented in the Scope and Methods
13    Plan included:6

14           •  PM size fractions: We planned to focus primarily on estimating risk associated
15              with exposure to PM2.5 with a more limited assessment of PMio-2.5. Regarding PM
16              components and ultrafine particles, we concluded that, based on review of
17              evidence in the first draft ISA, there was insufficient data to support quantitative
18              risk assessment at this time.

19           •  Selection of health effects categories (PM2.s): We planned to focus on categories
20              for which the evidence supports a judgment that there is at  least a likely causal
21              relationship.  However, consideration would be given to expanding the risk
22              assessment to cover additional categories for which evidence supports a judgment
23              that there is a suggestive causal relationship (e.g., reproductive, developmental
24              outcomes), if sufficient information was available to develop risk estimates for
25              these additional categories.

26           •  Selection of health effect categories (PMi0-2.5): We planned to build on the
27              limited risk assessment conducted in the last review (EPA,  2005) with a focus on
28              health effect categories that staff judged to be sufficiently suggestive of a causal
29              relationship with short-term exposure to warrant analysis.

30           •  Selection of study areas:  We planned to expand the number of urban study areas
31              to between 15 and 20, with selection of these study areas being based on
32              consideration of a number of factors (e.g., availability of location-specific C-R
33              functions and baseline incidence data, coverage  for geographic heterogeneity in
34              PM risk-related attributes, coverage for areas with more vulnerable populations).
35              We also discussed the possibility of including more refined risk assessments for
36              locations where more detailed exposure studies had been completed (e.g.,  L.A.,
      6 We have focused here on highlighting design details that have broader implications for the risk assessment and
      have not included some of the more detailed aspects of the planned approach. These more detailed factors were
      carefully considered in conducting the risk assessment described in this document, but they are not discussed as
      part of this discussion focusing on the overall scope of the analysis.


      September 2009                            11           Draft - Do Not Quote or Cite

-------
 1              where a zip code level analysis of long-term PM2.-exposure related mortality was
 2              presented in Krewski et al., 2009).

 3           •   Simulation of air quality levels that just meet either current or alternative
 4              suites of standards: We planned to consider the use of non-proportional air
 5              quality adjustment methods, together with the proportional approach that has been
 6              used previously. These non-proportional adjustment methods could be based on
 7              (a) historical patterns of reductions in urban areas, if these result in support for
 8              non-proportional reductions across monitors and/or (b) model-based (e.g.,  CMAQ)
 9              rollback designed to more realistically reflect patterns of PM reductions across
10              monitors in an urban area.

11           •   Characterization of policy relevant background (PRB): We planned to  use
12              modeling (combination of the global-scale circulation model, GEOS-Chem, with
13              the regional scale air quality model, CMAQ) as presented in the first draft  PM
14              ISA, rather than empirical data to characterize PRB levels for use in the risk
15              assessment model.

16           •   Selection of epidemiological studies to provide C-R functions: We planned to
17              include both multi- and single-city  studies (given advantages associated with both
18              designs) as well as multi- and single-pollutant studies.  However, we also proposed
19              placing greater weight on the use of C-R functions reflecting adjusted single-city
20              estimates obtained from multi-city  studies.

21           •   Shape of the functional form of the risk model: We planned to emphasize non-
22              threshold C-R functions in the risk assessment model, based on the first draft ISA
23              conclusion that there was little support in the literature for population thresholds
24              for mortality effects associated with either long-term or short-term PM2.5 ambient
25              concentrations.7 However, we stated that we may consider population thresholds
26              as part of the sensitivity analysis.

27           •    Modeling of risk down to PRB versus Lowest Measured Level (LML): We
28              planned to model risk down to LML for estimating risk associated with long-term
29              PM2.5 exposures and down to PRB  for estimated risks associated with short-term
30              PM2.5 exposure effects.

31           •   Characterization of uncertainty and variability: We planned to include  a
32              discussion in the risk assessment report on the degree to which the risk assessment
33              covers key sources of variability related to PM risk.  For uncertainty, we planned
34              to include a qualitative discussion of key sources of uncertainly and provide
35              ratings (low, medium and high) in terms of their potential impact on risk estimates.
36              We also described the use of sensitivity analysis methods planned both to
37              characterize the potential impact of sources of uncertainty on risk estimates and to
38              provide an alternative set of reasonable estimates to supplement the main ("core")
39              set of risk estimates generated for the urban study areas.
      7 Note, that the draft ISA in discussing short-term exposure mortality studies while indicating support for no-
      threshold log-linear models, acknowledges the "possible influence of exposure error and heterogeneity of shapes
      across cities remains to be resolved" (draft ISA, section 6.5.2.7).


      September 2009                           12           Draft - Do Not Quote or Cite

-------
 1          •   National-scale health impact analysis: We planned to conduct a national-scale
 2              health impact assessment focused on mortality associated with long-term exposure
 3              to PM2.s using a recent conditions scenario.

 4          •   Representativeness analysis for the urban study areas: We planned to conduct
 5              an analysis to evaluate the representativeness of the selected urban study areas
 6              against national distributions for key PM risk-related attributes to determine
 7              whether they are nationally-representative, or more focused on a particular portion
 8              of the distribution for a given parameter.

 9         2.2.2   Population exposure analysis
10           The Scope and Methods Plan also described a population exposure analysis based on
11    micro-environmental exposure modeling using the Air Pollution Exposure Model (APEX).
12    The planned analysis would have focused on PM2 5 and have involved a subset of the urban
13    study areas included in the risk assessment.  The results of this analysis were planned to focus
14    on providing insights on population exposure with respect to informing the interpretation of
15    available epidemiological studies. For reasons presented below in section 2.4.2, this analysis
16    will not be completed as part of the current PM NAAQS review. We have decided to
17    continue development of the population exposure analysis methodology with the goal of
18    considering any results from this exposure assessment in support of the next PM NAAQS
19    review.

20         2.3   CASAC COMMENTS PROVIDED ON THE SCOPE AND METHODS
21              PLAN
22          CASAC met on April 2, 2009 to conduct a consultation on the Scope and Methods
23    Plan for the PM NAAQS risk assessment. Following that meeting, CASAC provided
24    comments from the Panel, summarized below, as well as more detailed comments providing
25    individual views from CASAC PM Panel Members (Samet, 2009).8

26          •   Regarding the overall analysis, CASAC suggested that "priorities be established
27              quickly in developing the health risk and exposure assessment, giving emphasis to
28              those analyses that may be most informative for establishing particulate matter
29              standards."

30          •   With regard to the selection of health effects endpoints, CASAC recommended
31              that EPA "provide a transparent algorithm for selecting endpoints based on the
32              level of certainty and the relative and attributable risks." Furthermore, CASAC
33              suggested that "weight be given to the level of classification while still considering
34              the Administrator's obligation to set a standard with a 'margin of safety' as
35              described in the Clean Air Act."  By way of example, the letter stated that "several
36              CASAC members do not recommend a risk assessment based  on birth outcomes,
      8 See
      http://yosemite.epa.gov/sab/sabproduct.nsf/4620a620d0120f93852572410080d786/350899ecl345529485257466
      00691 de5! OpenDocument&TableRow=2.0#2.


      September 2009                           13           Draft - Do Not Quote or Cite

-------
 1              in part because the level of evidence is still at the suggestive level."  The letter
 2              went on to note that "one panel member proposed setting a higher priority for
 3              those health effects shown to have the highest risks in the epidemiological
 4              literature." (Samet, 2009, p. 1-2).

 5          •   Regarding the inclusion of PMio-2.5 in the risk assessment, the letter stateed that
 6              "there was support (among CASAC members) for doing a limited risk assessment
 7              for short-term exposure to PMi0-2.5 for appropriate outcomes such as
 8              hospitalization" (Samet, 2009, p.2).

 9          •   With regard to aspects of the plan related to air quality characterization,
10              particularly characterizing policy relevant background (PRB) and simulation of
11              ambient air levels associated with alternative standards, the letter stated that
12              "CASAC generally supports EPA's proposed approach for estimating PRB
13              levels." Furthermore, there was general support for the proportional rollback
14              approach that EPA had followed in previous risk assessments. Individual
15              comments did include support for considering a non-proportional rollback
16              approach as an alternative to the proportional approach although several Panel
17              members noted that there could be considerable uncertainty associated with the
18              non-proportional  approach.

19          •   With regard to the risk assessment component, CASAC generally agreed with the
20              planned approach to identifying C-R relationships.

21          •   There was also strong support for the planned national scale health impact
22              assessment for long-term exposure mortality related to PM2.5.  In fact, CASAC
23              stated that it believed such a national  assessment "should play a central role in the
24              overall risk assessment" (Samet, 2009, p.2).

25          •   Regarding the characterization of uncertainty, CASAC expressed support for the
26              general approach, but did emphasize the need to carefully separate sensitivity
27              analyses from uncertainty analyses. With regard to the approach for classifying
28              the degree of uncertainty, CASAC suggested that EPA explore the use of "various
29              structured approaches for describing uncertainty," noting that "recent examples
30              may be found in the work of the World  Health Organization and the
31              Intergovernmental Panel on  Climate Change" (Samet, 2009, p.2).

32          •   Regarding the population exposure analysis, CASAC welcomed its inclusion
33              noting that the planned analysis "rightly seeks to identify various personal and
34              building-related factors that may account for some of the variability in PM2.5-
35              related health risks" (Samet, 2009, p.2). However, CASAC also noted that "more
36              information is needed on how the results from the exposure assessment will be
37              integrated and used to interpret epidemiological  studies" (Samet, 2009, p.2). In
38              pointing out potential benefits of the analysis, CASAC also acknowledged that the
39              "Agency's time and resources are not unlimited" (Samet, 2009, p.2).
40          Comments from the CASAC PM Panel and individual Panel members were carefully
41    considered in finalizing the scope and methods for the risk assessment described in this
42    document.
      September 2009                           14           Draft - Do Not Quote or Cite

-------
 1         2.4   CURRENT SCOPE AND KEY DESIGN ELEMENTS
 2           The current scope and design of the risk assessment is based on consideration of the
 3    following: (1) lessons learned from the risk assessment conducted for the previous review, as
 4    summarized in section 2.1 (see also EPA, 2005, chapter 4;  Abt Associates, 2005); (2)
 5    consideration of CASAC's advice on the Scope and Methods Plan, as described in section
 6    2.3; and (3) public comments on the Scope and Methods Plan; and consideration of the new
 7    scientific evidence presented in the second draft ISA (EPA, 2009a). These considerations
 8    also led to our decision to continue development of the population exposure analysis
 9    methodology, rather than applying it for this review. Key design elements of the risk
10    assessment, as well as the rationale for the decision to  continue development of the population
11    exposure analysis for consideration in the next PM NAAQS review are presented below.

12         2.4.1  Risk Assessment
13            Key design elements, along with the rationale for any differences between the design
14    of the risk assessment as implemented and the approach described in the Scope and Methods
15    Plan, include:

16           •   PM size fractions: The risk assessment characterizes risk associated with PM2.5-
17              related exposures only.  Careful consideration of evidence provided in the draft
18              ISA regarding health effects potentially associated with short-term exposure to
19              PMio-2.5 as well as limited air quality data has resulted in the decision not to
20              quantitatively assess risk for this size fraction as part of the current  risk assessment
21              (see section 3.3.1 for additional discussion). Furthermore, EPA staff have
22              determined that data are too limited, at this time, to support a quantitative risk
23              assessment for specific PM components, including ultrafine particles (UFPs).  We
24              note, however, that the evidence for health  effects associated with thoracic coarse
25              particles, UFPs, and PM components will be addressed as part of the evidence-
26              based analysis that will  be presented in the  forthcoming draft PA.

27           •   Selection of health effects categories (PM2.s): The final set of health effects
28              categories included in the risk assessment for PM2.5 (see section 3.3.1) are
29              consistent with those outlined in the Scope  and  Methods plan for PM2.5 (i.e., those
30              classified as having a causal or likely causal relationship with PM2.5 exposure, as
31              presented in the draft ISA).  However, we decided not to include any of the health
32              effect categories classified as suggestive of a casual relationship in the draft ISA,
33              based on a number of considerations including: (1) CASAC Panel member views,
34              which did not express strong support for inclusion of these less-well supported
35              health effect categories  (see section 1.3); (2) limited information available to
36              support selection of C-R functions for specific endpoints within these health effect
37              categories; and/or (3) lack of available baseline incidence data for these other
38              health effect endpoints.

39           •   Selection of health effect categories (PMi0-2.5): As noted above, we have decided
40              not to model risk related to PMio-2.5 exposure (see section 3.3.1).
      September 2009                            15           Draft - Do Not Quote or Cite

-------
 1           •   Selection of urban study areas: We have included 15 urban study areas in the
 2               risk assessment, with the selection of these areas being based on a number of
 3               criteria including: (a) consideration of urban study areas evaluated in the risk
 4               assessment conducted to inform the previous PM NAAQS review; (b)
 5               consideration of locations evaluated in key epidemiological studies; (c) preference
 6               for locations with relatively elevated 24-hour and/or annual PM2.5 monitored levels
 7               so that the assessment can provide potential insights into the degree of risk
 8               reduction associated with alternative standard levels and (d) desire to include
 9               locations that would provide coverage for different regions across the country
10               where these regions are defined to reflect potential differences in PM sources,
11               composition and potentially other factors which might impact PM-related risk ( see
12               section 3.3.2).9  We note that, due to the time and resource limitations, we have
13               not included a specialized analysis of risk based on epidemiology studies using
14               more highly-refined exposure analysis (e.g.,  the study of L.A. involving zip code-
15               level effect estimates as presented in Krewski et al., 2009).  However, we have
16               included consideration of studies with more refined surrogate measures of
17               exposure in our discussion of uncertainty related to long-term mortality, since they
18               can inform our interpretation of the degree of potential bias associated with the
19               effect estimates used to model risks (see section 3.5.3).

20           •   Simulation of air quality levels that just meet either current or alternative
21               standard levels: For this analysis, we used a proportional rollback approach as the
22               basis for simulating current and alternative standard levels for the core risk
23               estimates that were generated.10  However, as part of the sensitivity analysis, we
24               also included application of a hybrid (non-proportional) adjustment procedure,
25               which simulated a combination of regional and local controls (see section 3.2.3).

26           •   Characterization of PRB: Consistent with the planned approach described in the
27               Scope and Methods Plan, we used regional PRB  estimates generated using a
28               combination of GEOS-Chem and CMAQ modeling (these estimates were obtained
29               directly from the assessments prepared for and summarized in the draft ISA - see
30               section 3.2.2).

31           •   Selection of epidemiological studies to provide C-R functions:  In line with the
32               planned approach outlined in the Scope and Methods Plan, in modeling risk
33               associated with short-term PM2.5 exposures, we focused on two large multi-city
34               studies based on our conclusion that these studies provided more defensible effect
35               estimates (see section 3.3.1 for additional details).  C-R functions selected from
      9 An error was identified in the approach used to simulate ambient PM2 5 levels just for the Pittsburgh study area
      for the scenarios involving just meeting the current and alternative sets of standards. This impacts risk estimates
      generated for these air quality scenarios, as well as sensitivity analysis results involving this urban study area.
      While we have removed discussion of these risk estimates (and sensitivity analysis results) form the body of this
      report, there was insufficient time after identifying this error to either generate corrected risk estimates or remove
      the erroneous risk estimates from the summary tables (presented in Appendix E and F). We will correct this
      error and release updated results for the Pittsburgh study area as soon as is practicable and will include the
      corrected results in the next version of this document.
      10 As described in section 3.1, the risk assessment includes a set of core risk estimates based on application of
      model inputs having the greatest support in the literature. The analysis also includes a reasonable alternative set
      of risk estimates generated as part of the sensitivity analysis, where these estimates, while not having as much
      support as the core risk results, are still based on inputs having a reasonable degree of support in the literature.


      September 2009                            16            Draft - Do Not Quote or Cite

-------
 1              several single city studies were also included in our analysis to provide coverage
 2              for additional health effect endpoints associated with short-term PM2 5 exposures
 3              (e-g-, emergency department visits). Modeling of long-term exposure-related
 4              mortality focused on the latest reanalysis of the ACS dataset (Krewski et al.,
 5              2009).  This study was published after the Scope and Methods Plan was the subject
 6              of a CAS AC consultation and public review and,  therefore, was not included in the
 7              preliminary set of studies under consideration (EPA, 2009b, Table 3-2). However,
 8              as discussed in section 3.3.3, use of this study is consistent with the planned
 9              approach presented in the Scope and Methods Plan and, extends and expands upon
10              previous publications presenting evaluations of the ACS long-term cohort study.

11          •   Characterization of uncertainty and variability: The approach for
12              characterizing uncertainty and variability in the current risk assessment closely
13              follows the planned approach as outlined in the Scope and Methods Plan.
14              However, reflecting consideration of comments received from CASAC, we have
15              considered: (a) the WHO Guidance on Characterizing and Communicating
16              Uncertainty In Exposure Assessment (WHO, 2008) to ensure that our approach is
17              consistent with the recommended step-wise process described in that document,
18              and (b) the interpretation of the results of our sensitivity analysis and the most
19              effective ways to communicate these results as a set of reasonable additional risk
20              estimates that supplement the core estimates, recognizing that they do not
21              represent a formal uncertainty distribution (see section 4.3).

22          •   Representativeness analysis for the urban study areas:  Consistent with the
23              approach described in the Scope and Methods Plan, EPA staff have completed a
24              representativeness analysis providing a comparison of the  15 urban study areas
25              against national distributions for key PM risk-related attributes to evaluate whether
26              they are more nationally-representative, or more representative of a particular
27              portion of the distribution for a given parameter (see section 4.4).

28          •   National-scale health impact analysis: Consistent with the approach  described in
29              the Scope and Methods Plan,  a national-scale PM2.5-related long-term exposure
30              mortality analysis, using recent air quality data for the continental U.S. has been
31              completed (see chapter 5).

32         2.4.2  Population exposure analysis
33          Following release of the Scope and Methods Plan, we continued development of the
34    approach for conducting a population exposure analysis, with the goal of completing the
35    analysis as part of the current PM review. However, this additional design work further
36    emphasized the need for clearly outlining the purpose of the  analysis, including specific ways
37    in which the results would be used to interpret the estimates generated from the risk
38    assessment (e.g., potentially identifying sources of exposure  measurement error associated
39    with the epidemiological studies providing the C-R functions and the  magnitude of the impact
40    of those sources of error on risk estimates). When combined with consideration for CASAC
41    comments on the Scope and Methods Plan which emphasized the same point regarding the
42    importance of clearly outlining how the estimates from the analysis would be used, we
      September 2009                           17           Draft - Do Not Quote or Cite

-------
 1    decided to continue methods development work, rather than attempt to complete a preliminary
 2    population exposure analysis as part of this review. Development of the population exposure
 3    analysis methodology is ongoing and we anticipate considering any results from this exposure
 4    assessment within the context of the next PM NAAQS review.

 5         2.5   ALTERNATIVE STANDARD LEVELS INCLUDED IN THE RISK
 6               ASSESSMENT
 7           As noted earlier, EPA staff has modified the scope of the risk assessment to focus on
 8    evaluating potential public health impacts of fine particles only and consequently
 9    consideration of alternative standards to be evaluated in the risk assessment were developed
10    exclusively for PM2.5. Specifically, we selected alternative levels for the annual and 24-hour
11    PM2.5 standards that we judged to be appropriate to evaluate in the context of this quantitative
12    risk assessment.  Alternative averaging times and forms were not considered in our analyses.
13    The averaging times and forms used in evaluating alternative levels were those associated
14    with the current 24-hour and annual standards.11  We note that all of the basic elements of the
15    standards (e.g., indicator, averaging time, level, and form) will be discussed in a forthcoming
16    draft Policy Assessment which will present staff conclusions based on both evidence-based
17    and risk-based approaches to inform judgments that the EPA Administrator must make in
18    deciding whether to retain or revise the existing PM standards.
19           In selecting alternative levels for the annual and 24-hour PM2.5 standards for the
20    purpose of inclusion in the quantitative risk assessment, we focused on the range of standard
21    levels likely to be discussed in the draft PA.  This range of alternative standard levels, in turn,
22    reflects consideration of ambient air quality levels associated with health effects as reflected
23    in key short- and long-term PM2.5 exposure epidemiological studies discussed in the  draft
24    ISA.
25           As discussed further in section 3.3.3,  in selecting alternative levels for consideration in
26    the risk assessment, we placed emphasis on effect estimates reported in multi-city studies
27    because these studies have a number of advantages compared to single-city studies including:
28    (1) multi-city studies reflect ambient PM2.5 levels and potential health impacts across a range
29    of diverse locations; (2)  multi-city studies "clearly do not suffer from potential omission of
30    negative analyses due to 'publication bias'" (EPA, 2004a, p. 8-30); and (3) multi-city studies
31    generally have higher statistical power.
      11 The "form" of a standard defines the air quality statistic that is compared to the level of the standard in
      determining whether an area attains the standard.  The form of the 24-hour PM2 5 standard is the 98th percentile of
      the distribution of 24-hour PM2 5 concentrations at each population-oriented monitor within an area, averaged
      over 3 years. The form of the annual PM2 5 standard is an annual arithmetic mean, averaged over 3 years, from
      single or multiple community-oriented monitors.

      September 2009                           18            Draft - Do Not Quote or Cite

-------
 1           Specifically, regarding alternative levels for the annual PM2.5 standard to be evaluated
 2    in this risk assessment, we first considered long-term average PM2.5 concentrations associated
 3    with health effects observed in long-term epidemiological studies as summarized in Figure 2-
 4    2 of the draft ISA. In general, the draft ISA concludes that the association between increased
 5    risk of mortality and long-term PM2.5 exposure "becomes more precise and consistently
 6    positive in locations with mean PM2.5 concentrations of 13.5 |ig/m3 and above." (EPA, 2009a,
 7    section 2.3.1.2).  The draft ISA also concludes that the strongest evidence for cardiovascular-
 8    related effects related to long-term PM2.5 exposures has been reported in large, multi-city
 9    U.S.-based studies and, specifically, one of these studies, the Women's Health Initiative
10    (WHI) Study, "reports associations between PM2.5 and cardiovascular effects among post-
11    menopausal women using a 1-yr average PM2.5 concentration (mean =13.5 |ig/m3)" (EPA,
12    2009a, section 2.3.1.2). In addition, we evaluated long-term average PM2.5 concentrations in
13    short-term exposure studies that reported statistically significant effects. More specifically, as
14    reported in the draft ISA, both cardiovascular and respiratory morbidity effects (e.g.,
15    emergency department visits, hospital admissions) have been observed and become "more
16    precise and consistently positive in locations with mean PM2.5 concentrations of 13 |ig/m3 and
17    above" (EPA, 2009a,  section 2.3.1; also see Figure 2-1).12
18           Based on the available epidemiological evidence indicating effects associated with a
19    range of annual averaged PM2.5 concentrations, as briefly described above, we selected levels
20    of 12 and  13  |ig/m3 as the alternative annual standard levels to be evaluated in the quantitative
21    risk assessment.
22           In  identifying  alternative levels for the 24-hour PM2.5 standard to be evaluated  in this
23    risk assessment, we considered the ambient PM2.5 levels associated with mortality and
24    morbidity effects as reported in key short-term epidemiological studies. We focused on the
25    98th percentile PM2.5 ambient levels reported in two multi-city studies that provided C-R
26    functions used in the core risk assessment, Zanobetti and Schwartz (2009) and Bell et  al.
27    (2008).  The focus on the 98th percentile of the 24-hour PM2.5 concentrations observed in the
28    epidemiological studies is consistent with the approach used in the prior PM NAAQS  review
29    and is consistent with the current form of the 24-hour PM2.5 standard.
30           The draft ISA presents 98th percentile 24-hour PM2.5  values for each of the 112 urban
31    areas included in the Zanobetti and Schwartz (2009) short-term mortality study (EPA, 2009a,
32    Figure 6-22). We evaluated the trend in these county-level 98th percentile 24-hour PM2.5
      12 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).

      September 2009                            19           Draft - Do Not Quote or Cite

-------
 1    levels in conjunction with the statistical significance of the associated county-lev el effect
 2    estimates. If we had found an association between the air quality levels and statistically
 3    significant effect estimates (i.e., higher 98th percentile PM2 5 levels were consistently
 4    associated with statistically significant effect estimates), then it would have been reasonable
 5    to consider the lowest 98th percentile PM2.5 level associated with the set of counties for which
 6    a statistically significant effect estimates was observed as the basis for selecting an alternative
 7    standard level for evaluation in this risk assessment. However, no  such association was
 8    observed. Rather, we observed mixed results with no clear correlation between 98th percentile
 9    air quality levels and statistically significant effect estimates. Therefore, we focused on the
10    overall range of 98th percentile values across the entire set of counties and considered the
11    lower quartile of that distribution as representative of a reasonably  cautious approach for
12    identifying alternative levels for consideration in the risk assessment.  The 10th and 25th
13    percentiles values were 25.5 and 29.8 |ig/m3, respectively (Zanobetti, 2009). We note that the
14    overall 98th percentile value across the entire set of urban areas analyzed in Zanobetti and
15    Schwartz. (2009) was 34.3 |ig/m3 (EPA, 2009a, Figure 2-1; Zanobetti and Schwartz, 2009)
16           Next, we completed a similar analysis of the county-level ambient air quality data
17    (Bell, 2009) for the 202 counties associated with the Bell et al. (2008) study. Analysis  of the
18    overall distribution of 98th percentile values across the entire dataset resulted in identifying
19    10th and 25th percentile values of about 24.4 and 29.3 |ig/m3, respectively.  We note that the
20    overall 98th percentile value across the entire set of counties analyzed in Bell et al.  (2008))
21    was 34.2 |ig/m3 (EPA, 2009a, Table 6-11; Bell, 2009).
22           Based on the available epidemiological evidence indicating effects associated with a
23    range of 98th percentile 24-hour PM2.5 concentrations, as briefly described above, we selected
24    levels of 25  and 30 |ig/m3 as the alternative 24-hour standard levels to be evaluated in this
25    quantitative  risk assessment.
26            Once alternative levels were identified for the annual and 24-hour PM standards, the
27    next step was to identify specific combinations of these standard levels to be considered in the
28    risk assessment.  In selecting the pairing of annual and 24-hour standard levels, we considered
29    which standard was predicted to be controlling across the set of 15  urban study areas (either
30    the annual or 24-hour standard will be the "controlling standard" at a given location,
31    depending on the design value associated with that location).13 Ultimately, for this risk
32    assessment,  the goal was to select combinations of annual and 24-hour levels that would result
33    in a mixture of behavior in terms of which standards would control across the various urban
34    study areas.  For example, with the 12/35 combination, the annual level of 12 |ig/m3 is the
      13 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.

      September 2009                            20           Draft - Do Not Quote or Cite

-------
 1   controlling standard for all 15 urban study areas, while with the 12/25 combination, the
 2   annual standard is the controlling standard at some locations and the 24-hour standard is the
 3   controlling standard at other locations.  Consideration of these factors resulted in a set of four
 4   alternative combinations of annual and 24-hour standards being identified for inclusion in the
 5   risk assessment.
 6          The full set of air quality scenarios included in the risk assessment, including the
 7   recent conditions air quality scenario and current standards scenario along with the four
 8   alternative sets of standards are as follows:
 9          •   Recent conditions (risk estimates based on ambient PM2.5 monitoring data for the
10              analysis period - 2005 to 2007)
11          •   Current PM2.5 NAAQS: annual 15 |ig/m3; 24-hour 35 |ig/m3
12          •   Alternative PM2.5 standards: annual 13 |ig/m3; 24-hour 35 |ig/m3
13          •   Alternative PM2.5 standards: annual 12 |ig/m3; 24-hour 35 |ig/m3
14          •   Alternative PM2.5 standards: annual 13 |ig/m3; 24-hour 30 |ig/m3
15          •   Alternative PM2.5 standards: annual 12 |ig/m3; 24-hour 25 |ig/m3
      September 2009                           21            Draft - Do Not Quote or Cite

-------
 1              3   METHODS USED IN URBAN CASE STUDY ANALYSIS

 2          This section provides an overview of the methods used in the risk assessment. Section
 3    3.1 discusses the basic structure of the risk assessment, identifying the modeling elements and
 4    related sources of input data needed for the analysis. Section 3.2 discusses air quality
 5    considerations.  Section 3.3 discusses the selection of health endpoints, urban study areas and
 6    C-R functions from key epidemiological studies used in modeling those endpoints. Section
 7    3.4 discusses baseline health effects incidence rates. Finally, section 3.5 describes how
 8    uncertainty and variability are addressed in the risk assessment.

 9         3.1  GENERAL APPROACH
10         3.1.1  Basic Structure of the Risk Assessment
11          The general approach used in both the prior and the current PM risk assessment relies
12    upon C-R functions which have been estimated in epidemiological studies.  Since these
13    studies estimate C-R functions using ambient air quality data from fixed-site, population-
14    oriented monitors, the appropriate application of these functions in a PM risk assessment
15    similarly requires the use of ambient air quality data at fixed-site, population-oriented
16    monitors.
17          The general PM health risk model, illustrated in Figure 3-1, combines information
18    about PM2.5 air quality for specific urban areas with C-R functions derived from
19    epidemiological studies, baseline  health incidence data for specific health endpoints, and
20    population estimates to derive estimates of the  annual incidence of specified health effects
21    attributable to ambient PM2.5 concentrations under different air quality scenarios.  The
22    analyses were conducted for recent air quality and for air quality simulated to reflect
23    attainment of current and alternative suites of PM2.5 ambient standards.
24          The PM2 5 risk assessment was implemented within TRIM.Risk, the component of
25    EPA's Total Risk Integrated Methodology (TRIM) model that estimates human health risks.14
26    In the first part of the risk assessment, we estimate health effects incidence associated with
27    recent PM2.5 levels. In the second part, we estimate the reduced health effects incidence
28    associated with those PM2.5 concentrations that would result if the current or alternative PM2.5
29    standards were just met in the assessment locations, as well as the percent reductions in
30    incidence from incidence under the current suite of standards. Both parts of the risk
31    assessment consider only the incidence of health effects associated with PM2.5 concentrations
32    in excess of either policy-relevant background  (PRB) for evaluating effects  associated with
      14 For more detailed information about TRIM.Risk, go to: http://www.epa.gov/ttn/fera/trim_risk.html

      September 2009                           22            Draft - Do Not Quote or Cite

-------
 1    short-term PM2.5 concentrations or the lowest measured level (LML) for evaluating effects
 2    associated with long-term PM2.5 concentrations.
 3          Consistent with past risk assessments for NAAQS reviews, the risk assessment is
 4    intended to estimate risks attributable to anthropogenic sources and activities, and thus risks
 5    are only estimated for concentrations in excess of PRB levels.  For all health endpoints
 6    associated with short-term exposure to PM2.5, the risk assessment considers only the incidence
 7    of health effects associated with PM2.5 concentrations in excess of PRB levels.  In the studies
 8    estimating a relationship between mortality and long-term exposure to PM2.5, however, the
 9    LMLs in the studies were substantially above PRB. Thus, estimating risk down to PRB
10    would have required substantial extrapolation of the estimated C-R functions below the range
11    of the data on which they were estimated.  We therefore estimated risk only down to the LML
12    to avoid extrapolating the estimated C-R functions too far below the range of the PM2.5 data
13    on which they were estimated. To provide consistency across the long-term exposure C-R
14    functions, however, and, in particular,  to avoid the choice of LML unduly influencing the
15    results of the risk assessment, we selected a single LML - 5.8 |ig/m3 from the later exposure
16    period used in Krewski et al. (2009) -  to be used in all cases involving long-term  exposure.
17
      September 2009                           23           Draft - Do Not Quote or Cite

-------
Figure 3-1.    Major components of particulate matter health risk assessment.
                       Air Quality
                        Ambient Population-
                        Oriented Monitoring and
                        Estimated Policy
                        Relevant Background
                        Levels for Selected Cities
                         Air Quality Adjustment
                         Procedures
                         Alternative Proposed
                         Standards
                     Concentration-Response
                        Human Epidemiological
                        Studies (various health
                        endpoints)
                        Estimates of City-specific
                        Baseline Health Effects
                        Incidence Rates
                        (various health
                        endpoints) and
                        Population Data
Concentration
Response
Relationships
                                       Risk Estimates:
                                       * Recent Air
                                         Quality
                                       • Alternative
                                         Scenarios
September 2009
     24
Draft - Do Not Quote or Cite

-------
       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 set 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 upper levels (e.g., PM2.5 concentrations that just meet a
specified set of standards) to specified lower levels (PRB levels or the LML).
       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 5 PRB concentrations appropriate to this location, and (3) a method
       for adjusting the air quality data to reflect patterns  of air quality change estimated to
       occur when the area just meets the specified 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 5_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, 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).
       As noted below (in section 3.2.1), the risk assessment was carried out using three
years of recent air quality data -2005,  2006, and 2007. 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. Because 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, 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.
September 2009                           25           Draft - Do Not Quote or Cite

-------
       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.
       For this risk assessment, we have 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, EPA does believe that 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 may have the greatest impact on risk estimates when acting alone, or in
combination with other sources of uncertainty.  In addition, 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. The potential
utility  of the sensitivity analysis-based risk estimates in informing consideration of
uncertainty in the core results is discussed in section 4.5.2.  A number of modeling elements
are 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.
September 2009                            26            Draft - Do Not Quote or Cite

-------
       Figure 3-2.    Flow diagram of risk assessment for short-term exposure studies.
                       Air Quality Data
                       Concentration-Response Functions


Identity
location-
specific
studies





Identify
Relative Risk
 or slope
coefficients (B)


i Convert RR |

> (if necessary j

Identify
functional form
                                                                                             | Speciyrolback
                                                                                                method  I
                                                                                               (for certain  I
                                                                                             !  analyses)  i
                        Baseline Health Incidence
                                                                              Compute
                                                                              annual
                                                                                n

                                                                             associated
                                                                               with
                                                                             change in PM
                                                                                                  Estimate of
                                                                                                  percent change in
                                                                                                  total incidence
                               Estimate of
                               PM-associated
                               incidence
September 2009
27
Draft - Do Not Quote or Cite

-------
       Figure 3-3.   Flow diagram of risk assessment for long-term exposure studies.
                       Air Quality Data
                       Concentration-Response Functions


Identify
studies




Identifyfunctional
form

Identify
Relative Risk
(RR) or slope
coefficients (B)
i
r 	
1 Convert RR
* 	 * toB
{if necessary
                        Baseline Health Incidence
I                        Specify roltisck
                          method
                         (for certain  I
                         analyses)   I
                                                                                                Estimate of
                                                                                                percent change in
                                                                                                total incidence
                                                                             Compute #

                                                                             associated
                                                                               with
                                                                            change in PM
                             Estimate of
                             PM-associated
                             incidence
September 2009
28
Draft - Do Not Quote or Cite

-------
 1         3.1.2  Calculating PM-Related Health Effects Incidence
 2                    3.1.2.1  General approach
 3           The C-R functions used in the risk assessment are empirically estimated relations
 4    between average ambient concentrations of PM2.5 and the health endpoints of interest (e.g.,
 5    mortality or hospital admissions associated with short- and long-term exposure to PM2 5)
 6    reported by epidemiological studies for specific locations. This section describes the basic
 7    method used to estimate changes in the incidence of a health endpoint associated with
 8    changes in PM, using a "generic" C-R function of the most common functional form.
 9           Although some epidemiological  studies have estimated linear C-R functions and some
10    have estimated logistic functions, most of the studies used a method referred to as "Poisson
11    regression" to estimate exponential (or log-linear) C-R functions in which the natural
12    logarithm of the health endpoint is a linear function of PM2 5:
13
14                                      y = Be^                                        (1)
15
16           where x is the ambient PM2 5 level, y is the incidence of the health endpoint of interest
17    at PM2.s level x, p is the coefficient of ambient PM2.s concentration, and B is the incidence at
18    x=0, i.e., when there is no ambient PM2 5. The relationship between a specified ambient PM2 5
19    level, XQ, for example,  and the incidence of a given health endpoint associated with that level
20    (denoted as yo) is then
21
22                                      y0=Beflc".                                      (2)
23
24           Because the log-linear form of a  C-R function (equation (1)) is by far the most
25    common form, we use this form to illustrate the "health impact function" used in the PM2 5
26    risk assessment.
27           If we let x0 denote the baseline (upper) PM2.5 level, and x} denote the lower PM2.5
28    level, and yo and yi denote the corresponding incidences of the health effect, we can derive the
29    following relationship between the change in x, Ax= (x0- Xi\ and the corresponding change in
30    y, Ay, from equation (I)15:
31
32                                      Aj = (j0-Jl) = Jo[l-^^].                      (3)
      15 If Ar < 0 - i.e., if Ar = (rr xg) - then the relationship between Ar 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.

      September 2009                            29           Draft - Do Not Quote or Cite

-------
 1          Alternatively, the difference in health effects incidence can be calculated indirectly
 2    using relative risk.  Relative risk (RR) is a measure commonly used by epidemiologists to
 3    characterize the comparative health effects associated with a particular air quality comparison.
 4    The risk of mortality at ambient PM2.5 level x0 relative to the risk of mortality at ambient
 5    PM2.5 level xj, for example, may be characterized by the ratio of the two mortality rates: the
 6    mortality rate among individuals when the ambient PM2.5 level is x0 and the mortality rate
 7    among (otherwise identical) individuals when the ambient PM2 5 level is Xj. This is the RR for
 8    mortality associated with the difference between the two ambient PM2.5 levels, XQ and xj.
 9    Given a C-R function of the form shown in equation (1) and a particular difference in ambient
10    PM2.5 levels, Ax, the RR associated with that difference in ambient PM2.5, denoted as RR-Ax, is
11    equal to e13^.  The difference in health effects incidence,  Ay, corresponding to a given
12    difference in ambient PM2.5 levels, Ax, can then be calculated based on this RR-Ax as:
13
14                                 Ay = (y0-yl) = y0[\-(\/RR^].                       (4)
15
16          Equations (3) and (4) are simply alternative ways of expressing the relationship
17    between a given difference in ambient PM2 5 levels, Ax > 0, and the corresponding difference
18    in health effects incidence, Ay.  These health impact equations are the key equations that
19    combine air quality information, C-R function information, and baseline health effects
20    incidence information to estimate ambient PM2 5 health risk.

21                    3.1.2.2   Short-term vs. long-term exposure
22          C-R functions that use as input annual average PM2.5 levels (or some function of these,
23    such as the average over a period of several years) relate these to the annual incidence of the
24    health endpoint - i.e., in such studies x in equation (1) above is the average PM2.5
25    concentration over a period of one or more years, meant to represent long-term exposure, and
26    y is the annual incidence of the health effect associated with that long-term exposure.
27          C-R functions that use as input 24-hour average PM2.5 levels (or some function of
28    these,  such as the average over one or more days) relate these to the daily incidence of the
29    health endpoint - i.e., in such studies x in equation (1) above is the average PM2 5
30    concentration over a period of one or a few days (short-term exposure), and y is the daily
31    incidence of the health effect associated with that short-term exposure.
32         There are several variants of the short-term (daily)  C-R function.  Some C-R functions
33    were estimated by using moving averages of ambient PM2.5 to predict daily health effects
34    incidence. Such a function might, for example, relate the incidence of the health effect on day
35    t to the average of PM2.5 concentrations on days t and (M).  Some C-R functions consider the
36    relationship between daily incidence and daily average PM2 5 lagged a certain number of days.

      September 2009                          30           Draft - Do Not Quote or Cite

-------
 1    For example, a study might estimate the C-R relationship between mortality on day t and
 2    average PM2.5 on a prior day (/-I).  A few studies have estimated distributed lag models, in
 3    which health effect incidence is a function of PM2 5 concentrations on several prior days - that
 4    is, the incidence of the health endpoint on day tis a function of the PM2.5 concentration on day
 5    t, day (/-I), day (t-2), and so forth.  Such models can be reconfigured so that the sum of the
 6    coefficients of the different PM2.5 lags in the model can be used to predict the changes in
 7    incidence on several days. For example, corresponding to a change in PM on day tin a
 8    distributed lag model with 0-day, 1-day, and  2- day lags considered, the sum of the
 9    coefficients of the 0-day, 1-day, and 2-day lagged PM2.5 concentrations can be used to predict
10    the sum of incidence changes on days t, (t+l) and (t+2).
11           Most daily time-series epidemiological studies estimated C-R functions in which the
12    PM-related incidence on a given day depends only on same-day PM concentration or
13    previous-day PM concentration (or some variant of those, such as a two-day average
14    concentration). Such models necessarily assume that the longer pattern of PM levels
15    preceding the PM concentration on a given day does not affect mortality on that day. To the
16    extent that PM-related  mortality on a given day is affected by PM concentrations over a
17    longer period of time, then these models would be mis-specified, and this mis-specification
18    would affect the predictions of daily incidence based on the model.
19           The extent to which time-series  studies using single-day PM2 5 concentrations may
20    misrepresent the relationship between short-term PM2 5 exposure and mortality is unknown.
21    However, there is some evidence, based on analyses  of PMio data, that mortality on a given
22    day is influenced by prior PM exposures up to more than a month before the date of death
23    (Schwartz, 2000). The extent to which  short-term exposure studies (including those that
24    consider distributed lags) may not capture the full impact of long-term exposures to PM2 5 is
25    similarly not adequately understood, although the current evidence (e.g., Krewski et al., 2009;
26    Krewski et al., 2000) suggests that there is a substantial impact of long-term exposures on
27    health effects that is not picked up in the short-term exposure studies.

28                    3.1.2.3   Calculating annual incidence
29           The risk assessment estimated health  effects incidence, and changes in incidence, on
30    an annual basis, for 2005, 2006, and 2007. For mortality, both short-term and long-term
31    exposure studies have reported estimated C-R functions.  As noted above, most short-term
32    exposure C-R functions estimated by daily time-series epidemiological studies relate daily
33    mortality to same-day PM2 5 concentration or previous-day PM2 5 concentration (or some
34    variant of those).
35           To estimate the daily health impacts of 24-hour average ambient PM2.5 levels above
36    PRB, C-R functions from short-term exposure studies were used together with estimated


      September 2009                           31           Draft - Do Not Quote or Cite

-------
 1    changes in 24-hour ambient PM2.5 concentrations to calculate the daily changes in the
 2    incidence of the health endpoint. After daily changes in health effects were calculated, an
 3    annual change was calculated by summing the daily changes.
 4          The mortality associated with long-term exposure is likely to include mortality related
 5    to short-term exposures as well as mortality related to longer-term exposures. As discussed
 6    previously, estimates of daily mortality based on the time-series studies also are likely
 7    influenced by prior PM exposures. Therefore, the estimated annual incidences of mortality
 8    calculated based on the short- and long-term exposure studies are not likely to be completely
 9    independent and should not be added together. While we can characterize the statistical
10    uncertainty surrounding the estimated PM2.5 coefficient in a reported C-R function, there are
11    other sources of uncertainty associated with  the C-R functions used in the risk assessment
12    that are addressed via sensitivity analyses and/or qualitatively discussed in section 3.5.3.

13         3.2   AIR QUALITY INPUTS
14         3.2.1   Characterizing recent conditions
15          As noted earlier, a major input to the PM2.5 risk assessment is ambient PM2.5 air
16    quality data for each assessment location. Twenty-four hour PM2.5 air quality data for 2005,
17    2006,  and 2007 were obtained for each of the urban study areas from monitors in EPA's Air
18    Quality System  (AQS). To characterize PM2.5 air quality in each risk assessment location as
19    accurately as possible, we used only those monitors that were located within the county or
20    counties that were analyzed in the epidemiological studies used to select C-R functions..  In a
21    few cases, an urban area was delineated differently by two  or more epidemiological studies
22    used in the risk assessment. For example, Birmingham, AL was defined as Blount, Jefferson,
23    Shelby, St.  Clair, and Walker Counties  in one study and as  only Jefferson County in another
24    study. In such cases, we matched our delineation of the urban study area to that used  in each
25    study, resulting in two or more different delineations of the urban study area and identified
26    them as, for example, Birmingham 1  and Birmingham 2. The counties and the number of air
27    quality monitors included within each urban area are given in Table 3-1.
28          In order to be consistent with the approach generally used in the epidemiological
29    studies that estimated PM2.5 C-R functions, the average ambient PM2.5 concentration on each
30    day for which measured data were available was deemed most appropriate for use in the risk
31    assessment. Consistent with the approach used in the prior PM risk assessment, a composite
32    monitor data set was created for each assessment location based on a composite of all
33    monitors located within each urban study area. Specifically, the value at the composite
34    monitor on a given day was calculated as the average of the values at those monitors in a
35    specific urban study area that reported a measured value for that day.
      September 2009                           32           Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
            There were some days on which none of the monitors in a risk assessment location
      reported PM2.5 concentrations.  The numbers of missing days at the composite monitors in the
      risk assessment locations are given in Table 3-1. We used 7-day moving averages to fill in
      missing values at composite monitors.
            To summarize, air quality data inputs for the risk assessment model were developed as
      follows: first we calculated the composite monitor value for each day. For any day that had a
      missing value at the composite monitor, we inserted the 7-day moving average centered on
      that day.  We then evaluated the new series of composite monitor values (with missing days
      filled in as described); if any  day still had a missing value, we filled it in with the 7-day
      moving average centered on the missing day, where the values in the 7-day moving average
      were calculated from the series created on the previous step. We repeated this process until
      all missing days were filled in.
      Table 3-1.    Numbers of Monitors and Numbers of Missing Days at Composite
                   Monitors in Risk Assessment Locations from 2005 Through 2007.
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
New York, NY - 2
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO - 1
St. Louis, MO - 2
Tacoma, WA
Counties
Cobb, De Kalb, Fulton, Gwinnett
Cobb, De Kalb, Fulton
20-County MSA**
Baltimore city, Baltimore county
Blount, Jefferson, Shelby, St. Clair,
Walker
Jefferson
Dallas
Wayne
Fresno
Harris
Los Angeles
Kings, New York City (Manhattan),
Queens, Richmond, Bronx
New York City (Manhattan)
Philadelphia
Maricopa
Allegheny
Salt Lake
Jefferson, Madison (IL), St. Louis, St.
Louis City, St. Clair (IL)
Madison (IL), St. Louis, St. Louis City,
St. Clair (IL)
Pierce
Number of
Monitors
8
7
10
8
10
8
6
9
3
6
10
12
5
7
5
12
7
15
14
1
Number of Missing
Days at Composite
Monitor Over the 3-
Year Period*
8
8
7
2
1
1
21
22
40
59
0
4
731***
14
710
1
4
0
1
741***
15
16
17
18
19
      *The value on a given day at the composite monitor is the average of all monitors reporting on that day.
      ** Barrow, Bartow, Carroll, Cherokee, Clayton, Cobb, Coweta, DeKalb, Douglas, Fayette, Forsyth, Fulton,
      Gwinett, Henry, Newton, Paulding, Pickens, Rockdale, Spalding, and Walton.
      *** Note, that the sets of monitors for New York (Manhatten) have 1 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
      September 2009
                                                33
Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 9
10
11
12
13
14
15
16
17
18
19
20
(i.e., 731) had no monitor coverage for the New York urban study area, resulting in a need to interpolate
estimates for these days (for the composite monitor) using the approach described above. Similarly, with
Tacoma, the single monitor at that location also has 1 in 3 day sampling, resulting again, in 2/3 of the days not
having data with interpolation being used to derive estimates for those days (for the composite monitor).

       Appendix A summarizes the PM2.5 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.

      3.2.2  Estimating policy relevant background
       PRB estimates used in the risk assessment model (see Table 3-2 below) were obtained
from the draft ISA (Table 3.7-6, draft ISA, EPA, 2009a).  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
3.6.1.2).  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.
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
21

22
23
24
25
26
27
28
     3.2.3  Simulating air quality to just meet current and alternative standards
       This section describes the methodology used to simulate ambient PM2.5 levels in an
area upon just meeting specified PM2.5 standards.  The form of the PM2.5 standards
promulgated in October 2006 requires that the 3-year average (rounded to the nearest 0.1
|ig/m3) of the annual means from single monitors 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
      September 2009
                                          34
Draft - Do Not Quote or Cite

-------
 1    24-hour standard.  In determining attainment of the annual average standard, an area may
 2    choose to use either the spatially averaged concentrations across all population-oriented
 3    monitors, subject to meeting certain criteria detailed in Part 50, Appendix N, of the CFR, or it
 4    may use the highest 3-year average based on individual monitors. The most realistic
 5    simulation of just meeting both the annual and the 24-hour PM2.5 standards in a location
 6    would require changing the distribution of 24-hour PM2.5 concentrations at each monitor
 7    separately, reflecting the specific mix of local and regional controls impacting that particular
 8    location.  This would require extensive analysis and assumptions about the nature of future
 9    control strategies that was considered beyond the scope of the previous risk assessment and is
10    similarly considered beyond the scope of the current risk assessment.
11           In the previous PM risk assessment, just meeting the current or alternative PM2 5
12    standards was simulated by changing 24-hour PM2.5 concentrations at a "composite monitor,"
13    which represented the average of the monitors in a location.  In the current PM risk
14    assessment, just meeting the current or alternative PM2.5 standards was simulated by changing
15    24-hour PM2.5 concentrations at each monitor separately. This change was made because the
16    current PM risk assessment considers two alternative approaches to simulating PM2 5
17    concentrations that just meet a given set of standards. One of these approaches (the
18    "proportional rollback" approach) was used in previous PM2.s risk assessments and involves
19    proportional  adjustments to monitor levels, in which PM2 5 concentrations  are reduced
20    ("rolled back") by the same percentage each day.  When this approach is used, it doesn't
21    matter whether (1) PM2 5 concentrations are first rolled back by the same percentage each day
22    at each monitor, and then the composite monitor values are calculated from these monitor-
23    specific values or (2) first the composite monitor values are calculated and then these are
24    rolled back by the same percentage each day - the results will be the same.
25           The second approach (the "hybrid rollback" approach) used in a sensitivity analysis in
26    some of the risk assessment locations (in comparison to the proportional rollback approach),
27    has two steps: (1) first PM2 5 concentrations are reduced  at a specific monitor location within
28    an urban study area and then additional monitors within that urban study area are adjusted to a
29    lesser extent (with the magnitude of adjustment based on a distance-decay function); then (2)
30    a proportional rollback of the adjusted PM2 5 concentrations  at all of the different monitors is
31    carried out. This two-step  approach is intended to simulate situations  in which it is likely that
32    attainment of a set of standards will be  achieved by first  implementing more localized
33    controls to target a specific monitor (with lesser reductions resulting at other near-by monitors
34    due to the influence of adopting more localized controls) followed by  regional controls that
35    result in a more universal (proportional) reduction across all the monitors in a study area.
36    Because the initial step reflecting localized controls  is non-proportional, this needs to be
37    completed on the monitor datasets (associated with a particular study area) prior to

      September 2009                           3 5            Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
construction of the composite monitor. However, once those non-proportional reductions have
been implemented, a composite monitor can then be constructed (as described earlier) and the
second step of conducting proportional adjustment to simulate the current or alternative set of
standard levels, can be calculated for the composite monitor.
       The percent reduction of 24-hour PM2.5 concentrations in the proportional rollback
approach (and in the second step of the hybrid rollback approach) at each monitor each day to
simulate just meeting current and alternative set of standard levels is determined by the PM2.5
annual and 24-hour design values. 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 is the
       annual design value, denoted dvmnuai\
The 98th percentile design value (in |ig/m3) was similarly calculated as follows:
   •   At each monitor, the 98th percentile 24-hour PM2.5 concentration was calculated for
       each of the years 2005, 2006, and 2007, and these three 98th percentile concentrations
       were then averaged.
   •   The maximum of these monitor-specific 3-year averages of 98th percentile
       concentrations is the 24-hour design value, denoted dv'daily 93 (note, we will refer to the
       98th percentile design value as the 24-hour design value throughout the rest of the
       document).
       The annual and 24-hour design values used in assessing the current and alternative
standards for PM2.5 are given in Table 3-3. Note, that monitors  that were closed in 2005 (and
therefore,  did not include monitoring data for the majority of the three year simulation
period), or which were missing an entire year's worth of monitoring data during any of the
three simulation years (2005, 2006 or 2007) were excluded from consideration as design
value monitors, although these monitors were still used to construct composite monitors for
purposes of estimating risks.
Table 3-3.    EPA Design Values for Annual and \24-hour PMi.s Standards for the
             Period 2005-2007.*
Location
Atlanta
Baltimore
Birmingham
Dallas
Detroit
Annual
(Hg/m3)
17
16
19
13
17
24-hour
(Hg/m3)
35
37
44
26
43
      September 2009
                                         36
Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19

20

21
22
23
Location
Fresno
Houston
Los Angeles
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
St. Louis
Tacoma
Annual
(Hg/m3)
17
16
20
16
15
13
20
12
17
10
24-hour
(Hg/m3)
63
31
55
42
38
32
60
55
39
43
       *The calculation of design values is explained in the text above.

       The percent reduction required to meet a standard (annual or 24-hour) was determined
by comparing the design value for that standard with the level of the standard. Because
pollution abatement methods are applied largely to anthropogenic sources of PM2.5, rollbacks
were applied only to PM2 5 above estimated PRB levels. The percent reduction was
determined by the controlling standard.  For example, suppose both an annual and a 24-hour
PM2.5 standard are being simulated. Suppose^ is the percent reduction required to just meet
the annual  standard (i.e., the percent reduction of daily PM2.s above background necessary to
get the annual design value down to the current or alternative annual standard).  Supposepd is
the percent reduction required to just meet the 24-hour standard (i.e., the percent reduction of
daily PM2 5 above background necessary to get the 24-hour PM2 5 design value down to the
24-hour standard). Ifpd is greater than/»«, then all 24-hour average PM2.5 concentrations
above background are reduced by pd percent. lfpa is greater thanp^, then all 24-hour average
PM2 5 concentrations are reduced by pa percent. The method of rollbacks to meet a set of
annual and 24-hour PM2.5 standards is summarized as follows:
   1.  The percent by which the above-PPJ3 portion of all daily PM2.5 concentrations (at the
       composite monitor) would have to be reduced to just meet the  annual standard
       (denoted stda) is
                                        i    (std.-PM)
              where PRBavg is the average of the daily PRB concentrations.
                                                                      16
      16 In the previous PM risk assessment, a constant PRB level was assumed for all days, and that constant PRB
      level was used in the formulas to calculate percent rollbacks necessary to just meet a standard. It can be shown
      that, if PRB levels vary from day to day, the average PRB level takes the place of the constant PRB level in the
      previous formula, as shown in the above equation.
      September 2009
                                          37
Draft - Do Not Quote or Cite

-------
 2       2.  The percent by which the above-PPJ3 portion of all 24-hour PM2.5 concentrations (at
 3           the composite monitor) would have to be reduced to just meet the current or
 4           alternative  24-hour standard (denoted std^s) is:
 5

 6                                      Pd9S = 1 -
                   a = maximum of (maximum of pa andp^s) and zero.17
 9
10
11       3.   Then ifPM0 denotes the original PM value on a given day (at the composite monitor),
12           the rolled back PM value on that day, denoted PMrb, is:
13
14                                      PMrb = PRB +(PM0 - PRB)*(\-pmax ).
15

16         3.3   SELECTION OF MODEL INPUTS
17         3.3.1  Health endpoints
18           As noted in the Scope and Methods Plan, selection of health effect endpoints reflects
19    consideration for a number of factors including: (a) the extent to which a particular health
20    effect endpoint is considered significant from a public health standpoint, (b) the overall
21    weight of evidence from the collective body of epidemiological, controlled  human exposure,
22    and toxicological studies and the determination made in the draft ISA as to whether there is a
23    causal or likely causal relationship between PM2.5 and the more general health effect category,
24    and (c) whether there are well-conducted studies providing estimated C-R functions for
25    specific health effect endpoints within the broader health effects category associated with
26    ambient PM2.5 levels (section  3.2.2, EPA, 2009b). An additional factor, not specifically
27    mentioned in the Scope and Methods Plan, that we considered is the availability of baseline
28    health effects incidence data that matches the study population(s) evaluated in the
29    epidemiological study(ies) from which  C-R function(s) were selected.
      17 If the percent rollback necessary to just meet the annual standard and the percent rollback necessary to just
      meet the 24-hour standard were both negative ~ i.e., if both standards were already met ~ then the percent
      rollback applied in the risk assessment was zero. That is, PM values were never increased, or "rolled up."

      September 2009                           3 8           Draft - Do Not Quote or Cite

-------
 1          Based on application of the above criteria, as outlined in the Scope and Methods Plan,
 2    we identified the following health effects endpoint categories as candidates for inclusion in
 3    the risk assessment:
 4          Health effect categories associated with short-term PM? 5 exposure:
 5              •  Cardiovascular effects (causal relationship)
 6              •  Respiratory effects (likely causal relationship)
 7              •  Mortality (likely causal relationship)
 8          Health effect categories associated with long-term PM? 5 exposure:
 9              •  Cardiovascular effects (causal relationship)
10              •  Respiratory effects (likely causal relationship)
11              •  Mortality (likely causal relationship)
12          In addition, as noted in the Scope and Methods Plan, we considered expanding the
13    focus of the risk assessment to include additional endpoints from health effects categories that
14    had been initially judged in the draft ISA to have evidence from the scientific evidence
15    evaluated that was suggestive of a casual relationship between ambient PM2.5 measurements
16    and the general category of health effects,  if those additional endpoints allowed us to address
17    potentially important policy issues related to the review of the PM NAAQS. In the Scope and
18    Methods Plan, we cited birth outcomes as  a potential  candidate for inclusion in the PM2.5 risk
19    assessment based on this additional  criterion, recognizing that, in considering endpoints
20    classified as suggestive of a causal relationship, it would be important to appropriately
21    characterize the additional uncertainty associated with the risk estimates (section 3.2.2, EPA,
22    2009).
23          In selecting the set of health effect endpoint categories (and associated endpoints) to
24    include in the PM2.5 risk assessment, we built upon the health effects evidence presented in the
25    draft ISA (EPA, 2009a), as well  as CASAC (Samet, 2009) and public comments received on
26    the Scope and Methods Plan.  Comments provided by CASAC regarding the selection of
27    health effects endpoints included providing "a transparent algorithm for selecting endpoints
28    based on the level of certainty and the relative and attributable risks" as well as the suggestion
29    "that weight be given to the level of classification while still considering the Administrator's
30    obligation to set a standard with  a "margin of safety"  as describe in the Clean Air Act" (Samet
31    2009, p. 1).  As an example, several  CASAC members did not recommend a risk assessment
32    based on birth outcomes, "in part because  the level of evidence is still at a suggestive level
33    (Samet 2009, p. 1-2).


      September 2009                           39           Draft - Do Not Quote or Cite

-------
 1           In reviewing the draft ISA in relation to PM2.5, we focused on the following sections:
 2    (a) section 2.3.1.1 (Effects of Short-Term Exposure to PM2 5), (b) section 2.3.1.2 (Effects of
 3    Long-Term Exposure to PM2 5), (c) section 2.3.2 (Integration of PM2 5 Health Effects), (d)
 4    6.2.12 (Summary and Causal Determinations [for effects related to short-term PM2.5
 5    exposure]), and (e) 7.3.9 (Summary and Causal Determinations [for  effects related to long-
 6    term PM2.5 exposure]). Our overall conclusions regarding the set of health effect endpoint
 7    categories to include in the risk assessment for PM2 5, based on review of these sections of the
 8    draft ISA, did not change from the provisional set provided in the Scope and Methods Plan.
 9           Consideration of information provided in the draft ISA (as referenced above) and
10    CAS AC and public comments on the Scope and Methods Plan, as well as review of the
11    available epidemiological studies for deriving C-R functions (see section 3.3.3) as well as
12    availability of baseline health effect incidence data to support risk modeling (see section 3.4),
13    has resulted in the following health effect categories (and associated health effect endpoints)
14    being selected for modeling PM2 5 in the risk assessment:
15           Short-term exposure
16              •   Premature mortality (non-accidental, respiratory, cardiovascular)
17              •   Respiratory effects (respiratory-related hospital admissions and asthma-related
18                  ED visits)
19              •   Cardiovascular effects (cardiovascular-related hospital admissions)
20            Long-term exposure
21              •   Mortality (all-cause, ischemic  heart disease (MD), cardiopulmonary, lung
22                  cancer)
23           In addition to estimating risk for PM2.5, in the Scope and Methods Plan, we also
24    outlined plans for modeling risk associated with short-term exposures to thoracic coarse
25    particles (PMio-2.s) (section 3.6, EPA, 2009).  Specifically, we identified a set of short-term
26    cardiovascular and respiratory morbidity endpoints as potential candidates for inclusion in the
27    risk assessment (cardiovascular and respiratory hospital admissions,  asthma-related ED visits
28    and respiratory symptoms), noting that support in the first draft ISA  for these endpoints did
29    not rise above being suggestive of a causal relationship. We noted in the Scope and Methods
30    Plan that the decision whether to include these thoracic coarse particle-related health
31    endpoints in a quantitative risk assessment would depend on review  of the scientific evidence
32    presented in the second draft ISA as well as consideration for  public and CASAC comments
33    on the Scope and Methods Plan. During a consultation on the Scope and Methods  Plan,
      September 2009                            40           Draft - Do Not Quote or Cite

-------
 1    CASAC expressed support for a limited risk assessment focusing on exposure to PMio-2.5, for
 2    appropriate outcomes such as hospitalizations (Samet 2009, p. 2).
 3           In the risk assessment conducted for the last PM NAAQS review, EPA concluded that
 4    the nature and magnitude of the uncertainties and concerns associated with this portion of the
 5    risk assessment weighed against use of these risk estimates as a basis for recommending
 6    specific standard levels (EPA, 2005, p. 5-69). In reviewing the evidence provided in the
 7    second draft ISA (US EPA, 2009, ref specific sections or chapters) specifically addressing
 8    effects associated with exposure to PMio-2.5, we recognize that the ISA concludes that there is
 9    suggestive support for an relationship between short-term exposure to PMio-2.5 and
10    cardiovascular and respiratory effects, as well as mortality.  However, we believe that the
11    underlying epidemiological evidence does not readily support derivation of C-R functions
12    applicable to urban case studies in the U.S without the introduction of significant uncertainty
13    into the risk estimates.  Further, we find that research to inform the uncertainties identified in
14    the last review have not fundamentally changed these uncertainties.  Therefore, we conclude
15    that additional  analyses quantifying PMi0-2.5-related risks would not provide additional
16    information beyond the assessment done in the last review and, therefore, no quantitative risk
17    assessment has been conducted for PMio-2.5 in this  document.

18         3.3.2   Selection and delineation of urban study areas
19           This  section describes the approach used in selecting the 15 urban study areas included
20    in this risk assessment (see Table 3-3 for a listing of the urban study areas). The approach for
21    selecting urban study areas considered  criteria from the prior risk assessment and adds two
22    new criteria.
23           Criteria used in the prior risk assessment include: (a) sufficient air quality data for at
24    least one year for the period 1999 or later (at least 11 observations per quarter for a one year
25    period and at least 122 observations per year), (b) coverage of the location by one of the key
26    epidemiology studies included in the risk assessment (at or close to the location where at least
27    one C-R function for one of the recommended health endpoints has been estimated by a study
28    satisfying the selection criteria used in the risk assessment), and (c) sufficient baseline
29    incidence data  for the location (see EPA, 2005, section 3.2 p. 37, for additional detail on these
30    three criteria). Regarding the first criteria (sufficiency of air quality), we assessed prospective
31    study areas by  insuring that there was at least one PM2.5 monitor within the boundaries of the
32    prospective study area that met completeness criteria for the period 2005 to 2007  (note, that
33    locations with more than one PM2.5 monitor meeting completeness criteria were favored,  since
34    this provided a better characterization of ambient air levels for that urban location). The two
35    remaining criteria from the prior risk assessment were largely addressed due to new
36    epidemiological studies and baseline incidence data that are now available. Specifically,


      September 2009                            41           Draft - Do Not Quote or Cite

-------
 1    regarding coverage by key epidemiology studies, because the current risk assessment
 2    primarily utilizes multi-city studies to evaluate risk for key short-term and long-term health
 3    endpoints (whereas the prior risk assessment used city-specific studies in modeling short-term
 4    endpoints), this criterion no longer applies for most prospective areas. Regarding sufficiency
 5    of baseline health effects incidence data, an ongoing effort by EPA to collect county-level
 6    hospital and emergency department admissions data from states to support this risk
 7    assessment (see section 3.5) has resulted in enhanced health effects baseline incidence data,
 8    largely addressing this criterion (i.e., most urban areas in the U.S. now have coverage with the
 9    updated baseline health effects incidence data).
10           In addition to these criteria from the prior risk assessment, as noted above, we have
11    also included consideration for two additional factors in selecting urban study areas. First, we
12    focused on those urban areas with PM2.5 monitoring levels suggesting the potential for risk
13    reduction under alternative (daily or annual) standards. Specifically, only those urban
14    locations with at least one monitor having an annual average above 12 |ig/m3 and/or a 24-hour
15    value above 25 |ig/m3 (the levels in the lowest alternative standard considered in the risk
16    assessment) were considered. Furthermore, locations with ambient PM2.5 level significantly
17    higher than these levels were favored (with several urban study areas selected having annual
18    and daily design values exceeding the current standard level, being selected - see Table 3-4).
19    Application of this criterion reflects a desire to include urban case studies that are
20    representative of areas likely to experience some degree of risk reduction under alternative
21    standard levels.
22           The second criterion we added  for study area selection, was the goal of providing
23    coverage for factors believed to play a role in influencing risk heterogeneity at the national-
24    level (e.g., PM source characteristics and composition, demographics, SES status, air
25    conditioner use). We implemented this criterion by using the 7 PM regions originally
26    identified in the 1996 PM Criteria Document (US EPA,  1996), to guide selection of urban
27    study areas. Specifically, we attempted to include several urban locations from  each of the
28    PM regions in our suite of 15 urban study areas, to insure that each of the regions was
29    represented by one or more of the urban study areas (see Table 3-4). Note, that ultimately,
30    consideration of the criteria described here resulted in an urban study area not being identified
31    for one of the PM regions (the Upper Midwest). However, the remaining six regions each
32    included at least one of the 15 urban study areas evaluated in the risk assessment. While the
33    1996 PM regions as defined (see footnote 2), focused primarily on differences in PM
34    composition, size and seasonality, by selecting urban study areas from regions that cover the
35    continental U.S., we also have a better potential for covering regional differences in other
36    factors related to risk heterogeneity (e.g., demographics, SES, and behavior related to PM
37    exposure such as air conditioner use).  Note, that the representativeness analysis (which is

      September 2009                           42           Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
discussed in 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.
       Table 3-4 presents the 15 urban study areas selected for the risk assessment, including
(a) which state it is located in, (b) whether the urban study area was included in the prior risk
assessment, (c) which PM region the urban study area is located in, and (d) the daily and
annual design values considered in selected the location. 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 3-4.    Urban Study Areas Selected for the Risk Assessment.
Urban study
area
Atlanta
Baltimore
Birmingham
Dallas
Detroit
Fresno
Houston
LA
New York
Philadelphia
Phoenix
Pittsburgh
Salt Lake City
Tacoma
St. Louis
State
GA
MD
AL
TX
MI
CA
TX
CA
NY
PA
AZ
PA
UT
WA
MO
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
NW
IM
Annual design
value (jig/m3)
16
16
19
13
17
17
16
20
16
15
13
20
12
10
17
Daily design
value (ug/m3)
35
37
44
26
43
63
31
55
42
38
32
60
55
43
39
10
11
12
13
* SE (Southeast), IM (industrial Midwest), SCA (Southern California), NE (Northeast), NW (Northwest), SW
(Southwest) (See, EPA, 1996, section 6.4 for description of these PM regions).
      September 2009
                                           43
Draft - Do Not Quote or Cite

-------
 2    Figure 3-4    15 urban study areas included in the risk assessment (including seven PM
 3                 regions used to guide selection of study areas).
 4
 5          Once the 15 urban study areas were selected, the next step was to identify the spatial
 6    template to use in defining each study area (i.e., the geographical area associated with each
 7    study area that would be used in identifying which counties and PM2.5 monitors were
 8    associated with a particular study area). For 12 of the 15 urban study areas, we either used a
 9    combined statistical area (CSA) as the basis for the spatial template, or if that was not
10    available, we used a core-based statistical area (CBSA).  The three remaining urban study
11    areas were special cases and were handled as follows: (a) for Baltimore we used counties in
12    the Baltimore CBSA (and not the Baltimore-DC CSA, even though this CSA was available
13    since we felt it unlikely that this entire larger CSA would behave similarly with regard to PM
14    reduction strategies), (b) for Philadelphia, we used the Philadelphia CSA, but excluded Berks
15    County (Reading), and (c) for Tacoma, we only used Pierce County (since we felt it unlikely
16    that efforts to reduce emissions at the "elevated" monitor in Pierce County, would
17    significantly impact monitors in Seattle).
18          As noted above, in a few instances, two or more epidemiological studies used different
19    geographic boundaries for determining which populations were included in their studies. For
20    example, in one study conducted in Birmingham, AL populations from Blount, Jefferson,
21    Shelby, St. Clair, and Walker Counties were included, while another study included the
22    population residing in only  Jefferson County. In such cases, we matched our delineation of
      September 2009
44
Draft - Do Not Quote or Cite

-------
 1    the urban area to that of each study, resulting in two or more different delineations of the
 2    urban area.
 3          As we discuss below, two of the studies on which we rely for our core analysis -
 4    Zanobetti and Schwartz (2009) and Bell et al.  (2008) - are multi-location studies. Zanobetti
 5    and Schwartz (2009) specified the county or counties included in each of the urban areas they
 6    included in their analysis. Bell et al. (2008), however,  did not focus on urban areas, but
 7    instead focused on counties with populations above a specified threshold number.  To limit
 8    the number of different "versions" of a risk assessment location, wherever possible we
 9    specified the counties in a risk assessment location for Bell et al. (2008) to match the set
10    specified for Zanobetti and Schwartz (2009).  This was possible in those cases in which
11    Zanobetti and Schwartz (2009) identified an urban area as a single county, and that county
12    was also included in Bell et al. (2008).  This was the case for several of the risk assessment
13    locations. In some cases, however, Zanobetti  and Schwartz (2009) used a multi-county
14    delineation of an urban area where at least one of the counties was not among those included
15    in Bell et al.  (2008).  In those cases, we had to delineate two definitions of the urban area -
16    one corresponding to Zanobetti and Schwartz  (2009) and the other corresponding to Bell et al.
17    (2008). This was the case for Atlanta, Birmingham, and St. Louis. In both Atlanta and New
18    York, other delineations by other studies forced additional delineation of these urban areas, as
19    shown in Table 3-1 above.
20          Finally, we applied the studies of mortality associated with long-term exposure to
21    PM2.5 to the urban areas as defined by the short-term exposure mortality study, Zanobetti and
22    Schwartz (2009), to  enable meaningful comparisons between estimates of premature morality
23    associated with short-term and long-term  exposure to PM2.5.

24         3.3.3   Selection of epidemiological studies and concentration-response (C-R)
25                functions within those studies
26          As discussed above, we included in the PM2.5 risk assessment only the better-
27    understood health effects for which the weight of the evidence supports a likely causal
28    inference.  Thus, in cases where the majority of the available studies did not report a
29    statistically significant relationship, the effect endpoint was not included. Once it had been
30    determined that a health endpoint would be included in the analysis, however, inclusion of a
31    study on that health  endpoint was not based on statistical significance.
32          A significant change since the previous PM risk assessment is the addition to the
33    relevant epidemiological literature of several multi-city studies.  This type of study has
34    several advantages over single-city studies. First, multi-city studies use the same study design
35    in each of the cities included in the study, so that city-specific results are readily comparable.
36    Second, when they are estimating a single C-R function based on several cities, multi-city
      September 2009                           45           Draft - Do Not Quote or Cite

-------
 1    studies also tend to have more statistical power and provide effect estimates with relatively
 2    greater precision than single city studies due to larger sample sizes, reducing the uncertainty
 3    around the estimated coefficient. Moreover, in a multi-city study the statistical power to
 4    detect an effect in any given city can be supplemented by drawing statistical power from data
 5    across all the cities included in the study (or all the cities in the same region) to adjust city-
 6    specific estimates towards the mean across all cities included in the analysis (or in the same
 7    region). This is particularly useful in those instances, where a city has relatively less data
 8    resulting in a larger standard error for the effect estimate. In this situation, the information on
 9    the C-R relationship in all the other cities included in a multi-city study can be used to help
10    inform an assessment of the C-R relationship in the city in question.  Finally, multi-city
11    studies tend to avoid the often-noted  problem of publication bias that single-city studies
12    confront (in which studies with statistically insignificant or negative results are less likely to
13    get published than those with positive and/or statistically significant results).
14           For this risk assessment, we selected what we considered to be the best study to assess
15    the C-R relationship between PM2.5 and a given health endpoint, and we included other
16    studies for that health endpoint only if they were judged to contribute something above and
17    beyond what we could learn from the primary study selected.
18           A primary study for a given health endpoint had to satisfy the study selection  criteria
19    that we have  used in past PM (and other) risk assessments.  In particular:
20       •   It had to be a published, peer-reviewed study that has been evaluated in the PM ISA
21           and judged adequate by EPA  staff for purposes of inclusion in this risk assessment
22           based on that evaluation.
23       •   It had to directly measure, rather than estimate, PM2.5 on a reasonable proportion of the
24           days in the study.
25       •   It had to either not rely on Generalized Additive Models (GAMs) using the S-Plus
26           software to estimate C-R functions or to appropriately have re-estimated these
27           functions using revised methods.18
28           Because of the advantages noted above, we selected multi-city studies as our primary
29    studies for assessing the risks of premature non-accidental, cardiovascular, and respiratory
30    mortality (Zanobetti and Schwartz, 2009) and cardiovascular and respiratory hospital
31    admissions (Bell et al., 2008) associated with short-term exposure to PM2.5 in our core
32    analysis. In each of these studies, the 15 urban areas selected for the PM risk assessment were
33    among the locations included in their analysis.
      18 The GAM S-Plus problem was discovered prior to the recent final PM risk assessment carried out as part of
      the PM NAAQS review completed in 2006. It is discussed in the 2004 PM Criteria Document (EPA, 2004), PM
      Staff Paper (EPA, 2005c), and PM Health Risk Assessment Technical Support Document (Abt Associates,
      2005).

      September 2009                            46           Draft - Do Not Quote or Cite

-------
 1           Studies often report more than one estimated C-R function for the same location and
 2    health endpoint.  Sometimes models including different sets of co-pollutants are estimated in
 3    a study; sometimes different lag structures are used.  Sometimes different modeling
 4    approaches are used to fit weather and temporal variables in the model. Once a study has
 5    been selected, the next step is to select one or more C-R functions from among those reported
 6    in the study.
 7           Zanobetti and Schwartz (2009) divided the United States into six regions, based on the
 8    Koppen climate classification (Kottek 2006; Kottek et al. 2006)(http://koeppen-
 9    eiger.vuwien.ac.at/).19 They estimated the coefficient of PM2.5 in single-pollutant log-linear
10    models using Poisson regression for each of 112 cities, as well as in two-pollutant models
11    with coarse PM.  They estimated annual models (which assume that the relationship between
12    mortality and PM2.5 is the same through the year), as well as four seasonal models per
13    location. They then used a random effects meta-analysis to combine the city-specific results
14    (Berkey et al.  1998).  Pooling of city-specific results was done at the national level as well as
15    at the regional level, and separately for each season as well as for the annual functions.
16           With respect to the multi-city study for short-term exposure mortality, at the  request of
17    EPA, the authors produced Empirical Bayes "shrunken" city-specific estimates, adjusted
18    towards the appropriate regional mean, using the approach described in Le Tertre et  al.
19    (2005).  This was done for the annual  estimates as well as for each season-specific estimate.20
20    The annual city-specific "shrunken" estimates were used in our core analysis.21  The seasonal
21    estimates were used in a sensitivity analysis. City-specific estimates have the advantage of
22    relying on  city-specific data; however, as noted above, such estimates can have large standard
23    errors (and thus be unreliable); "shrinking" city-specific estimates towards the regional mean
24    estimate is a more efficient use of the data.22  Such "shrinking" can be thought of as
25    combining the advantages of a single-city study (in which the  estimation of a city-specific
      19 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).
      20 These city-specific "shrunken" estimates were provided to EPA (see Zanobetti, 2009).
      21 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.
      22 The degree to which a city-specific estimate  is "shrunken" towards the regional mean depends on the size of
      the standard error of the city-specific estimate relative to that of the regional mean estimate.  The larger the city-
      specific estimate relative to the regional mean estimate, the less shrinkage toward the regional mean.

      September 2009                             47            Draft - Do Not Quote or Cite

-------
 1    coefficient is not influened by data from other locations) with the advantages of a multi-city
 2    study (in which there is much greater statistical power to detect small effects).
 3           Since all models with PM2.5 in Zanobetti and Schwartz (2009) used the same lag
 4    structure (an average of same-day and the previous day's PM2.5), there was no  selection from
 5    among different C-R functions with different lag structures from this study.  There were,
 6    however, both single-pollutant and two-pollutant models (with coarse PM).  We selected the
 7    single-pollutant models, in part to avoid collinearity problems, and in part to be consistent
 8    with most of the other studies used in the risk assessment, which were single-pollutant
 9    studies.
10           Bell  et al. (2008) estimated log-linear models relating short-term exposure to PM2.5
11    and hospital admissions for cardiovascular and respiratory illnesses among people 65 and
12    older, using  Poisson regression, for each of 202 counties in the United States.  They reported
13    both annual  and season-specific results, nationally and regionally (for four regions:
14    Northeast, Southeast, Northwest, and Southwest), but not at the local (city-specific) level.  All
15    cardiovascular hospital admissions models were single-pollutant, 0-day lag models; for
16    respiratory hospital admissions, both single-pollutant 0-day models and single-pollutant 2-day
17    models were estimated. We used the regional, annual C-R functions in our core analysis
18    (identifying  the appropriate region for each of our 15 risk assessment locations).23 For
19    respiratory hospital admissions (for the core analysis), we selected the 2-day lag models,
20    based on evidence that for respiratory effects the strongest associations with  PM exposure
21    may be associated with longer lag periods (on the order of 2 days or more).24 We used the
22    regional season-specific functions in a sensitivity  analysis.
23           We identified two studies that estimated C-R relationships between short-term
24    exposure to PM2.5  and emergency department (ED) visits for cardiovascular  and/or respiratory
25    illnesses. (There were no multi-city studies for this category of health endpoint.)  Tolbert et al.
26    (2007) examined both cardiovascular and respiratory ED visits in Atlanta, GA, using single-
27    pollutant log-linear models with a 3-day moving average (0-day, 1-day, and 2-day lags) of
28    PM2.5. Ito et al. (2007) estimated the relationship between short-term exposure to PM2.5 and
29    ED visits for asthma in New York City (Manhattan). They estimated two single-pollutant
30    models, one for the whole year and one for the period from April through August; in addition,
31    they estimated several two-pollutant models for the period from April through  August.  We
32    selected the  single-pollutant model for the whole year for the core analysis, and we explored
      23 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.
      24 The ISA states that, "Generally, recent studies of respiratory HAs that evaluate multiple lags, have found
      effect sizes to be larger when using longer moving averages or distributed lag models. For example, when
      examining HAs for all respiratory diseases among older adults, the strongest associations where observed when
      using PM concentrations 2 days prior to the HA." (EPA, 2009, section 2.4.2.2).

      September 2009                            48           Draft - Do Not Quote or Cite

-------
 1    the impacts of using the annual versus the April-through-August model, as well as the single-
 2    versus multi-pollutant models in sensitivity analyses.
 3           For the purpose of conducting a sensitivity analysis to show the impact of different lag
 4    structures, different modeling approaches, and single- versus two-pollutant models on
 5    estimates of the risks of premature mortality and hospital admissions associated with short-
 6    term exposure to PM2.5, we selected Moolgavkar (2003). This study reported results for
 7    premature non-accidental, cardiovascular, and respiratory mortality and for cardiovascular
 8    and respiratory hospital admissions associated with short-term exposures to PM2.5 in Los
 9    Angeles, using several different lag structures and several different approaches to modeling
10    the effects of weather and temporal variables.
11           We selected Krewski et al. (2009) as our primary study for assessing the risks of
12    premature mortality associated with long-term exposure to PM2.5 in our core analysis. This
13    study is an extension of the ACS prospective cohort study (Pope et al., 2002), used in the
14    previous PM risk assessment, extending the period of observation of the cohort to eighteen
15    years (1982 - 2000). Krewski et al. (2009) considered mortality from all causes,  as well as
16    cardiopulmonary mortality, mortality from ischemic heart disease, and lung cancer mortality.
17    They presented a variety of C-R functions, in  an effort to show how the results changed with
18    various changes to the method/model used.  It was not apparent from review of the HEI
19    report, that the authors of the study recommended any one of these as clearly superior to the
20    others.  For our core analysis, we selected what  appeared to be two reasonable "standard"
21    options - one corresponding to the earlier exposure period considered in the study, from 1979
22    - 1983, and the other corresponding to the later  exposure period, from 1999 - 2000.   Both C-
23    R functions were based  on follow-up of the cohort through 2000. Both used the standard Cox
24    proportional hazards model, with 44 individual and 7 ecologic covariates. The relative risks
25    for a  10 |ig/m3 change in PM2.5 from which the PM2.5 coefficients were back-calculated were
26    taken from Table 33 of Krewski et al. (2009).25
      25 Note, EPA corresponded with the authors of the Krewski et al., 2009 study to obtain additional clarification
      regarding specific aspects of the study and associated results as presented in the HEI report (Krewski, 2009). In
      response to the EPA's 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, they go 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). Note, that if the 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.
      However, a random effects model with ecological covariates was only provided for the more recent PM
      monitoring period. Therefore, we opted to use the standard Cox model with ecological covariates, since this

      September 2009                            49            Draft - Do Not Quote or Cite

-------
 1          We selected several additional C-R functions from Krewski et al. (2009) to use in
 2    sensitivity analyses carried out in two risk assessment locations (Los Angeles and
 3    Philadelphia).  These are described below.  In addition, we used Krewski et al. (2000)
 4    [reanalysis of the Six Cities Study].

 5         3.3.4  A summary  of selected health endpoints, urban areas, studies, and
 6                concentration-response (C-R) functions used in the risk assessment
 7         A summary of the selected health endpoints, urban areas, and epidemiological studies
 8    used in the risk assessment is given below in Tables 3-5 and 3-6 for short-term and long-term
 9    exposure studies, respectively. A more  detailed overview of the locations, health endpoints,
10    studies, and C-R functions included in the core analysis is given in Table 3-7. An overview of
11    the locations, health endpoints, studies,  and C-R functions included in sensitivity analyses is
12    given in Table 3-8.
      model form had been fitted for both PM monitoring periods. Note, however, that we did consider the random
      effects model form in the sensitivity analysis (see section 3.5.4).

      September 2009                           50           Draft - Do Not Quote or Cite

-------
       Table 3-5.    Locations, Health Endpoints, and Short-Term Exposure Studies Included in the PM2.5 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.
September 2009
51
Draft - Do Not Quote or Cite

-------
        Table 3-6.    Locations, Health Endpoints, and Long-Term Exposure Studies Included in the PM2.5 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.
September 2009
                                              52
                                                Draft - Do Not Quote or Cite

-------
      Table 3-7.    Summary of Locations, Health Endpoints, Studies and Concentration-Response Functions Included in
                   the Core Analysis.*
Risk
Assessment
Location
Atlanta
Baltimore
Counties
Cobb, De Kalb, Fulton,
Gwinnett
Cobb, DeKalb, Fulton,
Barrow, Bartow,
Carroll, Cherokee,
Clayton, Cobb, Coweta,
DeKalb, Douglas,
Fayette, Forsyth,
Fulton, Gwinett, Henry,
Newton, Paulding,
Pickens, Rockdale,
Spalding, Walton
Baltimore city,
Baltimore county
Study/C-R Function
Zanobetti and Schwartz (2009)1
Zanobetti and Schwartz (2009) l
Zanobetti and Schwartz (2009) l
Krewski et al. (2009) 2
Krewski et al. (2009) 2
Krewski et al. (2009) 2
Krewski et al. (2009) 2
Bell et al. (2008)3
Bell et al. (2008)3
Tolbert et al. (2007)
Tolbert et al. (2007)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Health Endpoint
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure Emergency room (ED) visits,
cardiovascular
Short-term exposure Emergency room (ED) visits,
respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Lag Structure
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-,1-day, and
2-day lags
Avg. of 0-,1-day, and
2-day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
September 2009
53
Draft - Do Not Quote or Cite

-------
Risk
Assessment
Location

Birmingham
Dallas
Counties

Blount, Jefferson,
Shelby, St. Clair,
Walker
Jefferson
Dallas
Study/C-R Function
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Health Endpoint
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Lag Structure
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
September 2009
54
Draft - Do Not Quote or Cite

-------
Risk
Assessment
Location

Detroit
Fresno
Counties

Wayne
Fresno
Study/C-R Function
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Health Endpoint
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Lag Structure
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
September 2009
55
Draft - Do Not Quote or Cite

-------
Risk
Assessment
Location

Houston
Los Angeles
Counties

Harris
Los Angeles
Study/C-R Function
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Health Endpoint
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Lag Structure
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
September 2009
56
Draft - Do Not Quote or Cite

-------
Risk
Assessment
Location
New York
Philadelphia
Phoenix
Counties
Kings, New York City
(Manhattan), Queens,
Richmond, Bronx
New York City
(Manhattan)
Philadelphia
Maricopa
Study/C-R Function
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Ito et al. (2007)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Health Endpoint
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure Emergency room (ED) visits,
asthma
Short-term exposure non-accidental mortality
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure non-accidental mortality
Lag Structure
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day andl-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
September 2009
57
Draft - Do Not Quote or Cite

-------
Risk
Assessment
Location

Pittsburgh
Salt Lake City
Counties

Allegheny
Salt Lake
Study/C-R Function
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Health Endpoint
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Lag Structure
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
September 2009
58
Draft - Do Not Quote or Cite

-------
Risk
Assessment
Location

St. Louis
Tacoma
Counties

Jeffferson, Madison
(IL), St. Louis, St.
Louis city, St. Clair (IL)
Madison (IL), St. Louis,
St. Louis city, St. Clair
(IL)
Pierce
Study/C-R Function
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Zanobetti and Schwartz (2009)
Zanobetti and Schwartz (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Krewski et al. (2009)
Bell et al. (2008)
Bell et al. (2008)
Health Endpoint
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Short-term exposure cardiovascular mortality
Short-term exposure respiratory mortality
Long-term exposure all-cause mortality
Long-term exposure cardiopulmonary mortality
Long-term exposure ischemic heart disease mortality
Long-term exposure lung cancer mortality
Short-term exposure HA (unscheduled), cardiovascular
Short-term exposure HA (unscheduled), respiratory
Lag Structure
NA
0-day lag
2-day lag
Avg. of 0-day and 1-
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
Avg. of 0-day and 1
day lags
Avg. of 0-day and 1-
day lags
NA
NA
NA
NA
0-day lag
2-day lag
*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-hr average; the exposure metric for all long-term exposure C-R functions is the annual average.
1 This is a multi-city study; city-specific estimates "shrunken" towards the mean across all cities in a region were supplied to EPA (Zanobetti, 2009).
2 Two C-R functions were used for the core analysis - one corresponding to the earlier exposure period, from 1979 - 1983, and the other corresponding to the
later exposure period, from 1999 - 2000.  Both C-R functions were based on follow-up of the cohort through 2000. Both used the standard Cox proportional
September 2009
59
Draft - Do Not Quote or Cite

-------
hazards model, with 44 individual and 7 ecologic covariates. The relative risks for a 10 ug/m3 change in PM2 5 from which the PM2 5 coefficients were back-
calculated were taken from Table 33 of Krewski et al. (2009).
3 This study estimated four regional C-R functions - for the Northeast, Southeast, Northwest, and Southwest - for each health endpoint. For each risk
assessment location, we used the regional C-R function for the region containing the risk assessment location. The designation of counties to each of these
four regions can be found at http://www.biostat.jhsph.edu/MCAPS/estimates-full.html.
September 2009                                                   60                                Draft - Do Not Quote or Cite

-------
      Table 3-8.    Summary of Locations, Health Endpoints, Studies and Concentration-Response Functions Included in
                   Sensitivity Analyses.
Sensitivity Analysis
Study/C-R Function
Health Endpoint**
Risk Assessment
Location(s)
Single-Factor Sensitivity Analyses:
Impact of using different model choices - fixed effects
log-linear vs. random effects log-linear vs. random
effects log-log C-R function*
Impact of 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
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)
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. (2009) - C-R
functions for each of two
exposure periods
Krewski et al. (2000) [reanalysis
of the Harvard Six Cities study]
Krewski et al. (2009)
Zanobetti and Schwartz (2009) -
seasonal functions vs. annual
function
Bell et al. (2008) - seasonal
functions vs. annual function
Ito et al. (2007)
Moolgavkar (2003)
All-cause, cardiopulmonary, ischemic
heart disease, and lung cancer mortality
associated with long-term exposure
Long-term exposure all-cause mortality
All-cause, cardiovascular, respiratory,
lung cancer mortality associated with
long-term exposure
All-cause mortality associated with
long-term exposure
Non-accidental mortality,
cardiovascular mortality, respiratory
mortality associated with short-term
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
Los Angeles and
Philadelphia
All 15 urban areas
Los Angeles and
Philadelphia
Baltimore, Birmingham,
Detroit, Los Angeles, New
York, Pittsburgh, and St.
Louis
All 15 urban areas
All 15 urban areas
New York
Los Angeles
September 2009
61
Draft - Do Not Quote or Cite

-------
Sensitivity Analysis

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
Study/C-R Function

Moolgavkar (2003)
Moolgavkar (2003)
Zanobetti and Schwartz (2009)
Health Endpoint**
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
Risk Assessment
Location(s)

Los Angeles
Los Angeles
Baltimore, Birmingham,
Detroit, Los Angeles, New
York, Pittsburgh, and St.
Louis
Multi-Factor Sensitivity Analyses:
Impact of using a fixed effects log-linear vs. a random
effects log-log model, estimating incidence down to the
lowest measured level (LML) in the study vs. down to
PRB, and using a proportional vs. hybrid rollback to
estimate incidence associated with long-term exposure
to PM2 5 concentrations that just meet the current
standards
Impact of using season-specific vs. all-year C-R
functions and proportional vs. hybrid rollbacks to
estimate incidence associated with short-term exposure
to PM2 5 concentrations that just meet the current
standards

Zanobetti and Schwartz (2009)
All-cause and ischemic heart disease
mortality associated with long-term
exposure
Non-accidental mortality associated
with short-term exposure
Los Angeles and
Philadelphia
Baltimore, Birmingham,
Detroit, Los Angeles, New
York, Pittsburgh, and St.
Louis
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 conbined 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.
September 2009
62
Draft - Do Not Quote or Cite

-------
 1         3.4  BASELINE HEALTH EFFECTS INCIDENCE DATA
 2         As noted in section 3.2.1 above, the form of C-R function most commonly used in
 3    epidemiological  studies on PM, shown in equation (1), is log-linear. To estimate the
 4    change in incidence of a health endpoint associated with a given change in PM2.5
 5    concentrations using this form of C-R function requires the baseline incidence (often
 6    calculated as the baseline incidence rate times the population) of the health endpoint, that
 7    is, the number of cases per unit time (e.g., per year) in the location before a change in
 8    PM2.5 air quality (denotedyo in equations 3 and 4).
 9          Incidence rates express the occurrence of a disease or event (e.g., asthma episode,
10    death, hospital admission) in a specific period of time, usually per year. Rates are
11    expressed either as a value per population group (e.g., the number of cases in
12    Philadelphia County) or a value per number of people (e.g., the number of cases per
13    10,000 residents in Philadelphia County), and may be age- and sex-specific. Incidence
14    rates vary among geographic areas due to differences in population characteristics (e.g,
15    age distribution) and factors promoting illness (e.g., smoking, air pollution levels).

16         3.4.1   Data sources
17                   3.4.1.1  Mortality
18          We obtained individual-level mortality data for 2006 for the whole United States
19    from the Centers for Disease Control (CDC), National Center for Health Statistics
20    (NCHS).  The data are compressed into a CD-ROM, which contains death information
21    for each decedent, including residence county FIPS, age at death, month of death, and
22    underlying causes (ICD-10  codes).  The detailed mortality data allow us to generate
23    cause-specific death counts at the county level for selected age groups.  Below we
24    describe how we generated  the county-level death counts.

25                   3.4.1.2  Hospital admission and emergency department visits
26          For hospital admissions (HA) and emergency  department (ED) visits, there are
27    multiple data sources:
28       •  Healthcare Cost and Utilization Project (HCUP) Central Distributor  HCUP
29          is a family of health care databases developed through a Federal-State-Industry
30          partnership and sponsored by the Agency for Healthcare Research and Quality
31          (AHRQ). The HCUP databases are based on the data collection efforts of data
32          organizations in participating states. We used  two HCUP databases: the State
33          Inpatient Database (SID) and the State Emergency Department Database (SEDD)
34          respectively. SID/SEDD include detailed HA/ED information for each discharge,
35          including patient county FIPS, age, admission type (e.g., emergent, urgent),

      September 2009                        63             Draft - Do Not Quote or Cite

-------
 1          admission/discharge season, and principle diagnosis (ICD-9 codes). The HCUP
 2          databases can be purchased from the HCUP Central Distributor, although not all
 3          participant states release the data to the Central Distributor.
 4       •  HCUP State Partners. For those HCUP participating states that don't release
 5          their data to the Central Distributor, we contacted the HCUP state partners to
 6          obtain the HA and/or ED data.
 7       •  Communication with the author(s) of selected epidemiological studies The
 8          ED data for Atlanta in 2004 were sent to EPA by one of the authors of Tolbert et
 9          al. (2007).
10          Table 3-9 shows the states for which we obtained data from the HCUP Central
11    Distributor and the HCUP State Partners. The data are at the discharge level if not
12    otherwise noted, and the data year is 2007 for all the states in the table. The column "PM
13    RA Location" indicates the selected risk assessment location(s) where the incidence rate
14    is applied.
15          The necessary baseline incidence data were not available for Atlanta,
16    Birmingham, Philadelphia, Pittsburgh and St. Louis.  Therefore, for each of these five
17    risk assessment locations EPA instead used the baseline incidence rate for a designated
18    surrogate location. Surrogate locations were chosen if they were deemed to be
19    sufficiently similar to the urban area whose baseline incidence data were not available.
20    Surrogate locations are noted in Table 3-9.
21    Table 3-9.
22
Sources of Hospital Admissions (HA) and Emergency Department
(ED) Visit Data.
States
Arizona
California
Illinois
Maryland
Michigan
New York
North Carolina
Texas
Utah
HCUP
Central
Distributor
HA data
NA*
NA
HA data
HA data
NA
HA data
NA
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
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.


      September 2009
                          64
Draft - Do Not Quote or Cite

-------
States

Washington
HCUP
Central
Distributor

HA data
HCUP State
Partner

-
PMRA
Location
City
Tacoma
Notes


 1    *NA denotes "not available, or not available with all variables required for our analysis.  If data were not
 2    available from the HCUP Central Distributor, we contacted the HCUP State Partner.
 3                    3.4.1.3  Populations
 4          To calculate baseline incidence rate, in addition to the health baseline incidence
 5    data we also need the corresponding population.  We obtained population data from the
 6    U.S. Census Bureau (http://www.census.gov/popest/counties/asrh/). These data, released
 7    on May 14, 2009, are the population estimates of the resident populations by selected age
 8    groups and sex for counties in each U.S. state from 2000 to 2008.  We used 2007
 9    populations for calculating most incidence rates except for the ED visit rate in Atlanta.
10    Because the ED visit data obtained from the authors of Tolbert et al. (2007) are for 2004,
11    we used 2004 population estimates for the 20-county Metropolitan area used in the
12    Tolbert et al. study  for the Atlanta area to calculate the ED incidence rates to be applied
13    when using that study in the risk assessment; we then applied the 2004 rates to the 2007
14    population, assuming the ED incidence rates in Atlanta didn't change significantly from
15    2004 to 2007. The  sizes of the populations in the assessment locations that are relevant
16    to the risk assessment (i.e., the populations for which the PM2.5 C-R functions are
17    estimated and to which the baseline incidences refer) are given in Table 3-10.
18
      September 2009
65
Draft - Do Not Quote or Cite

-------
      Table 3-10.   Relevant Population Sizes for PM Risk Assessment Locations.
City
Atlanta, GA - 1
Atlanta, GA - 2
Atlanta, GA - 3
Baltimore, MD
Birmingham, AL - 1
Birmingham, AL - 2
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY - 1
New York, NY - 2
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO - 1
Counties
Cobb, De Kalb, Fulton, Gwinnett
Cobb, De Kalb, Fulton
20-County MSA**
Baltimore city, Baltimore county
Blount, Jefferson, Shelby, St. Clair,
Walker
Jefferson
Dallas
Wayne
Fresno
Harris
Los Angeles
Kings, New York City (Manhattan),
Queens, Richmond, Bronx
New York city (Manhattan)
Philadelphia
Maricopa
Allegheny
Salt Lake
Jefferson, Madison (IL), St. Louis, St.
Louis city, St. Clair (IL)
Population (Year 2006 and 2007)*
All Ages
2006
3,126,000
2,376,000,
4,975,000
1,429,000
1,037,000
660,000
2,338,000
2,012,000
886,000
3,876,000
9,881,000
8,251,000
1,613,000
833,000
3,779,000
1,225,000
991,000
2,093,000
2007
3,198,000
2,421,000
5,123,000
1,426,000
1,044,000
659,000
2,367,000
1,985,000
899,000
3,936,000
9,879,000
8,275,000
1,621,000
1,450,000
3,880,000
1,219,000
1,010,000
2,091,000
Ages ^30
2006
1,817,000
1,400,000
2,831,000
849,000
619,000
397,000
1,285,000
1,176,000
444,000
2,097,000
5,544,000
4,940,000
1,061,000
833,000
2,103,000
790,000
504,000
1,259,000
2007
1,865,000
1,433,000
2,918,000
848,000
625,000
397,000
1,308,000
1,168,000
452,000
2,139,000
5,579,000
4,975,000
1,074,000
833,000
2,167,000
786,000
517,000
1,261,000
Ages £ 65
2006
236,000
191,000
391,000
190,000
131,000
88,000
195,000
236,000
86,000
299,000
1,011,000
1,004,000
201,000
189,000
417,000
208,000
83,000
274,000
2007
245,000
198,000
408,000
189,000
133,000
88,000
199,000
234,000
87,000
307,000
1,030,000
1,013,000
204,000
187,000
432,000
206,000
86,000
275,000
September 2009
66
Draft - Do Not Quote or Cite

-------
City
St. Louis, MO - 2
Tacoma, WA
Counties
Madison (IL), St. Louis, St. Louis city,
St. Clair (IL)
Pierce
Population (Year 2006 and 2007)*
All Ages
2006
1,879,000
764,000
2007
1,875,000
773,000
Ages ^30
2006
1,134,000
437,000
2007
1,134,000
444,000
Ages ^ 65
2006
253,000
79,000
2007
252,000
81,000
* Not all populations listed in the table were used for calculating the incidence rates. As noted above, the population year needs to match the year of the health
data and the population age group needs to match what is used in the epidemiological studies. In addition, 2004 population (all ages) is used for ED visits in
Atlanta-3, which is 4,663,946. Populations in this table are rounded to the nearest 1,000.
** The 20 counties are Barrow, Bartow, Carroll, Cherokee, Clayton, Cobb, Coweta, DeKalb, Douglas, Fayette, Forsyth, Fulton, Gwinett, Henry, Newton, Paulding, Pickens,
Rockdale, Spalding, and Walton.
September 2009
67
Draft - Do Not Quote or Cite

-------
 1         3.4.2  Calculation of baseline incidence rates
 2           To calculate a baseline incidence rate to be used with a C-R function from a given
 3    study, we matched the counties, age groupings, and ICD codes used in that study. For
 4    example, Bell et al. (2008) designated Dallas, TX as Dallas County and estimated a C-R
 5    function for ICD-9 codes 490-492, 464-466, and 480-487 (respiratory HA) among ages
 6    65 and up; we therefore selected only those HA records that had corresponding ICD
 7    codes for ages 65 and up in Dallas County and also selected the population for the same
 8    age group in the same county. The incidence rate is simply the ratio of the selected HA
 9    count to the population.  The same procedure was used to calculate baseline incidence
10    rates for all of the risk assessment locations.26
11           If a C-R function was estimated for a specific season, we selected only those HA
12    records within that season. The season definitions are:  winter (December, January,  and
13    February), spring (March, April, and May), summer (June, July, and August) and fall
14    (September, October, and November).  Note that the HA data for some states didn't
15    include information about admission season but only discharge season or discharge
16    quarter. The admission season was then approximated  using discharge season or
17    discharge quarter.27
18           Some studies (e.g., Bell et al., 2008) look at the unscheduled HAs only, so we
19    excluded scheduled admissions from the analyses to match the study. A HA is
20    unscheduled if the admission type is emergency or urgent.
21           The baseline mortality rates are given in Table 3-11.  The baseline HA and ED
22    visit rates are given in Table 3-12.
23
24
25
26
      26 For Atlanta, Birmingham, Philadelphia, Pittsburgh and St. Louis, the HA data are not available. We
      calculated the hospital admission rates for the surrogates cities. These cities are listed in Table 3-7.
      27 Based on communication with the HCUP state partner in Texas, patients are normally admitted and
      discharged in the same season.

      September 2009                          68             Draft - Do Not Quote or Cite

-------
Table 3-11.   Baseline Mortality Rates (Deaths per 100,000 Relevant Population per Year) for 2006 for PM Risk Assessment
             Locations.*
City
Atlanta, GA - 1
Atlanta, GA - 1
Atlanta, GA - 2
Atlanta, GA - 3
Baltimore, MD
Baltimore, MD
Birmingham, AL - 1
Birmingham, AL - 1
Birmingham, AL - 2
Dallas, TX
Dallas, TX
Detroit, MI
Detroit, MI
Fresno, CA
Fresno, CA
Houston, TX
Houston, TX
Los Angeles, CA
Los Angeles, CA
New York, NY - 1
Age Group
All ages
>30
—
—
All ages
>30
All ages
>30
—
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
Type of Mortality (ICD-10 or ICD-9 Codes)
All-Cause
—
860
—
—
—
1,700
—
1,600
—
—
1,020
—
1,500
—
1,300
—
920
—
1,030
—
Non-accidental
(AOO-R99)
480
—
—
—
950
—
920
—
—
540
—
850
—
620
—
480
—
560
—
630
Cardiovascular
(101-159)
120
—
—
—
270
—
260
—
—
150
—
300
—
190
—
130
—
190
—
270
Respiratory
(JOO-J99)
41
—
—
—
85
—
85
—
—
48
—
67
—
67
—
37
—
57
—
52
Cardio-
pulmonary
(401-440, 460-
519)
—
330
—
—
—
690
—
680
—
—
420
—
700
—
590
—
370
—
510
—
Ischemic
Heart
Disease
(410-414)
—
89
—
—
—
300
—
190
—
—
170
—
360
—
260
—
150
—
250
—
Lung
Cancer
(162)
—
51
—
—
—
110
—
104
—
—
66
—
107
—
66
—
57
—
55
—
COPD
(490-496)
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
29
—
—
September 2009
69
Draft - Do Not Quote or Cite

-------
City
New York, NY - 1
New York, NY - 2
Philadelphia, PA
Philadelphia, PA
Phoenix, AZ
Phoenix, AZ
Pittsburgh, PA
Pittsburgh, PA
Salt Lake City, UT
Salt Lake City, UT
St. Louis, MO - 1
St. Louis, MO - 1
St. Louis, MO - 2
Tacoma, WA
Tacoma, WA
National
National
Age Group
>30
—
All ages
>30
All ages
>30
All ages
>30
All ages
>30
All ages
>30
—
All ages
>30
All ages
>30
Type of Mortality (ICD-10 or ICD-9 Codes)
All-Cause
1,0800
—
—
1,700
—
1,100
—
1,800
—
980
—
1,500
—
—
1,200
810
1,300
Non-accidental
(AOO-R99)
—
—
970
—
600
—
1,090
—
480
—
870
—
—
660
—
750
1,300
Cardiovascular
(101-159)
—
—
280
—
160
—
330
—
110
—
270
—
—
190
—
220
370
Respiratory
(JOO-J99)
—
—
83
—
67
—
96
—
45
—
83
—
—
66
—
76
130
Cardio-
pulmonary
(401-440, 460-
519)
580
—
—
720
—
470
—
770
—
350
—
680
—
—
510
340
580
Ischemic
Heart
Disease
(410-414)
380
—
—
300
—
220
—
350
—
101
—
320
—
—
240
140
240
Lung
Cancer
(162)
56
—
—
120
—
68
—
120
—
37
—
106
—
—
88
53
90
COPD
(490-496)
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
42
71
* Figures in this table are rounded to a two-integer level of precision.
September 2009
70
Draft - Do Not Quote or Cite

-------
       Table 3-12.   Baseline Hospital Admission (HA) and Emergency Department (ED) Rates (Admissions/Visits per 100,000
                     Relevant Population per Year) for 2007 for PM Risk Assessment Locations.*
City
Atlanta, GA - 1
Atlanta, GA - 2
Atlanta, GA - 3
Baltimore, MD
Birmingham, AL - 1
Birmingham, AL - 2
Dallas, TX
Detroit, MI
Fresno, CA
Houston, TX
Los Angeles, CA
Los Angeles, CA
New York, NY - 1
New York, NY - 2
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO - 1
St. Louis, MO - 2
Tacoma, WA
Age Group
—
>65
All ages
>65
—
>65
>65
>65
>65
>65
All ages
>65
>65
All ages
>65
>65
>65
>65
—
>65
>65
Health Endpoints (ICD-9 Codes)
HA, cardio-
vascular (390-
429)
—
—
—
—
—
—
—
—
—
—
—
5,500
—
—
—
—
—
—
—
—
—
HA (unscheduled),
cardiovascular(426
-429, 430-438,
410-414, 440-449)
—
5,700
—
8,600
—
5,700
5,000
8,800
5,600
5,900
—
5,500
6,400
—
8,600
5,020
6,100
3,030
—
5,600
4,500
HA, COPD
(490-496)
—
—
—
—
—
—
—
—
—
—
223
—
—
—
—
—
—
—
—
—
—
HA (unscheduled),
respiratory (490-492,
464-466, 480-487)
—
2,020
—
2,600
—
2,020
2,000
3,000
2,100
2,200
—
2,000
2,030
—
2,600
1,600
1,900
1,200
—
2,600
1,600
ED visits,
cardiovascular (410-
414, 427, 428, 433-
437, 440, 443-445,
451-453)
—
—
690**
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
ED visits, respiratory
(460-465, 466.1, 466.11,
466.19, 477, 480-486, 491-
493, 496, 786.07, 786.09)
—
—
2600**
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
ED visits,
asthma
(493)
—
—
—
—
—
—
—
—
—
—
—
—
—
1,100
—
—
—
—
—
—
—
  * Figures in this table are rounded to a two-integer level of precision.
  ** These are 2004 incidence rates because Tolbert et al. (2007) provided 2004 ED visit data in a 20-county delineation of Atlanta. However, the 2004 rates were
  applied to the appropriate year population in the risk assessment.
September 2009
71
Draft - Do Not Quote or Cite

-------
 1         3.5   ADDRESSING UNCERTAINTY AND VARIABILITY
 2         3.5.1   Overview
 3           An important component of a population health risk assessment is the
 4    characterization of both uncertainty and variability.  Variability refers to the
 5    heterogeneity of a variable of interest within a population or across different populations.
 6    For example, populations in different regions of the country may have different behavior
 7    and activity patterns (e.g., air conditioning use, time spent indoors) that affect their
 8    exposure to ambient PM and thus the population health response.  The composition of
 9    populations in different regions of the country may vary in ways that can affect the
10    population response to exposure to PM - e.g., two populations exposed to the same levels
11    of PM might respond differently if one population is older than the other.  In addition, the
12    composition of the PM to which different populations are exposed may differ, with
13    different levels of toxicity and thus different population responses.  Variability is inherent
14    and cannot be reduced through further research. Refinements in the design of a
15    population risk assessment are often focused on more completely characterizing
16    variability in key factors affecting population risk - e.g., factors affecting population
17    exposure or response - in order to produce risk estimates whose distribution adequately
18    characterizes the distribution in the underlying population(s).
19            Uncertainty refers to the lack of knowledge regarding the actual values of inputs
20    to an analysis. Models are typically used in analyses, and there is uncertainty about the
21    true values of the parameters of the  model (parameter uncertainty) - e.g., the value of the
22    coefficient for PM2.5 in a C-R function.  There is also uncertainty about the extent to
23    which the model is an accurate representation of the underlying physical systems or
24    relationships being modeled (model uncertainty) - e.g., the shapes of C-R functions.  In
25    addition, there may be some uncertainty surrounding other inputs to an analysis due to
26    possible measurement error—e.g., the values of daily PM2.5 concentrations in a risk
27    assessment location, or the value of the baseline incidence rate for a health effect in a
28    population.   In any risk assessment, uncertainty is, ideally, reduced to the maximum
29    extent possible through improved measurement of key variables and ongoing model
30    refinement.  However, significant uncertainty often remains, and emphasis is then placed
31    on characterizing the nature of that uncertainty and its impact on risk estimates. The
32    characterization of uncertainty can be both qualitative and, if a sufficient knowledgebase
33    is available,  quantitative.
34           The selection of urban study areas for the PM2.5 risk assessment was designed to
35    cover the range of PM2.s-related risk experienced by the U.S. population and, in general,

      September 2009                         72             Draft - Do Not Quote or Cite

-------
 1    to adequately reflect the inherent variability in those factors affecting the public health
 2    impact of PM2.5 exposure. Sources of variability reflected in the risk assessment design
 3    are discussed in section 3.5.2, along with a discussion of those sources of variability
 4    which are not fully reflected in the risk assessment and consequently introduce
 5    uncertainty into the analysis.
 6           The characterization of uncertainty associated with risk assessment is often
 7    addressed in the regulatory context using a tiered approach in which progressively more
 8    sophisticated methods are used to evaluate and characterize sources of uncertainty
 9    depending on the overall complexity of the risk assessment (WHO, 2008). Guidance
10    documents developed by EPA for assessing air toxics-related risk and Superfund Site
11    risks (USEPA, 2004b and 2001, respectively) as well as recent guidance from the World
12    Health Organization (WHO, 2008) specify multi-tiered approaches for addressing
13    uncertainty.
14           The WHO guidance presents a four-tiered approach, where the decision to
15    proceed to the next tier is based on the outcome of the previous tier's assessment. The
16    four tiers described in the WHO guidance include:
17       •   Tier 0 - recommended for routine screening assessments, uses default uncertainty
18           factors (rather than developing site-specific uncertainty characterizations);
19       •   Tier 1 - the lowest level of site-specific uncertainty characterization, involves
20           qualitative characterization of sources of uncertainty (e.g., a qualitative
21           assessment of the general magnitude and direction of the effect on risk results);
22       •   Tier 2 - site-specific deterministic quantitative analysis involving sensitivity
23           analysis, interval-based assessment,  and possibly probability bound (high- and
24           low-end) assessment; and
25       •   Tier 3 - uses probabilistic methods to characterize the effects on risk estimates of
26           sources of uncertainty, individually and combined.
27           With this four-tiered approach, the WHO framework provides a means for
28    systematically linking the characterization of uncertainty to the sophistication of the
29    underlying risk assessment. Ultimately,  the decision as to which tier of uncertainty
30    characterization to include in a risk assessment will depend both on the overall
31    sophistication of the risk assessment and the availability of information for characterizing
32    the various sources of uncertainty. EPA staff has used the WHO guidance as a
33    framework for developing the approach used for characterizing uncertainty in this risk
34    assessment.
35           The overall analysis in the PM NAAQS risk assessment is relatively  complex,
36    thereby warranting consideration of a full probabilistic (WHO Tier 3) uncertainty
      September 2009                         73              Draft - Do Not Quote or Cite

-------
 1    analysis. However, limitations in available information prevent this level of analysis
 2    from being completed at this time. In particular, the incorporation of uncertainty related
 3    to key elements of C-R functions (e.g., competing lag structures, alternative functional
 4    forms, etc.) into a full probabilistic WHO Tier 3 analysis would require that probabilities
 5    be assigned to each competing specification of a given model element (with each
 6    probability reflecting a subjective assessment of the probability that the given
 7    specification is the "correct" description of reality). However, for many model elements
 8    there is insufficient information on which to base these probabilities. One approach that
 9    has been taken in such cases is expert elicitation; however, this approach is resource- and
10    time-intensive and consequently, it was not feasible to use this technique in the current
                                                          98
11    PM NAAQS review to support a WHO Tier 3 analysis.
12           For most elements of this risk assessment, rather than conducting a full
13    probabilistic uncertainty analysis, we have included qualitative discussions of the
14    potential impact of uncertainty on risk results (WHO Tierl) and/or completed sensitivity
15    analyses assessing the potential impact of sources of uncertainty on risk results (WHO
16    Tier 2).  Note, however, that in conducting sensitivity analyses, we have used both single-
17    and multi-factor approaches (to look at the individual and combined impacts of sources
18    of uncertainty on risk estimates). Also, as discussed below in section 3.5.4, in conducting
19    sensitivity analyses, we used only those alternative  specifications for input parameters or
20    modeling approaches that were deemed to have scientific  support in the literature (and so
21    represent alternative reasonable input parameter values or modeling options).  This means
22    that the alternative risk results generated in the sensitivity analyses represent reasonable
23    risk estimates that can be used to provide a context, with regard to uncertainty, within
24    which to assess the set of core (base case) risk results (see section 4.5.3).
25           In addition to the qualitative and quantitative treatment of uncertainty and
26    variability which are described here, we have also completed two additional analyses
27    intended to place the risk results generated for the 15 urban study areas  in a broader
28    national context.  The first is a representativeness analysis (described in section 4.4)
29    which evaluates the set of urban study areas against national-distributions of key PM
30    risk-related attributes (with the goal of determining the degree to which the study areas
31    are representative of national trends in these parameters).  The second is a national-scale
32    assessment of long-term mortality  related to PM2.5 exposures (discussed in chapter 5).  In
      28 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.
      September 2009                          74              Draft - Do Not Quote or Cite

-------
 1    addition to providing an estimate of the national impact of PM2.5 on long-term mortality,
 2    this analysis also evaluates whether the set of 15 urban study areas generally represents
 3    the broader distribution of risk across the U.S., or a more focused portion of the national
 4    risk distribution (e.g., the higher-end).
 5           The remainder of this section is organized as follows. Key sources of variability
 6    which are reflected in the design of the risk assessment, along with sources excluded
 7    from the design, are discussed in section 3.5.2. A qualitative discussion of key sources of
 8    uncertainty associated with the risk assessment (including the potential direction,
 9    magnitude and degree of confidence associated with our understanding of the source of
10    uncertainty - the knowledge base) is presented in section 3.5.3. The methods and results
11    of the single- and multi-factor sensitivity analyses completed for the risk assessment are
12    presented in section 3.5.4. An  overall summary of the methods used to address
13    uncertainty and variability for the 15 urban study areas (including the two assessments
14    intended to place the urban study areas in a broader national context) is presented in
15    section 3.5.5.

16         3.5.2  Key sources of variability
17           The risk assessment was designed to cover the key sources of variability related to
18    population exposure and exposure response, to the extent supported by available data.
19    However, as with all risk assessments, there are sources of variability which have not
20    been fully reflected  in the design of the risk assessment and consequently introduce a
21    degree of uncertainty into the risk estimates.  We note, in addition, that while different
22    sources of variability were captured in the risk assessment, it was generally not possible
23    to separate out the impact of each factor on population risk estimates, since many  of the
24    sources of variability are reflected collectively in a specific aspect of the risk model.  For
25    example, inclusion of urban study areas from different PM regions likely provides some
26    degree of coverage for a variety of  factors associated with PM2.5 risk (e.g., air conditioner
27    use, PM2.5 composition,  differences in population commuting and exercise patterns,
28    weather).  However, the model is not sufficiently precise or disaggregated to allow the
29    individual impacts of any one of these sources of variability on the risk estimates to be
30    characterized.
31           Key sources  of potential variability that are likely to affect population risks are
32    discussed below, including the degree to which they are (or are not) fully captured in the
33    design of the risk assessment:
34         •   PM2.s composition:  While the risk assessment did not include modeling of risk
35             associated with different components of PM2.5, the assessment did use effect
      September 2009                         75              Draft - Do Not Quote or Cite

-------
 1             estimates (for a number of the short-term exposure-related health endpoints)
 2             differentiated by region of the country, or differentiated for specific urban
 3             locations (see section 3.3.3 and 3.3.4). While many factors may contribute to
 4             differences in effect estimates (for the same health endpoint) across different
 5             locations, compositional differences in PM2.5 may be partially responsible.
 6             Therefore, while the analysis did not explicitly address compositional
 7             differences in generating risk estimates, potential differences in PM2.5
 8             composition may be reflected in those effect estimates that are differentiated by
 9             region and/or urban study area.  The effect estimates for mortality associated
10             with long-term exposure to PM2 5 are not regionally differentiated and instead, a
11             single national-scale estimate is used. This means that any differences in risks
12             of mortality associated with long-term exposure to PM2.s that are linked to
13             differences in PM2.5 composition (or to any other differences across regions or
14             locations) would not be discernable, since a single national-scale risk estimate
15             is generated for each mortality category. This remains an important limitation
16             of the analysis.  In addition to using region- or location-specific effect estimates
17             for health effects associated with short-term exposures, the selection of urban
18             areas to include in the risk assessment was designed in part to  ensure that areas
19             in different regions of the country, with different PM2.5 composition, were
20             included.

21         •   Intra-urban variability in ambient PMi.s levels: Several recent studies (e.g.,
22             Jerrett et al., 2005) have addressed the issue of heterogeneity of PM
23             concentrations within urban areas and its potential impact on the estimation of
24             premature mortality associated with long-term exposure to PM2.5. Most
25             recently, the HEI Reanalysis II (Krewski et al., 2009), focusing on the ACS
26             dataset, discusses epidemiological analyses completed for Los Angeles and
27             New York City which included more highly-refined (zip code level).
28             characterizations of spatial gradients in population exposure within each urban
29             area based on land-use regression methods and/or kriging. While both analyses
30             provide insights into the issue of intra-urban heterogeneity in PM2.5
31             concentrations and its potential implications for epidemiology-based health
32             assessments, due to the time and resource necessary to integrate them into the
33             risk assessment, we were not able to incorporate these studies  quantitatively.
34             The implications of these studies for interpretation of long-term mortality C-R
35             functions and potential exposure error associated with those functions, is
36             discussed below in section 3.5.3.

37         •   Demographics (e.g., greater concentrations of susceptible populations in
38             certain locations):  We have included multiple urban study areas reflecting
39             differences in demographics in different regions of the country to address this
40             issue.

41         •   Behavior affecting exposure to PM2.s (e.g., time spent outdoors, air
42             conditioning use):  We have incorporated, where available, region- and/or city-
43             specific effect estimates in order to capture behavioral differences across
44             locations that could affect population exposures to PM2.5. However, while
45             these location-specific effect estimates may be capturing differences in
      September 2009                         76              Draft - Do Not Quote or Cite

-------
 1             behavior, they may also be capturing other differences (e.g., differences in the
 2             composition of PM2 5 to which populations are exposed). As noted above, it
 3             was not possible to separate out the impact of these different factors, which
 4             may vary across locations and populations, on effect estimates.

 5         •   Baseline incidence of disease: We collected baseline health effects incidence
 6             data (for mortality and morbidity endpoints) from a number of different sources
 7             (see section 3.4). Often the data were available at the county-level, providing a
 8             relatively high degree of spatial refinement in characterizing baseline incidence
 9             given the overall level of spatial refinement reflected in the risk assessment  as a
10             whole.  Otherwise, for urban study areas without county-level data, either (a) a
11             surrogate urban study area (with its baseline incidence rates) was used, or (b)
12             less refined state-level incidence rate data were used.

13         •   Longer-term temporal variability in ambient PM2.s levels (reflecting
14             meteorological trends, as well as future changes in the mix of PM2.5 sources and
15             regulations impacting PM^.s): Risk estimates for the PM2.5 NAAQS review
16             have been generated using recent years of air quality data. In other words,
17             efforts have not been made to simulate potential future changes  in either the
18             concentrations or composition of ambient PM2 5 in the risk assessment locations
19             based on possible changes in economic activity, demographics or meteorology.
20             Actual risk levels potentially experienced in the future as a result of
21             implementing alternative standard levels may differ from those presented in this
22             report due, in part, to potential changes in these factors related to ambient
23             PM2.5.

24         3.5.3  Qualitative assessment of uncertainty
25           As noted in section 3.5.1, we have based the design of the uncertainty analysis
26    carried out for this risk assessment on the framework outlined in the WHO guidance
27    document (WHO, 2009).  That guidance calls for the completion of a Tier 1 qualitative
28    uncertainty analysis, provided the initial Tier 0 screening analysis suggests there is
29    concern that uncertainty associated with the analysis is sufficient to significantly impact
30    risk results (i.e., to potentially affect decision making based on those risk  results).  Given
31    previous sensitivity  analyses completed for prior PM NAAQS reviews, which have
32    shown various sources of uncertainty to have a potentially significant impact on risk
33    results, we believe that there is justification for conducting a Tier 1 analysis. In fact, as
34    argued earlier, given the complexity of the  overall risk assessment, a full Tier 3
35    uncertainty analysis is warranted for consideration under WHO's guidelines.
36           For the qualitative uncertainty analysis, We have described each source of
37    uncertainty and qualitatively assessed its potential impact (including both the magnitude
38    and direction of the  impact) on risk results, as specified in the WHO guidance.  As shown
39    in Table 3-13, for each source of uncertainty, we have (a) provided a  description, (b)
40    estimated the direction of influence (over, under, both, or unknown) and magnitude (low,


      September 2009                          77             Draft - Do Not Quote or Cite

-------
 1   medium, high) of the potential impact of each source of uncertainty on the risk estimates,
 2   (c) assessed the degree of uncertainty (low, medium, or high) associated with the
 3   knowledge-base (i.e., assessed how well we understand each source of uncertainty), and
 4   (d) provided comments further clarifying the qualitative assessment presented. Table 3-
 5   13 includes all key sources of uncertainty identified for the PM2.5 NAAQS risk
 6   assessment. A subset of these sources has been included in the Tier 2 quantitative
 7   assessment discussed in section 3.5.4.
 9
10
11
      September 2009                         78              Draft - Do Not Quote or Cite

-------
    1
    2
Table 3-13.   Summary of Qualitative Uncertainty Analysis of Key Modeling Elements in the PM NAAQS Risk
                Assessment.
    Source
         Description
                                                  Potential influence of
                                                   uncertainty on risk
                                                       estimates
Direction
Magnitude
Knowledge-
    Base
uncertainty*
                        Comments
  (KB: knowledge base, INF: influence of uncertainty on risk
                         estimates)
A.
Characterizing
ambient PM2 5
levels for study
populations
using the
existing
ambient
monitoring
network
If the set of monitors used in a
particular urban study area to
characterize population
exposure as part of an ongoing
risk assessment do not match
the ambient monitoring data
used in the original
epidemiological study, then
uncertainty can be introduced
into the risk estimates.
  Both
   Low-
  medium
Low-medium
KB and INF: In modeling risk, we focus on those counties that
were included in the epidemiological studies supplying the
underlying C-R functions. This means that, particularly for those
endpoints modeled using C-R functions obtained from more recent
studies, there is likely a close association between the monitoring
network used in the risk assessment and the network used in the
study supplying the C-R function(s). Note, however, that in those
instances where the networks are different (e.g., when older
studies are used, resulting in an increased potential for networks to
have changed), uncertainty may be introduced into the risk
assessment and it is challenging to evaluate the nature and
magnitude of the impact that that uncertainty would have on risk
estimates, given the complex interplay of factors associated with
mismatched monitoring networks (i.e., differences in the set of
monitors used in modeling risk and those used in the underlying
epidemiological study).	
B.
Characterizing
policy-relevant
background
(PRB)
For this analysis, we have used
modeling to estimate PRB
levels for each urban study
area. Depending on the nature
of errors reflected in that
modeling, uncertainty (in both
directions) may be introduced
into the analysis.
  Both
   Low
    Low
INF: Given that the risk assessment focuses primarily on the
reduction in risk associated with moving from the current NAAQS
to alternative standard levels, the impact of uncertainty in PRB
levels on the risk estimates is expected to be low. In addition, for
long-term exposure related mortality, we have based the core
analysis on modeling risk down to LML rather than PRB, which
reduces the significance of the PRB issue in the context of
modeling long-term exposure-related mortality.	
C. Procedure
for adjusting
air quality to
simulate
alternate
standard levels
Uncertainty is associated with
the manner in which monitor
levels are rolled back to
simulate just meeting
alternative standard levels (e.g.,
will localized sources be
addressed resulting in a
geographically focused	
  Both
   Low-
  medium
    Low
INF: There is uncertainty associated with projecting the nature of
strategies likely to be used to achieve alternate standard levels, as
well as the degree to which localized strategies would
disproportionately affect levels at particular (proximate) monitors.
However, the sensitivity analysis completed in support of this risk
assessment suggests that this source of uncertainty (as represented
through the use of proportional vs. hybrid non-proportional
rollback modeling) may have a relatively small impact on long-
         September 2009
                                                          79
                                                               Draft - Do Not Quote or Cite

-------
    Source
         Description
                                                   Potential influence of
                                                    uncertainty on risk
                                                         estimates
Direction     Magnitude
                 Knowledge-
                    Base
                uncertainty*
                                       Comments
                (KB: knowledge base, INF: influence of uncertainty on risk
                                       estimates)
                 reduction in ambient levels, or
                 will more generalized regional
                 strategies be used).	
                                                                               term mortality risk in some locations with other areas showing a
                                                                               slightly larger impact (see section 4.3 for results of the sensitivity
                                                                               analysis addressing this source of uncertainty).	
D.
Characterizing
intra-urban
population
exposure in the
context of
epidemiology
studies linking
PM2 5 to
specific health
effects
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
effect estimates obtained from
epidemiology studies.
  Under
(generally)
  Medium-
    high
  High
KB and INF: Recent analyses in Los Angeles and New York City
based on ACS data (as reported in Krewski et al., 2009)
demonstrate the relatively significant effect that this source of
uncertainty can have on effect estimates (and therefore on risk
results).  These analyses also illustrate the complexity and site-
specific nature of this source of uncertainty. The results of the Los
Angeles analysis suggest that exposure error may result in effects
estimates that are biased low and therefore result in the
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 draft 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"
(draft ISA, section 7.6.3).	
E. 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
   Both
• Low-
  medium
  (long-term
  health
  endpoints)
• Medium
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
         September 2009
                                                            80
                                                                 Draft - Do Not Quote or Cite

-------
    Source
         Description
                                                   Potential influence of
                                                    uncertainty on risk
                                                         estimates
Direction
Magnitude
Knowledge-
    Base
uncertainty*
                         Comments
  (KB: knowledge base, INF: influence of uncertainty on risk
                         estimates)
                 associated with the fitting of
                 statistical effect-response
                 models in epidemiological
                 studies.
                                                (short-term
                                                health
                                                endpoints)
                                              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
                                              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.	
F. 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
   Low-
  medium
Low-medium
INF: Regarding long-term mortality, the ISA suggests that a log-
linear non-threshold model is best supported in the literature for
modeling both short-term and long-term health endpoints.
Although consideration for alternative model forms (Krewski et
al., 2009) does suggest that different models can impact risk
estimates to a certain extent, generally this appears to be a
relatively modest source of overall uncertainty. Particularly if, as
is the case in this risk assessment, we are not extrapolating below
the lowest measured  levels found in the underlying
expidemiological studies. With regard to long-term mortality, the
         September 2009
                                                            81
                                                                 Draft - Do Not Quote or Cite

-------
    Source
         Description
                                                   Potential influence of
                                                    uncertainty on risk
                                                         estimates
Direction
Magnitude
 Knowledge-
    Base
uncertainty*
                         Comments
  (KB: knowledge base, INF: influence of uncertainty on risk
                         estimates)
                                                                                               ISA concludes that, "Using a variety of methods and models, most
                                                                                               of the studies evaluated support the use of a no-threshold, log-
                                                                                               linear model..." (section 2.4.3). Regarding short-term morbidity,
                                                                                               the  ISA states that, "Overall, the limited evidence from the studies
                                                                                               evaluated supports the use of a no-threshold, log-linear model,
                                                                                               which is consistent with the observations made in studies that
                                                                                               examined the PM-mortality relationship." (section 2.4.3).	
G. Addressing
co-pollutants
The inclusion or exclusion of
co-pollutants which may
confound, or in other ways,
effect the PM effect, introduces
uncertainty into the analysis.
  Both
   Low-
  medium
   Medium
INF: With regard to long-term health endpoints, the 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). The
ISA also notes that in some instances, consideration of
copollutants can have a significant impact on risk estimates (i.e.,
the  more refined study of lung cancer mortality in LA as reported
in Krewski et al., 2009 - see ISA, section 7.5.1.1).  With regard to
short-term mortality and morbidity, the ISA generally concludes
that observed associations are fairly robust to the inclusion of
copollutants (see ISA, sections 6.3.8, 6.3.9, and 6.3.10)	
H. 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 PM2 5 sources (both
natural and anthropogenic). If
these compositional differences
in fact translate into significant
differences in public health
impact (per unit exposure) 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.
I. Specifying
lag structure
(short-term
Different lags may have
varying degrees of association
with a particular 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
         September 2009
                                                            82
                                                                 Draft - Do Not Quote or Cite

-------
    Source
         Description
                                                   Potential influence of
                                                    uncertainty on risk
                                                         estimates
Direction     Magnitude
                Knowledge-
                    Base
                uncertainty*
                                          Comments
                   (KB: knowledge base, INF: influence of uncertainty on risk
                                          estimates)
exposure
studies)
endpoint and it may be difficult
to clearly identify a specific lag
as producing the majority of a
PM-related effect (recently,
distributed lags have been
recommended since they allow
for a distribution of the impact
across multiple days of PM
exposure prior to the health
outcome). A lack of clarity
regarding the specific lag(s)
associated with a particular
health endpoint adds
uncertainty into risk estimates
generated for that endpoint.
                                              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."  (ISA, section 2.4.2). This
                                              suggests that uncertainty remains concerning the identification of
                                              appropriate lags, and thus the etiologically relevant time period for
                                              exposure to PM for specific health endpoints.
J.
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.
K. Impact of
historical air
quality on
estimates of
health risk
from long-term
PM25	
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
  Both
  Medium
   Medium
INF: The latest HEI Reanalysis II study (HEI, 2009) which looked
at exposure windows (1979-1983 and 1999-2000) for long-term
exposure in relation to mortality, did not draw any conclusions as
to which window was more strongly associated with mortality.
However, the study did suggest that moderately different effects
estimates are associated with the different exposure periods (with
the more recent period having larger estimates). Overall, the	
         September 2009
                                                            83
                                                                 Draft - Do Not Quote or Cite

-------
    Source
         Description
                                                   Potential influence of
                                                    uncertainty on risk
                                                        estimates
Direction     Magnitude
              Knowledge-
                  Base
              uncertainty*
                                     Comments
              (KB: knowledge base, INF: influence of uncertainty on risk
                                     estimates)
exposures
window is most strongly
associated with mortality.
                                             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 two years (Draft ISA, section 7.6.4).	
L.
Characterizing
baseline
incidence rates
Uncertainty can be introduced
into the characterization of
baseline incidence in a number
of different ways (e.g., error in
reporting incidence for specific
endpoints, mismatch between
the spatial scale in which the
baseline data were captured
and the level of the risk
assessment).	
  Both
 Low-
medium
Low
INF: The degree of influence of this source of uncertainty on the
risk estimates likely varies with the health endpoint category under
consideration.  There is no reason to believe that there are any
systematic biases in estimates of the baseline incidence data. The
influence on risk estimates that are expressed as incremental risk
reductions between alternative standards should be relatively
unaffected by this source of uncertainty.
KB:  The county level baseline incidence and population estimates
at the county level were obtained from data bases where the
relative degree of uncertainty is low.	
    1     * Refers to the degree of uncertainty associated with our understanding of the phenomenon,
    2     (specifically in the context of modeling PM risk)
                                                                           in the context of assessing and characterizing its uncertainty
         September 2009
                                                           84
                                                                Draft - Do Not Quote or Cite

-------
 1          3.5.4   Single and multi-factor sensitivity analyses
 2           We quantitatively examined the impact of several inputs to the risk assessment in
 3    a series of single-factor sensitivity analyses summarized above in Table 3-8.  Rather than
 4    present results for each sensitivity analysis for all of the air quality scenarios considered
 5    in the core analysis, we selected a single air quality scenario - PM2.5 concentrations that
 6    just meet the current standards - to use for the sensitivity analyses.  The one exception to
 7    this was the sensitivity analyses examining the impact of an alternative approach to
                                                                      9Q
 8    simulating just meeting alternative standards (the hybrid rollback).
 9           In discussing the approach used in conducting the sensitivity analysis, we focus
10    first on methods used in assessing long-term exposure related health endpoints followed
11    by the methods  used in assessing short-term exposure related health endpoints. Note, that
12    the results of the sensitivity analyses (including both  single- and multi-factor analyses)
13    are presented and discussed in section 4.3.
14           Because Krewski et al. (2009) presented results based on alternative model
15    specifications only for the later exposure period (1999 - 2000), our sensitivity analyses
16    focusing on the estimates of health effects  incidence associated with long-term exposure
17    to PM2.5 similarly used the C-R functions based on this later exposure period.  Krewski  et
18    al. (2009) considered several alternative modeling approaches  to estimate the relationship
19    between mortality (both all cause and cause-specific) and long-term exposure to PM2.5,
20    providing us the opportunity to examine the impact of alternative modeling approaches
21    on the estimate  of mortality risk associated with long-term exposure.  In particular, we
22    examined the impact of using a random effects log-linear model and of using a random
23    effects log-log model30 (rather than the standard fixed effects log-linear model used in the
24    core analysis) to estimate the risks of all cause mortality,  cardiopulmonary mortality,
25    ischemic heart disease mortality, and lung  cancer mortality associated with long-term
26    exposure in Los Angeles and Philadelphia.31  The coefficient of PM2.5 in the random
      29 Sensitivity analyses focusing on the hybrid 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.
      30 In the log-log model, the natural logarithm of mortality is a linear function of the natural logarithm of
      PM25.
      31 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.
      September 2009                          85              Draft - Do Not Quote or Cite

-------
 1    effects log-linear model was back-calculated from the relative risk reported in Table 9
 2    ("Autocorrelation at MSA and ZCA levels" group - "MSA & DIFF" row) of Krewski et
 3    al. (2009).  The coefficient of PM2.5 in the random effects log-log model was back-
 4    calculated from the relative risks reported in Table 11 ("MSA and DIFF" rows) of
 5    Krewski et al. (2009).
 6           As noted above, for all health endpoints associated with long-term exposure to
 7    PM2.5 we estimated risk associated with PM2.s concentrations above 5.8 |ig/m3 (the LML
 8    for the later exposure period used in Krewski et al., 2009). In a sensitivity analysis we
 9    examined the impact of that limitation by comparing those mortality risk estimates to the
10    mortality risk estimates obtained when we estimated risk associated with PM2.5
11    concentrations above estimated PRB levels.  This sensitivity  analysis was carried out for
12    all cause mortality in all 15 risk assessment urban areas.
13           In addition, we compared the impact of using the primary C-R functions used in
14    the risk assessment, taken from Table 33 of Krewski et al. (2009), versus C-R functions
15    for mortality associated with long-term exposure reported in  another study, Krewski et al.
16    (2000), which was based  on a reanalysis of the Harvard Six Cities Study. The C-R
17    functions estimated in Krewski et al. (2000) from the Harvard Six Cities cohort were
18    estimated for ages 25 and up, while the C-R functions estimated in Krewski et al. (2009)
19    from the ACS cohort were for ages 30 and up.  For purposes  of consistency in the
20    comparison, however, we applied the C-R functions from Krewski et al. (2000) to ages
21    30 and up (and used the baseline incidence rates for that age group as well).32 This
22    sensitivity analysis was carried out for all cause mortality, cardiopulmonary mortality,
23    and lung cancer mortality in Los Angeles and Philadelphia.
24           Finally, we compared estimates of all-cause mortality associated with long-term
25    exposure when PM2.5 levels just meet the current and alternative standards, using the
26    proportional rollback approach versus the hybrid rollback approach. This sensitivity
27    analysis was carried out in Baltimore, Birmingham, Detroit, Los Angeles, New York, and
28    St. Louis.33
29           In all cases, in addition to calculating the incidence of the health effect when an
30    alternative approach  is taken, we calculated the percent difference in estimates from the
31    core analysis resulting from the change in analysis input.  So for example, when we
32    calculated the incidence of all cause mortality associated with long-term exposure to
      32 The baseline incidence rates for ages 25 and up and ages 30 and up are likely to be very similar.
      33 As noted earlier in section 2.4.1, an error was identified in the approach used to simulate ambient PM2 5
      levels just for the Pittsburgh study area for the scenarios involving just meeting the current and alternative
      sets of standards. Therefore, while Pittsburgh had initially been included in this sensitivity analysis, these
      results have been removed since they are based on the set of current standard levels.
      September 2009                         86              Draft - Do Not Quote or Cite

-------
 1    PM2.5 using a random effects log-log model (instead of the fixed effects log-linear model
 2    used in the core analysis), we calculated the percent difference in the result as (incidence
 3    estimated using a random effects log-log model - incidence estimated using a fixed
 4    effects log-linear model)/( incidence estimated using a fixed effects log-linear model).
 5           The primary studies selected to assess mortality risk and risk of hospitalization
 6    associated with short-term exposure to PM2.5 (Zanobetti and Schwartz, 2009, and Bell et
 7    al., 2008, respectively) both provided all-year C-R functions as well as season-specific C-
 8    R functions.  We examined the impact of using season-specific functions by applying
 9    these functions to each season, as defined by the study authors,34 and summing the
10    estimated season-specific incidences of mortality and hospitalizations. We compared
11    these estimates to the estimates obtained by applying the corresponding all-year C-R
12    functions to a year of air quality data.35  This sensitivity analysis was carried out for all
13    15 of the risk assessment urban areas.
14           In addition, Ito et al. (2007) estimated an annual C-R function as well as a
15    seasonal function for April through August for asthma ED visits in New York City. We
16    compared the results of applying the annual C-R function to a whole year of air quality
17    data to the results of applying the seasonal function to only those months (April through
18    August) for which it was estimated.
19           Moolgavkar (2003) estimated C-R functions for several health endpoints - non-
20    accidental and cardiovascular mortality; and cardiovascular and respiratory HAs -
21    associated with short-term exposures to PM2.5 in Los Angeles using different lag
22    structures, different modeling approaches to incorporating weather and temporal
23    variables, and single-pollutant versus multi-pollutant models. This study thus provided
24    an opportunity to show the impact of lag structure, modeling approach, and single- vs.
25    multi-pollutant models, individually, for several health endpoints associated with short-
26    term exposures, although it is difficult to generalize to other locations since the study was
27    only conducted in a single urban area.
      34 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.
      35 The mean season-specific incidence estimates can be summed to produce an all-year estimate of
      incidence. However, the 2.5th and 97.5thpercentile 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.
      September 2009                           87             Draft - Do Not Quote or Cite

-------
 1           Finally, we compared estimates of non-accidental mortality associated with short-
 2    term exposures to PM2.5 (using Zanobetti and Schwartz, 2009) when PM2.5 levels just
 3    meet the current and alternative standards, using the proportional rollback approach
 4    versus the hybrid rollback approach. This sensitivity analysis was carried out in
 5    Baltimore, Birmingham, Detroit, Los Angeles, New York, and St. Louis.
 6           In all cases except the ED visits sensitivity analysis, in addition to calculating the
 7    incidence of the health effect when an alternative approach is taken, we calculated the
 8    percent difference in estimates from the core analysis resulting from the change in
 9    analysis input.36
10           Each single-element sensitivity analysis shows how the estimates of PM2.s-related
11    health effects incidence change as we change a single element of the analysis (such as the
12    form of the C-R function or the way we simulate just meeting a set of standards).
13    Because each of the alternative modeling choices is considered to be a reasonable choice,
14    the results of these single-element sensitivity analyses provide a set of reasonable
15    alternative estimates that may  similarly be considered plausible (see section 4.3). The
16    results of the single-element sensitivity analysis are presented and discussed in section
17    4.3.1.
18           The single-element sensitivity analyses provide insight into which sources of
19    uncertainty may have the greatest impact on risk estimates when acting alone. However,
20    there are several sources of uncertainty in estimating PM2.5-related health effects. To
21    provide a more complete picture of the uncertainty surrounding estimates of PM2.5-
22    related health effects incidence - and to expand the set of reasonable alternative  estimates
23    - we next carried out multi-element sensitivity analyses. The results of the multi-factor
24    sensitivity analysis are presented and discussed in section 4.3.2.
25            The choice of uncertain analysis elements to include in the multi-element
26    sensitivity analyses was guided by the single-element sensitivity analyses. In particular,
27    we selected those modeling choices that had the greatest impacts on the estimates of
28    health effects incidence in the  single-element sensitivity analyses to provide insight into
29    the scope of possible estimates that, while perhaps not based on our first choice of
30    analysis elements, are nevertheless plausible alternative estimates.
31           We identified three analysis elements that substantially affected the estimates of
32    mortality associated with long-term exposure to PM2.5 — the model choice (fixed effects
      36 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.

      September 2009                          88              Draft - Do Not Quote or Cite

-------
 1    log linear vs. random effects log-log), whether effects are estimated associated with PM2.5
 2    concentrations down to the LML in the study (5.8 |ig/m3) or down to PRB, and whether a
 3    proportional or a hybrid rollback is used to simulate PM2.5 concentrations that just meet a
 4    given set of standards.  This resulted in 2x2x2 = 8 different estimates of mortality, all
 5    of which could be considered plausible, based on the fact that the underlying model
 6    choices are all considered reasonable.
 7           We identified two analysis elements that substantially affected the estimates of
 8    mortality associated with short-term exposure to PM2.5 - whether season-specific or all-
 9    year C-R functions were used and whether a proportional or a hybrid rollback approach
10    was used to simulate just meeting the current and alternative standards.

11         3.5.5  Summary of approach to addressing variability and uncertainty
12           The characterization of uncertainty and variability associated with the risk
13    assessment includes a number of elements, which have been discussed in detail above.
14    These include:

15         •  Identification of key sources of variability associated with PM2.5-related
16            population exposure and hazard response  and the degree to which they area
17            captured in the risk assessment (see section 3.5.2). When important sources of
18            variability in exposure and/or hazard response are not reflected in a risk
19            assessment, significant uncertainty can be introduced into the risk estimates that
20            are generated. While not explicitly referenced in the WHO guidance, this
21            assessment (focused on coverage for key sources of variability) could be
22            considered part of a Tier 1  analysis (i.e., the qualitative characterization of
23            sources of uncertainty).

24         •  Qualitative assessment of uncertainty, including both an assessment of the
25            magnitude of potential impact of each source on risk estimates (along with the
26            potential direction of that impact) as well  as an assessment of overall
27            confidence associated with our understanding of that source of uncertainty (see
28            section 3.5.3). This represents a WHO Tier 1 analysis.

29         •  Single-factor  sensitivity analysis intended to evaluate the impact of individual
30            sources of uncertainty on risk estimates (see section 3.5.4). The goal of this
31            assessment is to evaluate the relative importance of these sources of uncertainty
32            in impacting core risk estimates. The single-factor sensitivity analysis
33            represents a WHO  Tier 2 analysis. Note,  that in conducting these assessments,
34            we have used alternative representations of modeling elements that have
35            support in the literature to ensure that the  risk estimates that are generated
36            represent reasonable alternate estimates that can supplement the core risk
37            estimates generated in the analysis (see section 4.5.3).

38         •  Multi-factor sensitivity analysis intended to assess the combined impact of
39            multiple sources of uncertainty on risk estimates (see section 3.5.4). By
      September 2009                         89             Draft - Do Not Quote or Cite

-------
 1             considering the combined effect of multiple sources of uncertainty, this analysis
 2             has the potential to identify any non-linearities which can magnify the impact
 3             of uncertainty on risk estimates, especially if several non-linear factors act in
 4             concert. This also represents a WHO Tier 2 analysis.  Note, that as with the
 5             single-factor  sensitivity analysis results, these risk estimates are also generated
 6             using modeling inputs which have support in the literature and consequently,
 7             they  also represent reasonable alternate estimates that supplement the core risk
 8             estimates (see section 4.5.2).
 9           It is important to reiterate that, due to our inability to characterize overall levels of
10    confidence in alternative model inputs, the uncertainty characterization completed for
11    this risk assessment did not include a full probabilistic assessment of uncertainty and its
12    impact on core risk estimates (i.e., a WHO Tier 3 analysis was not completed).
13    Furthermore, the risk estimates generated using the single- and multi-factor sensitivity
14    analyses do not represent uncertainty distributions, but rather additional plausible point
15    estimates of risk (i.e., we do not know whether they represent risk estimates near the
16    upper or lower bounds  of a true but undefined uncertainty distribution and we do not
17    know the actual population percentiles that they represent). The appropriate use for these
18    reasonable alternate risk estimates in informing consideration of uncertainty in the core
19    risk estimates is discussed in section 4.5.3.
20           In addition to the specific analyses discussed above, we have also completed two
21    additional analyses intended to place the 15 urban study areas in a broader national
22    context with regard to risk. These include a representativeness analysis which evaluates
23    the way the 15 urban study areas compare to national distributions for key PM-related
24    risk attributes (discussed in section 4.4). We have also completed a national-scale
25    assessment of long-term mortality related to PM2.5 exposures (Chapter 5), which, in
26    addition to providing an estimate of the national impact of PM2.5 on long-term mortality,
27    also evaluates whether the set of 15 urban study areas generally represents the broader
28    distribution of risk across the U.S., or a more focused portion  of the national risk
29    distribution (e.g., the higher-end).
      September 2009                         90             Draft - Do Not Quote or Cite

-------
 1                                       4  RESULTS

 2           As discussed above in section 3.1.1, for this risk assessment, we have developed a
 3    core set of risk estimates supplemented by an alternative set of risk results generated
 4    using single-factor and multi-factor sensitivity analysis. The core set of risk estimates
 5    was developed using model inputs believed to have a relatively greater degree of support
 6    in the literature (compared with inputs used in the sensitivity analyses).  Therefore, we
 7    have emphasized the core set of risk estimates in presenting and discussing risk estimates
 8    in this section, with the results of the sensitivity analysis serving to augment the core
 9    estimates as discussed below.
10           The results of the sensitivity analysis allow us to evaluate and rank the potential
11    impact of key sources of uncertainty on  the core risk estimates. In addition, as noted in
12    section 3.5.4, because the sensitivity analysis was conducted using alternative modeling
13    inputs having some degree of support in the literature, the results of the sensitivity
14    analysis also represent a set of reasonable alternatives to the core set of risk estimates that
15    can be used in informing characterization of uncertainty in the core results (see section
16    4.3 below).
17           As discussed in section 2.2 and 3.2, this risk assessment includes consideration of
18    the following air quality scenarios:
19           •   Recent conditions: based on  PM2.5 concentrations characterized through
20              monitoring for the period 2005-2007 at each urban case study location;
21           •   Current NAAQS: based on rolling back PM2.5 concentrations to just meet the
22              current suite of standards in each urban study area (annual standard of 15
23              Hg/m3 and a 24-hour standard of 35 |ig/m3, denoted 15/35);
24           •   Alternative NAAQS: based on rolling back PM2.5 concentrations to just meet
25              an alternative set of standards in each urban study area:
26              o   annual  standard of 13 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted
27                  13/35);
28              o   annual  s
29                  12/35);
30              o   annual  s
31                  13/30);
32              o   annual  standard of 12 |ig/m3 and a 24-hour standard of 25 |ig/m3 (denoted
33                  12/25).
34           In simulating both current and alternative standard levels, for the core analysis,
35    we used a proportional roll-back approach (see section 3.2.3), while a hybrid roll-back
      September 2009                          91             Draft - Do Not Quote or Cite
28              o  annual standard of 12 |ig/m3 and a 24-hour standard of 35 |ig/m3 (denoted

30              o  annual standard of 13 |ig/m3 and a 24-hour standard of 30 |ig/m3 (denoted

-------
 1    approach reflecting the potential for local source control was used for a subset of urban
 2    study areas as part of the sensitivity analysis conducted for this assessment (see section
 3    3.2.3).
 4           As described in section 2.1 and 3.3.2, for this risk assessment, we assessed risk
 5    for 15 urban study areas chosen to provide coverage for the diversity of urban settings
 6    across the U.S. that reflect areas with elevated annual and/or daily PM2.5 concentrations.9
 7    At a minimum all areas selected had recent air quality levels at or above the most
 8    stringent annual and/or 24-hour standard analyzed. In addition, our goal was to select
 9    areas reflecting the heterogeneity in PM risk-related attributes such as sources,
10    composition, demographics, and population behavior.
11           Risk estimates were generated for the following health effects endpoints: (a) long-
12    term exposure-related mortality (all-cause, cardiopulmonary disease-related (CPD),
13    ischemic heart disease-related (IHD) and lung cancer-related), (b) short-term  exposure-
14    related mortality (non-accidental, cardiovascular disease-related (CVD), respiratory), and
15    (c) short-term exposure-related morbidity (hospital admissions (HA) for CVD and
16    respiratory illness and emergency department (ED) visits). Risk estimates are presented
17    separately for each of these 15 study areas, although in  certain circumstances, risk
18    estimates may be restricted to a subset of these locations if, for example, an endpoint is
19    modeled using a C-R function derived from an epidemiological study that was conducted
20    only in a subset of the urban areas.  For the core analysis, long-term exposure mortality
21    risk was modeled down to lowest measured level (LML), because the LML was higher
22    than estimated PRB and because there is substantial uncertainty as to the shape of the
23    concentration-response (C-R) function at concentrations below the LML. For long-term
24    exposure mortality a sensitivity analysis was conducted that estimated risk  down to
25    policy-relevant background (PRB). In contrast, all short-term exposure health effects
26    endpoints were modeled down to PRB, since this was higher than the LML across all
27    studies and for purposes of NAAQS decision making, EPA is focused on risks associated
28    with PM2.5 levels that are  due to anthropogenic sources that can be controlled by U.S.
29    regulations (or through international agreements with neighboring countries).
30           In modeling long-term exposure mortality, for the core analysis, we have based
31    estimates on the latest reanalysis of the American Cancer Society (ACS) dataset, with
32    two sets of risk estimates being generated; one using a C-R function derived by fitting
33    PM2.5 monitoring data from 1979-1983 and a second set based on fitting PM2.5
34    monitoring data from 1999-2000 (Krewski et al., 2009) (see section 3.3.3).  In presenting
35    core risk estimates for long-term mortality, both  sets of estimates are given equal weight.
      September 2009                          92             Draft - Do Not Quote or Cite

-------
 1           In modeling short-term exposure mortality and morbidity for the core analysis, we
 2    have used the latest multi-city studies (Zanobetti and Schwartz, 2009; Bell et al., 2008)
 3    (see section 3.3.3). In the case of short-term exposure mortality,  we obtained and used
 4    city-specific effects estimates derived using empirical Bayes methods from the study
 5    authors (Zanobetti, 2009).  Multi-city studies were favored for the core analysis, since
 6    these studies are not subject to publication bias and because they reflect a diverse set of
 7    locations with regard to the observed relationship between short-term PM2.5 exposure and
 8    health affect response in the population. Additional detail on the specific C-R functions
 9    and related modeling elements such as effects estimates and lag periods used in the core
10    analysis relative to the sensitivity analysis are presented in sections 3.3 and 3.4 and called
11    out where appropriate below as specific risk estimates are discussed.
12           Because the recent conditions air quality scenario spans three years (2005-2007),
13    risk estimates are generated for each of these years, reflecting the underlying air quality
14    data for a particular year. Risk metrics generated for the above health effects endpoints
15    include:

16       •   Annual incidence of the endpoint due to PM2.s exposure (annual incidence):
17           Generated for the population associated with a given urban study area (for a given
18           simulation year), in most cases, these risk estimates include both a point estimate
19           as well as a 95th percentile confidence interval, the latter reflecting sampling error
20           as characterized in the underlying epidemiological study.

21       •   Percent of total annual incidence for the health endpoint due to PM2.s
22           exposure (percent of total incidence attributable to PM2.s): Again, generated for
23           the population associated with a given urban study area (and simulation year), this
24           metric characterizes the fraction of total incidence that is associated with PM2.5
25           exposure.  As with the underlying PM-related incidence estimates, this risk metric
26           also typically includes a 95th percentile confidence interval reflecting sampling
27           error associated with the effects estimate.  Compared with the annual incidence
28           metric which reflects underlying population size for each study area, this risk
29           metric has the advantage of not being dependent on the size of the underlying
30           population, thereby allowing direct  comparison of the potential impact of PM2.5
31           for the health effect endpoint of interest across urban study area locations. For
32           this reason, in discussing risk estimates in this section, the percent of total
33           incidence attributable to PM2.s risk metric is given greater emphasis than the
34           absolute measure of annual incidence attributable to PM2.s.

35       •   Percent reduction in PM2.s-related health effect incidence for an alternative
36           set of standards or the recent conditions scenario, relative to the current
37           standards (percent change from the current set of standards). Also estimated
38           separately for each urban study area and simulation year,  this metric characterizes
39           the degree of risk reduction (for alternative standard levels) or increased risk (for
40           the recent conditions  scenario) relative to the current NAAQS. For this metric, a
      September 2009                         93             Draft - Do Not Quote or Cite

-------
 1           negative value represents an increase in risk (this is the case for the recent
 2           conditions scenario, where risks are actually higher than the current NAAQS).
 3           This metric is positive, or zero, for alternative NAAQS since they either produce
 4           no risk reduction (if ambient air levels under recent conditions are already at or
 5           below that alternative NAAQS level), or a positive risk reduction for alternative
 6           standards resulting in a reductions in ambient PM2 5 concentrations.  We note, that
 7           because this metric is incremental,  it was not possible to generate the 95th
 8           percentile confidence intervals included with the other two "absolute" risk metrics
 9           described above. As with the previous  risk metric, this metric is not dependent on
10           the underlying population size and therefore, allows direct comparison across
11           urban study areas.
12           Tables presenting estimates for these risk metrics are included in Appendix E and
13    referenced in the discussion of risk estimates presented in this section (the detailed results
14    tables themselves are not included in this section due to the large number of tables
15    generated).  In addition to the above risk metrics, we also generated a series of figures
16    based on the third metric describe above (the percent reduction from the current
17    standards). These figures allow the reader to quickly evaluate trends across air quality
18    scenarios and across the 15 urban study areas in terms of the degree of reduction in
19    PM2.s-related risk.  A subset of these figures (for 2007) are in this section, with the full
20    set of figures for all three simulation years being included in Appendix E.
21           Although risk estimates were generated for all three  simulation years, in this
22    chapter core risk estimates primarily from  2007 are presented and discussed for both the
23    recent conditions air quality  scenario and current and alternative standards.  This reflects
24    the observation that generally, 2007 represents a reasonable central year (in terms of the
25    magnitude of risk generated for the three simulated years), when considering results for
26    all modeled health effect endpoints across  the 15 study areas. In addition, 2007 is the
27    most recent year of the three simulated. We note, however, that while we do focus  on
28    2007 in presenting and discussing risk estimates, we include an assessment of general
29    trends across the three simulation years to  gain perspective on year-to-year variation in
30    PM2.s-related risk estimates as assessed here.
31           In presenting the results of the sensitivity analysis, as with the core risk results,
32    we also focus on 2007. This is done to increase compatibility between the core results
33    and the sensitivity analysis results (i.e., allow the sensitivity analysis  results to be used to
34    gain insights into uncertainty associated with the core results, without adding the
35    additional factor of year-to-year variation into the mix). As with the core risk estimates,
36    the full set of sensitivity analysis results for all three simulation years have been
37    generated and are presented in Appendix F.
      September 2009                          94             Draft - Do Not Quote or Cite

-------
 1           For a subset of the urban case studies (e.g., Dallas and Phoenix), incremental
 2    reductions across alternative standards are initially very low (or even zero) reflecting the
 3    fact that recent ambient PM2.5 concentrations for these study areas are well below the
 4    current annual standard levels. For these study areas, meaningful reductions in risk may
 5    not be seen until relatively lower alternative standards are assessed (and results in the
 6    percent reduction from the current set of standards tables and figures may be zero for
 7    several of the less stringent, alternative sets of standards).
 8           For a number of the urban study areas, confidence intervals (and in some
 9    instances, point estimates) for short-term mortality and morbidity incidence and related
10    risk metrics include values that fall below zero.  Population incidence estimates with
11    negative lower-confidence bounds (or point estimates) do not imply  that additional
12    exposure to PM2.5 has a beneficial effect, but only that the estimated PM2.5 effect estimate
13    in the C-R function was not statistically significantly different from zero.  In the case of
14    short-term exposure mortality, where study area-specific effects estimates were used (see
15    section 3.4), several of the urban locations have non-statistically significant effects
16    estimates; these result in incidence estimates with non-positive lower bounds and/or best
17    estimates (e.g., Birmingham, Detroit, and Los Angeles for non-accidental mortality). In
18    the case of short-term morbidity (e.g., HAs), where regional effects estimates were used,
19    one of the regional coefficients (for the southeast) is not statistically  significant,
20    producing incidence estimates including negative values in the confidence interval for
21    urban study areas falling within that region (e.g., Atlanta, Dallas, and Houston, for CV-
22    related HAs).  Lack of statistical significance could mean that there is no relationship
23    between PM2.5 and the health endpoint or it could mean that there was not sufficient
24    statistical power to detect a relationship that actually exists. In the case of PM2 5 and both
25    short-term exposure mortality and morbidity, given the available evidence in the
26    literature, which has resulted in the draft ISA concluding that there is likely to be a causal
27    relationship between short-term PM2 5 exposure and adverse health effects (see section
28    3.3.1), we believe it is reasonable to assume that instances where effects estimates are
29    not-statistically significant are likely to reflect insufficient sample size, rather than the
30    absence of an actual association. We note, however, that (as discussed in section 3.6.3)
31    many factors can potentially result in variations in the magnitude of  effect estimates. In
32    addition to  sample size, these include: source and compositional differences for PM2 5,
33    exposure error associated with the use of ambient monitors as a surrogate for actual
34    exposure, and  differences in population susceptibility and vulnerability.
35           The remainder of this section is organized as follows.  Core modeling results for
36    the recent conditions air quality scenario are presented in section 4.1. Core modeling

      September 2009                          95              Draft - Do Not Quote or Cite

-------
 1    results for just meeting the current NAAQS and just meeting alternative NAAQS are
 2    presented in section 4.2.  The results of the sensitivity analysis (including single-factor
 3    and multi-factor results) are presented in section 4.3.  The results of a representativeness
 4    analysis involving comparison of counties associated with the 15 urban study area
 5    locations against the national distribution of counties with regard to a set of PM-risk
 6    related attributes are presented in section 4.4. The chapter concludes in section 4.5 with
 7    an overall summary and presentation of key observations drawn from consideration of the
 8    core analysis results combined with the results of the sensitivity analysis results. We note
 9    that all of the risk estimates discussed in this chapter pertain to the 15 urban study areas.
10    The results  of the national-scale health impact analysis are presented separately in chapter
11    5, although implications for interpretation of risk estimates generated for the 15 urban
12    study areas  are discussed in section 4.5.4.

13         4.1   ASSESSMENT OF HEALTH RISK ASSOCIATED WITH RECENT
14               CONDITIONS (CORE ANALYSIS)
15          This section discusses core risk estimates generated for the recent conditions air
16    quality scenario,  focusing on the 2007 simulation year (although observations related to
17    the three simulated years are also presented). In discussing results for the recent
18    conditions air quality scenario, we have focused on absolute risk (either above PRB or
19    LML, depending on the health effect endpoint). This reflects the fact that this air quality
20    scenario represents recent conditions within the urban study areas and therefore, does not
21    lend itself well to an incremental  assessment.  The section is organized by health
22    endpoint category, with results discussed in the following order: long-term exposure
23    mortality, short-term exposure mortality and short-term exposure morbidity.

24         4.1.1   Long-term exposure mortality
25          This section summarizes core estimates for long-term exposure mortality (all
26    cause, IHD, CPD and lung cancer) generated for the recent conditions air quality scenario
27    for simulation year 2007.
28          •  Percent of total incidence and PMi.s-related incidence (all cause
29              mortality): the percent of total long-term exposure mortality incidence
30              associated with PM2.5 exposure under recent conditions for the 2007
31              simulation year is estimated to  range from 1.7% (Tacoma) to  5.2% (Fresno)
32              based on the C-R function derived using 1979-1983 PM2.5 monitoring data
33              (Appendix E, Table E-7). The  percent of total incidence for long-term
34              exposure mortality (for all-cause) based on the C-R function derived using
35              1999-2000 PM2.5 monitoring data are somewhat larger, ranging from 2.2%
36              (Tacoma) to 6.7% (Fresno) (Appendix E, Table E-15). Total  all-cause
      September 2009                          96             Draft - Do Not Quote or Cite

-------
 1              mortality incidence associated with long-term exposure to PM2.5 ranges from
 2              about 90 (Tacoma) to 2,200 (L. A.) (based on the C-R function derived using
 3              1979-1983 PM2.5 monitoring data - Appendix E, Table E-3) and from 120
 4              (Tacoma) to 2,900 (L.A.) (based on the C-R function derived using 1999-2000
 5              PM2.5 monitoring data - Appendix E, Table E-12).

 6          •   Percent total incidence and PMi.s-related incidence (IHD-related
 7              mortality): the percent of total long-term exposure mortality incidence (IHD)
 8              associated with PM2 5 exposure under recent conditions for the 2007
 9              simulation year is significantly higher than all-cause mortality, ranging from
10              6.6% (Tacoma) to 17.3% (Birmingham) based on a C-R function derived
11              using 1979-1983 PM2.5 data (Appendix E, Table E-24). Percent of total
12              incidence for this mortality endpoint is even higher when a C-R function
13              based on 1999-2000 PM2 5 monitoring data is  used; ranging from 8.4%
14              (Tacoma) to 23.6% (Fresno) (Appendix E, Table E-33).  We note, that while
15              the percent of total incidence is significantly larger for this health endpoint (3-
16              4 times as high as for total all-cause mortality), overall PM2.s-associated
17              incidence is significantly less than PM2.5-associated incidence for all-cause
18              mortality (on the order of about 50% lower, depending on the city - see
19              results in Appendix E, Table E-12 versus E-30). This reflects the fact that
20              baseline health effect incidence for all-cause mortality is significantly higher
21              than for IHD mortality, which compensates for the greater slope of the IHD
22              mortality effects estimate (see section 3.5).  Total IHD-related mortality
23              incidence associated with long-term exposure to PM2 5 ranges from about 70
24              (Tacoma) to 2,000 (L.A.) (based on the C-R function derived using 1979-1983
25              PM2.5 monitoring data - Appendix E, Table E-21) and from about 90
26              (Tacoma) to 2,600 (L.A.) (based on the C-R function derived using 1999-2000
27              PM2 5 monitoring data - Appendix E, Table E-30).

28          •   Percent total incidence and PM2.s-related incidence (CPD-related
29              mortality): overall percent of total incidence results for CPD mortality
30              associated with long-term PM2 5 exposure under recent conditions for the 2007
31              simulation year fall between those for all-cause and ischemic heart disease.37
32              Specifically, estimates based on the C-R function derived using 1979-1982
33              range from 3.6% (Tacoma) to  10.6% (Fresno) (Appendix E, Table E-42), with
34              results based on the 1999-2000 C-R function being somewhat higher
35              (Appendix E,  Table E-51).  Actual incidence estimates for this mortality
36              endpoint are slightly less than  for all-cause mortality, likely reflecting the fact
37              that, while the effect estimates for CPD mortality are significantly larger than
38              for all-cause mortality, the baseline health effects incidence rate is larger for
39              all-cause mortality, again compensating for the smaller effects estimate.  Total
40              CPD-related mortality incidence associated with long-term exposure to PM2 5
41              ranges from 81 (Tacoma) to 2,300 (L.A.) (based on the C-R function derived
42              using 1979-1983 PM2 5 monitoring data - Appendix E, Table E-39) and from
            37 This is expected, since IHD represents a subset of cardiopulmonary mortality.


      September 2009                         97             Draft - Do Not Quote or Cite

-------
 1              120 (Tacoma) to 3,200 (L.A.) (based on the C-R function derived using 1999-
 2              2000 PM2.5 monitoring data - Appendix E, Table E-48).

 3          •   Percent total incidence and PM2.s-related incidence (lung cancer
 4              mortality): overall percent of total incidence estimates for lung caner
 5              mortality associated with long-term PM2.5 exposure under recent conditions
 6              for the 2007 simulation year are similar to estimates for CPD mortality.
 7              Specifically, using a C-R function derived using 1979-1982 PM2 5 monitoring
 8              data, percent incidence estimates range from 3.5% (Tacoma) to 10.4%
 9              (Fresno) (Appendix E, Table E-60). As expected, estimates derived using the
10              C-R function based on 1999-2000 ambient monitoring data are larger (Seep
11              Appendix E, Table E-69). We note, however, that actual PM2.5-associated
12              lung cancer mortality incidence estimates are significant lower than those
13              generated for CPD mortality, reflecting the significantly lower baseline health
14              effects incidence rates associated with lung cancer mortality. Total lung
15              cancer-related mortality incidence associated with long-term exposure to
16              PM2.5 ranges from about 10 (Salt Lake City) to 240 (L.A.) (based on the C-R
17              function derived using 1979-1983 PM2 5 monitoring data - Appendix E, Table
18              E-57) and from 20 (Tacoma) to 340 (L.A.) (based on the C-R function derived
19              using 1999-2000 PM2.5 monitoring data - Appendix E, Table E-66).

20          •   Pattern in percent of total incidence across urban study areas:
21              Differences across urban study areas  for the different long-term mortality
22              categories can be significant (e.g., a factor of over 2 for all cause mortality
23              between Tacoma and Fresno, the two locations with the greatest disparity in
24              estimated incidence - see Appendix E, Table E-3).  This spread in incidence
25              estimates likely  results primarily from differences in underlying recent
26              conditions ambient air PM2.5 levels across the urban study areas (see
27              Appendix A), with differences in baseline health effects incidence data also
28              playing a role (see section 3.5). However, because the C-R functions used to
29              estimate incidence rates for long-term mortality are not differentiated by
30              region or urban study area (see section 3.4 - i.e., the same function is used for
31              all study areas for a given long-term mortality endpoint), differences in
32              incidence estimates generated across  the urban study areas do not reflect any
33              underlying differences in the potential response of populations (located in
34              different urban areas) to PM2.5 exposure. We note, that as discussed in section
35              3.6.2 and 3.6.3, there  is the potential  for regional differences in PM2 5 sources
36              and composition, as well as other factors related to long-term PM exposure
37              and risk, that may well produce differences in the response of populations to
38              PM2.5 exposure, however, we are not currently able to reflect this factor
39              quantitatively in modeling long-term exposure-related mortality.

40          •   Pattern of percent of total incidence across the three simulated years
41              (2005-2007):  A comparison of all-cause mortality incidence estimates (based
42              on the C-R function derived using 1979-1982 monitoring data)  across the
43              three years (see Appendix E, Tables E-19 through E-21) shows that, while
44              2007 does produce incidence estimates between those estimated for 2005 and
45              2006 for some urban areas (e.g., Tacoma, St. Louis, LA), results for 2007 can
      September 2009                         98             Draft - Do Not Quote or Cite

-------
 1              be the highest of the three years (e.g., Fresno) or the lowest (e.g., Baltimore)
 2              for some locations. Generally, results for the same urban study area across the
 3              three years are fairly similar (results for Birmingham vary by less than 3%
 4              across the years), although they can vary by as much as 30% or more in some
 5              locations (see results for Tacoma in 2005 and 2006).  All of this temporal
 6              variation results from year-to-year variation in the annual average PM2 5 levels
 7              for the study areas (see Appendix A).  This is because other candidate input
 8              parameters, which could also involve temporal variability (e.g., demographics
 9              and baseline incidence rates) were not modeled with year-specific values, but
10              rather using one representative year (see section 3.4.1.3 and 3.5 for
11              demographics and baseline health effects incidence rates, respectively).

12           •  Statistical significance of the effect estimates underlying the risk
13              estimates and implications for interpretation of those risk estimates: All
14              of the effects estimates used in modeling long-term mortality categories are
15              statistically significant and consequently none of the confidence intervals (or
16              associated best estimates) include negative incidence estimates. We note, that
17              this is in contrast to results for recent air quality generated for short-term
18              mortality and morbidity health effects endpoints, in which case some of the
19              effects estimates were not statistically significant,  as noted below.

20           •  Relationship between all-cause and IHD PM2.s-associated incidence
21              estimates for New York City:  The estimated incidence attributable to PM2.5
22              for all cause mortality in New York City (for recent conditions in 2007) is
23              1,500 (see Appendix E, Table E-3). However, estimated incidence for IHD
24              attributable to PM2.5 (for recent conditions in 2007) is  2,000 (Appendix E,
25              Table E-21). This set of estimates is counter to what would  be expected, since
26              IHD is a subset of all-cause mortality. This outcome likely  results from the
27              fact that we are applying national-scale effects estimates to  individual urban
28              study areas, which, in the case of New York City has significantly different
29              baseline incidence rates for all cause and critically, IHD mortality, compared
30              with national estimates.38 If (a) the fraction of all-cause mortality attributable
31              to IHD is greater in New York City, compared with the national average
32              (which is supported by the data cited here) and (b) the effect estimate for IHD
33              mortality is larger than for all-cause (see section 3.3.4), it would then follow
34              that the estimate for all-cause mortality generated  for New York City is biased
35              low.  Specifically, the effect estimate for all-cause mortality obtained from
36              Krewski et al., 2009 reflects a fraction of IHD-related  deaths that is
37              significantly lower than that seen in New York City, which  would result in a
38              downward-bias in the effects estimate for all-cause mortality for New York
39              City.  Consequently, if this explanation holds, then the all-cause PM2.5-
      38 Specifically, the baseline incidence rates for IHD mortality for New York City are 380 per 100,000 while
      national rate is 242 per 100,000 (see section 3.5, Table 3-9). This translates into New York City having
      approximately 1.5 times the rate of IHD deaths relative to the national average.  All-cause mortality
      baseline incidence also differs, although to a lesser extent, with New York City  having 1,077 per 100,000
      and the national-average being 1,327 per 100,000. This translates into New York City having a baseline
      incidence rate for all-cause mortality that is 23% lower than the national average.
      September 2009                          99              Draft - Do Not Quote or Cite

-------
 1              associated mortality estimate for New York City should be larger than what is
 2              presented. It is important to note that many other factors can potentially
 3              influence the magnitude of effects estimates for specific locations (e.g., PM2.5
 4              sources and composition, differences in population susceptibility and
 5              vulnerability).

 6         4.1.2  Short-term  exposure mortality
 7           This section summarizes core estimates for short-term exposure mortality (non-
 8    accidental, IHD, CVD and respiratory) generated for the recent conditions air quality
 9    scenario for simulation year 2007.

10       •   Percent of total incidence and PMi.s-related incidence (non-accidental
11           mortality): the percent of total short-term exposure (non-accidental) mortality
12           incidence associated with PM2.5 exposure under recent conditions for the 2007
13           simulation year is  estimated to range from 0.2% (Los Angeles) to 1.7%
14           (Baltimore) (Appendix E, Table E-78). This estimate is notably lower than that
15           generated for long-term exposure (all cause) mortality (see above).  Total PM2.5-
16           related incidence estimates for this mortality category, vary greatly across the
17           study areas, ranging from about 50 (Birmingham) to 790 (New York) - see
18           Appendix E., Table E-75). Of the 15 urban study areas modeled for this endpoint,
19           three locations had negative lower bound estimates of incidence, reflecting use of
20           non-statistically significant effects estimates (see section 4.0 for additional
21           discussion).

22       •   Percent total incidence (CVD mortality): the percent of short-term exposure
23           mortality incidence (CVD) associated with PM2.5 exposure under recent
24           conditions for the  2007 simulation year is similar to non-accidental mortality,
25           with the highest estimate being 2.3% (Philadelphia) (Appendix E, Tables E-87).
26           Total PM2.s-related incidence estimates for this mortality category also vary
27           greatly across the  study areas (ranging from -50 (Los Angeles) to 510 (New York)
28           - see Appendix E., Table E-84).  Of the 15 urban study areas modeled for this
29           endpoint, 12 locations had negative lower bound estimates of incidence  (and two
30           of these head negative point estimates), reflecting use of non-statistically
31           significant effects  estimates (see section 4.0 for additional discussion).

32       •   Percent total incidence and PM2.s-related incidence (respiratory mortality):
33           the percent of short-term exposure mortality incidence (respiratory) associated
34           with PM2.5 exposure under recent conditions for the 2007 simulation year is also
35           similar to estimates for non-accidental mortality, with the largest estimate being
36           2.9% (Fresno) (Appendix E, Table E-96).  Total PM2.5-related incidence estimates
37           for this mortality category, vary greatly across the study areas, ranging from 8
38           (Tacoma) to 110 (New York) - see Appendix E., Table E-93).  Of the 15 urban
39           study areas modeled for this endpoint, six locations had negative lower bound
40           estimates of incidence, reflecting use of non-statistically significant effects
41           estimates (see section 4.0 for additional discussion).
      September 2009                         100             Draft - Do Not Quote or Cite

-------
 1       •  Pattern in percent of total incidence across urban study areas:  Differences
 2          across urban study areas for the different short-term exposure mortality categories
 3          can be large (e.g., a factor of 2 or higher for non-accidental mortality between
 4          Detroit and Baltimore, the two locations with the greatest disparity in statistically-
 5          significant percent of total incidence estimates - see Appendix E, Table E-78).
 6          We note, that differences in this risk metric across urban study areas are
 7          substantially higher (up to an order of magnitude) if estimates are compared
 8          without consideration for statistical  significance (e.g., compare 0.2% for non-
 9          accidental mortality in Los Angeles with 1.7% in Baltimore - see Appendix E,
10          Table E-78)). The greater overall difference in this risk metric across urban study
11          areas for short-term mortality categories compared with long-term mortality
12          categories (see last section) is to be  expected given that both the  effect estimates
13          as well as patterns of daily ambient PM2 5 levels differ across the urban locations,
14          while with long-term mortality, only annual-average PM2.5 levels differ (as noted
15          earlier, the same effect estimates is used in modeling incidence for all urban study
16          areas for a given long-term exposure mortality category - see section 3.3.4).

17       •  Pattern of percent of total incidence across the three simulated years (2005-
18          2007):  A comparison of non-accidental mortality incidence estimates across the
19          three years (see Appendix E, Tables E-76 through E-78) shows that, while 2007
20          does produce incidence estimates between those estimated for 2005 and 2006 for
21          some urban areas (e.g., Dallas, Houston, Los Angeles, New York, Pittsburgh),
22          results for 2007 can be the highest of the three years (e.g., Atlanta, Fresno) or the
23          lowest (e.g., Detroit, Philadelphia) in some of the 15 urban study areas.
24          Generally, results for the same urban study area across the three years are fairly
25          similar (results for Atlanta vary by less than 2% across the three  years), although
26          they can vary by as much as 26% or more (compare results for Detroit in 2005
27          and 2007). As with the long-term mortality risk metrics, all of this temporal
28          variation results from year-to-year variation in the daily PM2.5 levels for the study
29          areas (see Appendix A), given that other candidate input parameters, which could
30          have temporal variability (e.g., demographics and baseline incidence rates) were
31          not modeled with year-specific values, but rather using one representative year
32          (see sections 3.4.1.3  and 3.5 for demographics and baseline health effects
33          incidence rates, respectively).

34       •  Comparison of long-term exposure and short-term exposure mortality
35          estimates: The PM25-related incidence for short-term exposure  (non-accidental)
36          mortality is notably smaller than the estimate for  long-term exposure (all  cause)
37          mortality, with short-term incidence on an aggregate annual basis generally
38          ranging from 25 to 50% of the long-term (annual) estimate for the same city.
39          Furthermore, the maximum value for short-term (non-accidental) mortality  is
40          about 790 for New York, while the largest value for long-term (all cause)
41          mortality is 2,900 for Los Angeles (both values are for simulation year 2007). It
42          is interesting to note that maximum  estimates for these two mortality endpoints
43          did not occur in the same urban study area.
44
      September 2009                         101             Draft - Do Not Quote or Cite

-------
 1         4.1.3  Short-term exposure morbidity
 2           This section summarizes core estimates for short-term exposure morbidity (HA
 3    for CVD and respiratory; ED visits for CV, respiratory and asthma) generated for the
 4    recent conditions air quality scenario for simulation year 2007.

 5       •   Percent of total incidence and PM2.s-related incidence (CVD HAs): the
 6           percent of total short-term exposure (CV) HAs associated with PM2.5 exposure
 7           under recent conditions for the 2007 simulation year is estimated to range from
 8           0.4% (Atlanta) to 1.65%  (Pittsburgh) (Appendix E, Table E-105).  Total PM2.5-
 9           related incidence estimates for this morbidity category, range from 20 (Salt Lake
10           City) to 810 (New York) (see Appendix E., Table E-102). Of the 15 urban study
11           areas modeled for this endpoint, five locations had negative  lower bound
12           estimates of incidence, reflecting use of non-statistically significant effects
13           estimates (see section 4.0 for additional discussion).

14       •   Percent of total incidence and PM2.s-related incidence (respiratory HAs):  the
15           percent of total short-term exposure (respiratory) HAs associated with PM2.5
16           exposure under recent conditions for the 2007 simulation year is estimated to
17           range from 0.17% (Tacoma) to 1.6% (Fresno) (Appendix E, Table E-l 14). This
18           estimate is similar to estimates generated for total short-term (CV) HAs.  Total
19           PM2.s-related incidence estimates for this morbidity category, range from <10
20           (Tacoma) to 270  (Los Angeles) (see Appendix E., Table E-l 10).  Of the 15 urban
21           study areas modeled for this endpoint, eleven locations had negative lower bound
22           estimates of incidence, reflecting use of non-statistically significant effects
23           estimates (see section 4.0 for additional discussion).

24       •   Percent of total incidence and PM2.s-related incidence (ED visits for CV,
25           respiratory and  asthma illness):  in contrast to the other short-term and long-
26           term exposure endpoints  discussed in this section (and included in the core
27           analysis), ED visit endpoints were assessed for specific urban case study
28           locations, using epidemiological studies (and associated effects estimates) derived
29           specifically for those locations. Percent of total incidence estimates for both
30           endpoints for the recent conditions air quality  scenario was 0.6% (see Appendix
31           E, Table E-123) with PM2.5-related incidence estimates being 220 (CV) and 830
32           (respiratory) for recent conditions in simulation year 2007.  The results for
33           Atlanta both included negative lower bound estimates, reflecting use of non-
34           statistically significant effects estimates (see section 4.1 for additional
35           discussion). Percent of total incidence attributable to PM2.5 for ED visits in New
36           York (for asthma) is 6.2% while the related total PM2.s-associated incidence
37           estimate is 1,100  (Appendix E, Table E-120).  The effect estimates used in
38           evaluating ED visits for New York is statistically significant.

39       •   Pattern in percent of total incidence across  urban study areas: Differences
40           across urban study areas  for the different short-term exposure HAs categories can
41           be significant (e.g., a factor of over 3 for CV HAs between Houston and
42           Pittsburgh, the two locations with the greatest disparity in percent of total
43           incidence estimates - see Appendix E, Table E-105).  We note, that variation in
      September 2009                         102             Draft - Do Not Quote or Cite

-------
 1          risk estimates across cities for this endpoint category is significantly smaller than
 2          that for short-term mortality.  This is not unexpected since location-specific
 3          effects estimates were not used in modeling short-term exposure HAs (regional
 4          estimates were used), while short-term exposure mortality, as noted earlier, was
 5          modeled using location-specific effects estimates.

 6       •  Pattern of percent of total incidence across the three simulated years (2005-
 7          2007):  A comparison of CV illness-related HAs incidence estimates across the
 8          three years (see Appendix E, Tables E-100 through E-102) shows that, while 2007
 9          does produce incidence estimates between those estimated for 2005 and 2006 for
10          six of the 15 urban study areas, for the remaining locations, estimates for 2007
11          sometimes represent the highest and sometimes the lowest estimates.  Generally,
12          results for the same urban study area across the three years are fairly similar
13          (results for Atlanta vary by less than 2% across the years), although they can vary
14          by as much as 17% or more (compare results for St Louis in 2005 and 2007). As
15          with the long-term and short-term exposure mortality risk metrics, all of this
16          temporal variation results from year-to-year variation in the daily PM2.5 levels for
17          the study areas (see Appendix A), given that other candidate input parameters,
18          which could have temporal variability (e.g., demographics and baseline incidence
19          rates) were not modeled with year-specific values, but rather using one
20          representative year (see section 3.4.1.3 and 3.5 for demographics and baseline
21          health effects incidence rates, respectively).

22         4.2  ASSESSMENT OF HEALTH RISK ASSOCIATED WITH JUST
23              MEETING THE  CURRENT AND ALTERNATIVE STANDARDS
24              (CORE ANALYSIS)
25          This section discusses core (i.e., base case) risk estimates generated for the
26    current set of standards and alternative sets of standard, focusing on the 2007 simulation
27    year (although  general trends in observations across the three simulated years are
28    discussed to a limited extent).  The results discussed below are based on results from 14
29    of the 15 study areas, with Pittsburgh being excluded at this time due to an error that was
30    identified in the approach used to simulate ambient PM2.5  levels just for the Pittsburgh
31    study area for the scenarios involving just meeting the current and alternative sets of
32    standards. As noted earlier in  section 2.4.1, there was insufficient time after identifying
33    this error to either generate corrected risk estimates or remove the erroneous risk
34    estimates from the summary tables (presented in Appendix E).  We will correct this error
35    and release updated results for the Pittsburgh study area as soon as is practicable and will
36    include the corrected results in the next version of this document.
37          In discussing risk estimates for the current and alternative standards, we have
38    emphasized the incremental reduction (delta) across sets of standards rather than focusing
39    on the absolute degree of risk (total PM2.5-related incidence) for a particular set of
40    standards. This reflects the fact that we have greater confidence in the ability of the risk
      September 2009                         103             Draft - Do Not Quote or Cite

-------
 1    models to differentiate risk between sets of standards, since this requires the models to
 2    estimate risk for ambient air PM2.5 levels likely near or within the range of ambient air
 3    quality data used in the underlying epidemiology studies.  By contrast, estimates of
 4    absolute risk (for a given air quality scenario) require the models to perform at the lower
 5    boundary of ambient air PM2.5 levels reflected in the studies (i.e., down to the LML
 6    reflected in the long-term  exposure mortality epidemiology studies or down to PRB
 7    levels in the short-term exposure morbidity and mortality studies). There is greater
 8    overall uncertainty in risk estimates generated based on the contribution to risk of
 9    exposures at these lower ambient air PM2.5 levels.
10           This section is organized by health endpoint category, with results discussed in
11    the following order:   long-term exposure mortality, short-term exposure mortality and
12    short-term exposure morbidity. We note, that observations presented in the last section
13    regarding the statistical significance of effects estimates used in generating risk estimates
14    and their implications for interpretation of those risk estimates also hold for estimates
15    presented in this section. Consequently, observations regarding risk results with
16    confidence intervals including negative estimates are not presented here and the reader is
17    referred back to the earlier discussion in section 4.1.
18           An important factor to consider in interpreting the risk estimates for both the
19    current set of standards and sets of alternative  standards is whether the annual or 24-hour
20    standard for a given pairing of standards is controlling for a particular area.39 This factor
21    can have a significant impact  on the pattern of risk reductions predicted for a given
22    location under the simulation  of just meeting a specific  set of standards.  A brief
23    overview of which urban study areas are predicted to have  risk reductions under the set of
24    current standards and alternative sets of standards included in the risk assessment is
25    presented below.
26        •   Current set  of standards: Two of the 14 urban study areas  (Dallas and Phoenix)
27           have both annual and 24-hr design values that are below the matching current
28           standard levels of 15 and 35 |ig/m3, respectively (see Table 3-3).40 This means
29           that these two urban study areas would not  have any reduction in long-term or
30           short-term exposure risk under the current standards, as reflected in the
      39 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 Table 3-3 for the annual and 24-hr
      design values for each of the urban study areas). 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.
      40 As noted earlier, Pittsburgh has been excluded due to an error that was identified in the approach used to
      simulate ambient PM2 5 levels just for the Pittsburgh study area for the scenarios involving just meeting the
      current and alternative  sets of standards. Consequently, the number of study areas discussed in relation to
      the current set of standards (as well as alternative sets of standards) is 14, rather than 15.
      September 2009                          104             Draft - Do Not Quote or Cite

-------
 1          summaries presented below (note, however, that both areas have predicted risk
 2          reductions under some of the alternative sets of standards).

 3       •  Alternative sets of standards focusing on lower annual standard levels (13/35
 4          and 12/35): Dallas and Phoenix had annual and 24-hour design values which
 5          were lower than the 13/35 set of alternative standards, but annual design values
 6          were above the 12/35  set of alternative standards (see Table 3-3). For two of the
 7          other study areas (Fresno and Los Angeles), where the 24-hr standard was
 8          controlling, simulating just meeting the current standard resulted in significant
 9          reductions in the annual PM2.5 levels, such that no risk reduction was  seen for the
10          13/35 alternative set of standards (i.e., adjusted annual PM2.5 levels for these
11          study areas under the current set of standards were already below 13 jig/m3). We
12          note, however, that Los Angeles did show risk reductions under the 12/35 set of
13          standards, while Fresno continued not to have predicted risk reductions under the
14          12/35 set of alternative standards due to the significant reduction in annual levels
15          associated with just meeting the current set of standards.  Because Tacoma and
16          Salt Lake City already had annual design values (10 and 12 |ig/m3, respectively)
17          at or below the 12 |ig/m3 associated with the lower of these two alternative sets of
18          standards, neither study area exhibited risk reductions.

19       •  Alternative sets of standards focusing on combinations of lower annul and
20          lower 24-hour standard levels (13/30,12/25): Because of the combination of
21          lower 24-hr and annual levels, 13 of the 14 urban study areas had risk reductions
22          under the 13/30 standard level (only Dallas continued not to have predicted risk
23          reductions), while all  14 urban study areas exhibited risk reductions under the
24          12/25 alternative set of standards.
25          Note, that this pattern of urban areas experiencing risk reductions under various
26    standard levels, is reflected in the detailed discussion of risk estimates provided below in
27    sections 4.2.1 through 4.2.3.

28         4.2.1   Long-term exposure mortality
29          This section summarizes core estimates for long-term exposure mortality (all
30    cause, IFID, CPD and lung cancer) generated for the current and alternative sets of
31    standards for simulation year 2007.

32       •  Comparison of recent conditions risk with risk associated with current
33          standards:  a shift in ambient air PM2 5 levels from recent conditions to the
34          current standards is predicted to result in notable reductions in the percent of total
35          incidence (for long-term exposure all cause mortality) attributable to PM2 5 across
36          most of the urban case study locations.  For example, Fresno is estimated to have
37          a -60% reduction in percent of total incidence attributable to PM2.5 under the
38          current standards for simulation year 2007 using C-R functions based on either
39          ambient monitoring period (relative to recent conditions - see Appendix E, Table
40          E-6  and E-15). However,  a subset of urban case studies with recent ambient air
41          PM2.5 levels at or below the current standards (e.g., Dallas, Phoenix - see
42          Appendix E, Table E-6 and E-15) are not  estimated to experience reductions in
      September 2009                         105            Draft - Do Not Quote or Cite

-------
 1          long-term exposure mortality-related risk.  A similar pattern of risk reduction  is
 2          seen for the other long-term mortality categories, with most locations having
 3          estimated reductions of up to -60% and a subset of locations not estimated to
 4          have any reductions because their recent conditions ambient air PM2.5 levels are
 5          already at or below the current standards. This pattern of risk changes associated
 6          with moving from recent conditions to the current standards is depicted
 7          graphically in Figures 4-1 through 4-8.  We note, however, that in these plots,  risk
 8          for each of the long-term mortality categories is expressed in terms of the percent
 9          change relative to the current standards, and thus, the values are depicted as
10          negative values reflecting the fact that these figures focus on reductions in risk for
11          alternative sets of standards, relative to the current set of standards.

12       •  Trends in risk reduction across  alternative sets of standards focusing on
13          lower annual levels (13/35 and 12/35 combinations): reducing ambient PM2 5
14          levels to meet alternative sets of standards with lower annual levels (i.e., 13/35
15          and 12/35) is estimated to produce a systematic reduction in PM25-related all
16          cause mortality for roughly half of the urban study areas. Specifically, in
17          simulation year 2007, 8 of the 14 study areas would see estimated reductions in
18          PM2.5-related all-cause mortality on the order of 5 to -25% under a 13/35
19          standard level combination with this reduction increasing to between 10 and 35%
20          under the  12/35 combination (see  Appendix E, Tables E-9 and E-18). These
21          percentage reductions hold for incidence estimates generated using C-R functions
22          based on ambient PM2.5 data from both ambient monitoring periods (see
23          Appendix E, Tables E-9 and E-18). The estimated degree of reduction in PM2 5-
24          related mortality across these standard combinations with reduced annual levels is
25          similar for the other long-term exposure mortality categories. This pattern of
26          estimated risk reductions associated with moving from the current standards to
27          alternative standard combinations (including lower annual levels) is depicted
28          graphically in Figures 4-1 through 4-8.

29       •  Trends in risk reduction across  alternative sets of standards focusing on
30          lower combinations of 24-hour and annual levels (13/30 and 12/25
31          combinations):  Under the 13/30  combination, PM2.5-related all cause mortality
32          is reduced by a slightly greater amount than under the 13/35 combination
33          discussed above with estimated percent reductions ranging from 14 to 44% - see
34          Appendix E, Tables E-9 and E-18. With the more stringent 24-hour level (but
35          same alternative annual level of 13), this standard level combination is estimated
36          to reduce long-term exposure mortality risk in 13 of the  14 urban study areas,
37          which is due to the fact that the  24-hour standard level is controlling in most of
38          the urban  study area locations for  this combination of standards.  When the most
39          stringent set of standards (12/25) included in this analysis is modeled, the overall
40          degree of reduction in PM2 5-attributable all cause long-term mortality is the
41          largest (relative to the current standards) with estimated percent reductions
42          ranging from 12 to 89% - see Appendix E, Tables E-9 and E-18 (and all of the
43          urban study areas are predicted to experience some degree of risk reduction).  We
44          note that estimates for the other long-term exposure mortality categories in terms
45          of risk reductions associated with  these two alternative standard combinations are
      September 2009                         106            Draft - Do Not Quote or Cite

-------
 1           similar to those presented above for all-cause.  This pattern of risk reductions
 2           associated with moving from the current standards to alternative standard
 3           combinations (including lower 24-hour and annual levels) is depicted graphically
 4           in Figures 4-1 through 4-8.

 5       •   Pattern of reduction of PM2.5-associated long-term mortality incidence across
 6           urban study areas under alternative combinations of standards:  Differences
 7           in the degree of reduction of PM25-related long-term exposure mortality risk
 8           across the 14 urban study  areas can be substantial. For example, considering
 9           reductions in all-cause mortality under the most stringent alternative standards
10           combination (12/25), percent reductions in incidence (relative to risk associated
11           with just meeting the current standards) range from 12% (Dallas) to 89%
12           (Tacoma) (Appendix E, Tables E-9 and E-18). In the case of Dallas, reductions
13           are more modest because of relatively low recent conditions ambient air PM2.5
14           levels (relative to the alternative standards considered - see Appendix A).  With
15           Tacoma, the greater degree of risk reduction results from having a single monitor
16           with an annual average PM2.5 estimate (under recent conditions) that is lower than
17           the alternative standards considered, but 24-hour average levels that are elevated
18           relative to the 24-hour standard. Therefore, for Tacoma, the most stringent
19           alternative standard combination (12/25), with a 24-hour level of 25 |ig/m3, has a
20           relatively large impact on overall risk, as reflected in the 89% reduction in the risk
21           estimate.  Both of these examples illustrate that, with regard to the degree of risk
22           reduction under alternative sets of standards, variation across urban study areas
23           not only reflects variation in baseline incidence rates, but also a number of factors
24           related to ambient air PM2.5 levels, including absolute levels in terms of both
25           annual-averaged values and the distribution of 24-hour levels, and how these
26           translate into design values for a given study area. Figures 4-1 through 4-8
27           provide a ready means for comparing the spread in the relative degree of risk
28           reduction for long-term exposure mortality resulting from alternative sets of
29           standards across the 14 urban study areas.

30       •   Pattern of percent reduction of total incidence across the three simulated
31           years (2005-2007):  Reductions in PM2.5-associated long-term exposure mortality
32           associated with alternative sets of standards across the simulation years (2005-
33           2007) do display significant variation, however, this variation is smaller than the
34           variation in risk reductions across urban study areas for the same simulation year
35           (as discussed in the previous bullet).  For example, for long-term PM2.5-related
36           all-cause mortality, the two urban study areas with the greatest degree of
37           reduction under the most stringent standard level combination (12/25) (Salt Lake
38           City and Tacoma), retain that status across all three years and have percent
39           reductions that vary by 12 and 32%, respectively, across the three simulation
40           years (based on comparison of results in Appendix E, Tables E-7 through E-9 and
41           Tables E-16 through E-18).  Variation across the three simulation years (for the
42           same urban study area) for other long-term mortality categories and for other
43           alternative suite of standards is  generally somewhat below this range, with greater
44           variation in degree of risk reduction (across the three years for a given study area)
45           seen with more stringent sets of standards, and with urban study areas modeled
      September 2009                         107             Draft - Do Not Quote or Cite

-------
1          using fewer PM2.5 monitors. It appears that, as a generality, urban locations with
2          a greater number of monitors, such as Los Angeles, are buffered somewhat from
3          year-to-year variations in annual levels, relative to locations with a smaller
4          number of monitors, such as Tacoma.
     September 2009                        108            Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
Figure 4-1.
 5
 6
 7
 8
 9
10
11
12
13
                Estimated Percent Reductions From the Current Standards to
                Alternative Set of Standards in All Cause Mortality Associated with
                Long- Term Exposure to PM2.5 (Exposure Period: 1979 - 1983):
                Based on 2007 Air Quality Data.*
                100%
             re
             •o
             c
                 20%
                -40%
       •g    0%
       g
       3
       o
       I
       c
       o
       'o  -80%
       3
       T3
       HI
       a:
             £  -140%
                -180%
                -200%



                       2007 air     15/35**
                        quality
                                            13/35
                                                      12/35
                                                               13/30
                                                                         12/25
                                         Alternative Standard
                             -»- Atlanta, GA 571 (370-769); 3.6% (2.3% -4.8%)
                             -m- Baltimore, MD 439 (284-592); 3.1% (2% -4.2%)
                             -*- Birmingham, AL 335 (217-451);  3.3% (2.2% -4.5%)
                             -*- Dallas, TX 311 (201 -420); 2.3% (1.5% -3.1%)
                             -*- Detroit, Ml 366 (236-494); 2.1% (1.3% -2.8%)
                             -•- Fresno, CA 112 (73-152); 2% (1.3% -2.6%)
                             -i- Houston, TX 687 (445-926); 3.5% (2.3% -4.7%)
                             - Los Angeles, CA 979 (632-1323); 1.7% (1.1% -2.3%)
                                 New York, NY 1067 (690-1442);  2% (1.3% -2.7%)
                             -»- Philadelphia, PA 413 (267-557); 2.8%  (1.8% -3.8%)
                             -•- Phoenix, AZ 451  (292-610); 1.8% (1.2% -2.5%)
                             -*- Pittsburgh, PA 236 (153-319); 1.7% (1.1% -2.3%)
                                 Salt Lake City, UT 52 (34-71); 1% (0.7% -1.4%)
                                 St. Louis, MO 545 (353-735); 2.9% (1.9% -3.9%)
                                 Tacoma.WA 53 (34-72); 1% (0.7% -1.4%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
       September 2009
                                              109
                                                                Draft - Do Not Quote or Cite

-------
 1    Figure 4-2.   Estimated Percent Reductions From the Current Standards to
 2                   Alternative Set of Standards in All Cause Mortality Associated with
 3                   Long-Term Exposure to PM2.s (Exposure Period: 1999 -
 4                   2000): Based on 2007 Air Quality Data.*
             •5
             re
             •O
             o
             I
             c
             o
             u
             •o
                        2007 air
                         quality
                                  15/35
                                             13/35
                                                       12/35
                                                                 13/30
                                                                            12/25
                                          Alternative Standard
 5
 6
 1
 8
 9
10
                              -»- Atlanta, GA 731 (467-990); 4.6% (2.9%-6.2%)
                              -•- Baltimore, MD 563 (359-764); 4% (2.5%-5.4%)
                              -*- Birmingham, AL 429 (274-581); 4.3% (2.7%-5.8%)
                                  Dallas, IX 399 (254-542); 3% (1.9%-4.1%)
                              -*- Detroit, Ml 469 (299-638); 2.7% (1.7%-3.6%)
                              -•- Fresno, CA 144 (92-196); 2.5% (1.6%-3.4%)
                              -(- Houston, TX 880 (562-1193); 4.5% (2.9%-6.1%)
                              	 Los Angeles, CA 1257  (799-1711);  2.2% (1.4%-3%)
                                  New York, NY 1370 (871 -1863); 2.6% (1.6%-3.5%)
                              -»- Philadelphia, PA 530 (338-719); 3.6% (2.3%-4.9%)
                              -m- Phoenix, AZ 580 (369 - 789); 2.3% (1.5% - 3.2%)
                              -*- Pittsburgh, PA 303 (193-413); 2.2% (1.4%-3%)
                                  Salt Lake City, UT  67 (43 - 91); 1.3%  (0.8% -1.8%)
                                  St. Louis, MO  698 (445-948); 3.7% (2.3%-5%)
                              -•- Tacoma.WA 69 (44-94); 1.3% (0.8% -1.8%)
*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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                            110
Draft - Do Not Quote or Cite

-------
 1    Figure 4-3.    Estimated Percent Reductions From the Current Standards to
 2                    Alternative Set of Standards in Ischemic Heart Disease Mortality
 3                    Associated with Long-Term Exposure to PM2.s (Exposure Period:
 4                    1979 - 1983): Based on 2007 Air Quality Data.*
                        100%
                       -200%
                               2007 air     15/35**
                                quality
                                                    13/35
                                                              12/35
                                                                        13/30
                                                                                  12/25
                                                 Alternative Standard
 5
 6
 1
 8
 9
10
11
                                    -»- Atlanta, GA 221 (181-260); 13.3% (10.9% -15.6%)
                                    -m- Baltimore, MD 296 (243-349); 11.7% (9.6%-13.7%)
                                    -*- Birmingham, AL 149 (122-175); 12.5% (10.2% -14.7%)
                                        Dallas, TX 191 (156-226); 8.8% (7.2%-10.4%)
                                    -*- Detroit, Ml 327 (267-386); 7.9% (6.4%-9.3%)
                                    -•- Fresno, CA 86 (70-102); 7.5% (6.1%-8.8%)
                                    -i- Houston, TX 417 (342-489); 13% (10.7% -15.3%)
                                    	 Los Angeles, CA 923 (752-1091); 6.5% (5.3%-7.7%)
                                      - New York, NY  1432 (1169-1692); 7.6% (6.2%-9%)
                                    -»- Philadelphia, PA 267 (218-314); 10.7% (8.7%-12.6%)
                                    -m- Phoenix, AZ 329 (269 - 389); 7% (5.7% - 8.3%)
                                    -*- Pittsburgh, PA 180 (147-213); 6.6% (5.3%-7.8%)
                                        Salt Lake City, UT 21 (17-25); 4%  (3.2%-4.7%)
                                        St. Louis, MO  429 (351 -505); 10.8% (8.8%-12.7%)
                                        Tacoma, WA 41 (33 - 48); 3.9% (3.2% - 4.6%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                             111
Draft - Do Not Quote or Cite

-------
1
2
3
4
Figure 4-4.
                      Estimated Percent Reductions From the Current Standards to
                      Alternative Set of Standards in Ischemic Heart Disease Mortality
                      Associated with Long-Term Exposure to PM2.s (Exposure Period:
                      1999 - 2000): Based on 2007 Air Quality Data.*
                100%
             •g
             CO
             3
             o
             2
             1
             1!  -
             0
             Q_
                -200%
                       2007 air     15/35*
                        quality
                                            13/35
                                                      12/35
                                                                13/30
                                                                          12/25
                                          Alternative Standard
 5
 6
 7
 8
 9
10
11
                             -»- Atlanta, GA 278 (228-326); 16.8% (13.7% - 19.6%)
                             -«- Baltimore, MD 374 (306-440); 14.7% (12% -17.3%)
                             -*- Birmingham, AL 188 (154-220); 15.7%  (12.9% - 18.4%)
                                Dallas, IX 243 (198-286); 11.2% (9.1% -13.2%)
                             -*- Detroit, Ml 415 (338-490); 10% (8.1% -11. 8%)
                             -•- Fresno, CA 109 (89-129); 9.5% (7.7% -11. 2%)
                             -i- Houston, IX 524 (430-615);  16.4% (13.5% - 19.2%)
                             - Los Angeles, CA 1173 (954-1388);  8.3% (6.7% -9.8%)
                              - New York, NY 1818 (1481 -2148); 9.6%  (7.8% -11. 4%)
                             -»- Philadelphia, PA 337 (276-397); 13.5% (11% -15.9%)
                             -m- Phoenix, AZ 418 (341 -495);  8.9% (7.2% -10.5%)
                             -*- Pittsburgh, PA 229 (187-271); 8.3% (6.8% -9.9%)
                                Salt Lake City, UT 27 (22-31); 5.1%  (4.1% -6.1%)
                                St. Louis, MO 541 (443-637); 13.7% (11.2% - 16.1%)
                                Tacoma.WA 52 (42-62); 5% (4.1% -5.9%)
*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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                             112
                                                                    Draft - Do Not Quote or Cite

-------
1
2
3
4
Figure 4-5.
                      Estimated Percent Reductions From the Current Standards to
                      Alternative Set of Standards in Cardiopulmonary Disease Mortality
                      Associated with Long-Term Exposure to PM2.s (Exposure Period:
                      1979-1983):  Based on 2007 Air Quality Data.*
                100%
                -200%
                       2007 air
                        quality
                            15/35**     13/35      12/35

                                    Alternative Standard
                                                                13/30
                                                                          12/25
 5
 6
 7
 8
 9
10
                             -•- Atlanta, GA 452 (345-558); 7.3% (5.6% -9%)
                             --- Baltimore, MD 375 (286-463); 6.4% (4.9% -7.9%)
                             -*- Birmingham, AL 290 (221 -358); 6.8% (5.2% -8.4%)
                                Dallas, IX 264 (201 -326); 4.8% (3.6% -5.9%)
                             -*- Detroit, Ml 347 (264-430); 4.3% (3.2% -5.3%)
                             -•- Fresno, CA 108 (82-133); 4% (3.1% -5%)
                             -i- Houston, IX 563 (429 - 694);  7.2% (5.5% - 8.8%)
                             - Los Angeles, CA 1002  (761 -1241); 3.5% (2.7% -4.4%)
                              - New York, NY 1188 (903-1471); 4.1% (3.1% -5.1%)
                             -+- Philadelphia, PA 349 (266-431); 5.8% (4.4% -7.2%)
                             -m- Phoenix, AZ 381 (289-471);  3.8% (2.9% -4.7%)
                             -*- Pittsburgh, PA 214 (162-265); 3.5% (2.7% -4.4%)
                                Salt Lake City, UT 39 (30-48); 2.1% (1.6% -2.7%)
                                St. Louis, MO 505 (384 - 623); 5.9% (4.5% - 7.3%)
                                Tacoma.WA 48 (36-60); 2.1% (1.6% -2.6%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                             113
                                                                    Draft - Do Not Quote or Cite

-------
 2
 3
 4
 5
Figure 4-6.
Estimated Percent Reductions From the Current Standards to
Alternative Set of Standards in Cardiopulmonary Disease Mortality
Associated with Long-Term Exposure to PMi.s (Exposure Period:
1999 - 2000):  Based on 2007 Air Quality Data.*
                100%
                 80%
                 60%
             •O
             ra
             c
             3
             O
             E
             o
             c
             o
          -40%
             01
             OL
             1:
             s
             <5
             Q.
          -60%
          -80%
         -100%
         -120%
         -140%
         -160%
         -180%
         -200%
                       2007 air
                        quality
                           15/35**     13/35      12/35

                                   Alternative Standard
                                                                13/30
                                                                          12/25
                             -»- Atlanta, GA 640 (503-774); 10.4% (8.1%-12.5%)
                             -m- Baltimore, MD 533 (418-645); 9.1%  (7.1%-11%)
                             -*- Birmingham, AL 411 (322-498); 9.7% (7.6%-11.7%)
                             -x- Dallas, TX 376 (294 - 456); 6.8% (5.3% - 8.3%)
                             -*- Detroit, Ml 495 (387-602); 6.1% (4.7%-7.4%)
                             -•- Fresno, CA 154 (120-187); 5.8% (4.5%-7%)
                             -i- Houston, TX 797 (625-964);  10.1% (8%-12.3%)
                             	 Los Angeles, CA 1431 (1116-1741);  5%  (3.9%-6.1%;
                              - New York, NY 1694 (1323-2060); 5.9% (4.6%-7.1%)
                             -•- Philadelphia, PA 496 (388-601); 8.3% (6.5%-10%)
                             -m- Phoenix, AZ 543 (424-661);  5.4% (4.2%-6.6%)
                             -*- Pittsburgh, PA 305 (238-372); 5.1%  (4%-6.2%)
                                Salt Lake City, UT 56 (44-68); 3.1%  (2.4%-3.7%)
                                St. Louis, MO 717 (561 -869); 8.4% (6.6%-10.2%)
                             -•- Tacoma.WA 69 (53-84); 3% (2.3%-3.7%)
 9
10
11
*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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                             114
                                               Draft - Do Not Quote or Cite

-------
 1    Figure 4-7.   Estimated Percent Reductions From the Current Standards to
 2                   Alternative Set of Standards in Lung Cancer Mortality Associated
 3                   with Long-Term Exposure to PM2.5 (Exposure Period:  1979 - 1983):
 4                   Based on 2007 Air Quality Data.*
             "S
             re
             •O
             o
             &
             o
             S
             3
             •D

             'c
                        2007 air    15/35**
                        quality
                                             13/35
                                                       12/35
                                                                 13/30
                                                                           12/25
                                          Alternative Standard
 5
 6
 7
 8
 9
10
                              -•-Atlanta, GA 68 (26-108); 7.2% (2.7%-11.4%)
                              -•- Baltimore, MD 60 (23-96);  6.3% (2.4%-10%)
                              -*- Birmingham, AL  44 (17-70); 6.7% (2.5%-10.7%)
                                  Dallas, TX 41 (15-66); 4.7% (1.8%-7.5%)
                              -*- Detroit, Ml 52 (20-84); 4.2% (1.6%-6.7%)
                              -•- Fresno, CA 12 (4-19); 4%  (1.5%-6.4%)
                              -i— Houston, TX 86  (33-137); 7% (2.7%-11.2%)
                              	 Los Angeles, CA 105 (39-170); 3.5% (1.3%-5.6%)
                                  New York, NY 112 (42-180); 4% (1.5%-6.5%)
                              -»- Philadelphia, PA 56 (21 -89); 5.7% (2.2%-9.1%)
                              -m- Phoenix, AZ 55  (21 -88); 3.7% (1.4%-6%)
                              -+- Pittsburgh, PA 32 (12-52);  3.5% (1.3%-5.6%)
                                  Salt Lake City, UT 4 (2-7);  2.1% (0.8%-3.4%)
                                  St. Louis, MO 77 (29-123);  5.8% (2.2%-9.3%)
                              -•- Tacoma.WA 8 (3-13); 2.1% (0.8%-3.3%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                            115
Draft - Do Not Quote or Cite

-------
 1    Figure 4-8.   Estimated Percent Reductions From the Current Standards to
 2                   Alternative Set of Standards in Lung Cancer Mortality Associated
 3                   with Long-Term Exposure to PM2.5 (Exposure Period: 1999 - 2000):
 4                   Based on 2007 Air Quality Data.*
                     re
                     •O
                     o
                     &
                     o
                               2007 air
                                quality
                                          15/35**
                                                    13/35
                                                               12/35
                                                                         13/30
                                                                                   12/25
                                                  Alternative Standard
 5
 6
 7
 8
 9
10
                                      -•-Atlanta, GA 98 (43-149); 10.4% (4.6%-15.8%)
                                      -•- Baltimore, MD 87 (38-133); 9.1% (4%-13.8%)
                                      -*- Birmingham, AL 63 (28-96); 9.7% (4.3%-14.8%)
                                        - Dallas, TX 59 (26-91); 6.8% (3%-10.5%)
                                      -*- Detroit, Ml 76 (33-117); 6.1% (2.7%-9.4%)
                                      -•- Fresno, CA 17 (7-26); 5.8% (2.5%-8.9%)
                                      -i— Houston, TX 125 (55-190); 10.1% (4.5%-15.4%)
                                      	 Los Angeles, CA 153 (67-238); 5% (2.2%-7.8%)
                                          New York, NY 163 (71 -252); 5.9% (2.6%-9%)
                                      -»- Philadelphia, PA 80  (35-123); 8.3% (3.6%-12.7%)
                                      -m- Phoenix, AZ 80 (35-123);  5.4% (2.4%-8.3%)
                                      -+- Pittsburgh, PA 47 (21 -73); 5.1% (2.2%-7.8%)
                                          Salt Lake City, UT 6  (3-9); 3.1% (1.3%-4.8%)
                                          St. Louis, MO  112 (49 -171); 8.4% (3.7% -12.8%)
                                          Tacoma.WA 12 (5-18); 3%  (1.3%-4.7%)
*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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                            116
Draft - Do Not Quote or Cite

-------
 1

 2         4.2.2  Short-term exposure mortality
 3           This section summarizes core estimates for short-term exposure mortality (non-
 4    accidental, IHD, CVD and respiratory) generated for the current and alternative sets of
 5    standards for simulation year 2007.

 6       •   Comparison of recent conditions risk with risk associated with current
 7           standards:  A shift in ambient air PM2.5 levels from recent conditions to just
 8           meeting the current set of standards results in notable reductions in the estimates
 9           of percent of total incidence for short-term exposure non-accidental mortality
10           attributable to PM2.5  across most of the urban case studies.  For example, Fresno,
11           Los Angeles and Salt Lake City are estimated to have reductions in the range of
12           up to 40 to 50% of total incidence attributable to PM2.5 under the current
13           standards for simulation year 2007 - see Appendix E, Table E-78).  However, a
14           subset of urban case  studies with recent ambient air PM2.5 levels at or below the
15           current standards (e.g., Dallas, Phoenix - see Appendix E, Table E-78) do not
16           experience any reductions in estimated short-term mortality-related risk, as would
17           be expected.  A similar pattern of risk reduction is seen for the other short-term
18           exposure mortality categories, although overall reductions are somewhat lower for
19           these categories compared with non-accidental mortality. As noted earlier, in
20           presenting results for recent conditions, the point estimate for Los Angeles, for
21           CV mortality, is negative. As discussed earlier (see section 4.1), negative point
22           estimates reflect the use of non-statistically significant effects estimates, which
23           can in turn, reflect a number of factors, including insufficient sample size in the
24           epidemiological study that provided the effects estimate.  The pattern of risk
25           reductions associated with moving from recent conditions to just meeting the
26           current set of standards is depicted graphically for short-term exposure mortality
27           in Figures 4-9 through 4-11.

28       •   Trends in risk reduction across alternative sets of standards focusing on
29           lower  annual levels  (13/35 and 12/35 combinations): reducing ambient PM2.5
30           levels to meet alternative sets of standards with lower annual levels (i.e., 13/35
31           and 12/35) is estimated to produce a systematic reduction in PM2.5-related non-
32           accidental mortality in roughly half of the urban study areas. Specifically, for
33           simulation year 2007, 8 of the  14 study areas have estimated reductions in PM2.5-
34           related non-accidental mortality on the order of 3 to 15% under a 13/35 set of
35           standards with this reduction increasing to between 5  and 23% under the 12/35 set
36           of standards (see Appendix E, Table E-81). The degree of estimated reduction in
37           PM2.s-related short-term exposure mortality across these  sets of standards with
38           reduced annual levels is similar for the other short-term exposure mortality
39           categories. The degree of reduction in short-term exposure mortality is somewhat
40           lower than what is estimated for long-term mortality (see section 4.2.1),  which is
41           expected since these  two alternative standards focus on lower annual levels and
42           consequently would impact long-term exposure mortality risk more than short-
43           term. This pattern of risk reductions associated with moving from just meeting
      September 2009                         117             Draft - Do Not Quote or Cite

-------
 1          the current set of standards to just meeting alternative sets of standards (including
 2          lower annual levels) is depicted graphically in Figures 4-9 through 4-11.

 3       •  Trends in risk reduction across alternative sets of standards focusing on
 4          lower combinations of 24-hour and annual levels (13/30 and 12/25
 5          combinations): Under the 13/30 standards combination, PM2.5-related non-
 6          accidental mortality is reduced by a slightly greater amount than under the 13/35
 7          combination discussed above (with percent reductions relative to just meeting the
 8          current set of standards ranging from 6 to 15% - see Appendix E, Table E-81).
 9          When the most stringent set of standards (12/25) included in this analysis is
10          modeled, the overall degree of reduction in PM2.5-attributable non-accidental
11          short-term mortality is increased to between 7 and 29% (and all of the urban study
12          areas are predicted to  experience some degree of risk reduction). We note that
13          results for the other short-term mortality categories (in terms of risk reductions
14          associated with these two alternative standard combinations) are similar to those
15          presented above for all non-accidental mortality. This pattern of risk reductions
16          associated with moving from just meeting the current set of standards to just
17          meeting  alternative sets of standards (including lower 24-hour and annual levels)
18          is depicted graphically in Figures 4-9 through 4-11.

19       •  Pattern  of reductions of PMi.s-associated short-term exposure mortality
20          incidence across urban study areas under alternative sets of standards:
21          Differences in the degree of reduction of PM2.s-related short-term exposure
22          mortality risk across the 14 urban study areas relative to just meeting the current
23          set of standards are moderate for the alternative sets of standards reflecting lower
24          annual standards (i.e., 13/35 and 12/35), with risk reductions ranging from 3 to
25          15% (for 13/35) and 5 to 23% (for 12/35) (Appendix E, Table E-81). Differences
26          in risk reduction across urban study areas are notably  small, for the alternative
27          sets of standards  reflecting lower 24-hour standards (i.e.,  13/30 and 12/25)
28          (Appendix E, Table E-81). Specifically, for these two alternative standard levels,
29          12 of the 14 urban study area locations have estimated percent reductions that
30          vary from each other by only 1% (or are the same). In the case of all four sets of
31          alternative standard levels, the range of risk reductions across urban study areas is
32          notably lower for short-term mortality than it is for long-term morality (see
33          section 4.3.1). We believe this reflects, in part, the fact that short-term exposure
34          mortality is modeled down to PRB, which represents a larger span in absolute
35          ambient  air levels relative to the annual standard which is modeled to a relatively
36          higher ambient concentration (i.e., the LML). Modeling risk down to PRB for
37          short-term exposure health impacts has a dampening effect on estimates of the
38          percent risk reduction associated with reductions in ambient air levels, since these
39          changes  represent a relatively smaller fraction of the total  range in  ambient air
40          levels being considered. Conversely, for long-term exposure mortality, which is
41          modeled down to LML, incremental reductions in ambient air levels and the
42          associated risk reductions represent a larger relative change, since the overall
43          spread being modeled is less (i.e., only down to LML and not PRB).

44       •  Pattern  of percent reduction of total incidence across the three simulated
45          years (2005-2007): Reductions in PM2 5-associated long-term mortality under
      September 2009                         118             Draft - Do Not Quote or Cite

-------
1          alternative sets of standards across the simulation years (2005-2007) are also
2          relatively uniform (e.g., for non-accidental mortality, see Appendix F, Tables E-
3          79 through E-81). As with the relatively uniform pattern across urban study
4          areas, we believe this consistency across the simulation years reflects, in part, the
5          fact that short-term exposure endpoints are modeled down to PRB, which can
6          have a dampening effect on the percent reduction risk estimates.
     September 2009                         119             Draft - Do Not Quote or Cite

-------
 2
 3
 4
 5
Figure 4-9.
Estimated Percent Reductions From the Current Standards to
Alternative Sets of Standards in Non-Accidental Mortality Associated
with Short-Term Exposure to PM2.5:  Based on 2007
Air Quality Data.*
                        10%
                         o%
                       -10%
                       -20%
                       -30%
                       -40%
                       -50%
                       -60%
                       -70%
                       -80%
                       -90%
                               2007 air
                                quality
                                    15/35"      13/35      12/35

                                            Alternative Standard
                                                                          13/30
                                                                                    12/25
 6
 7
 8
 9
10
                                      -»- Atlanta, GA 177  (34-319); 1.2% (0.2%-2.1%)
                                      -m- Baltimore, MD 225 (91 -357); 1.7% (0.7%-2.6%)
                                      -*- Birmingham, AL 37 (-58-131); 0.4% (-0.6% -1.4%)
                                         Dallas, TX 137 (33-240); 1.1% (0.3% -1.9%)
                                      -*- Detroit, Ml 112 (-20-242); 0.7%  (-0.1%-1.4%)
                                      -•- Fresno, CA 51 (7-94); 0.9% (0.1%-1.7%)
                                      -i— Houston, TX 240 (49 - 429);  1.3% (0.3% - 2.3%)
                                      	 Los Angeles, CA 79 (-113-270); 0.1%  (-0.2%-0.5%)
                                         New York, NY  659 (387-930); 1.3% (0.7% -1.8%)
                                      -•- Philadelphia, PA 206  (76-334); 1.5% (0.5%-2.4%)
                                      -m- Phoenix, AZ 242 (40-441);  1% (0.2%-1.9%)
                                      -*- Pittsburgh, PA 123 (36-209); 0.9% (0.3%-1.6%)
                                         Salt Lake City, UT 36  (7 - 65); 0.7% (0.2% -1.3%)
                                         St. Louis, MO  222 (64-378); 1.2% (0.4%-2.1%)
                                         Tacoma.WA 42 (7-76); 0.8% (0.1%-1.5%)
*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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                            120
                                              Draft - Do Not Quote or Cite

-------
 2    Figure 4-10.   Estimated Percent Reductions From the Current Standards to
 3                    Alternative Sets of Standards in Cardiovascular Mortality Associated
 4                    with Short-Term Exposure to PM2.s:  Based on 2007
 5                    Air Quality Data.*
                -90%
                       2007 air
                        quality
                            15/35**      13/35      12/35

                                    Alternative Standard
                                                                 13/30
                                                                            12/25
 6
 7
 8
 9
10
11
                              -»- Atlanta, GA 32 (-33-95); 0.8% (-0.8% - 2.4%)
                              -m- Baltimore, MD 61 (-4-125); 1.6% (-0.1%-3.2%)
                              -*- Birmingham, AL  -1  (-46-44); 0% (-1.7%-1.6%)
                                  Dallas, TX 29 (-19-75); 0.8% (-0.5%-2.1%)
                              -*- Detroit, Ml 55 (-7-117); 0.9% (-0.1%-2%)
                              -•- Fresno, CA 13 (-9-35); 0.8% (-0.5%-2.1%)
                              -i— Houston, TX 52  (-36-139); 1% (-0.7%-2.7%)
                              	 Los Angeles, CA -30 (-136-74); -0.2% (-0.7%-0.4%)
                                - New York, NY 425 (248-600); 1.9% (1.1%-2.7%)
                              -»- Philadelphia, PA 83 (22-143); 2.1% (0.5%-3.6%)
                              -m- Phoenix, AZ 84  (-4-170); 1.3%  (-0.1%-2.7%)
                              -*- Pittsburgh, PA 37 (-7-80); 0.9%  (-0.2%-2%)
                                  Salt Lake City, UT 10 (-2 - 21); 0.8% (-0.2% -1.8%)
                                  St. Louis, MO  104 (23-184); 1.8% (0.4%-3.2%)
                                  Tacoma.WA 11 (-6-27); 0.7% (-0.4%-1.8%)
* 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                             121
Draft - Do Not Quote or Cite

-------
 2    Figure 4-11.   Estimated Percent Reductions From the Current Standards to
 3                    Alternative Sets of Standards in Respiratory Mortality Associated
 4                    with Short-Term Exposure to PM2.5:  Based on 2007
 5                    Air  Quality Data.*
                        30%
                        20% -

                        10% -

                         0%

                       -10% -

                       -20% -

                       -30% -

                       -40% -

                       -50% -

                       -60% -

                       -70% -
                       -80%
                               2007 air    15/35**
                               quality
                                                    13/35
                                                              12/35
                                                                        13/30
                                                                                   12/25
                                                 Alternative Standard
                                       -»- Atlanta, GA 20 (-8-47); (-0.6%-3.6%) 2.6%
                                       -•- Baltimore, MD 31 (6-56); (0.5%-4.6%) 1.1%
                                       -+- Birmingham, AL 10  (-8-28); (-0.9%-3.1%) 0.9%
                                          Dallas, IX 10 (-9-29); (-0.8%-2.5%) 1.6%
                                       -«- Detroit, Ml 22 (1 -42); (0.1%-3.2%) 1.6%
                                       -•- Fresno, CA 10 (0-19); (0.1%-3.2%) 2.6%
                                       -i- Houston, IX 38 (6-68); (0.4%-4.7%) 1%
                                       	 Los Angeles,  CA 56 (5-105); (0.1%-1.9%) 2.1%
                                          New York, NY 90 (32-147); (0.7%-3.4%)  1.8%
                                       -»- Philadelphia,  PA 22 (-2-45); (-0.2%-3.8%) 1.8%
                                       -•- Phoenix, AZ 47 (4-90); (0.1%-3.5%) 1.3%
                                       -+- Pittsburgh, PA 16 (-2-32);  (-0.1%-2.7%)  1.4%
                                          Salt Lake City, UT 6 (1 -12); (0.2%-2.5%) 1.5%
                                          St. Louis, MO 27 (-7 - 59); (-0.4% - 3.4%) 1.3%
                                       -•- Tacoma.WA  6 (0-13); (0%-2.5%)
 9
10
11
* 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                             122
Draft - Do Not Quote or Cite

-------
 1

 2         4.2.3   Short-term exposure morbidity
 3          This section summarizes core estimates for short-term exposure morbidity (HA
 4    for CVD and respiratory causes; ED visits for CV, respiratory and asthma) generated for
 5    the current and alternative sets of standards for simulation year 2007.

 6       •  Comparison of recent conditions risk with risk associated with current
 7          standards:  A shift in ambient air PM2.5 levels from recent conditions to just
 8          meeting the current set of standards results in notable estimated reductions in the
 9          percent of total incidence (for short-term CV HAs) attributable to PM2.5 across
10          most of the urban case studies. For example, Fresno and Salt Lake City have
11          estimated reductions in the range of up to 40 to 50% in the percent of total
12          incidence attributable to PM2.5 under the current standards for simulation year
13          2007 - see Appendix E, Table E-105). However, as expected, a subset of urban
14          case studies with recent ambient air PM2 5 levels near or below the current
15          standards (e.g., Dallas, Phoenix - see Appendix E, Table E-105) do not show any
16          estimated reductions in short-term mortality-related risk. A similar pattern of
17          risk reduction (in comparing just meeting the current set of standards to recent
18          conditions) is seen for respiratory-related HAs, although overall reductions are
19          somewhat lower for these categories compared with CV HAs for most urban
20          study areas. The pattern of risk reductions associated with moving from recent
21          conditions to just meeting the current set of standards is depicted graphically for
22          short-term exposure morbidity in Figures 4-12 and 4-13.  Since these figures
23          show the change in risk for each of the short-term exposure morbidity categories
24          expressed in terms of the percent change relative to just meeting the current set of
25          standards, the change from recent conditions relative to just meeting the current
26          set of standards is often negative, reflecting higher risks under recent conditions.

27       •  Trends in risk reduction across alternative sets of standards focusing on
28          lower annual levels (13/35 and 12/35 combinations): reducing ambient PM2 5
29          levels to simulate just meeting alternative suites of standards with lower annual
30          levels (i.e., 13/35 and 12/35) produces a systematic reduction in estimated PM2.5-
31          related CV HAs for roughly half of the urban study areas. Specifically, in
32          simulation year 2007, 8  of the  14 study areas are estimated to have reductions in
33          PM2.5-related CVHAs on the order of 3% to 15% under a 13/35 suite of standards
34          with this reduction increasing to between 5 and 23% under the 12/35 set of
35          standards (see Appendix E, Table E-107). The degree of reduction in PM25-
36          related mortality across these sets of standards with reduced annual levels is
37          similar for respiratory-related HAs. This pattern of risk reductions associated
38          with moving from the current set of standards to alternative sets of standards
39          (including lower annual levels) is depicted graphically in Figures 4-12 and 4-13.

40       •  Trends in risk reduction across alternative sets of standards focusing on
41          lower 24-hour and annual levels (13/30 and 12/25 combinations): Under the
42          13/30 set of standards, PM2.5-related CV HAs are reduced by a slightly greater
43          amount than under the 13/35 set discussed above (with percent reductions ranging
      September 2009                         123            Draft - Do Not Quote or Cite

-------
 1          from 6 to 15% - see Appendix E, Table E-107).  When the most stringent set of
 2          standards (12/25) included in this analysis is modeled, the overall degree of
 3          reduction in PM2.5-attributable estimated CV HAs increases to between 7 and
 4          30% (and all of the urban study areas are predicted to experience some degree of
 5          risk reduction).  We note, that results for respiratory HAs (in terms of estimated
 6          risk reductions associated with these two alternative standard combinations) are
 7          similar to those presented above for non-accidental mortality.  This pattern of risk
 8          reductions associated with moving from just meeting the current set of standards
 9          to just meeting alternative sets of standards  (including lower annual and 24-hour
10          levels) is depicted graphically in Figures 4-12 and 4-13.

11       •  Pattern of the reduction of PM2.s-associated short-term exposure morbidity
12          incidence across urban study areas under alternative sets  of standards: As
13          with short-term exposure mortality, differences in the degree  of reduction of
14          PM2.5-related short-term morbidity risk across the 14 urban study areas are
15          moderate for those sets of alternative standards focusing on lower annual levels
16          (13/35 and 12/35) and notably smaller for those alternative  sets of standards  with
17          lower annual and 24-hour levels (13/30 and 12/25) (see Appendix E, Table E-
18          107). We believe the same factors discussed above for short-term exposure
19          mortality are responsible for this trend in estimated incidence for short-term
20          exposure morbidity (see section 4.2.2).

21       •  Pattern of percent reduction of total incidence across the three simulated
22          years (2005-2007): Reductions in PM2.5-associated short-term exposure
23          morbidity under alternative sets of standards across the simulation years (2005-
24          2007) are also relatively uniform (see Appendix E, Tables E-106 through E-108).
25          As with the relatively uniform pattern across urban study areas, we believe this
26          consistency across the simulation years reflects the combination of using a
27          proportional reduction approach to simulate alternative 24-hour standard levels,
28          combined with modeling of short-term exposure morbidity.
      September 2009                         124             Draft - Do Not Quote or Cite

-------
 2    Figure 4-12.  Estimated Percent Reductions From the Current Standards to
 3                    Alternative Sets of Standards in Cardiovascular Hospital Admissions
 4                    Associated with Short-Term Exposure to PM2.5: Based on
 5                    2007 Air Quality Data.*
                        30%
                        20%

                        10% -

                         0%

                        -10%

                        -20%

                        -30%

                        -40%

                        -50%

                        -60%

                        -70%
                        -80%
                        -90%
                               2007 air
                               quality
                                   15/35**      13/35      12/35

                                           Alternative Standard
                                                                        13/30
                                                                                   12/25
                                    -»- Atlanta, GA 41 (-27-108); 0.36% (-0.24% - 0.96%)
                                    -m- Baltimore, MD 215 (158-271); 1.32% (0.97% -1.67%)
                                    -^- Birmingham, AL 17 (-11 -46); 0.35% (-0.23% - 0.92%)
                                        Dallas, TX 27 (-18-72);  0.28% (-0.18%-0.73%)
                                    -*- Detroit, Ml 215 (158-272); 1.04% (0.77% -1.32%)
                                    -•- Fresno, CA 24 (0-48); 0.5% (0.01%-0.99%)
                                    -i- Houston, TX 64 (-42-169); 0.36% (-0.23% - 0.94%)
                                    — Los Angeles, CA 265 (3-526); 0.47% (0.01%-0.93%)
                                      - New York, NY 676 (496-855); 1.04%  (0.76% -1.31%)
                                    -»- Philadelphia, PA  201 (148-254); 1.25% (0.92% -1.58%)
                                    -•- Phoenix, AZ 108 (1 -215); 0.5% (0.01%-0.99%)
                                    -*- Pittsburgh, PA 120 (88-152); 0.96% (0.7%-1.21%)
                                        Salt Lake City, UT 10 (0-20); 0.38% (0%-0.75%)
                                        St. Louis, MO 176 (129-222); 1.25%  (0.92% -1.58%)
                                    -•- Tacoma.WA 19  (-46-82); 0.52% (-1.28%-2.26%)
 7
 8
 9
10
* Based on Bell et al. (2008). 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
      September 2009
                                             125
Draft - Do Not Quote or Cite

-------
 2
 3
 4
 5
Figure 4-13.   Estimated Percent Reductions From the Current Standards to
                Alternative Sets of Standards in Respiratory Hospital Admissions
                Associated with Short-Term Exposure to PM2.5:  Based
                on 2007 Air Quality Data.*
 6
 7
 8
 9
10
11
                        30%
                        20%
                        10%
                     -  -10%
                        -20%
                     3
                     o
                     E
                     |  -40%
                     3
                     |  -50%

                     §  -60%
                     01
                     Q_
                        -70%
                        -80%
                        -90%
                               2007 air     15/35"
                                quality
                                                     13/35
                                                               12/35
                                                                         13/30
                                                                                    12/25
                                                  Alternative Standard
                                     -•-Atlanta, GA 17 (-22-56);  0.44% (-0.55% - 1.42%)
                                     -m- Baltimore, MD 17 (-10-44); 0.35% (-0.2% - 0.89%)
                                     -*- Birmingham, AL 7 (-9-24); 0.42% (-0.53% - 1 .36%)
                                         Dallas, IX 13 (-16-42); 0.33% (-0.42% - 1.08%)
                                     -*- Detroit, Ml 19 (-11 -49); 0.27% (-0.16% -0.7%)
                                     -•- Fresno, CA 16 (4-29); 0.89% (0.21% - 1.55%)
                                     -i- Houston, IX 29 (-36-93); 0.43% (-0.54% - 1 .39%)
                                     - Los Angeles, CA 171  (40-301); 0.83%  (0.19% - 1.47%)
                                       - New York, NY 56 (-33-144); 0.27% (-0.16% - 0.7%)
                                     -»- Philadelphia, PA 16 (-9-42); 0.33% (-0.19% -0.84%)
                                     -m- Phoenix, AZ 61 (14-108); 0.88% (0.21% - 1.55%)
                                     -*- Pittsburgh, PA 10 (-6-25); 0.25% (-0.15% -0.64%)
                                         Salt Lake City, UT 7 (2-13); 0.67% (0.16% - 1.18%)
                                         St. Louis, MO  21 (-12-54); 0.33% (-0.19% - 0.85%)
                                         Tacoma.WA 2 (-24-27);  0.14% (-1.87% -2.05%)
* Based on Bell et al. (2008). 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 24-hour standard of 35 ug/m3.
Combinations of an annual standard (n) and a 24-hour standard (m) are denoted n/m in this figure.
figure.
      September 2009
                                             126
Draft - Do Not Quote or Cite

-------
 1         4.3   SENSITIVITY ANALYSIS RESULTS
 2           As noted in section 3.6.4 and section 4.1 the sensitivity analysis was conducted in
 3    order to gain insights into which of the identified sources of uncertainty in the risk
 4    assessment model may have significant impacts on risk estimates. A second goal of the
 5    sensitivity analysis was to generate an additional set of reasonable risk estimates to
 6    supplement the core set of risk estimates and which can be used to inform staffs
 7    characterization of uncertainty associated with those core estimates.
 8           The first goal can be achieved by considering the magnitude of the impact of
 9    individual modeling elements based on results from the sensitivity analysis and
10    identifying those elements which have the greatest impact on the core risk estimates.
11    Regarding the second goal, given the design of the sensitivity analysis, staff judges that
12    the results of this analysis represent a reasonable set of alternate risk estimates, that fall
13    within an overall set of plausible risk estimates surrounding the core estimates.  While
14    not representing a formal uncertainty distribution, the output of the sensitivity analysis,
15    when combined with the core risk estimates, represent a set of plausible risk estimates,
16    which reflect consideration for uncertainty in various elements of the risk assessment
17    model.  Therefore, later in section 4.5, when discussing risk estimates for a particular
18    scenario, we focus on the core estimates as representing higher-confidence estimates and
19    then use the output of the sensitivity analysis to provide some perspective on the potential
20    range of uncertainty about the risk estimates given consideration of key sources of
21    uncertainty. While we do not know what confidence interval is captured by this range of
22    estimates,  or what specific percentiles of the risk distribution are represented by points
23    within that range, the output of the sensitivity analysis does provide a set  of plausible risk
24    estimates and, therefore provides some perspective on the magnitude of potential
25    uncertainty associated with the core estimates.
26           In conducting the sensitivity analysis we modeled 2 of the 15 urban study areas
27    (Philadelphia and Los Angeles - representing east and west coast urban areas,
28    respectively) for most simulations. We note, however, that for some modeling elements
29    (e.g., the hybrid rollback approach), we included a larger number of the urban study areas
30    that were applicable to the topic being assessed.41 In conducting the sensitivity analysis,
31    we have also focused on long-term exposure mortality and short-term exposure mortality
32    and have included short-term morbidity to a lesser extent because we expect similar
      41 As noted earlier in section 2.4.1, an error was identified in the approach used to simulate ambient PM2 5
      levels just for the Pittsburgh study area for the scenarios involving just meeting the current and alternative
      sets of standards. Therefore, while Pittsburgh had initially been included in this sensitivity analysis, these
      results have been removed since they are based on the set of current standard levels.
      September 2009                         127             Draft - Do Not Quote or Cite

-------
 1    patterns with that observed for short-term exposure mortality based on our review of the
 2    core estimates.
 3           The results of the sensitivity analysis are summarized in Table 4-1 (detailed
 4    results tables are presented in Appendix F).  In presenting the results of the sensitivity
 5    analysis, we have compared the risk estimates for the particular simulation to the core set
 6    of risk estimates generated for the same health effect endpoint/urban study area
 7    combination. Specifically, we have calculated a percent difference between the
 8    sensitivity analysis result and the associated core risk estimate to foster comparisons of
 9    the results of the sensitivity analysis across the different modeling elements that were
10    considered.  These percent difference results are emphasized in Table 4-1 and in the
11    discussion presented below.
12           In discussing the results of the sensitivity analysis, we have developed four
13    descriptive categories, based on the general magnitude of the percent difference estimate
14    generated for a particular modeling element:
15         •  Small contributors to uncertainty  in the core risk estimates: Modeling
16            elements estimated to have percent differences of 20% or smaller (i.e., they
17            produced risk estimates that differed from the core risk estimates by no more
18            than 20%) are classified as having a  small contribution to uncertainty in the
19            core risk estimates.
20         •  Moderate contributors to uncertainty in the core risk estimates: Modeling
21            elements estimated to have percent difference estimates in the range of 20 to
22            50% are classified as having a moderate contribution to uncertainty in the core
23            risk estimates.
24         •  Moderate-Large contributors to uncertainty in the core risk estimates:
25            Modeling elements estimated to have percent difference estimates in the range
26            of 50 to 100%  are classified as having a moderate-large contribution to
27            uncertainty in the core risk estimates.
28         •  Large contributors to uncertainty  in the core risk estimates:  Modeling
29            elements estimated to have percent difference results >100% are classified as
30            having a large contribution to uncertainty in  the core risk estimates.
31           These categories are used in discussing the results of the sensitivity  analysis,
32    particularly in section 4.5.
33           In discussing the results of the sensitivity analysis, results of the single-factor
34    simulations are presented first, followed by the results  of the multi-factor simulations
35    (within these categories, results are organized  by health effect endpoint with results for
36    long-term exposure mortality discussed first and then short-term exposure mortality,
37    followed by short-term exposure morbidity).
      September 2009                         128            Draft - Do Not Quote or Cite

-------
 1           The sensitivity analysis based on Moolgavkar's (2003) study in Los Angeles
 2    addressing model specifications for both short-term mortality and morbidity (e.g., model
 3    selection, lag structure and co-pollutant models) are discussed together as a group, rather
 4    than being divided up and discussed with other sensitivity analyses. This reflects the fact
 5    that the Moolgavkar-based simulations were all based on the same underlying dataset and
 6    focused on Los Angeles. In addition, the discussion of the Moolgavkar-based sensitivity
 7    analysis results presented below (and the summary of results presented in Table 4-1)
 8    focus on the difference in spread of risk results across the Moolgavkar-based model
 9    specifications (for a particular endpoint), rather than the percent difference results (based
10    on comparison against the core result) that are emphasized with the other sensitivity
11    analyses.42
12           Although the sensitivity analysis was completed for all three simulation years, we
13    have focused on results  for 2007 in this presentation to foster comparability with the core
14    results discussed in sections 4.1 and 4.2.
15           The discussion provided below for each of the simulations conducted as part of
16    the sensitivity analysis,  first gives a brief overview of the purpose of the simulation (i.e.,
17    what aspect of the risk assessment model is being evaluated). Next the results of the
18    simulation are discussed in the context of the first goal described above (i.e., the relative
19    impact of a given source of uncertainty on the  core risk estimates).  An overall conclusion
20    regarding which of the factors included in the sensitivity analysis represent potentially
21    significant sources of uncertainty impacting the core risk estimates is presented at the end
22    of each sub-section.  Use of the alternative risk estimates generated through the
23    sensitivity analysis in informing characterization of uncertainty associated with the core
24    risk estimates is discussed in section 4.5.
      42 Comparison of the Moolgavkar-based risk estimates against the core risk estimates consistently produce
      percent difference estimates that range to levels well above +100%, resulting in a blanket 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.
      September 2009                          129             Draft - Do Not Quote or Cite

-------
1
2
Table 4-1.    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 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 potential for
greater use of localized
controls in some study
areas(evaluated across
current and alternative
standard levels)
• All-cause, CPD,
IHD
• Los Angeles and
Philadelphia
• All-cause, CPD,
IHD
• Los Angeles and
Philadelphia
• All cause
• All 15 urban
study areas
• All-cause, CPD,
lung cancer
• Los Angeles,
Philadelphia
• All-cause
mortality
• Baltimore,
Birmingham,
Detroit, Los
Angeles, New
York and St.
Louis
Random effects log-linear C-R
model:
• all-cause: +23%
• IHD: +12%
Random effects log-log C-R
model:
• All-cause: +122 to 155%
• CPD: +49 to 71%
• IHD: +79 to 107%
• Lung Cancer: +68 to 92%
• All-cause : +47 to + 1 97%
• All-cause: -25%
• CPD: -55 to -56%
• Lung cancer: -192 to - 195%
• 4 urban locations had
impacts of <20% across all
alternative standard levels
(with most of these in the
<10% range).
• 1 urban locations had
impacts of up to +54%
Table F-3
Table F-3
Table F-6
Table F-9
Table F-12
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
• Non-accidental
mortality, CV,
respiratory
• All 15 urban
study areas
• Non-accidental
mortality
• Non-accidental: -114 to
+186%
• CV: -92 to +600%
• Respiratory: -47 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
Table F-15
Table F-18
Table F-21
Table F-36
    September 2009
                                     130
Draft - Do Not Quote or Cite

-------
Sensitivity Analysis1
reflecting potential for
greater use of localized
controls in some study areas
Health Endpoint and
Risk Assessment
Location
• Baltimore,
Birmingham,
Detroit, Los
Angeles, New
York and St.
Louis
Summary of Results
(percent difference in risk
estimate relative to the core
estimate)
current and alternative
standard levels) do not
exceed +20%, with most
<+10%.
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)
• HA
(unscheduled),
CVand
respiratory
• All 15 urban
study areas
• Asthma ED visits
• New York
• HA (CV): -105 to +10%
• HA (respiratory): -54 to
+71%
(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).
Table F-24
Table F-27
Table F-30
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)
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)
• Mortality (non-
accidental, CV);
HA(CV)
• Los Angeles
• Mortality (non-
accidental)
• Los Angeles
• Mortality (CV);
HA(CV)
• Los Angeles
• Non-accidental mortality:
+79%
• CV mortality: +49
• CVHA:+36%
• Non-accidental mortality:
+55%
• CV mortality: +105%
• CV HA: +142%
Table F-33
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
• All-cause, IHD
long-term
mortality
• Los Angeles and
Philadelphia
• All-cause : +26 to + 1 ,020%
• IHD: +25 to +627%
F-39
September 2009
131
Draft - Do Not Quote or Cite

-------
Sensitivity Analysis1
rollback to estimate
incidence associated with
long-term exposure to PM2 5
concentrations that just meet
the current standards
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)

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): -109 to +119%
F-42
      : Unless otherwise noted, sensitivity analysis results are based on the scenario reflecting just meeting the
 2     current set of PM2 5  standards. 1
 3    2 Note: this metric is the percent spread in risk estimates across the Moolgavkar-based model specifications
 4     (not the percent difference estimates - see text discussion above).
 5         4.3.1   Single-factor sensitivity analysis results
 6           This section presents the results of the single-factor sensitivity analysis, which
 7    involved consideration of alternate model inputs on the core risk estimates, when those
 8    alternate inputs are considered one at a time (consideration for the combined effect of
 9    several model inputs being varied is covered by the multi-factor sensitivity analysis
10    discussed in section 4.3.2). Note, that the results of the single-factor sensitivity analysis
11    are characterized qualitatively using the four-category approach described above (i.e.,
12    low, moderate, moderate-large and large, with each of these representing a defined range
13    of percent difference from the core risk estimates, as detailed above).

14                    4.3.1.1   Sensitivity analysis results associated with long-term
15                             exposure mortality
16           This section summarizes the results of the sensitivity analysis focused on long-
17    term exposure-related mortality endpoints (see Table  5-1 for the specific modeling
18    elements considered in the sensitivity analysis).  The results of individual sensitivity
19    analysis simulations are presented below, with overall observations presented at the end
20    of the section.
21       •   Impact of  using different model choices for C-R function - fixed effects log-
22           linear (the core approach) vs. random  effects log-linear or random effects
23           log-log models: This simulation considered two alternative C-R model forms
24           obtained from Krewski et al., 2009 for modeling all-cause, CPD, IHD and lung
      September 2009
132
Draft - Do Not Quote or Cite

-------
 1          cancer mortality, including (a) random effects log-linear model and (b) a random
 2          effects log-log model (note, the core effect estimate was derived using a fixed
 3          effects log-linear model obtained from Krewswki et al., 2009). The results of the
 4          simulation suggest that the use of a random effects log-linear model, rather than
 5          the core fixed effects model has a relatively small effect on risk, raising it by 12 to
 6          23% across the mortality categories and urban study areas modeled (Appendix F,
 7          Table F-3). However, use of a random effects log-log model has a larger impact
 8          on risk estimates, raising them by 49 to 155% (Appendix F, Table F-3). The
 9          greater impact of the log-log model results from this function having an
10          incrementally steeper slope at lower PM levels, which quickly increases incidence
11          estimates compared with the core log-linear model (whose slope has a much more
12          gradual incremental increase in slope at lower PM levels).

13       •  Impact of estimating risks down to PRB rather than down to LML: This
14          simulation compared long-term exposure mortality incidence associated with
15          modeling risk down to PRB (which varies by region - see section 3.2.1) with the
16          core approach of modeling down to LML (5.8 |ig/m3 for long-term mortality - see
17          section 3.1).  This simulation involved all 15 urban study areas, given that PRB is
18          stratified by region and therefore, results of the simulation could differ
19          significantly across the 15 urban study areas (or at least across the six PM regions
20          represented by those study areas). The results of this simulation suggest that
21          modeling risk down to PRB could have a moderate to large impact on  long-term
22          exposure mortality incidence, with estimates ranging from 47 to 197% higher than
23          the core estimates (for matching urban locations) (Appendix F, Table F-6). Note,
24          however, that risk metrics based on considering the incremental reduction in risk
25          between two  standard levels would not be impacted by this source of uncertainty,
26          since it only effects estimates of absolute risk (for a particular standard level).

27       •  Impact of C-R function from alternative long-term exposure mortality study:
28          This simulation considered use of alternative C-R functions (and effect estimates)
29          based on the reanalysis of the Six Cities study (Krewski et al., 2000).  The results
30          suggest that use of the alternative C-R function could have a moderate to
31          moderate-large effect on all cause and CPD mortality (-25% for all-cause and
32          about -55% for CPD) (Appendix F, Table F-9).  The effect of using the alternative
33          function on risk estimates for lung cancer mortality was much larger (-192 to -
34          195%) (Appendix F, Table F-9).  The results of this simulation suggest that (at
35          least with regard to application of C-R functions obtained from the Six Cities
36          study), the potential impact of functions from alternative studies on long-term
37          exposure mortality depends on the mortality category being considered. In this
38          analysis, use  of the alternative C-R functions was shown to have a significant
39          impact on lung cancer mortality, but a much smaller impact on the other long-
40          term exposure mortality categories.

41       •  Impact of using alternative hybrid rollback approach to simulate just
42          meeting the current and alternative suite of standards: This simulation
43          evaluated the potential impact of simulating just meeting the current and
44          alternative sets of standards using a hybrid rollback approach that employs a
45          combination of localized controls (resulting in non-proportional rollbacks of
      September 2009                        133             Draft - Do Not Quote or Cite

-------
 1          monitored PM2.5 concentrations) with a second phase of proportional adjustments
 2          (see section 3.3.3).  We note that the core analysis utilized proportional
 3          adjustment exclusively in simulating conditions for the current and alternative sets
 4          of standards. In discussing these results, we focus on risk estimates generated for
 5          the most stringent alternative set of standards (12/25) since this is the level where
 6          the greatest differences between the core and hybrid rollback approach were seen
 7          (for all relevant urban study areas).  The results of this simulation suggest that the
 8          use of the hybrid rollback strategy can have widely varying impacts on long-term
 9          mortality incidence. For five of the seven urban study areas for which this
10          procedure was  evaluated, the results were relatively small (e.g., percent
11          differences ranged from +2% to +17%) (Appendix F, Table F-12). For the urban
12          study area demonstrating the greatest impact (Los Angeles), percent differences
13          were moderate-large (+54%) (Appendix F, Table F-12).  The results of this
14          simulation suggest that for many of the urban study areas, the use of an alternative
15          rollback strategy  employing non-proportional adjustment, did not produce a very
16          large impact on long-term mortality risk estimates.  However, for a subset of the
17          study areas, the use of this rollback approach was shown to have moderate
18          impact.
19          Based on the simulations discussed above covering potential sources of
20    uncertainty impacting long-term mortality, we conclude that the following factors
21    contribute potentially large sources of uncertainty to the core risk estimates (i.e., they
22    produce risk estimates that are at least 100% different from the core risk estimate):  (a)
23    use of alternative form of the C-R function, specifically use of a random-effects log-log
24    model form obtained from the updated ACS study (Krewski et al.,  2009) (b) use of an
25    alternative C-R function  with effects estimates obtained from the reanalysis of the Six
26    Cities study (Krewski et  al. 2000) (specifically in modeling lung-cancer risk), and (c)
27    estimation of risk down to PRB. We note, that the use of the hybrid (non-proportional
28    adjustment) approach for simulating conditions for alternative suites of standards was
29    shown to have a moderate-large impact (+90%) in one of the urban study areas.  Other
30    factors considered in the sensitivity analysis had smaller impacts on core risk estimates.

31                    4.3.1.2   Sensitivity analysis results associated with short-term
32                            exposure mortality
33          This section summarizes the results of the sensitivity analysis focused on short-
34    term exposure-related mortality endpoints (see Table 5-1 for the specific modeling
35    elements considered in the sensitivity analysis). The results of individual sensitivity
36    analysis simulations are presented below, with overall  observations presented at the end
37    of the section.

38       •  Impact of using  season-specific C-R functions (vs. an annual C-R function):
39          This simulation considered the impact on short-term exposure mortality  risk of
      September 2009                         134             Draft - Do Not Quote or Cite

-------
 1           using seasonally-differentiated effects estimates rather than the core approach of
 2           using a single C-R function for the whole year (note, that the seasonal models
 3           were based on the same study as the model used in the core analysis - Zanobetti
 4           and Schwartz, 2009). The results of the simulation suggest that this source of
 5           uncertainty can have a wide range of effects across urban study areas (including
 6           not only variation in the magnitude of effect, but also in the direction). Percent
 7           changes compared with the core risk estimate were large, ranging from -114%
 8           (Los Angeles) to +186% (Birmingham) (these results are for non-accidental
 9           mortality - see Appendix F, Table F-15). We note that these two locations also
10           have relatively low overall incidence estimates, which does raise concerns over
11           the degree of stability in the sensitivity analysis estimates.  Furthermore, for 9 of
12           the 15 urban study areas (for non-accidental morality), percent changes from the
13           core were small, with absolute values of 12% or less (Appendix F, Tables F-15).
14           The results for CV and respiratory mortality also demonstrate considerable
15           variation across locations, but are generally smaller than results cited above for
16           non-accidental, with one exception. Birmingham is estimated to have short-term
17           CV mortality that is +600% higher using seasonal effects estimates compared
18           with the core results (We note, however, that this endpoint category also has very
19           small incidence, again raising concerns over the stability of the sensitivity
20           analysis results) (see Table F-18). The  results for respiratory-related mortality
21           also demonstrate considerable variability with results that could suggest a
22           moderate to large impact (i.e., -47 to +162% -  see Appendix F, Table F-21).  We
23           note, however, that small incidence estimates again raise concerns regarding the
24           stability of these percent difference results.

25       •   Impact of using  alternative hybrid rollback approach:  This simulation
26           evaluates the potential impact of using the hybrid (non-proportional) approach for
27           simulating just meeting current and alternative sets of standards to reflect the
28           potential greater use of more localized  controls in some urban study areas (see
29           section 3.3.3), in place of the proportional approach used in the core analysis.
30           The results of this simulation (as contrasted with the impact of using the hybrid
31           approach on long-term exposure mortality) suggest that use of the hybrid rollback
32           approach has relatively little effect on short-term mortality risk (e.g., percentage
33           differences relative to the core risk estimates were in the low single digits for
34           most locations, with one location having a difference of+20% - see Appendix F,
35           Table F-36). These results suggest that the issue of how alternative standard
36           levels are simulated appears to introduce relatively little uncertainty into the
37           modeling of short-term exposure mortality.
38           The sensitivity analysis results discussed above, result in a number of overall
39    observations regarding sources of uncertainty potentially impacting short-term exposure
40    morality endpoints.  The results of using the seasonally-differentiated effect estimates in
41    modeling short-term exposure mortality appear to generally have a relatively small
42    impact (e.g., <15%) in most study areas. For some study areas, the impact does appear to
43    be much larger, with results including both substantial negative and positive percent
44    differences from the core estimates. However, in all of these cases, the total incidence
      September 2009                        13 5             Draft - Do Not Quote or Cite

-------
 1    estimates involved are very small, raising concerns over the stability of the risk estimates
 2    generated as part of this particular sensitivity analysis (in many of these instances, the
 3    estimates include negative lower bounds, reflecting the use of non-statistically significant
 4    effects estimates). For these reasons, the results of this sensitivity analysis, while initially
 5    appearing to be notable in terms of magnitude in some study areas, need to be interpreted
 6    with care. At this point, we are not able to reach a definitive conclusion as to how
 7    important this source of uncertainty is in the context of short-term exposure mortality
 8    estimation (due to the limitations regarding small incidence estimates and the stability of
 9    the relative differences in percent incidence estimates). Regarding the use of the
10    alternative hybrid (non-proportional) approach for simulating conditions under alternative
11    standard levels, the results suggest that this factor has a modest impact on short-term
12    exposure mortality (significantly less impact than with the use of the hybrid approach in
13    estimating long-term exposure mortality).  With the exception of factors examined using
14    the Moolgavkar et al., (2003) study in Los Angeles (see section 4.3.1.4), it would appear
15    that the factors examined  here do not have a large impact on risk estimates generated for
16    short-term exposure mortality.

17                    4.3.1.3  Sensitivity analysis results associated with short-term
18                            exposure  morbidity
19          This section summarizes the results of the sensitivity analysis focused on short-
20    term exposure-related morbidity endpoints (see Table 5-1 for the specific modeling
21    elements considered in the sensitivity analysis). The  results of individual  sensitivity
22    analysis simulations are presented below, with overall observations presented at the end
23    of the section.
24       •  Impact of using season-specific C-R functions (vs. an annual C-R function):
25          This simulation considered the impact on short-term exposure morbidity (HAs) of
26          using seasonally-differentiated effects estimates rather than the core approach of
27          using a single C-R function for the whole year (we note that the seasonal models
28          were obtained from the same study as the model used in the core analysis - Bell et
29          al, 2008).  The results of the simulation suggest that, as with short-term exposure
30          mortality this source of uncertainty can have a wide range of impacts on the risk
31          estimates across urban study  areas (including  not only variation in the magnitude
32          of risk, but also in the direction) depending on the specific health endpoint
33          examined. We note, however, that the magnitude of impact appears to be less for
34          short-term morbidity than for short-term mortality. Percent changes for most of
35          the 15 urban study areas were small for CV HAs (generally less than a 20%
36          difference in either direction, although there was a large impact for Tacoma (-
37          105%)) (see Appendix F, Table F-24).  This source of uncertainty has a moderate
38          to moderate-large  impact for  respiratory-related HAs with most locations having
39          greater than a 40% to 70% absolute effect (see Appendix F, Table F-27).

      September 2009                         136             Draft - Do Not Quote or Cite

-------
 1       •  Impact of using a seasonal function for April through August (applied only
 2          to that period) in modeling asthma-related ED visits in New York, relative to
 3          the core approach of using a single annual effect estimate (and applying that
 4          to the whole year):  This sensitivity analysis compared the core approach of
 5          using a single effect estimate to model asthma-related ED visits in New York
 6          (based on Ito et al., 2007) to the approach of using a season-specific estimate to
 7          model incidence for the period April through August (also obtained from Ito et
 8          al., 2007).  Because the simulation periods used with the two approaches do not
 9          match (i.e., a seasonal estimate versus annual estimate), we have not directly
10          compared the two to generate a percent difference estimate (as is done with the
11          other sensitivity analysis  simulations). While we do not draw a conclusion
12          regarding the importance of this factor, the results do argue for further research to
13          more fully characterize the impact of using seasonally-differentiated estimates in
14          modeling this endpoint).  As part of ongoing efforts to refine the sensitivity
15          analysis, we are considering compare risk estimates (for April-August) generated
16          using the single (annual)  effect estimates, to estimates generated for this same
17          time period using the seasonally-differentiated effect estimates.
18          Given the results  of the set of simulations completed for short-term exposure
19    morbidity (both of which focused on the use of seasonally-differentiated effects
20    estimates), it would appear that this factor does not have a substantial impact on risk
21    estimates.  Additional factors potentially impacting short-term exposure morbidity are
22    addressed below in relation to the sensitivity analysis based on alternative models from
23    Moolgavkar et al. (2003).

24                   4.3.1.4   Single-factor sensitivity analysis addressing model
25                            selection, lags, and co-pollutant models
26          As noted earlier in the introduction to section 4.3, the results of sensitivity
27    analysis based on Moolgavkar et al., (2003) include percent difference estimates based on
28    considering the range of risk estimates generated using alternative model specifications
29    from this study for a given health endpoint and it is these results that are discussed below.
30    However, sensitivity analysis results reflecting comparison of Moolgavkar-based risk
31    estimates against the core risk estimates, while included in the detailed results tables in
32    Appendix F, are not discussed here due to their high degree of uncertainty.

33       •  Impact of model selection (e.g., log-linear GAM with 30df, log-linear GAM
34          with lOOdf, and log-linear GLM with lOOdf) on estimating short-term
35          exposure mortality and  morbidity:  Application of models obtained from
36          Moolgavkar et al., (2003) with various formulations related to model selection
37          (degrees of freedom, GLM vs. GAM) to the Los Angeles urban case study
38          location results in a range of short-term exposure mortality estimates (for non-
39          accidental and CV) that differ by 79% and 49%, respectively (see Appendix F,
40          Table F-33).  In the case of short-term exposure morbidity (specifically, CV-
      September 2009                         137            Draft - Do Not Quote or Cite

-------
 1           related HAs), incidence estimates differ by 36% (see Appendix F, Table F-33).
 2           These results suggest that these elements of model specification represent a
 3           moderate source of uncertainty in estimating short-term mortality and morbidity.
 4       •   Impact of lag structure (0-day through 5-day) on estimating short-term
 5           exposure mortality: Consideration of the range of risk estimates for non-
 6           accidental mortality  generated using different lag structures (and associated effect
 7           estimates) provided in Moolgavkar et al., (2003), suggest that this factor could
 8           have a moderate impact on risk (in the range of 55% when comparing the lowest
 9           and highest positive  incidence estimates generated), (see Appendix F, Table F-
10           33).
11       •   Impact of considering multi-pollutant models on estimating short-term
12           exposure mortality and morbidity: The results of the Moolgavkar-based
13           simulations (when considering the spread in risk estimates specifically across
14           these simulations) suggest that the multi-pollutant versus single-pollutant model
15           issue (i.e., including CO in addition to PM2.5), could have a large impact on the
16           estimation of short-term exposure mortality (105% for all-cause) and morbidity
17           (142% for CV-related HAs).
18           Overall observations regarding key sources of uncertainty impacting short-term
19    exposure mortality and morbidity risk estimates (based on the Moolgavkar et al., 2003
20    study) include the following. The spread in risk estimates generated across the
21    Moolgavkar-based model specifications (for a particular endpoint), suggests that C-R
22    model selection factors may have a moderate to large impact. More specifically,
23    specification of lag structure has a moderate impact on risk and use of single versus
24    multi-pollutant models could have a potentially large impact on risk. Note, however,  that
25    as discussed earlier, the relevance of these sensitivity analysis results to the interpretation
26    of risk estimates generated using C-R functions derived from alternative epidemiology
27    studies that used different underlying  datasets and analytical approaches (i.e., Bell et al.,
28    2008 and Zannobetti and Schwartz et  al., 2009) is not clear and may be relatively low.

29         4.3.2  Multi-Factor  Sensitivity Analysis
30           The results of the multi-factor sensitivity analyses are intended to support both
31    goals of the sensitivity analysis:  (a) identify which factors (now in combination), appear
32    to have a significant impact on uncertainty in the core estimates  and (b) to derive a set of
33    reasonable alternative risk estimates for use in considering uncertainty associated with
34    the core risk estimates. Regarding the second goal, given that these sensitivity analysis
35    results reflect the combined impact of multiple uncertainty factors, they will likely
36    provide high- and low-bounds on the range of reasonable alternative risk estimates,
37    thereby providing key  information in the consideration of uncertainty associated with the
38    core risk estimates.  Consequently, in  providing an overall observations section here, we
      September 2009                         13 8             Draft - Do Not Quote or Cite

-------
 1    emphasize the potential reasonableness of each of the multi-element combinations
 2    included in the sensitivity analysis, since this factor is key to determining the degree to
 3    which these risk estimates can inform consideration of uncertainty related to the core risk
 4    estimates.
 5          Note, that as discussed in section 4.4.3, while we have included sensitivity
 6    analyses involving estimation of risk down to PRB for long-term exposure mortality,
 7    overall uncertainty associated with these estimates is considered higher than other
 8    estimates (due to the need to extrapolate the behavior of the C-R function below the
 9    LML). Consequently, results of the simulations involving estimation of risk down to
10    PRB will not be included in the set of reasonable alternative risk estimates used in
11    supporting consideration of uncertainty in the core risk estimates,  although these results
12    are discussed below in the context of understanding key sources of uncertainty associated
13    with the core risk estimates.

14                   4.3.2.1  Multi-factor sensitivity analyses - long-term exposure
15                            mortality
16          This section summarizes the results of the sensitivity analysis focused on long-
17    term exposure-related mortality endpoints (see Table 4-1 for the specific modeling
18    elements considered in the sensitivity analysis). The results of individual sensitivity
19    analysis simulations are presented below, with overall observations presented at the end
20    of the section.
21         •  Impact of using log-linear vs. log-log C-R model with fixed or random
22            effects, estimating incidence down to the LML vs. PRB, and using
23            proportional vs. hybrid rollback to estimate long-term exposure mortality:
24            This multi-factor sensitivity analysis focused on a number of model design
25            choices related to modeling long-term exposure mortality (all-cause and IHD).
26            Modeling elements reflected in the simulations included: model form (log-
27            linear vs log-log and random vs fixed effects), modeling risk down to PRB (vs
28            LML), and use of an alternative hybrid rollback approach (vs proportional
29            rollback) to simulate just meeting the current and alternative sets of standards.
30            Various permutations of these various design elements choices (relative to the
31            elements selected for the core analysis) were considered. Percent difference
32            estimates (for all-cause mortality) ranged from 147% (for a model estimating
33            risk down to PRB and use of the hybrid rollback approach) to 1,020% (for a
34            model with  random effects log-log model, risk estimated down to PRB, and use
35            of the hybrid rollback approach).
36          We believe that application of a log-log model with random effects is a
37    reasonable alternative to the core model (fixed-effects log-linear model), based on our
38    review of the discussion in Krewski et al. (2009). Similarly, the use of a hybrid rollback
      September 2009                         139            Draft - Do Not Quote or Cite

-------
 1    approach involving non-proportional adjustment where there is the potential for greater
 2    use of local control strategies to address local-sources, is a reasonable alternative to
 3    solely using a proportional rollback approach in all study areas.  Therefore, we believe
 4    that the combinations of modeling elements including these alternative choices are
 5    reasonable. However, there is more concern in predicting risk down to PRB. This is not
 6    because there is evidence for a threshold, but rather because we do not have data to
 7    support characterization of the nature of the C-R function in the vicinity of PRB.
 8    Specifically, there is increasing uncertainty in predicting the nature of the C-R function as
 9    you move below the LML. So, while we believe it is reasonable conceptually to estimate
10    risk down to PRB, the quantitative process of doing this requires use of a function with
11    very high uncertainty. Therefore, we concluded  that those alternative risk estimates
12    generated using risk estimated down to PRB should not be used in creating the reasonable
13    alternative set of risk estimates in considering uncertainty associated with the core  risk
14    estimates.
15                    4.3.2.2  Multi-factor sensitivity analyses - short-term exposure
16                            mortality
17          This section summarizes the results of the sensitivity analysis focused on short-
18    term exposure-related mortality endpoints (see Table 4-1 for the specific modeling
19    elements considered in the sensitivity analysis).  The  results of individual sensitivity
20    analysis simulations are presented below, with overall observations presented at the end
21    of the section.
22         •   Impact of using season-specific vs. annual effect estimates and
23             proportional vs. hybrid rollback approaches in modeling short -term
24             exposure mortality: This multi-factor sensitivity analysis focused on a
25             number of model design choices related to modeling short-term mortality (non-
26             accidental). Modeling elements included in this sensitivity analysis were use of
27             seasonal vs. annual effects estimates and use of hybrid vs proportional rollback
28            to simulate just meeting current and alternative standard levels. Percent
29             difference estimates (for non-accidental mortality) across the 7 urban study
30             areas included in the simulation ranged from -109% (LA) to +119%
31             (Birmingham) (see Appendix F, Table  F-42).  However, we note that the total
32             incidence estimates associated with these higher-impact locations were
33             relatively low, again raising the concern for the stability in relative differences
34            with the core estimates.
35          We believe that the application of both alternative model formulations reflected in
36    this multi-factor sensitivity analysis (seasonally-differentiated C-R functions and the
37    hybrid rollback approach) are reasonable, and consequently the risk estimates that are
38    generated do represent reasonable alternatives to the core estimates.

      September 2009                         140             Draft - Do Not Quote or Cite

-------
 1         4.4   EVALUATING THE REPRESENTATIVENESS OF THE URBAN
 2               STUDY AREAS IN THE NATIONAL CONTEXT
 3           The goal in selecting the 15 urban study areas included in this risk assessment was
 4    two fold: (a) to choose urban locations with relatively elevated ambient PM levels (in
 5    order to evaluate risk for locations likely to experience some degree of risk reduction
 6    under alternative standards) and (b) to include a range of urban areas reflecting
 7    heterogeneity in other PM risk-related attributes across the country. To further support
 8    interpretation of risk estimates generated in this analysis, we are assessing the degree to
 9    which urban study areas represent the range of key PM2.5 risk-related  attributes that
10    spatially vary across the nation. We have partially addressed this issue by selecting urban
11    study areas that provide coverage for different PM regions of the country (see section
12    3.3.2).  In addition, we are considering how well the selected urban areas represent the
13    overall U.S. for a set of spatially-distributed PM2.5 risk related variables (e.g., PM2 5
14    composition, weather, demographics including SES, baseline health incidence rates).
15    This analysis will help to inform how well the urban study areas reflect national-level
16    variability in these key PM risk-related variables.  Based on generally available data (e.g.
17    from the 2000 Census, Centers for Disease Control (CDC), or other sources),
18    distributions for risk-related variables across U.S.  counties and for the specific counties
19    represented in the urban study areas are generated. The specific values of these variables
20    for the  selected urban  study areas are then plotted  on these distributions, and an
21    evaluation is conducted of how representative the  selected study areas are with respect to
22    these individual variables, relative to the national distributions.
23           Estimates  of risk (either relative or absolute, e.g. number of cases) within our risk
24    assessment framework are based on four elements: population, baseline incidence rates,
25    air quality, and the coefficient relating air quality and the health outcome (i.e., the PM2.5
26    effect estimates).  Each of these elements can contribute to heterogeneity in risk across
27    urban locations, and each is variable across locations. In addition, there may be
28    additional identifiable factors that contribute to the variability of the four elements across
29    locations.  In this  assessment, we examine the representativeness of the selected urban
30    area locations for the four main elements, and also provide additional  assessment of
31    factors that have been identified as influential in determining the magnitude of the C-R
32    function across locations.
33           The specific choice of variables which may affect the PM2.5 effect estimates for
34    which we will examine urban study area representativeness is informed by an assessment
35    of the epidemiology literature. We particularly focused on meta-analyses and multi-city
36    studies which identified variables that influence heterogeneity in PM2.5 effect estimates,


      September 2009                         141             Draft - Do Not Quote or Cite

-------
 1    and exposure studies which explored determinants of differences in personal exposures to
 2    ambient PM2.5.  While personal exposure is not incorporated directly into PM
 3    epidemiology studies, differences in the PM2.5 effect estimates between cities clearly is
 4    impacted by differing levels of exposure and differences in exposure are clearly related to
 5    a number of exposure determinants. Broadly speaking, determinants of PM2.5 effect
 6    estimates can be grouped into three areas: demographics, baseline health conditions, and
 7    climate and air quality.  Based on a review of these studies, we identified the following
 8    variables within each group as potentially determining the PM2.5 effect estimates:
 9          •   Demographics:  education (see Zeka et al, 2006; Ostro et al, 2006), age and
10              gender (see Zeka et al, 2006), population density (see Zeka et al, 2005),
11              unemployment rates (see Bell and Dominici, 2008), race (see Bell and
12              Dominici, 2008), public transportation use (see Bell and Dominici, 2008),
13          •   Baseline health  conditions:  disease prevalence (diabetes - Bateson and
14              Schwartz, 2004; Ostro et al, 2006; Zeka et al, 2006; pneumonia - Zeka et al,
15              2006; stroke - Zeka et al,  2006; heart and lung disease - Bateson and
16              Schwartz, 2004; acute myocardial infarction - Bateson and Schwartz, 2004).
17          •   Climate and air  quality: PM2.5 levels (average, 98th percentiles, and numbers
18              of days over the level of the 24-hour standard, e.g. 35 jig/m3), co-pollutant
19              levels, PM composition (see Bell  et al, 2009; Dominici et al, 2007; Samet,
20              2008; Tolbert, 2007), temperatures (temp) (days above 90 degrees, variance of
21              summer temp, mean summer temp,  98th percentile temp, mean winter temp —
22              see Roberts, 2004; Medina-Ramon et al, 2006; Zeka et al., 2005), air
23              conditioning prevalence (see Zanobetti and Schwartz, 2009; Franklin et al,
24              2007; Medina-Ramon et al,  2006), ventilation (see  Sarnat et al, 2006), percent
25              of primary PM from traffic (see Zeka et al., 2005),
26          Based on these identified potential risk  determinants, we identified possible
27    datasets that could be used  to generate nationally  representative distributions for each
28    parameter. We were not able to identify readily available national datasets for all
29    variables.  In these cases, if we were able to identify a broad enough dataset covering a
30    large enough portion of the U.S., we used that dataset to generate the parameter
31    distribution.  In addition, we were not able to find exact matches for all of the variables
32    identified through our review of the literature.  In cases where  an exact match was not
33    available, we identified proxy variables to serve as surrogates.  For each parameter, we
34    report the source of the dataset, its degree of coverage, and whether it is a direct measure
35    of the parameter or a proxy measure.  The target variables and  sources for the data are
36    provided in Table 4-2.  Summary statistics for the most relevant variables are provided in
37    Table D-3.
      September 2009                         142            Draft - Do Not Quote or Cite

-------
2   Table 4-2.    Data Sources for PM NAAQS Risk Assessment Risk Distribution
3                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
CnHe
\^-LJUt
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 -
Monitored Ann Mean
PM2 5 Levels ~
2007 AQS
2007 AQS
617 Monitored
counties
617 Monitored
    September 2009
143
Draft - Do Not Quote or Cite

-------

Potential Risk
Determinant


PM2 .5 Levels

PM2 .5 Levels

Copollutant Levels

Roadway
emissions/Exposure
Temperature

Temperature

Relative Humidity

Ventilation


Metric
Monitored 98th %ile

Average MCAPS

% days exceeding 35
, 3
ug/m
Ozone

% of primary emissions
from traffic
Annual Average

Mean July Temp 1941-
1970

Mean July RH 1941-
1970

Air conditioning
prevalence


Year








1999






2005


Source


MCAPS website 204 counties

MCAPS website 204 counties

AQS

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
additional processing as in Reid et
al (2009)
Degree of
National
Coverage
counties
204 MCAPS

counties
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

Obsesity
Level of exercise



Level of exercise

Respiratory Risk
Factors

Smoking
All Cause
Non Accidental
Cardiovascular
Respiratory

AMI prevelence

Diabetes Prevalence
Pneumonia Prevalence

Stroke Prevalence

CHD Prevelence
COPD Prevalence

BMI
vigorous activity 20
minutes
moderate activity 30

minutes or vigorous
activity 20 minutes
Current Asthma


Ever Smoked





2007

2007


2007

2007


2007
2007



2007

2007


2007
CDC Wonder 1999-2005
CDC Wonder 1999-2006
CDC Wonder 1999-2007
CDC Wonder 1999-2008

BRFSS MSA estimates

BRFSS MSA estimates


BRFSS MSA estimates

BRFSS MSA estimates


BRFSS MSA estimates
BRFSS MSA estimates



BRFSS MSA estimates

BRFSS MSA estimates


BRFSS MSA estimates
All counties
All counties
All counties
All counties
184 BRFSS
MSA
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
C-R Estimates
Mortality Risk

Mortality Risk
All Cause

Respiratory
2009

2009
Zanobetti and Schwartz (2009) 212
cities
Zanobetti and Schwartz (2009) 212
cities
212 cities

212 cities
September 2009
144
Draft - Do Not Quote or Cite

-------
Potential Risk
Determinant
Mortality Risk
Metric
Cardiovascular
Year
2009
Source
Zanobetti and Schwartz (2009) 212
cities
Degree of
National
Coverage
212 cities
September 2009                       145            Draft - Do Not Quote or Cite

-------
1   Table 4-3.    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
    September 2009
146
Draft - Do Not Quote or Cite

-------
Risk Attributes
July Temperature Long Term Average (°F)
July Relative Humidity Long Term Average (°F)
C-R Estimates
All Cause Mortality PM2.5 Risk Estimate
Respiratory Mortality PM2.5 Risk Estimate
Cardiovascular Mortality PM2.5 Risk Estimate
Average
Urban study U.S.
areas counties
78.1 75.9
58.2 56.2

0.000971 0.000974
0.001606 0.001670
0.001013 0.000842
Standard Deviation
Urban study U.S.
areas counties
4.5 5.4
14.0 14.6

0.000340 0.000216
0.000419 0.000305
0.000586 0.000324
Maximum
Urban study U.S.
areas counties
91.2 93.7
70.0 80.0

0.001349 0.001508
0.002157 0.002221
0.001958 0.001958
Minimum
Urban study U.S.
areas counties
64.8 55.5
19.0 14.0

0.000159 -0.000099
0.000931 -0.000346
-0.000180 -0.000180
Sample Size
Urban study U.S.
areas (number
(number of of
counties) counties)
31 3104
31 3104

15 112
15 112
15 112
September 2009
147
Draft - Do Not Quote or Cite

-------
 1           Formal comparisons of parameter distributions for the set of urban study areas
 2    and the national parameter distributions are conducted using standard statistical tests, e.g.
 3    the Kolmogorov-Smirnov non-parametric test for equality of distributions. In addition,
 4    visual comparisons are made using cumulative distribution functions, and boxplots.
 5           The formal Kolmogorov-Smirnov test results are provided in Table 4-4.  The K-S
 6    tests the hypotheses that two distributions are not significantly different. A high p-value
 7    indicates a failure to reject the null hypotheses that the case-study and national
 8    distributions are the same. We used a rejection criterion of p<0.05, which is a standard
 9    rejection criteria.  It should be noted that the K-S test provides a good overall measure of
10    fit, but will not provide a test of how well specific percentiles of the distributions are
11    matched.  As such, the K-S test results will not be sufficient to determine whether the
12    urban study areas adequately capture the tails of the distributions of specific risk related
13    variables. Additional visual analyses are used  to assess representativeness for the tails of
14    the distributions.  Overall, the K-S test results show that for many of the important risk
15    variables such as population, air quality, age, and baseline mortality rates, the urban study
16    areas are not representative of the distributions of these variables for the U.S. as a whole.
17    However, for some important potential risk determinants, such as prevalence of
18    underlying hear and lung diseases, the case study areas are representative of the national
19    distributions. However, for these specific variables, the national distribution is
20    represented primarily by large urban areas, so it is more accurate in these cases to suggest
21    that the urban study areas are representative of the overall distribution across urban areas.
22           Figures 4-14 through 4-17 show for the four critical risk function elements
23    (population, air quality, baseline incidence, and the PM2.5 effect estimate) the cumulative
24    distribution functions plotted for the nation, as well as for the urban study areas.  These
25    four figures focus on critical variables representing each type of risk determinant, e.g. we
26    focus on all-cause mortality rates, but we also have conducted analyses for cardiovascular
27    and respiratory mortality separately. The complete set of analyses is provided in
28    Appendix D.  The vertical black lines in each graph show the values of the variables for
29    the individual urban study areas. These figures show that the selected urban study areas
30    represent the upper percentiles of the distributions of population and air quality, while not
31    representing lower population locations with lower 24-hour PM2.5 levels. This is
32    consistent with the objectives of our case study selection process, e.g. we are
33    characterizing risk in areas that are likely to be experiencing excess risk due to PM levels
34    above alternative  standards.  The urban case study locations represent the full distribution
35    of PM2.5 risk coefficients, but do not capture the upper end of the distribution of baseline
36    all-cause mortality. The interpretation of this is that the case study risk estimates may not

      September 2009                         148             Draft - Do Not Quote or Cite

-------
 1    capture the additional risk that may exist in locations that have the highest baseline
 2    mortality rates.
 3          Figures 4-18 through 4-21 shows for several selected potential risk attributes the
 4    CDF plotted for the nation as well as for the urban study areas. These potential risk
 5    attributes do not directly enter the risk equations, but have been identified in the literature
 6    as potentially affecting the magnitude of the PM2.5 C-R functions reported in the
 7    epidemiological literature.  The selected urban study areas do not capture the higher end
 8    percentiles of several risk characteristics, including populations over 65, income, and
 9    baseline cardiovascular disease prevalence.  Comparison graphs for other risk attributes
10    are provided in Appendix D. Summarizing the analyses of the other risk attributes, we
11    conclude that the urban study areas provide adequate coverage across population,
12    population density, annual  and 24-hour PM2.5 levels, ozone co-pollutant levels,
13    temperature and relative humidity, unemployment rates, percent non-white population,
14    asthma prevalence, obesity prevalence, stroke prevalence, exercise prevalence, and less
15    than high school education. We also conclude that while the urban study areas cover a
16    wide portion of the distributions, they do not provide coverage for the upper end of the
17    distributions of age (all case study locations are below the 85th %ile), % of population 65
18    and older (below 85th %ile), percent of primary PM  emissions from mobile sources
19    (below 8oth %ile), prevalence of angina/coronary heart disease (below 85th %ile),
20    prevalence of diabetes (below 85th %ile), prevalence of heart attack (below 80th %ile),
21    prevalence of smoking (below 85th %ile), all-cause mortality rates (below 90th %ile),
22    cardiovascular mortality rates (below 90th %ile) and respiratory mortality rates (below
23    90th %ile). In addition, all of the case study locations were above the 25th percentile of
24    the distribution of personal income.
25          Based on the above analyses, we can draw several inferences regarding the
26    representativeness of the urban case studies. First, the case  studies represent urban areas
27    that are among the most populated and most densely population in the U.S.  Second, they
28    represent areas with relatively higher levels of annual mean and 24-hour 98th percentile
29    PM2.5. Third, they capture well the range of effect estimates represented in the Zanobetti
30    and Schwartz (2009) study. These three factors would suggest that the urban study areas
31    should capture well overall risk for the nation, with a potential for better characterization
32    of the high end of the risk distribution.  However, there are several other factors that
33    suggest that the urban study areas may not be representing areas that may have a high risk
34    per microgram of PM2.5. The analysis suggests that the urban study areas are not
35    capturing areas with the highest baseline mortality risks, nor those with the oldest
36    populations.  These areas may have higher risks per microgram of PM2.5, and thus the

      September 2009                         149            Draft - Do Not Quote or Cite

-------
 1   high end of the risk distribution may not be captured, although the impact on
 2   characterization of overall PM risk may not be as large, for the following reasons.
 3          It should be noted that several of the factors with underrepresented tails, including
 4   age and baseline mortality (R=0.81) are spatially correlated, so that certain counties
 5   which have high proportions of older adults also have high baseline mortality and high
 6   prevalence of underlying chronic health conditions. Because of this, omission of certain
 7   urban areas with higher percentages of older populations, for example, cities in Florida,
 8   may lead to underrepresentation of high risk populations.  However, with the exception
 9   of areas in Florida, most locations with high percentages of older populations have low
10   overall populations, less than 50,000 people in a county. And even in Florida, the
11   counties with the highest PM2.5 levels do not have a high percent of older populations.
12   This suggests that while the risk per exposed person per microgram of PM2.5 may be
13   higher in these locations, the overall risk to the population is likely to be within the range
14   of risks represented by the urban case study locations.
15   Table 4-4     Results of Kolomogrov-Smirnoff Tests for Equality Between National
16                 and Urban Study Area Distributions for  Selected National Risk
17                        Characteristic Variables
18                (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
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
      September 2009
150
Draft - Do Not Quote or Cite

-------
1


2



3

5


7


9
Risk Attributes
Air Quality and Climate
AQ - PM25 Annual Mean
AQ - PM25 98th %ile 24-hour Average
AQ - PM25 % of days above 35 ug/m3
AQ - O3 4th High Maximum 8-hour
Average
% Mobile Source PM Emissions
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?
p-value

Y
Y
Y

Y
Y
Y

N
0.0001
0.0001
0.0248

0.0003
0.0133
0.0003

0.0614

N
N
N
0.1585
0.2864
0.1161
Figure 4-14.  Comparison of distributions for key elements of the risk equation:

              total population.




         Comparison of Urban Case Study Area Population with U.S. Distribution of

                             Population (all U.S. Counties)
      100%
       90% -
    i/)
    .0)
    '^
    c

    o
    o
          Urban case study areas are
              all above the 65th Percentile
           100
                    1000
10000         100000


     Population
1000000
10000000
                   All Counties CDF ^—Case Study Counties CDF   •  Case Study Counties
     September 2009
                                        151
                     Draft - Do Not Quote or Cite

-------
1   Figure 4-15   Comparison of distributions for key elements of the risk equation:
2
3
                   r>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%
 70%
    «,
    0)
   1
   O  60%
    2j  50% -
    §  40%
   •g  30%
   ^  20% -
       10% -
        0%

                   10
                     20
30      40      50      60
  98th Percentile Daily PM2.5
70
80
90
                    All Counties CDF	Case Study Counties CDF  •  Case Study Counties
     September 2009
                                     152
                       Draft - Do Not Quote or Cite

-------
1
2
          Figure 4-16.  Comparison of distributions for key elements of
          the risk equation:  all cause mortality rate.

Comparison of Urban Case Study All Cause Mortality Rate to U.S. Distribution of All
                          Cause Mortality Rate
                          (3143 U.S. Counties)
1 UU /O -
90% -
80% -
8 70%-
•*=
1 60%
o
° 50%
CO
2 40%
0
£ 30%
20%
10%
n%
;





























































^^^^^



/
/



/









/


















**









--









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
     September 2009
                             153
Draft - Do Not Quote or Cite

-------
1
2
Figure 4-17.  Comparison of distributions for key elements of the risk equation:
             mortality risk effect estimate from Zanobetti and Schwartz (2008).

           Comparison of Urban Case Study PM All-cause Mortality Risk ((3) to
                    U.S. Distribution of PM All-cause Mortality Risk
                               (212 U.S. Urban Areas)
3
4
5
IUU/0 -
90% -
% 80% -
CD
< 70% -
ra 60% -
§ 50% -
$ 40% -
^ 30% -
^ 20% -
10% -
n% -



^^^^^—\



x 	 '
i — i 	 1



^
S^
i — i 	 1 — i



>
i 	 1 	 1



H







1 	 1



1 — 1

^

1^ 	 1



1 	 1



H



1 	 1 	
                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
    September 2009
                                     154
Draft - Do Not Quote or Cite

-------
1   Figure 4-18.  Comparison of distributions for selected variables expected to

2                 influence the relative risk from PMi.s: long term average July

3                 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%
4

5
       in
       0)
       '^

       3
       O
       o

       ttj
       =)
              50
60             70             80             90

   July 30 Year Average Temperature, 1941-1970
                            100
                       All Counties CDF ^—Case Study Counties CDF  •  Case Study Counties
     September 2009
              155
Draft - Do Not Quote or Cite

-------
2    Figure 4-19.  Comparison of distributions for selected variables expected to
3                  influence the relative risk from PM2.s: percent of population 65 and
4                  older.

                Comparison of Urban Case Study Area % 65 and Older to U.S. Distribution of % 65
                                               and Older
                                          (3141 U.S. Counties)
        o
       O
       CO
       Z)
              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
     September 2009
156
Draft - Do Not Quote or Cite

-------
2    Figure 4-20.  Comparison of distributions for selected variables expected to
3                 influence the relative risk from PM2.s: per capita annual personal
4                 income.

                  Comparison of Urban Case Study Area Per Capita Personal Income to U.S.
                                Distribution of Per Capita Personal Income
                                        (3141 U.S. Counties)
1 UU /O
90% -
80% -
w 70%
1 60%
0
ri 50%
2 40%
o
£ 30%
20%
1 0%
no/, -,





^/




I — 1 1




1 1 l-l 1




1 H




1 	 1




l-l 1




H 1 	 1




l-l
X



i — i

Urban case study areas are
all above the 25th percentile
of county per capita income


1, -n 	 1 1—
            $10,000  $20,000  $30,000  $40,000  $50,000  $60,000  $70,000  $80,000  $90,000

                                      Per Capita Personal Income, 2005
                       All Counties CDF ^—Case Study Counties CDF  •  Case Study Counties
     September 2009
157
Draft - Do Not Quote or Cite

-------
 2   Figure 4-21.  Comparison of distributions for selected variables expected to
 3                 influence the relative risk from PM2.s:  cardiovascular disease
 4                 prevalence days.

                Comparison of Urban Case Study Area Angina/CHD Prevalence to U.S. Distribution
                                       of Angina/CHD Prevalence
                                            (183 U.S. MSA)
        CO
        2
        CO
        CO
        U-
        o:
        m
100% -,
 90%
 80% -«
 70%
 60%
 50%
 40%
 30%
 20%
 10%
  0%
     1
Urban case study areas are
all below the 85th percentile
of MSA angina or CHD
prevalence
                                   34567
                                   % Prevalence of Angina or CHD, BRFSS, 2007
                     All BRFSS MSA CDF ^—Case Study County MSA CDF  •  Case Study County MSA
 6         4.5   OVERALL SUMMARY AND KEY OBSERVATIONS
 7          This section provides a summary of the key observations related to the core risk
 8   results, including both the recent conditions analysis results and estimates generated for
 9   the current and alternative sets of standards (section 4.5.1).  Next, observations resulting
10   from the sensitivity analyses related to the first goal for the analyses (what they tell us
11   about the key sources of uncertainty potentially impacting the core risk estimates) are
12   presented (section 4.5.2). Then, observations related to the second goal of the sensitivity
13   analysis  (how the output of the sensitivity analysis can be used to gain some perspective
14   on the magnitude of uncertainty related to the core risk estimates) are presented (section
15   4.5.3). The final two subsections end by discussing the results of two national-scale
16   assessments designed to evaluate the representativeness of the  15 urban study areas (and
17   the 31 counties comprising those urban areas) in the national context with regard to PM-
18   related risk.  Specifically, section 4.5.4 discuses observations from an assessment of how
      September 2009
                                   158
Draft - Do Not Quote or Cite

-------
 1    the 15 urban study areas (and their 31 counties) compare against a set of national
 2    distributions for key PM risk-related attributes.  Section 4.5.5 discusses observations
 3    resulting from seeing how the set of 15 urban study areas (and their 31 counties) are
 4    distributed across a cumulative plot of national-scale long-term PM2.5-related mortality
 5    generated for all counties  in the U.S.

 6          4.5.1  Core risk results from the recent conditions, current NAAQS, and
 7                alternative NAAQS analyses
 8           This section provides an overall summary and key observations related to the core
 9    risk estimates (see section 4.1 and 4.2 for a detailed discussion of these results). As
10    discussed in section 4.1, the core risk estimates are generated using model inputs believed
11    to have the greatest support in the literature and therefore, they have received greater
12    focus in the discussions of risk estimates generated for the urban study areas. The results
13    discussed below related to just meeting the current and  alternative sets of standards are
14    based on results from 14 of the 15 study areas, with Pittsburgh being excluded at this
15    time due to an error that was identified in the approach  used to simulate ambient PM2.5
16    levels just for the Pittsburgh study area for the scenarios involving just meeting the
17    current and alternative sets of standards. As  noted earlier in sections 2.4.1 and 4.2, there
18    was insufficient time after identifying this  error to either generate corrected risk estimates
19    or remove the erroneous risk estimates from the summary tables (presented in Appendix
20    E).  We will correct this error and release updated results  for the Pittsburgh study area as
21    soon as is practicable and will include the corrected results in the next version of this
22    document.
23           This overview of risk results is organized by air quality scenario, beginning with
24    recent conditions followed by just meeting the current set of standards and each of the
25    four sets of alternative standards considered.  An important factor to consider in
26    interpreting the risk estimates for both the current  set of standards and set of alternative
27    standards is whether the annual or 24-hour standard for a given pairing of standards is
28    controlling for a particular area.43 This factor can have a significant impact on the pattern
29    of risk reductions predicted for a given location under the simulation of just meeting a
30    specific set of standards.  As such, particular attention is given to this factor when
31    discussing risk reductions associated with the current and alternative sets of standards.
      43 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 Table 3-3 for the annual and 24-hr
      design values for each of the urban study areas). 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.

      September 2009                          159             Draft - Do Not Quote or Cite

-------
 1    Recent conditions analysis: Because this air quality scenario represents recent
 2    conditions, risk estimates are described in absolute terms, not in terms of a change or
 3    reduction relative to another air quality scenario.

 4         •   Long-term exposure mortality:  The recent conditions analysis for long-term
 5             exposure mortality suggests that across the 15 urban study areas, for the
 6             simulation year 2007, from 1.7 to 6.7% of all cause mortality incidence is
 7             estimated to be associated with PM2 5 exposure. Risk estimates are notably
 8             higher for specific mortality categories, with IHD having the highest percent
 9             contribution from PM2 5, ranging from 6.7 to 23.5% across the 15 urban study
10             areas. We note that a significant portion of these ranges results from
11             application of two sets of C-R functions - one based on monitoring data from
12             1979-1983 and a second set derived using monitoring data from 1999-2000
13             from the same cohort study (i.e., the extended analysis of the ACS).

14         •   Short-term exposure mortality and morbidity: Recent conditions estimates
15             for short-term exposure mortality suggest that from 0.2 to 1.7 percent of all
16             non-accidental mortality is associated with PM2 5.  As with long-term mortality,
17             estimates for individual short-term exposure mortality categories (e.g.,
18             respiratory) are higher, ranging up to 2.9%. Estimates for short-term exposure
19             morbidity (HAs) are  similar in magnitude to short-term mortality.  The analysis
20             of emergency department visits in New York and Los Angeles produced
21             estimates nearer the lower end of this range (e.g., 0.6% of incidence attributable
22             to PM2.5).
23    Simulating just meeting the current  standards:  Risk estimates for this air quality
24    scenario in 14 study areas are described by comparing estimates of risk attributable to
25    PM2.5 under the current set of standards to risk estimates under recent conditions.44  Note
26    that two of the 14 urban study areas (Dallas and Phoenix) have both annual and 24-hr
27    design values that are below the matching current standard levels of 15 and 35 (ig/m3,
28    respectively (see Table 3-3). This means that these two urban study areas would not have
29    any reduction in long-term or short-term risk under the current standards, as reflected in
30    the summaries presented below (note, however, that both areas have predicted risk
31    reductions under some of the alternative sets of standards).

32         •   Long-term exposure mortality:  Of the 12 urban study areas with at least one
33             design value exceeding the current standard levels, we estimated that the
34             percent of total all-cause mortality (associated with PM2.5) would be reduced by
35             5 to 15% (for 5 urban study areas), by 30 to 40% (for 4 urban study areas), and
36             by -60% (for 3 urban study areas) relative to risk estimated under recent
37             conditions.  The magnitude of risk reductions is similar for the other long-term
38             cause-specific mortality categories.
      44 As noted earlier, Pittsburgh has been excluded due to an error that was identified in the approach used to
      simulate ambient PM2 5 levels just for the Pittsburgh study area for the scenarios involving just meeting the
      current and alternative sets of standards. Consequently, the number of study areas discussed in relation to
      the current standard (as well as alternative standards) is 14, rather than 15.


      September 2009                         160             Draft - Do Not Quote or Cite

-------
 1         •   Short-term exposure mortality and morbidity:  Short-term exposure non-
 2             accidental mortality is estimated to have a slightly smaller overall degree of risk
 3             reduction under the current set of standards compared with reductions estimated
 4             for long-term exposure mortality.  With regard to the 12 urban study areas
 5             predicted to have risk reductions under the current standards, we estimated that
 6             total non-accidental mortality associated with PM2.5 would be reduced by less
 7             than 10% (for 4 areas), by 10 to 30% (for 5  areas), and by 40 to -70% (for 4
 8             areas) relative to risk estimated under recent conditions.  Other short-term
 9             exposure mortality categories, as well as short-term exposure morbidity
10             categories (HAs), display a similar distribution of risk reduction across the
11             urban study areas for the current set of standards.
12    Simulating just meeting alternative sets of standards focusing on lower annual
13    standard levels (13/35,12/35): Risk reductions  for alternative sets of standards were
14    calculated relative to risk estimated under the current set of standards.44  We note that
15    Dallas and Phoenix had annual and 24-hour design values which were lower than the
16    13/35 set of alternative standards , but the annual design values were above the  12/35 set
17    of alternative standards (see Table 3-3). For two of the other study areas (Fresno and
18    Los Angeles), where the 24-hr standard was controlling, simulating just meeting the
19    current standard resulted in significant reductions in the annual PM2 5 levels, such that no
20    risk reduction was seen for the 13/35 alternative set of standards (i.e., adjusted annual
21    PM2.s levels for these study areas under the current set of standards were already below
22    13 |ig/m3).  We note, however, that Los Angeles did show risk reductions under the 12/35
23    set of standards , while Fresno continued not to have predicted risk reductions under the
24    12/35 set of alternative standards  due to the  significant reduction in annual levels
25    associated with just meeting the current set of standards. Because Tacoma and Salt Lake
26    City already had annual design values (10 and 12 |ig/m3, respectively) at or below the 12
27    ng/rn3 associated with the lower of these two alternative sets of standards, neither study
28    area exhibited risk reductions. Specific ranges of risk reductions for those urban study
29    areas predicted to have reductions under these two alternative sets of standards are
30    summarized below:

31         •   Long-term exposure mortality: For the 8 study areas predicted to have risk
32             reductions under the 13/35  set of standards, one area had reductions in all-cause
33             mortality (attributable to PM^.s) of about 5%, while the remaining 7 had
34             reductions from over 10  to -25% relative to risk estimated under the current set
35             of standards. Under the 12/35 set of standards, 11 of the urban study areas were
36             predicted to have risk reductions, with 4 areas having 10 to 22% reductions and
37             7 locations having  25 to 35% reductions relative to risk estimated under the
3 8             current set of standards.  The magnitude of risk reductions is similar for the
39             other long-term exposure cause-specific mortality categories.
40    Short-term exposure mortality and morbidity: For the 8 study areas predicted to have
41    risk reductions for short-term exposure mortality and morbidity endpoints under the
42    13/35 set of standards, reductions range from 3 to 15% across all endpoints relative to
43    risk estimated under the current set of standards.  For the 11 urban study areas predicted
44    to have risk reductions for short-term exposure mortality and morbidity endpoints under
45    the  12/35 set of standards, reductions range from 5 to 23%  across all endpoints.
      September 2009                         161             Draft - Do Not Quote or Cite

-------
 1    Simulating just meeting  alternative sets of standards focusing on combinations of
 2    lower annul and lower 24-hour levels (13/30,12/25): Because of the combination of
 3    lower 24-hr and annual levels, 13 of the  14 urban study areas had risk reductions under
 4    the 13/30 standard level (only Dallas continued not to have predicted risk reductions),
 5    while all 14 urban study areas exhibited  risk reductions under the 12/25 alternative set  of
 6    standards.
 7         •   Long-term exposure mortality:  Under the 13/30 set of standards, reductions
 8             in all-cause mortality attributable to PM2.5 ranged from 14 and 44% relative to
 9             risk under the current set of standards across 13 study areas.  Under the  12/25
10             set of standards,  the degree of risk reduction in all-cause mortality was between
11             12 and 89% across all 14 study areas . The magnitude of risk reductions was
12             similar for the other long-term cause-specific mortality categories.
13         •   Short-term exposure mortality and morbidity:  Under the 13/30 set  of
14             standards, reductions in non-accidental mortality attributable to PM2.5 ranged
15             from 6 to 15% relative to the current set of standards. Under the 12/25 set of
16             standards, risk reductions ranged from 7 to 29%. Other short-term exposure
17             mortality categories, as well as short-term exposure morbidity categories
18             (HAs), display a similar distribution of risk reduction across the urban study
19             areas.
20         4.5.2  Use of the reasonable set  of alternative risk estimates generated
21                through the sensitivity analysis
22           As discussed in the introduction to this chapter, the use of alternative inputs to the
23    risk assessment model that have some degree of support in the literature (for the single
24    and multi-element sensitivity analyses) results in a set of reasonable risk estimates that
25    supplement the core risk estimates (the full set of sensitivity analysis-related risk
26    estimates are presented in the detailed tables presented in Appendix F and summarized in
27    section 4.3). Specifically, these additional risk estimates represent a set of alternative
28    realizations around the core set of risk results and while these results collectively do not
29    represent a formal uncertainty distribution, they can be used to gain some perspective
30    regarding the range  of uncertainty around the core risk results.
31           After careful consideration of the results of the sensitivity analysis completed for
32    the three categories  of health endpoints included in the risk assessment (long-term
33    exposure mortality,  short-term exposure mortality, and short-term exposure  morbidity),
34    we have concluded that only the sensitivity analysis results generated for long-term
35    exposure mortality are robust enough to  support using those estimates to inform
36    consideration of uncertainty  associated with the core risk estimates.45 In the case of both
      45 Note, however, that as discussed earlier in section 4.3.1, uncertainty associated with modeling risk down
      to PRB is such that we have not included these risk estimates as part of the reasonable set of risk estimates
      and therefore they are not discussed here. However, the PRB-based risk estimates can be used generally, to

      September 2009                         162             Draft - Do Not Quote or Cite

-------
 1    short-term exposure mortality and morbidity, the sensitivity analyses were not as
 2    comprehensive (the number of modeling elements considered was smaller), reflecting
 3    information readily available in the underlying epidemiological studies. As noted above
 4    (sections 4.3.1 and 4.3.2), the results of the sensitivity analysis conducted for these two
 5    short-term health impact categories, resulted in a combination of inconclusive results (in
 6    the case of seasonally-differentiated effects estimates) and fairly small impacts on risk
 7    (for the application of the hybrid rollback approach for simulating just meeting
 8    alternative suites of standards). Therefore, there would be little utility in using these
 9    sensitivity analysis results to generate an additional set of risk estimates to inform
10    consideration of uncertainty associated with the core risk results.46
11           We also considered the use of the Moolgavakar-based sensitivity analysis results,
12    reflecting the range of risk across alternative model choices based on this single study.
13    However, use of this study to generate alternative sets of risk estimates to supplement the
14    core estimates is problematic given the questionable relevance of the Moolgavkar-based
15    sensitivity analysis findings to the core risk results (i.e., core risk estimates are based on
16    large multi-location studies with different design elements relative to the Moolgavkar et
17    al., 2003 study conducted only for Los Angeles). Therefore, we concluded that the
18    Moolgavkar-based results should not be used to generate an additional set of risk
19    estimates to supplement the core estimates in considering uncertainty.  This means that
20    only the long-term mortality endpoints have been included in this effort to use the results
21    of the sensitivity analysis to inform consideration of uncertainty related to the core risk
22    estimates.
23           For purposes  of illustrating how the results of the sensitivity analysis (specifically
24    the multi-factor analysis) can be used to inform consideration of the range of uncertainty
25    associated with the core risk estimates involving long-term mortality, we have provided
26    one example below, focusing on Los Angeles for long-term (all cause) mortality.  We
      gain perspective for the overall magnitude of uncertainty that modeling risk down to PRB (rather than
      LML) might produce and for that reason they are discussed in the context of identifying key sources of
      uncertainty impacting the core risk estimates.
      46 As part of ongoing refinements to the risk assessment, we may consider additional sensitivity analyses
      focused on short-term exposure mortality and morbidity endpoints. One potential area of exploration is the
      C-R functions used, although in the case of short-term exposure mortality, it may not be possible to gain
      further insights into alternate model specifications (from a quantitative standpoint) since we obtained a set
      of location-specific effects estimates from the authors and various model designs provided in the original
      study would have little applicability to assessing sensitivity of the location-specific effect estimates.
      September 2009                          163             Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
note that the approach illustrated here for this urban study area location/health endpoint
combination can be readily repeated for other combinations (based on application of data
in Appendices E and F). We also note that the  sensitivity analyses were conducted based
exclusively for the scenario where the current set of standards is just met, so our
observations here are limited to that scenario. However, we believe that observations
made based on these results can be extrapolated with care to other air quality scenarios,
for the same health endpoint/urban study area location.
       The process of generating the additional set of reasonable risk estimates based on
the sensitivity analysis results involves  using the results of the sensitivity analysis (which
are in the form of percent difference compared to the core estimates) to adjust the core
risk estimates. This procedure results in the additional set of reasonable risk estimates
which can be used to inform consideration of uncertainty in the underlying core results.
Table 4-5 provides the underlying core  risk estimates, the percent difference results (from
the sensitivity analysis) and the resulting adjusted risk estimates that represent the
additional set of reasonable risk results  (for Los Angeles/all-cause mortality  combination
for simulation year 2007).
Table 4-5     Derivation of a set of reasonable alternative risk estimates to
              supplement the core  risk estimates (Los Angeles, current standards,
              for long-term all cause mortality).
Core risk estimate
Percent of total
incidence for all cause
mortality (current suite
of standards): 1.7 to
2.2%
(range reflects two C-
R functions based on
different periods of
ambient data - see
Appendix F, Tables F-
6 and F- 15)
Sensitivity analysis
Description simulation
Results
(percent difference:
sensitivity analysis
versus core estimate)
Adjusted set of risk
estimate to
supplement core
risk estimates1
Single-element sensitivity analysis results
Impact of using different
model choices: random effects
log-linear model
Impact of using different
model choices: random effects
log-log model
Impact of C-R function from
alternative long-term exposure
study (Krewski et al., 2000)
Impact of using alternative
hybrid roll-back approach to
simulate just meeting
alternative standards
+23%
+155%
-25%
+24%
2.1 to 2.7%
4.3 to 5.6%
1.3 to 1.7%
2.1 to 2.7%
Multi-element sensitivity analysis results
Random effects log-log &
hybrid non-proportional
rollback
+203%
5.2 to 6.7%
20
21
1 Percent of total incidence that is PM2 5- related.
      September 2009
                                        164
Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
       The reasonable set of alternative risk estimates is presented along with the core
set of risk estimates for the example scenario in Figure 4-22.  As noted earlier in
discussing the results of the single-factor sensitivity analysis (and in Table 4-5), 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 (i.e., for long-term ambient PM2.5
levels below the levels for any of the urban areas included in the underlying
epidemiological  studies). Therefore, results of modeling risk down to PRB are not
included in Figure 4-22.
Figure 4-22.  Comparison of core risk estimates with reasonable alternative set of
              risk estimates generated as part of sensitivity analyses.
Los Angeles for current conditions air quality scenario (2007 simulation year)
Core risk estimates L-
Additional set of "~"|
reasonable risk r"~
estimates — J
0
• «
•
• •



•

• •

•


II I I I I I I
01 234567
Percent total incidence attributable to PM2 5 (all cause mortality)
       The additional sets of risk estimates presented in Table 4-5 (and illustrated in
Figure 4-22) can be used to inform consideration for the range of uncertainty associated
with the core risk estimates.  Given consideration for alternative model specifications that
have some degree of support in the literature, the range of risk estimates around the core
estimates of 1.7 to 2.2% extends from 1.3 to 6.7% for total all cause mortality attributable
to PM2.5. It is important to reiterate that this set of alternative realizations presented in
Table 4-5  and depicted in Figure 4-22, 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 see in Figure 4-22. Furthermore, 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 range of risk estimates reflecting the impact of modeling element uncertainties
      September 2009
                                        165
Draft - Do Not Quote or Cite

-------
 1    does provide information that helps to inform our characterization of uncertainty related
 2    to the core risk estimates.

 3          4.5.3  Representativeness of the urban study areas in the national context
 4                based on consideration for coverage of PMi.s risk-related parameters
 5           The representativeness analysis (discussed in detail in section 4.4) considered
 6    how well the 15 urban study areas represent the overall U.S. for a set of spatially-
 7    distributed PM2.5 risk related variables (e.g., PM2.5 composition, weather, demographics
 8    including SES, baseline health incidence rates). In doing so, this analysis helps inform
 9    how well the urban study areas reflect national-level variability in these key PM risk-
10    related variables.
11           The results of the representativeness analysis suggest that, in relation to three of
12    the factors considered (coverage for: more heavily populated urban areas, urban areas
13    with higher ambient PM2.5 levels, and the range of effects  estimates for short-term
14    mortality from Zannobetti  and Schwartz., 2009), the 15 urban study areas would appear
15    to capture well the overall  risk for the nation, with a potential for better characterization
16    of the higher end of the national risk distribution.  However, consideration for other
17    factors (coverage for areas with higher baseline mortality rates and with the oldest
18    populations) suggests that the 15 urban study areas may not capture areas having the
19    highest risk per microgram of PM2.5 (i.e., they may not be capturing the high-end of the
20    national risk distribution).  The limitations in coverage for high baseline mortality rates
21    and older populations, are tempered by the fact that many  locations with these types of
22    "at risk" populations do not have elevated PM2.5 levels (e.g., urban locations in Florida).

23          4.5.4  Use of the results of the national-scale long-term mortality analysis to
24                evaluate degree of coverage of the 15 urban study areas for national-
25                distribution  of risk
26           As part of the national-scale long-term mortality analysis related to PM2.5
27    exposure, which is discussed in detail in chapter 5, we developed a cumulative
28    distribution of county-lev el long-term exposure mortality risk for the U.S. (see Figure 5-
29    4).  This then allowed us to determine where along this cumulative distribution the 31
30    counties associated with the 15 urban study areas modeled in the risk assessment fell
31    (note, the location of the 31 counties along the cumulative distribution is also shown in 5-
32    4).
33           Comparison of the  31 counties (from the 15 urban  study areas) against the
34    national county-level distribution of long-term exposure mortality risk results in the
35    observation that most of the 31 counties fall towards the upper end of the national
      September 2009                         166            Draft - Do Not Quote or Cite

-------
1    distribution with 23 of the 31 counties falling within the upper 5th percentile of the
2    distribution. These observations suggest that the 15 urban study areas included in the PM
3    NAAQS risk assessment capture well the upper-end of the national distribution of long-
4    term PM2.5 exposure-related risk.
5
     September 2009                         167             Draft - Do Not Quote or Cite

-------
 1             5   NATIONAL-SCALE ASSESSMENT OF LONG-TERM
 2                       MORTALITY RELATED TO PM2 5  EXPOSURE

 3         5.1   OVERVIEW
 4           In this section we present the estimated nationwide premature mortality resulting
 5    from recent exposures to ambient PM2.5.  The goal of this assessment is twofold: (1)
 6    estimate the incidence of premature mortality within the U.S. related to long-term PM2.5
 7    exposure; and (2) identify where the subset of counties assessed in the urban case study
 8    areas analysis fall along the distribution of national county-level  risk.47 To perform this
 9    assessment we use 2005 PM2.5 fused air quality estimates from the Community Model for
10    Air Quality (CMAQ) (Byun and Schere,  2006) in conjunction with the environmental
11    Benefits Mapping and Analysis  Program (BenMAP) to estimate long-term PM2.5-related
12    premature mortality nationwide.
13           To address the first goal  of the assessment, we estimate excess PM2.s-related long-
14    term mortality by applying two estimates of all-cause mortality risk found in the Krewski
15    et al. (2009) PM2.5 mortality extended analysis of the American Cancer Society (ACS)
16    cohort, and an estimate of all-cause mortality risk found in the Laden et al. (2006) PM2.5
17    mortality extended analysis of the Six-Cities cohort. We estimate that total PM2.5-related
18    premature mortality ranges from 63,000 (39,000—87,000) (95th percentile confidence
19    interval) and 88,000 (49,000—130,000), respectively; in each case we estimated deaths
20    per year down to the lowest measured levels (LMLs) in each epidemiological study.
21           In addressing the second goal of this assessment, we observe that the subset of 31
22    counties for the 15 urban study areas considered in the urban case study fall toward the
23    upper end of the national distribution.  Specifically, all of the 31  counties were above the
24    median of the national risk distribution and 23 of the 31 fell within the upper 5th
25    percentile of the national distribution. Therefore, according to this analysis, we appear to
26    be capturing high-end percentiles of the national risk distribution with the set of urban
27    case study areas we  are evaluating in the PM2.5 NAAQS risk assessment.
28
29
30
             47 We do not directly compare the estimated county-level risks generated in the urban case study
      assessment and the county-level risks generated in the national-scale analysis. Rather, we identify where
      the 31 counties modeled for urban case study fell along the national risk distribution.  This assessment
      revealed whether the baseline PM2 5 mortality risks in the 3 Icounties modeled in the urban case study areas
      represented more typical or higher-end risk relative to the national risk distribution.
      September 2009                        168             Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
 8
10
12
14
16
18
19
20
21
22
23
24
25
26
27

28
29
30
31
      5.2   METHODS
       This assessment combines information regarding estimated PM2.5 air quality
levels, population projections, baseline mortality rates, and mortality risk coefficients to
estimate PM2.5-related premature mortality. Figure 5-1 below provides a conceptual
diagram, detailing each of the key steps involved in performing this BenMAP-based
health impact assessment.  Appendix G contains additional information regarding the
data inputs to this analysis.
Figure 5-1    Conceptual diagram of data inputs and outputs for national long-term
              mortality risk assessment
               Census Population
                   Data
                       Air Quality
                       Monitoring
                    Mortality Functions
                                            Population
                                             Estimates
                                            Population
                                             Exposure
                                        Premature Mortality
    Population
    Projections
                                                           Air Quality
                                                            Modeling
                                                          Incidence and
                                                         Prevalence Rates
      5.2.1  Population Estimates
       The starting point for estimating the size and demographics of the potentially
exposed population is the 2000 census-block level population, which BenMAP
aggregates up to the same grid resolution as the air quality model. Using county-level
growth factors based on economic projections (Woods and Poole Inc., 2001), BenMAP
projects this 2000 population to the analysis year of 2005; we selected this population
year because it matches both the year in which the emissions inventory was developed for
the air quality modeling and the year to which the baseline mortality rates were projected
(see below).
      5.2.2  Population Exposure
       Having first estimated the size and geographic distribution of the potentially
exposed population, BenMAP then matches these population projections with estimates
of the ambient levels of PM2.5. In contrast to the urban study areas analysis, the national-
      September 2009
                                        169
Draft - Do Not Quote or Cite

-------
 1   scale analysis employed a data fusion approach, which joined 2005 monitored PM2.5
 2   concentrations with 2005 CMAQ-modeled air quality levels using the Voronoi Neighbor
 3   Averaging (VNA) technique (Abt, 2003). CMAQ was run at a horizontal grid resolution
 4   of 12km for the east and 36km in the west using 2005 estimated emission levels and
 5   meteorology. More information on this model run can be found in Appendix G of this
 6   document. Figure 5-2 shows the geographic distribution of baseline annual mean PM2.5
 7   concentrations across the continental U.S. The maximum predicted value within the U.S.
 8   is 31 |ig/m3, the mean PM2.5 value is 8.7 |ig/m3, median is 8.8 |ig/m3 and the 95th
 9   percentile value is about 14 |ig/m3.
11   Figure 5-2    2005 fused surface baseline PM2.s concentrations
                                                 2005 Fused Surface Baseline Concentrations (ug/m3)
                                                 ^B 1.03 to 4.2
                                                 ^H 4.3 to 6.5
                                                     6.6 to 9.34
                                                     9.35 to 12,30
                                                 ^H 1131 to 20.57
                                                   • 20.58 to 59.42
12
13
14
15
       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
      September 2009
                                       170
Draft - Do Not Quote or Cite

-------
 1    concentrations at each grid cell. These grid-level annual average concentrations were
 2    then input to BenMAP.

 3         5.2.3  Premature Mortality Estimates
 4          In this assessment of PM2.s-related premature mortality we considered risk
 5    estimates drawn from studies based on two prospective cohorts.  The first study is the
 6    recently published Krewski et al. (2009) extended reanalysis of the ACS cohort.  To
 7    remain consistent with the urban study areas analysis, we applied the two log-linear all-
 8    cause mortality risk coefficients based on the 1979-1983 and the 1999-2000 time periods
 9    that control for 44 individual and 7 ecologic covariates. We also applied a log-linear all-
10    cause mortality risk coefficient drawn from the extended analysis of the Six Cities cohort
11    as reported by Laden et al. (2006).  When estimating premature mortality using these
12    functions we considered air quality levels down to the lowest measured levels (LML) in
13    each study;  for the Krewski et al. (2009) study this is 5.8 |ig/m3 and for the Laden et al.
14    (2006) study this is 10 |ig/m3. In general, we place a higher degree of confidence in
15    health impacts estimated at air quality levels at or above the LML because the portion of
16    the concentration-response curve below this point is extrapolated beyond the observed
17    data.  We also estimated health impacts down to Policy Relevant Background (PRB)
18    levels  (EPA, 2008). The second draft ISA presents estimates of annual mean PRB  for
19    each of 7 Health Effects Institute PM regions; this value ranges from 0.62 |ig/m3 in the
20    southwest to 1.72 |ig/m3 in the southeast.
21          BenMAP contains baseline age-, cause- and  county-specific mortality rates  drawn
22    from the CDC-WONDER. Current baseline mortality estimates are an average of a three
23    year period  from 1996-1998. EPA is in the process of updating these rates with 2006-
24    2008 data; a sensitivity analysis suggests that the results reported here are largely
25    insensitive to the use of more current mortality rates.

26         5.3   RESULTS
27          Table 5-1 and figures 5-3 through 5-4 below summarize the results of the
28    national-scale analysis. Table 5-1 summarizes the total PM2.s-related premature mortality
29    associated with modeled 2005 PM2.5 levels.
30          Estimated PM2.5-Related Premature Mortality Associated with Incremental Air
31    Quality Differences Between 2005 Ambient Mean PM2.5 Levels and LML from the
32    Epidemiology Studies or PRB (90th percentile confidence interval)
33
34
      September 2009                        171             Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
Table 5-1     Estimated PM2.s-related premature mortality associated with
              incremental air quality differences between 2005 ambient mean
              pm2.5 levels and lowest measured level from the epidemiology studies
              or policy relevant background (90th percentile confidence interval)
Air Quality
Level
10 ug/m3 (LML
for Laden et al.,
2006)
5.8 ug/m3 (LML
for Krewski et
al., 2009)
Policy-Relevant
Background
Estimates Based on Krewski et al. (2009)
'79- '83 estimate
(90th percentile
confidence interval)
26,000
(16,000—36,000)
63,000
(39,000—87,000)
110,000
(68,000—150,000)
'99- '00 estimate
(90th percentile
confidence interval)
33,000
(22,000^4,000)
80,000
(54,000—110,000)
140,000
(94,000—180,000)
Estimates Based on
Laden et al. (2006)
(90th percentile
confidence interval)
88,000
(49,000—130,000)
210,000
(120,000—300,000)
360,000
(200,000—500,000)
Bold indicates that the minimum air quality level used to calculate this estimate corresponds to the
lowest measured level identified in the epidemiological study
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
       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.48  Figure 3 illustrates the percentage of baseline mortality
attributable to PM2.5 exposure in each of the grid cells according to the 2005 PM2.5 air
quality levels, using the Krewski et al. (2009) estimate based on 1999-2000 air quality
levels.
       Note, that as stated in Section 4.3.2, modeling of risk down to PRB is subject to considerable uncertainty.
      While there is no evidence for a threshold (which conceptually supports estimation of risk below LML), we
      do not have information characterizing the nature of the C-R function for long-term mortality below the
      LML and consequently estimates of mortality based on incremental exposure below LML (and down to
      PRB) is subject to greater uncertainty.
      September 2009
                                         172
Draft - Do Not Quote or Cite

-------
 1    Figure 5-3
 3
               Percentage of premature mortality attributable to PMi.5 exposure
               at various 2005 annual average PM2.s levels*
 4
 5
 9
10
11
12
13
14
15
16
17
18
19
20
                                                                               n=i2
                                        n=15,082
                     n=JL3.JL48
                   3tolOue/m3
                                     IGto 15 ug/m3           15to 20 ug/m3
                                     2005 Average Ambient Baseline PM,5 Levels
                                                                        20ug/i7i3 and above
'Attributable mortality calculated using Krevvski 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                                       "''
        This figure illustrates the number of deaths attributable to PM2.5 according to the
 baseline level of ambient average PM2.5 levels down to 5.8 |ig/m3 (the LML for the
 Krewski et al. (2009) analysis). Each of four box plots characterizes the range of
 premature mortality attributable to PM2.5 according to the baseline level of annual mean
 PM2.5 levels in that model grid cell.  Note that while the lower whisker of the box plots
 for the baseline air quality values of 5.8 |ig/m3 to 10 |ig/m3 appear to extend to zero, the
 minimum value is greater than zero. The number above each box plot indicates the
 number of grid cells summarized by that plot.
        Figure 5-4 displays the cumulative distribution of total mortality attributable to
 PM2.5 exposure at the county level developed as part of the national-scale analysis. The
 location of the  31 counties included in the urban case study analysis is then superimposed
 on top of the cumulative distribution.
      September 2009
                                          173
Draft - Do Not Quote or Cite

-------
 1    Figure 5-4    Cumulative distribution of county-level percentage of total
 2                  mortality attributable to PMi.s for the U.S. with markers
 3                  identifying where along that distribution the urban case study area
 5                  analysis fall*
                                                                              'ere, and
                                                                 how many, counties considered
                                                                 by the Risk and Exposure
                                                                 Assessmentfallalongthe
                                                                 distribution of national baseline
                                                                 mortality risk.
                                  Percentage of mortality attributable t
                                   dll*! jl.|20M)rh)
 6
 7           Counties considered in the urban scale analysis that are located toward the lower
 8    end of the distribution of all counties nationwide include Maricopa County, Arizona and
 9    Salt Lake City, Utah.  Counties assessed in the urban scale analysis that are located
10    toward the upper end of the distribution of all counties include Jefferson County,
11    Alabama and Los Angeles County, California.  The results of this analysis indicate that
12    most of the 31 counties included in the urban case study counties fall toward the upper
13    end of the national risk distribution and that 23 of these counties fall within the upper 5th
14    percentile of the risk distribution—suggesting that the PM2.5 mortality risk estimates
15    included  in the urban case study analysis generally represent the upper end of urban area
16    mortality risks within the nation.
      September 2009
174
Draft - Do Not Quote or Cite

-------
                                        REFERENCES
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.

American Farm Bureau Federation v. EPA, 559 F. 3d 512 (D.C. Cir. 2009).

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.

Franklin, M.; Zeka, A.; Schwartz, J. (2007). Association between PM2.5 and all-cause and specific-cause
        motrality in 27 US communities. J Expo Sci Environ Epidemiol. 17:279-287.

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.
September 2009                            175              Draft - Do Not Quote or Cite

-------
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, FJ. (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.

Hopke, P. (2002). Letter from Dr. Phil Hopke, Chair, Clean Air Scientific Advisory Committee (CASAC)
        Paniculate Matter Review Panel, to Honorable Christine Todd Whitman, Administrator, U.S.
        EPA. Final advisory review report by the CASAC Paniculate 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) Paniculate 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-series 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, 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].

Krewski (2009). Letter from Dr. Daniel Krewski to HEI's 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.
September 2009                             176               Draft - Do Not Quote or Cite

-------
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.

Moolgavkar, S. (2000). Air pollution and hospital admissions for chronic obstructive pulmonary disease in
        three metropolitan areas in the United States. Inhal Toxicol. 12(Supplement 4): 75-90.

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(1):29-
        33.

Pope, C.A. Ill; Burnett, R.T. ; Thun, MJ. ; 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.

Roberts, 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, 2009. 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).

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.

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:S29-S35.

U.S. Environmental Protection Agency. (1996).  Air Quality  Criteria 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
September 2009                             177               Draft - Do Not Quote or Cite

-------
        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. (2001). Risk assessment guidance for Superfund. Vol. Ill, Part A.
        Process for conducting probabilistic risk assessment (RAGS 3 A). Washington, DC, United States
        Environmental Protection Agency, EPA 540-R-02-002; OSWER 9285.7-45; PB2002 963302;
        http://www.epa.gov/oswer/riskassessment/rags3adt/index.htm.

US Environmental Protection Agency. (2004a). Air Quality Criteria 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. (2004b) EPA's risk assessment process for air toxics: History and
        overview. In: Air toxics risk assessment reference library. Vol. 1. Technical resource manual.
        Washington, DC, United States Environmental Protection Agency, pp. 3-1 - 3-30, EPA-453-K-
        04-001A; http://www.epa.gov/ttn/fera/data/risk/vol I/chapter 03.pdf.

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 Triangle 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 jm_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, Research Triangle Park, NC. EPA-452/P-09/005. September 2009.
        Available: http://www.epa.gOv/ttn/naaqs/standards/pm/s jm_index.html.

Woods & Poole Economics Inc. (2001). Population by Single Year of Age CD. CD-ROM. Woods & Poole
        Economics, Inc. Washington, D.C.
September 2009                             178               Draft - Do Not Quote or Cite

-------
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.
September 2009                            179              Draft - Do Not Quote or Cite

-------

-------
               APPENDIX A: AIR QUALITY ASSESSMENT
                               A-l
September 2009                                            Draft - Do Not Quote or Cite

-------
1                                 Appendix A. Air Quality Assessment
2
3          This Appendix describes the PM data for the 15 urban study areas evaluated in the risk
4    assessment, including summaries of PM2.5 monitoring data associated with each study area as
5    well as the composite monitor estimates generated for each study area based on that monitoring
6    data (see section 3.2 for additional detail regarding selection of monitors and derivation of
7    composite monitor values).
                                               A-2
     September 2009                                                           Draft - Do Not Quote or Cite

-------
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
Percent! le
(ug/m3)
2005
130630091 P)
130670003 a'2'3'
1 30670004 (1'2'3)
1 30890002 (12'3)
130892001 f1'2'3'
131210032 a'2'3)
131210039 °'2'3)
13121004811'2'3'
131350002 f'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
13.30
13.28
13.15
16.83
17.39
17.17
15.72
15.10
16.00
18.35
...
14.62
15.52
16.01
16.05
15.93
21.22
18.57
18.03
18.81
20.44
19.43
17.97
...
20.39
18.62
19.86
19.85
19.95
15.92
15.62
13.98
14.56
14.83
14.38
16.56
...
15.16
12.99
14.62
14.60
14.57
16.65
16.33
15.54
15.45
15.80
15.86
16.98
...
16.13
14.63
15.95
15.95
15.90
36.09
34.94
30.28
32.82
36.72
33.40
30.29
...
31.66
34.52
35.80
35.80
35.80
2006
130630091 P)
1 30670003 °'2'3)
1 30670004 (1'2'3)
130890002 f'2'3'
130892001 °'2'3)
131210032 a'2'3)
1312100390'2'3'
13121004811'2'3'
131350002 °'3)
1 32230003 P)
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
12.30
12.30
12.19
17.91
17.88
17.75
16.09
15.75
15.99
19.15
—
18.98
15.20
16.33
16.33
16.28
21.32
21.52
21.04
19.86
18.31
19.28
20.88
15.25
20.31
18.90
19.29
19.27
19.38
14.49
14.20
12.39
13.43
12.18
13.74
...
15.00
12.93
10.77
13.26
13.27
13.17
16.67
16.46
15.82
15.41
14.54
15.37
...
—
16.86
13.95
15.30
15.29
15.26
30.84
32.66
33.34
31.65
28.89
31.44
...
—
30.64
32.28
30.62
30.62
30.74
September 2009
                                                                        A-2
Draft - Do Not Quote or Cite

-------
Table A-1 cont'd.  Air Quality Data for Atlanta
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m )
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(ug/m3)
2007
130630091 P)
1 30670003 "^
1 30670004 (1'2'3)
1 30890002 "^
130892001 (1'2'3)
1 31 21 0032 "^
1312100390'2'3'
1 31 21 0048 "^
131 350002 (1'3)
1 32230003 P)
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
12.60
12.58
12.60
16.51
17.03
17.47
15.54
17.11
17.95
—
18.97
14.03
17.12
17.23
17.34
17.22
18.83
19.49
18.77
19.38
20.04
19.64
-..
18.24
17.97
18.95
19.32
19.40
19.29
13.02
13.41
11.39
12.15
12.38
13.08
—
12.83
11.68
10.64
12.38
12.44
12.32
15.56
15.85
15.03
14.96
15.50
15.91
—
15.87
14.18
14.73
15.38
15.44
15.36
36.04
35.51
33.54
34.22
37.42
35.10
—
37.52
30.19
33.82
35.07
35.04
35.07
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 1306700031s 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
September 2009
Draft - Do Not Quote or Cite

-------
         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
16.76
11.86
12.60
12.47
12.50
14.16
13.59
14.68
13.19
13.58
20.66
18.27
20.18
20.05
20.99
20.24
19.40
20.62
19.19
12.34
13.44
11.67
13.00
14.80
14.12
13.42
12.77
13.41
14.91
15.10
15.02
15.41
17.20
16.38
16.49
16.07
15.74
33.76
35.77
33.17
35.27
39.16
37.49
39.45
36.43
35.38
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
...
13.14
11.37
11.79
11.62
11.59
13.34
12.57
12.58
...
12.28
15.73
18.51
16.24
15.19
16.88
19.27
18.64
...
18.41
11.09
13.90
11.61
12.03
12.97
14.14
14.73
...
13.74
12.55
14.25
13.17
12.86
14.50
14.82
14.94
...
14.39
32.06
34.25
32.67
32.27
35.21
36.74
35.93
...
34.89
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
—
12.33
13.54
12.95
12.83
13.20
14.68
14.03
13.66
—
13.39
15.53
16.93
16.28
15.84
16.90
17.23
16.32
—
16.42
12.04
13.70
11.16
12.44
14.79
13.23
13.35
—
13.28
13.30
14.03
13.09
13.39
14.97
14.15
14.19
—
13.85
31.46
34.01
31.55
33.31
35.25
33.77
34.39
—
33.41
         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
September 2009
Draft - Do Not Quote or Cite

-------
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
13.25
13.34
20.49
16.70
16.12
16.91
18.48
16.46
16.33
16.42
15.67
15.31
18.28
18.44
26.42
22.61
20.26
22.77
23.75
21.11
21.08
21.94
19.60
18.86
23.98
24.19
17.27
14.33
11.87
15.51
15.03
13.79
12.61
12.74
12.92
12.17
15.50
15.56
19.63
16.32
14.52
16.73
17.94
15.72
15.21
15.43
14.85
14.18
17.75
17.88
49.68
35.06
37.68
36.46
44.41
33.98
36.23
39.20
32.86
33.17
45.90
45.90
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
13.13
13.21
20.57
18.84
17.16
18.63
20.48
18.08
17.15
17.42
16.37
17.49
19.64
19.75
22.35
19.59
17.78
18.71
21.62
20.02
19.61
18.84
18.38
17.38
21.07
21.22
17.02
13.38
10.02
12.37
15.67
12.33
10.60
11.31
11.65
11.83
15.07
15.15
18.39
15.58
13.44
15.32
18.05
15.30
14.31
14.48
14.09
14.14
17.23
17.33
39.55
33.14
31.69
32.28
40.18
31.69
33.16
33.22
29.79
34.53
38.03
38.35
September 2009
                                                                   A-6
Draft - Do Not Quote or Cite

-------
Table A-3cont'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
14.47
14.52
18.69
21.38
19.29
19.16
18.99
18.27
17.81
21.23
21.40
19.31
19.18
18.53
18.41
17.83
17.52
17.72
19.97
20.06
13.63
12.42
10.93
10.40
10.38
10.84
10.95
12.40
12.45
16.54
17.10
15.25
15.03
14.75
14.90
14.61
17.02
17.11
37.92
44.02
39.92
37.90
38.56
38.52
34.91
44.24
44.30
Note 1: The monitors marked with* are used for Birmingham - 2. All monitors shown in this table are used for Birmingham -1.
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-7
September 2009
Draft - Do Not Quote or Cite

-------
Table A-4. Air Quality Data for Dallas
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m )
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percentile
(ug/m3)
2005
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
11.05
15.16
15.01
16.07
13.80
13.32
13.58
13.87
13.90
15.64
14.41
14.03
13.45
12.82
14.07
—
12.47
—
11.11
10.18
—
11.07
—
13.77
—
12.50
11.70
—
12.51
—
28.55
...
27.44
24.55
—
26.25
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
—
10.05
—
12.53
—
12.15
11.66
—
12.19
—
12.98
—
11.73
10.89
—
11.77
—
10.68
—
9.26
8.45
—
9.27
—
11.79
—
10.78
10.05
—
10.82
—
22.16
—
21.99
19.55
—
21.97
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
—
10.43
—
11.76
—
10.91
11.16
—
11.04
—
15.42
—
13.78
12.70
—
13.91
—
—
—
10.14
9.30
—
9.66
—
—
—
11.24
10.78
—
11.26
...
...
—
23.24
20.03
—
23.82
 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
September 2009
Draft - Do Not Quote or Cite

-------
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
—
18.25
13.87
14.73
14.78
14.48
11.74
16.57
14.92
14.60
—
13.86
17.15
18.73
16.62
17.43
17.45
18.22
18.58
17.66
18.20
16.58
14.38
15.18
13.70
14.20
12.68
17.90
15.19
—
14.25
13.87
15.96
17.21
16.01
16.48
14.94
18.55
16.41
—
—
15.64
42.31
48.27
47.80
51.37
39.50
48.69
46.22
—
—
43.36
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
13.61
11.89
12.26
11.58
10.39
11.23
12.85
10.95
11.10
11.71
11.85
13.68
14.93
12.58
11.78
10.01
15.56
13.69
14.34
14.20
13.69
13.65
14.56
14.97
13.46
12.70
17.30
11.94
11.98
11.84
13.67
13.22
14.68
13.04
12.71
11.86
16.13
12.92
—
13.13
13.20
32.82
35.89
35.49
35.67
30.00
42.43
32.91
—
32.32
32.74
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
13.00
10.28
13.06
12.12
11.16
10.59
15.20
11.96
12.85
12.98
10.28
14.00
15.12
14.74
14.36
13.76
16.02
14.60
15.35
14.65
13.78
14.08
14.82
14.61
13.31
14.42
17.49
13.47
14.23
13.86
14.00
12.82
14.54
13.86
13.01
12.75
16.89
13.45
14.01
13.83
12.76
31.19
32.73
33.72
31.09
32.49
36.60
28.48
33.38
33.97
31.04
 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
September 2009
Draft - Do Not Quote or Cite

-------
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
50190008
50195001
50195025
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
18.97
7.19
7.55
8.29
7.24
11.42
10.78
11.24
11.34
28.65
29.95
27.92
28.07
16.70
16.35
16.92
16.41
67.64
64.56
71.90
65.72
2006
50190008
50195001
50195025
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
21.04
9.10
9.47
9.81
9.27
12.39
12.99
13.66
12.49
23.85
24.96
26.87
24.95
16.79
16.45
17.62
16.94
50.06
53.69
57.60
49.72
2007
50190008
50195001
50195025
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
26.39
8.32
7.16
8.73
8.37
10.70
9.91
9.65
10.41
28.71
24.91
24.10
27.80
18.84
16.42
16.85
18.24
66.95
61.01
57.53
66.00
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
September 2009
Draft - Do Not Quote or Cite

-------
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
12.25
14.39
13.10
12.31
12.99
15.36
16.59
14.97
17.17
14.47
12.97
14.40
14.49
18.41
17.69
11.83
—
...
12.19
—
15.47
14.90
13.79
—
...
12.88
—
15.89
14.95
26.00
—
...
24.61
—
30.10
29.99
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
13.65
11.66
—
—
12.34
—
18.15
17.32
15.97
—
—
9.04
—
17.38
16.15
12.58
—
—
9.82
—
14.48
13.56
12.78
—
—
10.24
—
16.00
15.17
23.80
—
—
21.93
—
32.01
29.12
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
13.59
12.82
...
—
10.96
...
17.02
15.87
14.64
...
—
11.84
...
16.62
15.95
—
...
—
11.75
...
14.50
13.79
—
...
—
10.99
...
15.64
14.80
—
...
—
25.48
...
32.00
28.87
Note: The information on the composite monitors in this table is based on the composite monitors after missing values have been filled in.
                                                                      A-ll
September 2009
Draft - Do Not Quote or Cite

-------
Table A-8. Air Quality Data for Los Angeles
Monitor
Quarterly Counts
Q1
Q2
Q3
Q4
Annual
Total
Quarterly Averages (ug/m3)
Q1
Q2
Q3
Q4
Annual
Average
(ug/m3)
98th
Percent! le
(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
13.24
13.97
13.75
13.78
11.97
13.28
11.63
12.95
11.54
10.83
8.27
12.20
20.71
18.55
19.62
15.01
18.15
17.13
17.15
16.21
15.63
9.96
17.68
21.78
21.95
22.48
16.18
21.75
22.31
17.28
22.56
19.59
9.00
20.44
16.96
17.82
17.79
13.86
17.46
16.99
15.09
15.93
14.67
8.85
15.89
51.56
50.47
52.91
35.69
47.18
52.65
42.71
40.11
37.44
15.96
43.30
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
13.72
16.17
18.34
14.69
14.21
14.76
13.92
14.64
12.27
11.99
7.27
13.56
16.95
15.87
16.34
12.95
15.11
17.19
13.46
13.53
14.21
8.36
15.52
15.87
16.66
16.80
13.00
19.26
18.57
12.51
15.57
17.22
8.00
16.32
15.40
16.55
15.58
12.84
16.69
16.63
13.37
14.14
14.44
7.44
14.78
36.83
43.21
38.55
30.42
43.98
42.34
31.95
33.89
34.17
12.86
34.90
September 2009
                                                                    A-12
Draft - Do Not Quote or Cite

-------
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! le
(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
14.33
17.11
15.96
16.05
13.24
14.05
14.01
15.60
12.42
12.26
7.67
14.37
14.68
15.36
14.62
12.55
13.00
15.18
14.02
11.50
11.30
9.00
13.01
17.47
22.47
20.19
17.72
19.99
20.45
15.24
19.04
17.31
8.67
17.99
15.71
16.86
16.78
13.25
16.00
16.60
14.37
14.60
13.68
8.02
14.93
48.71
45.32
49.41
28.90
45.22
49.40
43.62
39.96
33.25
19.28
40.34
 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
September 2009
Draft - Do Not Quote or Cite

-------
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
14.46
16.89
14.78
12.21
12.17
13.74
15.51
13.84
13.12
14.11
—
10.44
12.49
10.75
11.94
14.02
18.42
16.90
15.38
17.31
19.16
18.34
17.03
18.37
—
15.21
17.81
16.17
15.44
18.11
15.68
12.71
12.30
14.13
15.17
13.54
12.56
15.21
—
10.84
12.91
10.41
12.29
14.20
16.87
13.90
13.69
15.31
17.07
15.71
14.33
16.36
—
12.38
14.53
12.48
13.53
15.80
37.50
36.05
36.58
35.94
39.93
38.96
36.18
37.66
—
34.28
33.37
33.00
35.73
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
12.67
15.09
13.17
11.06
11.15
12.49
14.03
13.00
12.08
13.07
—
10.67
12.07
10.49
11.11
13.22
13.95
13.34
14.49
14.75
14.41
13.82
13.32
14.39
—
13.68
14.06
12.60
14.33
14.00
11.88
10.33
11.40
9.00
12.59
9.86
10.59
12.64
—
10.91
10.56
8.54
11.45
11.70
13.89
12.04
12.53
12.81
14.41
12.75
12.53
13.97
—
11.61
12.24
10.41
12.39
13.50
38.89
34.80
36.51
37.06
40.60
35.73
36.92
37.84
—
33.10
35.89
31.85
35.91
33.78
September 2009
                                                                    A-14
Draft - Do Not Quote or Cite

-------
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
12.70
16.77
13.49
11.72
11.64
12.82
14.73
...
12.48
13.83
14.12
10.66
12.37
10.49
11.26
13.55
16.20
13.91
14.22
15.92
15.99
...
14.92
14.63
16.43
12.30
14.55
14.29
13.64
15.67
15.43
12.87
12.31
13.00
15.29
...
12.89
14.76
14.08
11.35
11.91
10.54
12.22
14.10
15.64
13.16
12.77
13.85
16.11
...
13.60
15.58
13.29
11.41
12.97
11.48
12.46
15.02
36.16
32.50
33.92
33.38
36.12
...
33.86
37.01
33.66
30.81
31.58
28.56
33.41
30.59
 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
September 2009
                                                                 Draft - Do Not Quote or Cite

-------
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
Percentile
(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
14.09
—
13.06
—
10.76
12.04
—
11.40
11.81
—
17.26
—
16.26
18.91
—
19.06
16.76
14.35
13.28
—
12.02
12.31
—
12.91
12.33
—
14.21
—
12.93
15.06
—
14.23
13.75
—
35.83
—
34.57
37.70
—
31.13
33.58
2006
421010003
421010004
421010020
421010024
421010047
421010057
421010136
Composite Monitor for 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
12.35
8.74
11.85
—
10.56
14.57
—
12.06
12.32
—
17.23
—
16.17
18.04
—
16.29
16.95
—
12.41
—
11.34
15.04
—
12.25
12.76
—
13.56
—
12.40
15.52
—
13.14
13.59
—
38.08
—
34.60
35.91
—
36.36
36.39
2007
421010003
421010004
421010020
421010024
421010047
421010057
421010136
Composite Monitor for 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
12.98
—
13.19
...
12.76
13.05
—
13.38
12.93
—
15.15
...
14.88
16.33
10.96
14.36
15.32
—
12.96
...
11.73
13.43
13.13
12.99
12.82
—
13.73
...
12.85
14.32
—
13.33
13.51
—
34.61
...
33.42
35.07
—
31.53
33.43
 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
September 2009
Draft - Do Not Quote or Cite

-------
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
Annual
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
9.69
10.78
8.77
13.04
8.08
8.69
10.00
11.11
8.26
10.40
7.72
7.58
9.03
18.37
9.72
16.98
9.46
13.56
13.82
12.83
...
12.84
—
9.72
10.64
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
11.38
13.58
9.52
10.34
7.98
8.66
10.03
8.07
8.92
9.31
7.14
7.46
8.28
17.82
11.33
17.58
9.12
14.04
13.92
13.41
9.66
12.69
8.08
10.22
10.90
28.51
20.07
28.38
15.35
24.29
21.92
2007
40130019
40131003
40134003
40137020
40139997
Composite Monitor for 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
8.95
8.85
10.45
11.76
7.81
8.12
9.30
8.63
9.50
11.32
7.35
8.21
9.01
15.42
11.27
15.45
8.21
12.75
13.05
10.79
9.72
12.27
7.31
9.48
10.08
26.63
18.20
27.33
13.44
22.02
19.69
 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
September 2009
Draft - Do Not Quote or Cite

-------
Table A-12. Air Quality Data for Pittsburgh
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
14.68
15.29
14.99
22.26
13.95
13.83
14.49
16.42
12.62
15.60
16.86
14.25
14.01
17.81
20.72
22.00
25.94
20.35
23.66
21.55
21.68
20.51
21.90
23.90
24.36
21.32
22.54
13.40
11.49
21.10
10.26
9.63
9.83
12.66
9.51
13.52
13.37
12.71
11.25
15.91
15.80
15.35
21.40
14.22
14.44
14.67
16.15
14.04
15.95
17.13
16.36
14.88
17.73
42.23
35.01
69.46
33.87
41.68
36.09
38.72
27.32
40.11
38.22
30.68
37.93
51.14
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
—
12.54
13.28
...
17.89
9.52
9.85
10.97
...
—
14.30
14.55
13.51
—
15.10
20.19
...
22.78
16.39
16.38
18.22
...
—
18.32
19.89
19.16
—
21.11
12.54
...
20.97
9.06
9.41
10.31
...
—
11.63
13.11
12.36
—
15.93
14.40
...
19.13
11.14
11.50
12.38
...
—
14.03
15.03
14.49
—
16.17
37.44
...
55.70
28.04
29.46
36.70
...
—
37.54
37.73
34.73
—
45.21
September 2009
                                                                      A-18
Draft - Do Not Quote or Cite

-------
Table A-12. Air Quality Data for Pittsburgh
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
—
12.36
14.72
—
18.64
13.40
10.50
9.89
—
—
14.55
15.18
13.03
—
16.25
20.30
—
25.16
19.46
19.35
20.79
—
—
19.68
21.90
21.19
—
22.33
12.74
—
17.57
10.73
12.57
12.90
—
—
13.23
15.16
13.85
—
14.49
14.89
—
18.88
13.47
13.02
13.64
—
—
15.06
16.56
15.11
—
16.36
39.35
—
54.67
40.80
32.56
32.40
—
—
39.60
43.57
34.74
—
47.20
 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
September 2009
Draft - Do Not Quote or Cite

-------
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
—
14.77
6.58
9.59
5.47
6.27
7.40
6.03
—
7.43
8.98
12.68
8.61
9.56
10.57
7.76
—
10.47
14.49
17.24
11.35
14.17
16.36
7.45
—
14.93
11.06
14.06
9.32
10.99
11.99
7.79
—
11.90
41.66
43.36
36.25
43.23
39.37
26.61
—
41.98
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
—
10.54
6.98
11.22
5.65
7.21
7.18
6.85
—
8.60
9.41
14.19
8.65
8.54
11.56
9.26
—
10.80
13.58
14.91
9.29
12.37
13.61
7.09
—
12.72
10.18
13.03
7.88
9.68
10.61
7.33
—
10.67
38.67
37.93
27.72
37.54
35.69
21.97
—
36.89
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
—
18.77
6.97
11.45
6.44
6.11
7.17
6.06
7.68
8.06
10.99
—
10.08
9.42
11.53
9.66
11.62
10.72
13.89
—
9.71
12.05
13.42
7.09
13.00
12.35
12.49
—
9.41
11.44
12.46
8.21
—
12.48
55.65
—
29.84
54.28
50.13
23.02
—
50.64
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
September 2009
Draft - Do Not Quote or Cite

-------
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
...
16.24
16.44
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
...
14.64
14.59
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
...
19.17
18.95
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
...
12.97
13.10
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
...
15.76
15.77
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
...
39.59
39.83
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
...
12.77
12.97
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
...
12.21
12.27
19.40
—
20.18
13.49
16.47
15.72
15.20
15.46
...
13.87
15.89
15.26
15.48
16.17
...
16.20
16.28
12.11
—
14.05
12.92
10.87
14.48
12.00
11.49
...
11.00
12.51
12.68
10.90
13.18
...
12.56
12.82
16.02
—
16.32
13.08
13.11
14.53
13.38
12.59
...
11.73
13.12
13.36
12.47
13.65
...
13.44
13.58
32.81
—
36.24
27.28
27.54
29.18
27.92
30.20
...
27.61
29.39
28.52
30.46
29.60
...
28.60
28.87
September 2009
                                                                       A-21
Draft - Do Not Quote or Cite

-------
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
Percentile
(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
—
12.42
12.58
—
—
15.31
16.02
14.84
17.65
13.95
14.44
—
12.96
14.50
13.79
14.50
14.43
—
14.15
14.13
—
15.07
17.61
15.66
17.39
15.94
14.83
16.23
—
15.25
16.13
16.09
—
16.61
17.26
16.22
16.14
—
14.94
13.23
13.51
12.32
13.79
10.90
12.13
—
12.49
12.97
13.30
—
13.10
13.82
12.96
13.09
—
—
15.11
14.88
14.24
15.58
13.26
13.68
—
13.09
14.04
13.94
—
14.34
—
13.94
13.98
—
—
35.86
34.98
34.45
33.08
32.27
31.92
—
30.28
31.61
32.06
—
33.72
—
31.94
31.90
 Note 1: The monitors marked with * are used for St Louis - 2. All
 Note 2:  The information on the composite monitors in this table
monitors shown in the table are used for St Louis -1.
is based on the composite monitors after missing values have been filled in.
                                                                         A-22
September 2009
                                                                       Draft - Do Not Quote or Cite

-------
Table A-15. Air Quality Data for Tacoma
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
530530029
Composite Monitor for Tacoma
29
90
30
91
30
92
31
92
120
365
16.46
16.01
5.34
5.38
7.13
7.07
17.07
17.21
11.50
11.42
40.42
39.61
2006
530530029
Composite Monitor for Tacoma
30
90
30
91
31
92
26
92
117
365
8.92
8.91
5.89
5.94
7.45
7.40
15.93
15.13
9.55
9.35
39.82
37.05
2007
530530029
Composite Monitor for Tacoma
29
90
28
91
31
92
29
92
117
365
13.76
14.74
5.94
6.00
5.23
5.25
13.76
13.47
9.67
9.87
45.11
44.07
 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
September 2009
Draft - Do Not Quote or Cite

-------
       APPENDIX B: HYBRID (NON-PROPORTIONAL) ROLLBACK
                             APPROACH
                                B-l
September 2009                                            Draft - Do Not Quote or Cite

-------
 1    Appendix B.        Methodology for Rolling Back PM2.5 Concentrations Due to Local
 2                        Source Impacts (hybrid non-proportional approach)
 O
 4
 5           During the last review of the Particulate Matter National Ambient Air Quality Standards
 6    (NAAQS), a technique was employed to simulate fine paniculate concentrations under a series
 7    of attainment scenarios to determine the risk associated with each.  The "rolling back" of the
 8    concentrations consisted of simply using a proportional rollback calculation where every
 9    measured concentration value was multiplied by a constant to obtain a set of concentrations
10    which would meet alternative standard levels. This technique was reviewed by the Clean Air
11    Scientific Advisory Committee (CASAC) and was considered to be a satisfactory way to
12    simulating alternative PM2.5 distributions.  The rolled  back values, however, only constituted a
13    regional reduction in PM concentrations without accounting for in any way emission reductions
14    at local point sources.
15           For the current review, an alternative rollback approach reflecting the combined effect of
16    both local  and regional reduction strategies was considered (this alternative approach is referred
17    to as the hybrid non-proportional approach in the risk  assessment).  In addition to utilizing a
18    traditional proportional rollback to represent the regional PM reductions, a distance weighted
19    rollback was conducted on a subset of the 15 study areas which contain source oriented monitors
20    measuring concentrations higher than those observed at other sites within a particular area. *
21           Unique sites with high design values exceeding the NAAQS were further investigated to
22    determine  if they were in close proximity to a large source of PM2.5 (Figure 1). The presence of
23    possible source oriented  sites in each  area was visually determined using satellite photographs
24    provided by Google Earth.  Areas where source oriented adjustments were made include Detroit
25    MI, Pittsburgh PA, St. Louis MO-IL,  Baltimore MD, New York NY, Los Angeles CA and
26    Birmingham AL.
27
             1 In the risk assessment, as outlined in Section 3.1, the proportional rollback approach was used in
      generating the core risk estimates, while the hybrid non-proportional approach described here, was considered as
      part of the sensitivity analysis.

                                                 B-2
      September 2009                                                           Draft - Do Not Quote or Cite

-------
                       Detroit,  Ml (261630033)
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
                                                              Dearborn monitor
                                                              (marked by blue
                                                              circle) is located
                                                              adjacent to a large
                                                              rail yard and
                                                              Ford River
                                                              Rouge Plant
                                                              (encircled in red)
      Figure B-l.  Example of a PM2.5 source oriented monitoring site
      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
2).
                                             B-2
     September 2009
                                                                  Draft - Do Not Quote or Cite

-------
                                                         Drtnt. Ml
                                             U        «         IS        IS
                                                                                            1 Concentration
                                                                                              values ofthe site
                                                                                              in the blue circle
                                                                                              get reduced
                                                                                              so that eiher
                                                                                              the 24-hour or
                                                                                              annual average
                                                                                              design value
                                                                                              matches the
                                                                                              design value
                                                                                              ofthe sie in
                                                                                              the red circle.

                                                                                              The reduction
                                                                                              amount is
                                                                                              dependent on
                                                                                              the difference
                                                                                              between the
                                                                                              two sites' 24-hour
                                                                                              and annual design
                                                                                              values and
                                                                                              concentrations are
                                                                                              adjusted by the
                                                                                              larger of the two
                                                                                              values.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15









Figure B-2



Plot of the 24-hour versus the annual average PM2.5 design values for
individual sites in Detroit MI




The fractional reduction made to the site near
inverse distance in


the point source
kilometers between the source oriented site and all


was then weighted by the
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.
ofthe source oriented site received the same amount
Sites within one kilometer
of reduction as the source oriented site. An
example ofthe effect of this reduction technique for Detroit is presented in Table I. For Detroit,
adjustments were based on the difference between the two sites' annual design values.

Table B-l.

Site ID


260490021
260990009
261150005
261250001



Comparison ofthe original and adjusted design values for Detroit, MI
Original
Annual Design
Value (2005-
2007)
11.6
12.5
13.8
13.6
Adjusted
Annual Design
Value (2005-
2007)
11.5
12.4
13.7
13.5
Original
24-hour Design
Value (2005-
2007)
29
35
38
40
Adjusted
24-hour Design
Value (2005-
2007)
29
35
38
40
September 2009
                                                               B-4
Draft - Do Not Quote or Cite

-------
Site ID
261470005
261610005
261610008
261630001
261630015
261630016
261630019
261630025
261630033
261630036
261630038
261630039
Original
Annual Design
Value (2005-
2007)
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)
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)
41
39
39
36
40
41
40
34
43
36
40
37
Adjusted
24-hour Design
Value (2005-
2007)
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
 1
 2
 3           Reduction of the concentrations of the source oriented site reduced either the 24-hour or
 4    annual design value of the site to either the maximum non-source oriented site's 24-hour or
 5    annual design value. This did not necessarily mean that the adjusted values at the source
 6    oriented site met either the 24-hour or annual standard after the reduction.  Since, the adjusted
 7    design values were calculated using the same data handling rules as contained within 40 CFR
 8    Part 50 Appendix N, truncation or rounding of the adjusted concentrations could sometimes give
 9    adjusted design values at the source oriented site that were not exactly the same value as the
10    original design value at the reference site. However, they were usually within 1 ug/m3 for the
11    24-hour standard and a few tenths of a microgram per cubic meter for the annual standard.
                                                 B-5
      September 2009
Draft - Do Not Quote or Cite

-------
              APPENDIX C: EPIDEMIOLOGY STUDY-SPECIFIC
                INFORMATION FOR PM RISK ASSESSMENT
                               C-l
September 2009                                          Draft - Do Not Quote or Cite

-------
1            Appendix C. Epidemiology Study-Specific Information for PM2.s Risk Assessment
2
3          This Appendix provides detailed summary information for the epidemiological studies
4   used to obtain the concentration-response (C-R) functions used in the risk assessment. For
5   additional details on selection of epidemiological studies and specification of the C-R functions,
6   see section 3.3.3.
                                               C-2
     September 2009                                                           Draft - Do Not Quote or Cite

-------
Table C-l. 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 Long-Term Exposure to PM2S:
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
September 2009
                                                                           C-3
Draft - Do Not Quote or Cite

-------
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
September 2009
                                                                            C-4
Draft - Do Not Quote or Cite

-------
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
September 2009
                                                                            C-5
Draft - Do Not Quote or Cite

-------
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
September 2009
                                                                            C-6
Draft - Do Not Quote or Cite

-------
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
September 2009
                                                                           C-7
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
     APPENDIX D: SUPPLEMENT TO THE REPRESENTATIVENESS
             ANALYSIS OF THE 15 URBAN STUDY AREAS
                              D-l
September 2009                                         Draft - Do Not Quote or Cite

-------
 1   Appendix D.  Supplement to the Representativeness of the 15 Urban Study Areas
 2                 (additional graphical comparisons of distributions for key contributors to
 3                 PM2.5 risk)
 4
 5          Following the analysis discussed in Section 4.4, this appendix provides graphical
 6   comparisons of the empirical distributions of components of the risk function, and additional
 7   variables that have been identified as potentially influencing the risk associated with PM
 8   exposures.
 9                 In each graph, the orange line represents the empirical cumulative distribution
10   function (CDF) for the complete set of data available for the variable.  In some cases,  this may
11   encompass all counties in the U.S., while in others, it may be based on a subset of the U.S.,
12   usually for large urban areas. The green line in each graph represents the empirical cumulative
13   distribution function for the variable based only on the data available for the set of urban case
14   study locations.  The black squares at the bottom of each graph represents the specific value of
15   the variable for one of the case study locations, with the line showing where that value intersects
16   the two empirical CDFs.
                                                D-2
     September 2009                                                           Draft - Do Not Quote or Cite

-------
       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
September 2009
                                           D-3
                                               Draft - Do Not Quote or Cite

-------
Figure D-2.   Comparison of Distributions for Key Elements of the Risk Equation:
             Percent of Population Under 15 Years of Age
Comparison of Urban Case Study Area % Under 1 5 to U.S. Distribution of % Under

100% -i
90% -
80% -
£ 70% -
1 60% -
0
O
2 40% -
o
£ 30% -
20% -
10% -
0% -
1









^^^^^^M
0

15
(3141 U.S. Counties)








^^







•-I—I
25



30
% of Population Under Age 15, 2005
All Counties CDF ^— Case Study Counties • Case Study Counties









35

September 2009
                                        D-4
Draft - Do Not Quote or Cite

-------
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
September 2009
                                            D-5
                                                                   Draft - Do Not Quote or Cite

-------
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

September 2009
                                         D-6
Draft - Do Not Quote or Cite

-------
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

September 2009
                                         D-7
Draft - Do Not Quote or Cite

-------
Figure D-6.  Comparison of Distributions for Key Elements of the Risk Equation: 98
             %ile Daily Average PM2.5
                                                                                 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


September 2009
                                         D-8
Draft - Do Not Quote or Cite

-------
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

September 2009
                                         D-9
Draft - Do Not Quote or Cite

-------
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

September 2009
                                         D-10
Draft - Do Not Quote or Cite

-------
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
September 2009
                                            D-ll
Draft - Do Not Quote or Cite

-------
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)
          50
1 UU YO
90% -
80% -
8 70%-
1 60% -
o
ri 50%
2 40% -
0
£ 30% -
20% -
1 0% -
no/, ,






	 --"






r<
























































x






x^







Urban case study areas are
all below the 90th percentile
of county cardiovascular
mortality




150
250     350     450     550     650     750
    Cardiovascular Mortality per 100,000 Population
850
950
                    All Counties CDF	Case Study Counties CDF  •  Case Study Counties
September 2009
                                            D-12
                                                       Draft - Do Not Quote or Cite

-------
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







September 2009
                                         D-13
Draft - Do Not Quote or Cite

-------
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
September 2009
Draft - Do Not Quote or Cite

-------
       Figure D-13.  Comparison of Distributions for Key Elements of the Risk Equation:

                     Cardiovascular Mortality Risk Effect Estimate from Zanobetti and

Schwartz            (2008)
   c
   (0
  .0
  CO
  08
  N
100%


 90% -


 80% -


 70% -


 60% -


 50% -


 40%


 30% -


 20% -


 10% :


  0%
             Comparison of Urban Case Study PM Cardiovascular Mortality Risk (p) to U.S.
                         Distribution of PM Cardiovascular Mortality Risk
                                   (212 U.S. Urban Areas)

           0.0002  0.0004  0.0006  0.0008  0.001   0.0012  0.0014  0.0016  0.0018

               Bayesian Shrunken PM Cardiovascular Mortality Risk Coefficient (p)
                                                                                   0.002
         All Z&S Urban Areas CDF ^—Case Study Urban Area CDF  •  Case Study Urban Areas
                                           D-15
September 2009
                                                                  Draft - Do Not Quote or Cite

-------
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
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
September 2009
Draft - Do Not Quote or Cite

-------
D.2. Variables Expected to Influence the Relative Risk from PM2.5


              D.2.1. Demographic Variables


       Figure D-15. Comparison of Distributions for Selected Variables Expected to
Influence the        Relative Risk from PM2.5: Population Density
           Comparison of Urban Case Study Area Population Density with U.S. Distribution of
                              Population Density (all U.S. Counties)
              Urban case study areas are
              all above the 65th Percentile
              of county population density
                         10             100            1000
                                Population Density (persons per sq mile)
10000
100000
                   All Counties CDF ^—Case Study Counties CDF ^^Case Study Counties
September 2009
                                           D-17
   Draft - Do Not Quote or Cite

-------
       Figure D-16. Comparison of Distributions for Selected Variables Expected to
                    Influence the Relative Risk from PM2.5:  Unemployment Rate
           Comparison of Urban Case Study Area Unemployment Rate to U.S. Distribution of
                                   Unemployment Rates
                                   (3141 U.S. Counties)
     100%
                                        8      10     12
                                      Unemployment Rate (%)
14
16
18
20
                   All Counties CDF	Case Study Counties CDF  •  Case Study Counties
September 2009
                                         D-18
       Draft - Do Not Quote or Cite

-------
Figure D-17. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  % 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







September 2009
                                         D-19
Draft - Do Not Quote or Cite

-------
Figure D-18.  Comparison of Distributions for Selected Variables Expected to Influence the
              Relative Risk from PM2.5:  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
September 2009
                    Draft - Do Not Quote or Cite

-------
Figure D-19. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  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 0.2 0.4 0.6 0.8 1



All
% Air Conditioning Prevalence
Urban Areas CDF ^— Case Study Urban Areas CDF • Case Study Urban Areas
                                         D-21
September 2009
Draft - Do Not Quote or Cite

-------
Figure D-20. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  % 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
September 2009
Draft - Do Not Quote or Cite

-------
D.2.2.  Health Conditions

       Figure D-21.  Comparison of Distributions for Selected Variables Expected to
                    Influence the Relative Risk from PM2.5: 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 PM2.5: Asthma Prevalence



Comparison of Urban Case Study Area Current Asthma Prevalence to U.S.
Distribution of Current Asthma Prevalence
(183 U.S. MSA)



1 nn°/
I UU 70
90%
80%
< 70%
CO

5 60%
CO

£ 50%
a:

2 40%
o

^ 30%


20%
1 0%

















	 <:
f)0/ U 	 i












'

i i










i



i_i 1 1














1 1 i_i














i 	 i

x^
s











i i
^













i i
i
\J /O ^^^^ ^^^r^ ™ i 	 ^-^^^^^ ^— ^^ ^ | ^ |
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
September 2009
Draft - Do Not Quote or Cite

-------
Figure D-23. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  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
September 2009
Draft - Do Not Quote or Cite

-------
Figure D-24. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  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
September 2009
Draft - Do Not Quote or Cite

-------
Figure D-25.  Comparison of Distributions for Selected Variables Expected to Influence the
              Relative Risk from PM2.5: 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
September 2009
Draft - Do Not Quote or Cite

-------
Figure D-26. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  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
September 2009
Draft - Do Not Quote or Cite

-------
Figure D-27. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  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
September 2009
Draft - Do Not Quote or Cite

-------
Figure D-28. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  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
September 2009
Draft - Do Not Quote or Cite

-------
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
September 2009
                                          D-31
                                     Draft - Do Not Quote or Cite

-------
Figure D-30. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5:  % Mobile Source Direct PM2.5 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
£ 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







September 2009
                                         D-32
Draft - Do Not Quote or Cite

-------
       Figure D-31. Comparison of Distributions for Selected Variables Expected to
                    Influence the Relative Risk from PM2.5: 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
September 2009
                                         D-33
                                            Draft - Do Not Quote or Cite

-------
Figure D-32. Comparison of Distributions for Selected Variables Expected to Influence the
             Relative Risk from PM2.5: 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

September 2009
                                        D-34
Draft - Do Not Quote or Cite

-------
           APPENDIX E: RISK ESTIMATES (CORE ANALYSIS)
                                 E-l
September 2009                                             Draft - Do Not Quote or Cite

-------
 1                                Appendix E. Risk Analysis (core analysis)
 2
 3           This Appendix provides detailed risk estimates generated for the core analysis for the 15
 4   urban study areas. The tables cover all of the air quality scenarios modeled, including recent
 5   conditions, the current standard, and alternative standard levels. For additional detail on the types
 6   of risk metrics (and figures summarizing key metrics) presented in this Appendix, see section
 7   4.0.
 8           We have identified an error in the approach used to simulate ambient PM2.5 levels for the
 9   Pittsburgh study area for the scenarios involving just meeting the current and alternative sets of
10   standards. Consequently, this error impacts the risk estimates generated for the current and
11   alternative standards but does not impact the risk estimates for the recent conditions scenario.
12   There was insufficient time  after identifying this error to either generate corrected risk estimates
13   or remove the erroneous risk estimates from the summary tables presented in this Appendix. We
14   will correct this error and release updated results for the Pittsburgh study area as soon as is
15   practicable and will include the corrected results in the next version of this document. Note, that
16   this error does not impact risk estimates for any of the other urban study areas.
                                                 E-2
      September 2009                                                            Draft - Do Not Quote or Cite

-------
Table E-l. 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 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of All-Cause Mortality Associated with Long-Term Exposure to PM2
PM2.s Concentrations that Just Meet the Current and Alternative Annual (n
Combination Denoted n/m):
Recent PM25
Concentrations
645
(419-869)
592
(384 - 796)
494
(321 - 663)
366
(237 - 494)
754
(490-1015)
249
(161 -334)
721
(468-971)
2443
(1585-3286)
1725
(1118-2326)
490
(318-661)
476
(308 - 643)
698
(454 - 938)
123
(80-166)
791
(513-1064)
121
(79-164)
15/352
572
(371 - 770)
543
(352 - 732)
352
(228 - 474)
366
(237 - 494)
544
(352 - 733)
86
(55-116)
663
(430 - 893)
1116
(721 -1507)
1247
(807-1684)
427
(276 - 576)
476
(308 - 643)
285
(184-386)
42
(27 - 57)
669
(433-901)
79
(51 -107)
13/35
449
(290 - 605)
438
(284 - 590)
275
(178-370)
366
(237 - 494)
469
(303 - 633)
86
(55-116)
517
(335 - 697)
1116
(721 -1507)
1186
(767-1601)
379
(245-512)
476
(308 - 643)
285
(184-386)
42
(27 - 57)
530
(343-715)
79
(51 -107)
12/35
386
(250 - 522)
378
(245-510)
236
(152-318)
324
(210-437)
400
(259 - 540)
86
(55-116)
443
(287 - 599)
985
(636-1332)
998
(645-1349)
324
(209 - 437)
427
(276 - 577)
285
(184-386)
42
(27 - 57)
454
(294-614)
79
(51 -107)
5 Concentrations in a Recent Year and
and Daily (m) Standards (Standard
13/30
449
(290 - 605)
422
(273 - 569)
275
(178-370)
366
(237 - 494)
411
(266 - 555)
56
(36 - 76)
517
(335 - 697)
779
(503-1053)
903
(583-1220)
320
(207 - 432)
414
(267 - 559)
201
(130-272)
22
(14-29)
515
(334 - 695)
52
(34 - 70)
12/25
377
(244 - 509)
299
(194-404)
199
(129-269)
324
(210-437)
277
(179-375)
26
(17-35)
443
(287 - 599)
440
(284 - 596)
556
(359 - 752)
213
(137-287)
257
(166-348)
117
(75-158)
1
(1-1)
361
(233 - 487)
25
(16-34)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-3
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
624
(405 - 840)
515
(334 - 693)
478
(310-642)
281
(181 -379)
573
(371 - 773)
264
(172-355)
762
(494-1026)
2191
(1421 -2951)
1489
(964 - 2009)
481
(311 -648)
520
(336 - 703)
605
(393-814)
102
(66-138)
613
(397 - 826)
79
(51 -106)
15/352
551
(357 - 743)
470
(305 - 634)
339
(220 - 457)
281
(181 -379)
396
(256 - 534)
94
(60-127)
701
(455 - 945)
950
(614-1285)
1047
(677-1415)
418
(270 - 563)
520
(336 - 703)
231
(149-312)
27
(18-37)
508
(329 - 686)
44
(28 - 59)
13/35
430
(278 - 580)
374
(242 - 504)
263
(170-355)
281
(181 -379)
332
(215-449)
94
(60-127)
548
(355 - 739)
950
(614-1285)
991
(640-1339)
371
(240-501)
520
(336 - 703)
231
(149-312)
27
(18-37)
389
(252 - 526)
44
(28 - 59)
12/35
368
(238 - 497)
319
(206-431)
225
(145-304)
244
(158-330)
275
(177-371)
94
(60-127)
471
(304 - 635)
829
(535-1121)
818
(528-1106)
316
(204 - 426)
468
(303 - 633)
231
(149-312)
27
(18-37)
325
(210-439)
44
(28 - 59)
13/30
430
(278 - 580)
359
(232 - 485)
263
(170-355)
281
(181 -379)
284
(183-384)
63
(40 - 85)
548
(355 - 739)
636
(410-861)
729
(471 - 987)
312
(202 - 422)
454
(294-614)
155
(100-210)
8
(5-11)
377
(244-510)
22
(14-29)
12/25
359
(232 - 485)
247
(159-333)
189
(122-256)
244
(158-330)
171
(110-232)
31
(20 - 42)
471
(304 - 635)
320
(206 - 433)
409
(264 - 554)
206
(133-279)
288
(186-389)
78
(51 -106)
0
(0-0)
245
(158-331)
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).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-4
September 2009
Draft - Do Not Quote or Cite

-------
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 PM;
PM2.s Concentrations that Just Meet the Current and Alternative Annual (n
Combination Denoted n/m):
Recent PM25
Concentrations
646
(419-869)
482
(312-649)
473
(307 - 636)
311
(201 - 420)
536
(347 - 723)
299
(194-402)
748
(485-1007)
2240
(1453-3016)
1514
(980 - 2043)
476
(308-641)
451
(292-610)
613
(398 - 825)
142
(92-192)
653
(423 - 880)
91
(59-123)
15/352
571
(370 - 769)
439
(284 - 592)
335
(217-451)
311
(201 - 420)
366
(236 - 494)
112
(73 - 1 52)
687
(445 - 926)
979
(632-1323)
1067
(690-1442)
413
(267 - 557)
451
(292-610)
236
(153-319)
52
(34-71)
545
(353 - 735)
53
(34 - 72)
13/35
445
(288 - 600)
346
(224 - 467)
259
(168-350)
311
(201 - 420)
305
(197-412)
112
(73-152)
535
(346-721)
979
(632-1323)
1010
(652 - 1 364)
367
(237 - 495)
451
(292-610)
236
(153-319)
52
(34-71)
421
(272 - 569)
53
(34 - 72)
12/35
382
(247-516)
294
(190-397)
221
(143-299)
272
(176-367)
250
(161 -337)
112
(73-152)
458
(296-618)
855
(552-1156)
834
(539-1128)
312
(202-421)
402
(260 - 543)
236
(153-319)
52
(34-71)
354
(229 - 479)
53
(34 - 72)
5 Concentrations in a Recent Year and
and Daily (m) Standards (Standard
13/30
445
(288 - 600)
332
(215-449)
259
(168-350)
311
(201 - 420)
258
(167-349)
78
(51 -106)
535
(346-721)
659
(425 - 892)
745
(481 -1008)
308
(199-417)
388
(251 - 525)
160
(103-216)
29
(19-40)
408
(264-551)
30
(19-40)
12/25
372
(241 - 503)
224
(145-303)
186
(120-251)
272
(176-367)
150
(97 - 204)
44
(28 - 60)
458
(296-618)
338
(218-457)
421
(272 - 570)
203
(131 -274)
230
(148-311)
83
(53-112)
6
(4-9)
271
(175-366)
6
(4-8)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-5
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM25
Concentrations
4.3%
(2.8% - 5.8%)
4.2%
(2.7% - 5.6%)
5%
(3.3% - 6.8%)
2.9%
(1.8% -3. 8%)
4.2%
(2.7% - 5.7%)
4.5%
(2.9% - 6%)
3.9%
(2.5% - 5.2%)
4.3%
(2.8% - 5.8%)
3.3%
(2.1% -4.4%)
3.4%
(2.2% - 4.5%)
2.1%
(1.3% -2. 8%)
5%
(3.3% - 6.7%)
2.6%
(1.7% -3.5%)
4.2%
(2.7% - 5.6%)
2.4%
(1.5% -3.2%)
15/352
3.8%
(2.5% -5.1%)
3.8%
(2.5% - 5.2%)
3.6%
(2.3% - 4.8%)
2.9%
(1.8% -3. 8%)
3%
(2% -4.1%)
1 .5%
(1%-2.1%)
3.6%
(2.3% - 4.8%)
2%
(1.3% -2.7%)
2.4%
(1.5% -3.2%)
2.9%
(1.9% -4%)
2.1%
(1.3% -2. 8%)
2.1%
(1.3% -2. 8%)
0.9%
(0.6% - 1 .2%)
3.6%
(2.3% -4.8%)
1 .6%
(1%-2.1%)
13/35
3%
(1.9% -4%)
3.1%
(2% - 4.2%)
2.8%
(1.8% -3. 8%)
2.9%
(1.8% -3.8%)
2.6%
(1.7% -3. 5%)
1.5%
(1%-2.1%)
2.8%
(1.8% -3.7%)
2%
(1.3% -2.7%)
2.3%
(1.5% -3%)
2.6%
(1.7% -3.5%)
2.1%
(1.3% -2. 8%)
2.1%
(1.3% -2. 8%)
0.9%
(0.6% - 1 .2%)
2.8%
(1.8% -3.8%)
1 .6%
(1%-2.1%)
12/35
2.6%
(1.7% -3.5%)
2.7%
(1.7% -3.6%)
2.4%
(1.6% -3.2%)
2.5%
(1.6% -3. 4%)
2.2%
(1.5% -3%)
1.5%
(1%-2.1%)
2.4%
(1.5% -3. 2%)
1 .7%
(1.1% -2.3%)
1.9%
(1.2% -2.6%)
2.2%
(1.4% -3%)
1.9%
(1.2% -2. 5%)
2.1%
(1.3% -2. 8%)
0.9%
(0.6% -1.2%)
2.4%
(1.6% -3.3%)
1 .6%
(1%-2.1%)
13/30
3%
(1.9% -4%)
3%
(1.9% -4%)
2.8%
(1.8% -3. 8%)
2.9%
(1.8% -3. 8%)
2.3%
(1.5% -3.1%)
1%
(0.6% - 1 .4%)
2.8%
(1.8% -3. 7%)
1.4%
(0.9% -1.9%)
1 .7%
(1.1% -2.3%)
2.2%
(1.4% -3%)
1.8%
(1.2% -2. 4%)
1 .4%
(0.9% - 2%)
0.5%
(0.3% - 0.6%)
2.7%
(1.8% -3.7%)
1%
(0.7% - 1 .4%)
12/25
2.5%
(1.6% -3.4%)
2.1%
(1.4% -2.9%)
2%
(1.3% -2. 7%)
2.5%
(1.6% -3. 4%)
1 .6%
(1%-2.1%)
0.5%
(0.3% - 0.6%)
2.4%
(1.5% -3. 2%)
0.8%
(0.5% -1%)
1.1%
(0.7% -1.4%)
1.5%
(0.9% - 2%)
1.1%
(0.7% - 1 .5%)
0.8%
(0.5% -1.1%)
0%
(0% - 0%)
1.9%
(1.2% -2.6%)
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., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
September 2009
Draft - Do Not Quote or Cite

-------
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 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 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):
Recent PM25
Concentrations
4%
(2.6% - 5.4%)
3.6%
(2.4% - 4.9%)
4.8%
(3.1% -6. 5%)
2.1%
(1.4% -2. 9%)
3.2%
(2.1% -4.3%)
4.7%
(3% - 6.3%)
4%
(2.6% - 5.3%)
3.8%
(2.5% - 5.2%)
2.8%
(1.8% -3.8%)
3.3%
(2.1% -4. 5%)
2.2%
(1.4% -2. 9%)
4.4%
(2.8% - 5.9%)
2.1%
(1.3% -2. 8%)
3.2%
(2.1% -4.4%)
1.5%
(1%-2.1%)
15/352
3.5%
(2.3% - 4.8%)
3.3%
(2.2% - 4.5%)
3.4%
(2.2% - 4.6%)
2.1%
(1.4% -2. 9%)
2.2%
(1.4% -3%)
1.7%
(1.1% -2. 2%)
3.6%
(2.4% - 4.9%)
1.7%
(1.1% -2.2%)
2%
(1.3% -2.7%)
2.9%
(1.9% -3. 9%)
2.2%
(1.4% -2. 9%)
1 .7%
(1.1% -2. 3%)
0.6%
(0.4% - 0.8%)
2.7%
(1.7% -3.6%)
0.8%
(0.5% -1.1%)
13/35
2.8%
(1.8% -3.7%)
2.6%
(1.7% -3.6%)
2.6%
(1.7% -3. 6%)
2.1%
(1.4% -2. 9%)
1.9%
(1.2% -2.5%)
1 .7%
(1.1% -2.2%)
2.8%
(1.8% -3. 8%)
1 .7%
(1.1% -2.2%)
1.9%
(1.2% -2.5%)
2.6%
(1.7% -3.4%)
2.2%
(1.4% -2. 9%)
1.7%
(1.1% -2. 3%)
0.6%
(0.4% - 0.8%)
2.1%
(1.3% -2.8%)
0.8%
(0.5% -1.1%)
12/35
2.4%
(1.5% -3.2%)
2.3%
(1.5% -3%)
2.3%
(1.5% -3.1%)
1.9%
(1.2% -2. 5%)
1 .5%
(1%-2.1%)
1 .7%
(1.1% -2. 2%)
2.4%
(1.6% -3. 3%)
1.5%
(0.9% -2%)
1.5%
(1%-2.1%)
2.2%
(1.4% -2. 9%)
2%
(1.3% -2.6%)
1 .7%
(1.1% -2. 3%)
0.6%
(0.4% - 0.8%)
1 .7%
(1.1% -2. 3%)
0.8%
(0.5% -1.1%)
13/30
2.8%
(1.8% -3.7%)
2.5%
(1.6% -3.4%)
2.6%
(1.7% -3.6%)
2.1%
(1.4% -2. 9%)
1 .6%
(1%-2.2%)
1.1%
(0.7% -1.5%)
2.8%
(1.8% -3. 8%)
1.1%
(0.7% -1.5%)
1.4%
(0.9% -1.9%)
2.1%
(1.4% -2. 9%)
1.9%
(1.2% -2. 6%)
1.1%
(0.7% -1.5%)
0.2%
(0.1% -0.2%)
2%
(1.3% -2.7%)
0.4%
(0.3% - 0.6%)
12/25
2.3%
(1.5% -3.1%)
1 .7%
(1.1% -2.4%)
1.9%
(1.2% -2. 6%)
1.9%
(1.2% -2. 5%)
1%
(0.6% -1.3%)
0.6%
(0.4% - 0.8%)
2.4%
(1.6% -3. 3%)
0.6%
(0.4% - 0.8%)
0.8%
(0.5%-1%)
1 .4%
(0.9% -1.9%)
1.2%
(0.8% -1.6%)
0.6%
(0.4% - 0.8%)
0%
(0% - 0%)
1.3%
(0.8% -1.8%)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-7
September 2009
Draft - Do Not Quote or Cite

-------
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 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
Percent of Total Incidence of All-Cause 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):
Recent PM25
Concentrations
4%
(2.6% - 5.4%)
3.4%
(2.2% - 4.6%)
4.7%
(3.1% -6.3%)
2.3%
(1.5% -3.1%)
3%
(2% -4.1%)
5.2%
(3.4% - 7%)
3.8%
(2. 5% -5.1%)
3.9%
(2.5% - 5.2%)
2.8%
(1.8% -3.8%)
3.3%
(2.1% -4.4%)
1.8%
(1.2% -2. 5%)
4.5%
(2.9% - 6%)
2.8%
(1.8% -3. 8%)
3.4%
(2.2% - 4.6%)
1 .7%
(1.1% -2.3%)
15/352
3.6%
(2.3% - 4.8%)
3.1%
(2% - 4.2%)
3.3%
(2.2% - 4.5%)
2.3%
(1.5% -3.1%)
2.1%
(1.3% -2.8%)
2%
(1.3% -2. 6%)
3.5%
(2.3% - 4.7%)
1 .7%
(1.1% -2.3%)
2%
(1.3% -2.7%)
2.8%
(1.8% -3. 8%)
1.8%
(1.2% -2. 5%)
1 .7%
(1.1% -2. 3%)
1%
(0.7% - 1 .4%)
2.9%
(1.9% -3.9%)
1%
(0.7% - 1 .4%)
13/35
2.8%
(1.8% -3.8%)
2.5%
(1.6% -3.3%)
2.6%
(1.7% -3. 5%)
2.3%
(1.5% -3.1%)
1.7%
(1.1% -2.3%)
2%
(1.3% -2. 6%)
2.7%
(1.8% -3. 7%)
1.7%
(1.1% -2.3%)
1.9%
(1.2% -2.5%)
2.5%
(1.6% -3. 4%)
1.8%
(1.2% -2. 5%)
1.7%
(1.1% -2. 3%)
1%
(0.7% - 1 .4%)
2.2%
(1.4% -3%)
1%
(0.7% - 1 .4%)
12/35
2.4%
(1.5% -3.2%)
2.1%
(1.3% -2.8%)
2.2%
(1.4% -3%)
2%
(1.3% -2. 8%)
1 .4%
(0.9% -1.9%)
2%
(1.3% -2.6%)
2.3%
(1.5% -3.1%)
1.5%
(1%-2%)
1 .6%
(1%-2.1%)
2.1%
(1.4% -2.9%)
1 .6%
(1.1% -2.2%)
1 .7%
(1.1% -2. 3%)
1%
(0.7% - 1 .4%)
1.9%
(1.2% -2.5%)
1%
(0.7% - 1 .4%)
13/30
2.8%
(1.8% -3.8%)
2.4%
(1.5% -3.2%)
2.6%
(1.7% -3. 5%)
2.3%
(1.5% -3.1%)
1.5%
(0.9% - 2%)
1 .4%
(0.9% -1.8%)
2.7%
(1.8% -3. 7%)
1.1%
(0.7% - 1 .6%)
1 .4%
(0.9% -1.9%)
2.1%
(1.4% -2. 9%)
1 .6%
(1%-2.1%)
1 .2%
(0.7% - 1 .6%)
0.6%
(0.4% - 0.8%)
2.2%
(1.4% -2.9%)
0.6%
(0.4% - 0.8%)
12/25
2.3%
(1.5% -3.2%)
1.6%
(1%-2.1%)
1.9%
(1.2% -2. 5%)
2%
(1.3% -2. 8%)
0.9%
(0.5% -1.2%)
0.8%
(0.5% -1%)
2.3%
(1.5% -3.1%)
0.6%
(0.4% - 0.8%)
0.8%
(0.5% -1.1%)
1.4%
(0.9% -1.9%)
0.9%
(0.6% -1.3%)
0.6%
(0.4% - 0.8%)
0.1%
(0.1% -0.2%)
1 .4%
(0.9% -1.9%)
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., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-S
September 2009
Draft - Do Not Quote or Cite

-------
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.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):
Recent PM2 5
Concentrations
-13%
(-13% --13%)
-9%
(-9% - -9%)
-40%
(-40% - -41 %)
0%
(0% - 0%)
-39%
(-38% - -39%)
-191%
(-189% --192%)
-9%
(-9% - -9%)
-1 1 9%
(-11 8% --120%)
-38%
(-38% - -39%)
-15%
(-15% --15%)
0%
(0% - 0%)
-145%
(-143% --146%)
-193%
(-192% --194%)
-18%
(-18% --18%)
-54%
(-54% - -54%)
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%)
13/35
22%
(21 % - 22%)
19%
(19% -20%)
22%
(22% - 22%)
0%
(0% - 0%)
14%
(14% -14%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
21%
(21 % - 21 %)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
30%
(30% - 31 %)
33%
(33% - 33%)
11%
(11% -12%)
26%
(26% - 27%)
0%
(0% - 0%)
33%
(33% - 33%)
12%
(12% -12%)
20%
(20% - 20%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
32%
(32% - 32%)
0%
(0% - 0%)
13/30
22%
(21 % - 22%)
22%
(22% - 22%)
22%
(22% - 22%)
0%
(0% - 0%)
24%
(24% - 25%)
35%
(35% - 35%)
22%
(22% - 22%)
30%
(30% - 30%)
28%
(28% - 28%)
25%
(25% - 25%)
13%
(13% -13%)
29%
(29% - 30%)
49%
(49% - 49%)
23%
(23% - 23%)
34%
(34% - 34%)
12/25
34%
(34% - 34%)
45%
(45% - 45%)
43%
(43% - 44%)
11%
(11% -12%)
49%
(49% - 49%)
69%
(69% - 70%)
33%
(33% - 33%)
61%
(60% -61%)
55%
(55% - 56%)
50%
(50% - 50%)
46%
(46% - 46%)
59%
(59% - 59%)
98%
(98% - 98%)
46%
(46% - 46%)
68%
(68% - 68%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009). Percents are rounded to the nearest
whole number.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-9
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2.5 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
-13%
(-13% --13%)
-9%
(-9% - -9%)
-41 %
(-41% --41%)
0%
(0% - 0%)
-45%
(-45% - -45%)
-182%
(-181% --184%)
-9%
(-9% - -9%)
-131%
(-130% --131%)
-42%
(-42% - -42%)
-15%
(-15% --15%)
0%
(0% - 0%)
-162%
(-161% --163%)
-274%
(-273% - -275%)
-21 %
(-20% - -21 %)
-80%
(-80% - -80%)
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%)
13/35
22%
(22% - 22%)
21%
(20% -21%)
22%
(22% - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
33%
(33% - 33%)
32%
(32% - 32%)
34%
(33% - 34%)
13%
(13% -13%)
31%
(31 % - 31 %)
0%
(0% - 0%)
33%
(33% - 33%)
13%
(13% -13%)
22%
(22% - 22%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
36%
(36% - 36%)
0%
(0% - 0%)
13/30
22%
(22% - 22%)
24%
(24% - 24%)
22%
(22% - 22%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
22%
(22% - 22%)
33%
(33% - 33%)
30%
(30% - 30%)
25%
(25% - 25%)
13%
(13% -13%)
33%
(33% - 33%)
69%
(69% - 69%)
26%
(26% - 26%)
50%
(50% - 50%)
12/25
35%
(35% - 35%)
48%
(47% - 48%)
44%
(44% - 44%)
13%
(13% -13%)
57%
(57% - 57%)
67%
(66% - 67%)
33%
(33% - 33%)
66%
(66% - 66%)
61%
(61% -61%)
51%
(50% -51%)
45%
(45% - 45%)
66%
(66% - 66%)
100%
(100% -100%)
52%
(52% - 52%)
100%
(100% -100%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009). Percents are rounded to the nearest
whole number.
 The current primary PM2 5 standards include an annual standard set at 15 ug/m and a daily standard set at 35 ug/m .
                                                                      E-10
September 2009
Draft - Do Not Quote or Cite

-------
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 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):
Recent PM2 5
Concentrations
-13%
(-13% --13%)
-10%
(-10% --10%)
-41 %
(-41% --42%)
0%
(0% - 0%)
-47%
(-46% - -47%)
-166%
(-165% --168%)
-9%
(-9% - -9%)
-129%
(-128% --130%)
-42%
(-42% - -42%)
-15%
(-15% --15%)
0%
(0% - 0%)
-160%
(-158% --161%)
-173%
(-172% --174%)
-20%
(-20% - -20%)
-71 %
(-70% - -71 %)
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%)
13/35
22%
(22% - 22%)
21%
(21 % - 21 %)
23%
(22% - 23%)
0%
(0% - 0%)
17%
(17% -17%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
33%
(33% - 33%)
33%
(33% - 33%)
34%
(34% - 34%)
12%
(12% -13%)
32%
(32% - 32%)
0%
(0% - 0%)
33%
(33% - 33%)
13%
(13% -13%)
22%
(22% - 22%)
25%
(24% - 25%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
13/30
22%
(22% - 22%)
24%
(24% - 24%)
23%
(22% - 23%)
0%
(0% - 0%)
29%
(29% - 29%)
30%
(30% - 30%)
22%
(22% - 22%)
33%
(33% - 33%)
30%
(30% - 30%)
25%
(25% - 25%)
14%
(14% -14%)
32%
(32% - 33%)
44%
(44% - 44%)
25%
(25% - 25%)
44%
(44% - 44%)
12/25
35%
(35% - 35%)
49%
(49% - 49%)
44%
(44% - 45%)
12%
(12% -13%)
59%
(59% - 59%)
61%
(61% -61%)
33%
(33% - 33%)
66%
(65% - 66%)
61%
(60% -61%)
51%
(51% -51%)
49%
(49% - 49%)
65%
(65% - 65%)
88%
(88% - 88%)
50%
(50% - 50%)
89%
(89% - 89%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009). Percents are rounded to the nearest
whole number.
 The current primary PM2 5 standards include an annual standard set at 15 ug/m and a daily standard set at 35 ug/m .
                                                                      E-ll
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-l. Estimated Percent Reductions From the Current Standard to Alternative
Standards in All Cause Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983): Based on 2005 Air Quality Data*
       ra
       •c
       I
       OT
       3
       O
       O
       O
       I
       •c
       01
       a:
       +•»
       C
       ai
       Si
       ai
       a.
 -80%
-100%
-120%
-140%
-160%
-180%
-200%
                   2005 air     15/35**
                   quality
                               13/35
12/35
13/30
12/25
                                      Alternative Standard
                         -»- Atlanta, GA 572 (371 -770); 3.8% (2.5%-5.1%)
                         -•- Baltimore, MD  543 (352 - 732); 3.8% (2.5% - 5.2%)
                         -*- Birmingham, AL 352  (228-474);  3.6% (2.3%-4.8%)
                             Dallas, TX 366 (237 - 494); 2.9% (1.8% - 3.8%)
                         -*- Detroit, Ml 544 (352 - 733); 3% (2% - 4.1 %)
                         -+- Fresno, CA 86 (55-116); 1.5% (1%-2.1%)
                         -i- Houston, TX 663 (430 - 893);  3.6% (2.3% - 4.8%)
                         	 Los Angeles, CA 1116 (721 -1507); 2%  (1.3%-2.7%)
                             New York, NY 1247 (807-1684);  2.4% (1.5%-3.2%)
                         -•- Philadelphia, PA 427 (276-576); 2.9% (1.9%-4%)
                         -•- Phoenix, AZ 476 (308 - 643);  2.1 % (1.3% - 2.8%)
                         -*- Pittsburgh, PA  285 (184-386); 2.1% (1.3%-2.8%)
                             Salt Lake City,  UT 42 (27 - 57); 0.9% (0.6% -1.2%)
                             St. Louis, MO 669 (433-901); 3.6% (2.3%-4.8%)
                         -•- Tacoma.WA 79 (51 -107); 1.6% (1%-2.1%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                               E-12
September 2009
                                                                     Draft - Do Not Quote or Cite

-------
Figure E-2. Estimated Percent Reductions From the Current Standard to Alternative
Standards in All Cause Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983): Based on 2006 Air Quality Data*
           100%
            80%
        S
        V)
        'c
        S>
        S  -20%
        O
        I
        •s
        1
        a:
 -40%
 -60%
 -80%
-100%
-120%
-140% -
-160%
          -180%
          -200%
                   2006 air     15/35"
                    quality
                              13/35
12/35
13/30
12/25
                                      Alternative Standard
                         -•- Atlanta, GA 551 (357 - 743); 3.5% (2.3% - 4.8%)
                         -•- Baltimore, MD  470 (305 - 634); 3.3% (2.2% - 4.5%)
                         -*- Birmingham, AL 339 (220-457); 3.4% (2.2%-4.6%)
                             Dallas, TX 281 (181 -379); 2.1% (1.4%-2.9%)
                         -*- Detroit, Ml 396 (256-534); 2.2% (1.4%-3%)
                         -•- Fresno, CA 94 (60-127);  1.7% (1.1%-2.2%)
                         -i- Houston, TX 701  (455 - 945); 3.6% (2.4% - 4.9%)
                         — Los Angeles, CA  950 (614-1285);  1.7% (1.1%-2.2%)
                             New York, NY 1047 (677-1415); 2% (1.3%-2.7%)
                         -»- Philadelphia, PA 418 (270-563); 2.9%  (1.9%-3.9%)
                         -•- Phoenix, AZ 520  (336 - 703); 2.2% (1.4% - 2.9%)
                         -+- Pittsburgh, PA  231 (149-312); 1.7% (1.1%-2.3%)
                             Salt Lake City, UT 27 (18-37); 0.6% (0.4%-0.8%)
                             St. Louis, MO 508 (329 - 686); 2.7% (1.7% - 3.6%)
                         -•- Tacoma.WA 44  (28-59); 0.8% (0.5%-1.1%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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 reduction for 2006 air quality in Salt Lake City is -274%.
                                               E-13
September 2009
                                                                     Draft - Do Not Quote or Cite

-------
Figure E-3. Estimated Percent Reductions From the Current Standard to Alternative
Standards in All Cause Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983): Based on 2007 Air Quality Data*
       c
       S
       OT
          100%
           80%
           60%
           40%
 20%
       ^   -20%
       O
       p
 -40%
       I
       s  ,
       01
       o
       8.
 100%
-120%
-140%
-160%
-180%
-200%
 -60%
 -80%
                  2007 air
                   quality
                   15/35**     13/35       12/35

                            Alternative Standard
13/30
12/25
                        -•- Atlanta, GA 571 (370 - 769); 3.6% (2.3% - 4.8%)
                        -m- Baltimore, MD 439 (284-592); 3.1% (2%-4.2%)
                        -*- Birmingham, AL 335 (217-451);  3.3% (2.2%-4.5%)
                            Dallas, TX 311 (201 -420); 2.3% (1.5%-3.1%)
                        -*- Detroit, Ml 366 (236-494); 2.1% (1.3%-2.8%)
                        -•- Fresno, CA 112 (73-152); 2% (1.3%-2.6%)
                        -i- Houston, TX 687 (445 - 926); 3.5% (2.3% - 4.7%)
                        	 Los Angeles, CA 979 (632-1323); 1.7% (1.1%-2.3%)
                            New York, NY 1067 (690-1442);  2% (1.3%-2.7%)
                        -•- Philadelphia, PA 413 (267-557); 2.8%  (1.8%-3.8%)
                        -m- Phoenix, AZ 451  (292 - 610); 1.8% (1.2% - 2.5%)
                        -*- Pittsburgh, PA 236 (153-319); 1.7% (1.1%-2.3%)
                            Salt Lake City, UT  52 (34 - 71); 1 % (0.7% -1.4%)
                            St. Louis, MO 545 (353 - 735); 2.9% (1.9% - 3.9%)
                            Tacoma.WA  53 (34-72); 1% (0.7%-1.4%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                               E-14
September 2009
                                                                     Draft - Do Not Quote or Cite

-------
Table E-10.  Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM25 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):
Recent PM2.5
Concentrations
826
(528-1117)
757
(484-1024)
631
(404 - 852)
469
(299 - 637)
965
(617-1306)
318
(204 - 430)
923
(590-1250)
3125
(1999-4227)
2211
(1411 -2998)
628
(401 - 852)
611
(389-831)
892
(572-1205)
158
(101 -214)
1012
(647-1369)
156
(99-212)
15/352
732
(468 - 992)
696
(444 - 942)
451
(288-611)
469
(299 - 637)
697
(445 - 946)
110
(70-150)
849
(542-1151)
1432
(911 -1948)
1600
(1019-2174)
547
(349 - 743)
611
(389-831)
366
(233 - 498)
54
(34 - 74)
857
(547-1161)
101
(64-138)
13/35
575
(367 - 780)
561
(358-761)
352
(224 - 478)
469
(299 - 637)
601
(383-817)
110
(70-150)
663
(422 - 900)
1432
(911 -1948)
1521
(969 - 2068)
487
(310-661)
611
(389-831)
366
(233 - 498)
54
(34 - 74)
679
(433 - 922)
101
(64-138)
12/35
496
(316-673)
485
(309 - 659)
302
(193-411)
416
(265 - 565)
514
(327 - 698)
110
(70-150)
569
(362 - 773)
1265
(804-1722)
1282
(815-1744)
415
(264 - 564)
548
(349 - 746)
366
(233 - 498)
54
(34 - 74)
583
(371 - 792)
101
(64-138)
13/30
575
(367 - 780)
541
(345 - 734)
352
(224 - 478)
469
(299 - 637)
528
(336-717)
72
(46 - 98)
663
(422 - 900)
1001
(636-1363)
1159
(737-1578)
411
(261 - 558)
531
(338 - 723)
259
(164-352)
28
(18-38)
661
(421 - 897)
67
(42-91)
12/25
484
(308 - 657)
384
(245 - 522)
256
(163-348)
416
(265 - 565)
356
(226 - 485)
34
(21 - 46)
569
(362 - 773)
566
(359 - 772)
714
(454 - 974)
273
(174-372)
331
(210-451)
150
(95 - 205)
1
(1-2)
463
(295 - 630)
32
(21 - 44)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-15
September 2009
Draft - Do Not Quote or Cite

-------
Table E-ll.  Estimated Annual Incidence of All-Cause Mortality Associated with Long-Term Exposure to Ambient PM2 5 Concentrations
         in a Recent Year (2006) and PM2.5 Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting
          2006 PM25 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM25from 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):
Recent PM2.5
Concentrations
798
(510-1081)
659
(421 - 893)
611
(391 - 825)
360
(229 - 490)
735
(469 - 996)
338
(216-457)
976
(624-1321)
2805
(1793-3799)
1910
(1217-2593)
616
(393 - 835)
668
(425 - 908)
774
(496-1047)
131
(83-178)
785
(501 -1065)
101
(64-138)
15/352
706
(451 - 957)
603
(385-817)
434
(277 - 589)
360
(229 - 490)
508
(323 - 690)
120
(76-164)
898
(574-1217)
1221
(776-1662)
1345
(855-1829)
536
(341 - 727)
668
(425 - 908)
297
(189-404)
35
(22 - 48)
652
(415-885)
56
(36 - 77)
13/35
551
(351 - 748)
479
(305-651)
337
(215-458)
360
(229 - 490)
427
(271 -581)
120
(76-164)
702
(448 - 953)
1221
(776-1662)
1272
(809-1731)
476
(303 - 646)
668
(425 - 908)
297
(189-404)
35
(22 - 48)
500
(318-680)
56
(36 - 77)
12/35
473
(301 - 642)
409
(261 - 557)
289
(184-392)
314
(199-427)
353
(224 - 480)
120
(76-164)
604
(384 - 820)
1064
(676-1450)
1050
(668-1430)
405
(258-551)
601
(382-818)
297
(189-404)
35
(22 - 48)
417
(265 - 568)
56
(36 - 77)
13/30
551
(351 - 748)
460
(293 - 626)
337
(215-458)
360
(229 - 490)
364
(232 - 496)
80
(51 - 110)
702
(448 - 953)
817
(519- 1114)
937
(595-1277)
401
(255 - 545)
583
(371 -793)
199
(126-272)
11
(7-15)
484
(308 - 659)
28
(18-38)
12/25
461
(293 - 626)
317
(201 -431)
243
(155-331)
314
(199-427)
220
(140-300)
40
(26 - 55)
604
(384 - 820)
411
(261 -561)
526
(334-718)
265
(168-361)
370
(235 - 504)
101
(64-138)
0
(0-0)
315
(200 - 429)
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).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-16
September 2009
Draft - Do Not Quote or Cite

-------
Table E-12.  Estimated Annual Incidence of All-Cause Mortality Associated with Long-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 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):
Recent PM2.5
Concentrations
826
(528-1118)
617
(394 - 837)
605
(387-817)
399
(254 - 542)
687
(438 - 932)
382
(245-516)
957
(612-1297)
2867
(1833-3883)
1942
(1238-2636)
609
(389 - 826)
580
(369 - 789)
784
(502-1061)
183
(116-248)
837
(534-1134)
117
(74-159)
15/352
731
(467 - 990)
563
(359 - 764)
429
(274-581)
399
(254 - 542)
469
(299 - 638)
144
(92- 196)
880
(562-1193)
1257
(799- 1711)
1370
(871 - 1863)
530
(338-719)
580
(369 - 789)
303
(193-413)
67
(43-91)
698
(445 - 948)
69
(44 - 94)
13/35
571
(364 - 775)
444
(283 - 604)
332
(212-452)
399
(254 - 542)
392
(249 - 533)
144
(92-196)
686
(437-931)
1257
(799-1711)
1296
(824-1764)
470
(300 - 639)
580
(369 - 789)
303
(193-413)
67
(43-91)
540
(344 - 735)
69
(44 - 94)
12/35
490
(312-666)
377
(240-513)
284
(181 -386)
349
(222 - 475)
321
(204 - 437)
144
(92- 196)
588
(374 - 799)
1098
(698- 1496)
1072
(681 - 1459)
400
(255 - 544)
516
(328 - 703)
303
(193-413)
67
(43-91)
455
(289-619)
69
(44 - 94)
13/30
571
(364 - 775)
426
(271 - 579)
332
(212-452)
399
(254 - 542)
332
(211 -452)
101
(64-137)
686
(437-931)
847
(538-1155)
957
(608-1304)
396
(252 - 538)
499
(317-679)
205
(130-280)
38
(24 - 52)
524
(334-713)
38
(24 - 52)
12/25
478
(304 - 649)
288
(183-393)
239
(152-325)
349
(222 - 475)
193
(123-264)
57
(36 - 78)
588
(374 - 799)
434
(275 - 593)
541
(343 - 739)
261
(166-355)
296
(188-403)
106
(67-145)
8
(5-11)
348
(221 - 474)
8
(5-10)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-17
September 2009
Draft - Do Not Quote or Cite

-------
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 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 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 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):
Recent PM25
Concentrations
5.5%
(3.5% - 7.4%)
5.4%
(3.4% - 7.3%)
6.4%
(4.1% -8.7%)
3.7%
(2.3% - 5%)
5.4%
(3.5% - 7.3%)
5.7%
(3.7% - 7.7%)
5%
(3.2% - 6.7%)
5.5%
(3.5% - 7.4%)
4.2%
(2.7% - 5.7%)
4.3%
(2.8% - 5.8%)
2.6%
(1.7% -3.6%)
6.4%
(4.1% -8.7%)
3.3%
(2.1% -4. 5%)
5.4%
(3.4% - 7.3%)
3.1%
(1.9% -4.2%)
15/352
4.9%
(3.1% -6. 6%)
4.9%
(3.1% -6. 7%)
4.6%
(2.9% - 6.2%)
3.7%
(2.3% - 5%)
3.9%
(2.5% - 5.3%)
2%
(1.3% -2. 7%)
4.6%
(2.9% - 6.2%)
2.5%
(1.6% -3.4%)
3%
(1.9% -4.1%)
3.8%
(2. 4% -5.1%)
2.6%
(1.7% -3.6%)
2.6%
(1.7% -3.6%)
1.1%
(0.7% - 1 .6%)
4.5%
(2.9% - 6.2%)
2%
(1.3% -2.7%)
13/35
3.8%
(2.4% - 5.2%)
4%
(2.5% - 5.4%)
3.6%
(2.3% - 4.9%)
3.7%
(2.3% - 5%)
3.4%
(2.1% -4.6%)
2%
(1.3% -2.7%)
3.6%
(2.3% - 4.8%)
2.5%
(1.6% -3.4%)
2.9%
(1.8% -3. 9%)
3.3%
(2.1% -4. 5%)
2.6%
(1.7% -3.6%)
2.6%
(1.7% -3.6%)
1.1%
(0.7% - 1 .6%)
3.6%
(2.3% - 4.9%)
2%
(1.3% -2.7%)
12/35
3.3%
(2.1% -4. 5%)
3.4%
(2.2% - 4.7%)
3.1%
(2% - 4.2%)
3.2%
(2.1% -4.4%)
2.9%
(1.8% -3.9%)
2%
(1.3% -2. 7%)
3.1%
(1.9% -4.1%)
2.2%
(1.4% -3%)
2.4%
(1.5% -3. 3%)
2.8%
(1.8% -3. 9%)
2.4%
(1.5% -3.2%)
2.6%
(1.7% -3.6%)
1.1%
(0.7% -1.6%)
3.1%
(2% - 4.2%)
2%
(1.3% -2.7%)
13/30
3.8%
(2.4% - 5.2%)
3.8%
(2.4% - 5.2%)
3.6%
(2.3% - 4.9%)
3.7%
(2.3% - 5%)
3%
(1.9% -4%)
1.3%
(0.8% -1.8%)
3.6%
(2.3% - 4.8%)
1.8%
(1.1% -2.4%)
2.2%
(1.4% -3%)
2.8%
(1.8% -3. 8%)
2.3%
(1.5% -3.1%)
1 .9%
(1.2% -2.5%)
0.6%
(0.4% - 0.8%)
3.5%
(2.2% - 4.8%)
1 .3%
(0.8% -1.8%)
12/25
3.2%
(2% - 4.4%)
2.7%
(1.7% -3. 7%)
2.6%
(1.7% -3. 5%)
3.2%
(2.1% -4.4%)
2%
(1.3% -2.7%)
0.6%
(0.4% - 0.8%)
3.1%
(1.9% -4.1%)
1%
(0.6% - 1 .4%)
1.4%
(0.9% -1.8%)
1.9%
(1.2% -2. 6%)
1.4%
(0.9% - 2%)
1.1%
(0.7% -1.5%)
0%
(0% - 0%)
2.5%
(1.6% -3. 3%)
0.6%
(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., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-18
September 2009
Draft - Do Not Quote or Cite

-------
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 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 All-Cause 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):
Recent PM25
Concentrations
5.1%
(3.3% - 7%)
4.7%
(3% - 6.3%)
6.1%
(3.9% - 8.3%)
2.7%
(1.7% -3.7%)
4.1%
(2.6% - 5.6%)
6%
(3.8% -8.1%)
5.1%
(3.2% - 6.9%)
4.9%
(3.1% -6.7%)
3.6%
(2.3% - 4.9%)
4.2%
(2.7% - 5.7%)
2.8%
(1.8% -3.8%)
5.6%
(3.6% - 7.6%)
2.7%
(1.7% -3. 6%)
4.1%
(2.6% - 5.6%)
2%
(1.2% -2.7%)
15/352
4.5%
(2.9% - 6.2%)
4.3%
(2.7% - 5.8%)
4.4%
(2.8% - 5.9%)
2.7%
(1.7% -3.7%)
2.8%
(1.8% -3.9%)
2.1%
(1.4% -2. 9%)
4.7%
(3% - 6.3%)
2.1%
(1.4% -2.9%)
2.5%
(1.6% -3. 4%)
3.7%
(2.3% - 5%)
2.8%
(1.8% -3.8%)
2.1%
(1.4% -2.9%)
0.7%
(0.5%-1%)
3.4%
(2.2% - 4.7%)
1.1%
(0.7% -1.5%)
13/35
3.5%
(2.3% - 4.8%)
3.4%
(2.2% - 4.6%)
3.4%
(2.2% - 4.6%)
2.7%
(1.7% -3.7%)
2.4%
(1.5% -3.3%)
2.1%
(1.4% -2. 9%)
3.6%
(2. 3% -4. 9%)
2.1%
(1.4% -2.9%)
2.4%
(1.5% -3. 3%)
3.3%
(2.1% -4. 4%)
2.8%
(1.8% -3.8%)
2.1%
(1.4% -2.9%)
0.7%
(0.5%-1%)
2.6%
(1.7% -3. 6%)
1.1%
(0.7% - 1 .5%)
12/35
3%
(1.9% -4.1%)
2.9%
(1.8% -3. 9%)
2.9%
(1.8% -3. 9%)
2.4%
(1.5% -3.3%)
2%
(1.3% -2.7%)
2.1%
(1.4% -2. 9%)
3.1%
(2% - 4.3%)
1.9%
(1.2% -2.5%)
2%
(1.3% -2. 7%)
2.8%
(1.8% -3. 8%)
2.5%
(1.6% -3.4%)
2.1%
(1.4% -2.9%)
0.7%
(0.5%-1%)
2.2%
(1.4% -3%)
1.1%
(0.7% -1.5%)
13/30
3.5%
(2.3% - 4.8%)
3.3%
(2.1% -4. 4%)
3.4%
(2.2% - 4.6%)
2.7%
(1.7% -3.7%)
2%
(1.3% -2. 8%)
1 .4%
(0.9% -1.9%)
3.6%
(2.3% - 4.9%)
1 .4%
(0.9% - 2%)
1.8%
(1.1% -2.4%)
2.8%
(1.8% -3. 7%)
2.4%
(1.5% -3.3%)
1 .4%
(0.9% - 2%)
0.2%
(0.1% -0.3%)
2.6%
(1.6% -3. 5%)
0.5%
(0.3% - 0.7%)
12/25
3%
(1.9% -4%)
2.2%
(1.4% -3%)
2.4%
(1.6% -3. 3%)
2.4%
(1.5% -3.3%)
1 .2%
(0.8% -1.7%)
0.7%
(0.5%-1%)
3.1%
(2% - 4.3%)
0.7%
(0.5%-1%)
1%
(0.6% -1.3%)
1.8%
(1.2% -2. 5%)
1.5%
(1%-2.1%)
0.7%
(0.5% -1%)
0%
(0% - 0%)
1 .7%
(1.1% -2. 3%)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-19
September 2009
Draft - Do Not Quote or Cite

-------
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 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 All-Cause 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):
Recent PM2 5
Concentrations
5.2%
(3. 3% -7%)
4.4%
(2.8% - 5.9%)
6%
(3.9% - 8.2%)
3%
(1.9% -4.1%)
3.9%
(2.5% - 5.3%)
6.7%
(4.3% - 9%)
4.9%
(3.1% -6. 6%)
5%
(3.2% - 6.8%)
3.6%
(2.3% - 4.9%)
4.2%
(2.7% - 5.7%)
2.3%
(1.5% -3.2%)
5.7%
(3.6% - 7.7%)
3.6%
(2.3% - 4.9%)
4.4%
(2.8% - 6%)
2.2%
(1.4% -3%)
15/352
4.6%
(2.9% - 6.2%)
4%
(2.5% - 5.4%)
4.3%
(2.7% - 5.8%)
3%
(1.9% -4.1%)
2.7%
(1.7% -3.6%)
2.5%
(1.6% -3. 4%)
4.5%
(2. 9% -6.1%)
2.2%
(1.4% -3%)
2.6%
(1.6% -3. 5%)
3.6%
(2.3% - 4.9%)
2.3%
(1.5% -3.2%)
2.2%
(1.4% -3%)
1 .3%
(0.8% -1.8%)
3.7%
(2.3% - 5%)
1 .3%
(0.8% -1.8%)
13/35
3.6%
(2.3% - 4.9%)
3.1%
(2% - 4.3%)
3.3%
(2.1% -4. 5%)
3%
(1.9% -4.1%)
2.2%
(1.4% -3%)
2.5%
(1.6% -3. 4%)
3.5%
(2.2% - 4.7%)
2.2%
(1.4% -3%)
2.4%
(1.5% -3. 3%)
3.2%
(2.1% -4. 4%)
2.3%
(1.5% -3.2%)
2.2%
(1.4% -3%)
1.3%
(0.8% -1.8%)
2.9%
(1.8% -3. 9%)
1.3%
(0.8% -1.8%)
12/35
3.1%
(2% - 4.2%)
2.7%
(1.7% -3.6%)
2.8%
(1.8% -3. 9%)
2.6%
(1.7% -3.6%)
1.8%
(1.2% -2.5%)
2.5%
(1.6% -3. 4%)
3%
(1.9% -4.1%)
1.9%
(1.2% -2.6%)
2%
(1.3% -2.7%)
2.8%
(1.8% -3.7%)
2.1%
(1.3% -2.8%)
2.2%
(1.4% -3%)
1.3%
(0.8% -1.8%)
2.4%
(1.5% -3. 3%)
1.3%
(0.8% -1.8%)
13/30
3.6%
(2.3% - 4.9%)
3%
(1.9% -4.1%)
3.3%
(2.1% -4. 5%)
3%
(1.9% -4.1%)
1.9%
(1.2% -2.6%)
1.8%
(1.1% -2. 4%)
3.5%
(2.2% - 4.7%)
1.5%
(0.9% -2%)
1.8%
(1.1% -2. 4%)
2.7%
(1.7% -3. 7%)
2%
(1.3% -2.8%)
1.5%
(0.9% - 2%)
0.7%
(0.5%-1%)
2.8%
(1.8% -3. 8%)
0.7%
(0.5% -1%)
12/25
3%
(1.9% -4.1%)
2%
(1.3% -2. 8%)
2.4%
(1.5% -3.2%)
2.6%
(1.7% -3.6%)
1.1%
(0.7% -1.5%)
1%
(0.6% - 1 .4%)
3%
(1.9% -4.1%)
0.8%
(0.5% -1%)
1%
(0.6% -1.4%)
1.8%
(1.1% -2.4%)
1.2%
(0.8% -1.6%)
0.8%
(0.5% -1.1%)
0.2%
(0.1% -0.2%)
1.8%
(1.2% -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., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-20
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2 5 Concentrations that Just Meet the Current and
Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
-13%
(-13% --13%)
-9%
(-9% - -9%)
-40%
(-39% - -40%)
0%
(0% - 0%)
-38%
(-38% - -39%)
-189%
(-187% --191%)
-9%
(-9% - -9%)
-118%
(-11 7% --11 9%)
-38%
(-38% - -38%)
-15%
(-15% --15%)
0%
(0% - 0%)
-144%
(-142% --145%)
-192%
(-191% --193%)
-18%
(-18% --18%)
-54%
(-53% - -54%)
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%)
13/35
21%
(21% -22%)
19%
(19% -19%)
22%
(22% - 22%)
0%
(0% - 0%)
14%
(14% -14%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
21%
(21% -21%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
30%
(30% - 30%)
33%
(33% - 33%)
11%
(11% -11%)
26%
(26% - 26%)
0%
(0% - 0%)
33%
(33% - 33%)
12%
(12% -12%)
20%
(20% - 20%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
32%
(32% - 32%)
0%
(0% - 0%)
13/30
21%
(21% -22%)
22%
(22% - 22%)
22%
(22% - 22%)
0%
(0% - 0%)
24%
(24% - 24%)
35%
(35% - 35%)
22%
(22% - 22%)
30%
(30% - 30%)
28%
(27% - 28%)
25%
(25% - 25%)
13%
(13% -13%)
29%
(29% - 29%)
49%
(49% - 49%)
23%
(23% - 23%)
34%
(34% - 34%)
12/25
34%
(34% - 34%)
45%
(45% - 45%)
43%
(43% - 43%)
11%
(11% -11%)
49%
(49% - 49%)
69%
(69% - 70%)
33%
(33% - 33%)
60%
(60% -61%)
55%
(55% - 55%)
50%
(50% - 50%)
46%
(46% - 46%)
59%
(59% - 59%)
98%
(98% - 98%)
46%
(46% - 46%)
68%
(68% - 68%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-21
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM25
Concentrations
-13%
(-13% --13%)
-9%
(-9% - -9%)
-41%
(-40% --41%)
0%
(0% - 0%)
-45%
(-44% - -45%)
-181%
(-179% --183%)
-9%
(-9% - -9%)
-130%
(-129% --131%)
-42%
(-42% - -42%)
-15%
(-15% --15%)
0%
(0% - 0%)
-161%
(-159% --163%)
-273%
(-272% - -274%)
-20%
(-20% --21%)
-80%
(-80% - -80%)
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%)
13/35
22%
(22% - 22%)
20%
(20% -21%)
22%
(22% - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
33%
(33% - 33%)
32%
(32% - 32%)
34%
(33% - 34%)
13%
(13% -13%)
31%
(30% -31%)
0%
(0% - 0%)
33%
(33% - 33%)
13%
(13% -13%)
22%
(22% - 22%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
36%
(36% - 36%)
0%
(0% - 0%)
13/30
22%
(22% - 22%)
24%
(23% - 24%)
22%
(22% - 22%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
22%
(22% - 22%)
33%
(33% - 33%)
30%
(30% - 30%)
25%
(25% - 25%)
13%
(13% -13%)
33%
(33% - 33%)
69%
(69% - 69%)
26%
(26% - 26%)
50%
(50% - 50%)
12/25
35%
(35% - 35%)
47%
(47% - 48%)
44%
(44% - 44%)
13%
(13% -13%)
57%
(57% - 57%)
66%
(66% - 67%)
33%
(33% - 33%)
66%
(66% - 66%)
61%
(61% -61%)
51%
(50% -51%)
45%
(44% - 45%)
66%
(66% - 66%)
1 00%
(100% -100%)
52%
(52% - 52%)
100%
(100% -100%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-22
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM25
Concentrations
-13%
(-13% --13%)
-10%
(-10% --10%)
-41%
(-41% --41%)
0%
(0% - 0%)
-46%
(-46% - -47%)
-165%
(-163% --167%)
-9%
(-9% - -9%)
-128%
(-127% --129%)
-42%
(-41% --42%)
-15%
(-15% --15%)
0%
(0% - 0%)
-158%
(-157% --160%)
-172%
(-171% --174%)
-20%
(-20% - -20%)
-70%
(-70% --71%)
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%)
13/35
22%
(22% - 22%)
21%
(21% -21%)
22%
(22% - 23%)
0%
(0% - 0%)
17%
(16% -17%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
33%
(33% - 33%)
33%
(33% - 33%)
34%
(34% - 34%)
12%
(12% -13%)
32%
(32% - 32%)
0%
(0% - 0%)
33%
(33% - 33%)
13%
(13% -13%)
22%
(22% - 22%)
24%
(24% - 25%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
13/30
22%
(22% - 22%)
24%
(24% - 24%)
22%
(22% - 23%)
0%
(0% - 0%)
29%
(29% - 29%)
30%
(30% - 30%)
22%
(22% - 22%)
33%
(33% - 33%)
30%
(30% - 30%)
25%
(25% - 25%)
14%
(14% -14%)
32%
(32% - 32%)
44%
(44% - 44%)
25%
(25% - 25%)
44%
(44% - 44%)
12/25
35%
(34% - 35%)
49%
(49% - 49%)
44%
(44% - 45%)
12%
(12% -13%)
59%
(59% - 59%)
61%
(61% -61%)
33%
(33% - 33%)
65%
(65% - 66%)
60%
(60% -61%)
51%
(51% -51%)
49%
(49% - 49%)
65%
(65% - 65%)
88%
(88% - 88%)
50%
(50% - 50%)
89%
(89% - 89%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-23
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-4. Estimated Percent Reductions From the Current Standard to Alternative
Standards in All Cause Mortality Associated with Long-Term Exposure to PMi.s (Exposure
Period: 1999 - 2000): Based on 2005 Air Quality Data*
          100%
       •E
       ra
       •c
       c
       53
       =
       o
       E
       •c
       £
       01
       a.
          -200%
                  2005 air
                   quality
15/35**      13/35       12/35

        Alternative Standard
13/30
12/25
                        -•-Atlanta, GA 732 (468-992);  4.9% (3.1%-6.6%)
                        -m- Baltimore, MD 696 (444-942); 4.9% (3.1%-6.7%)
                        -*- Birmingham, AL 451  (288-611); 4.6% (2.9%-6.2%)
                            Dallas, TX 469 (299 - 637); 3.7% (2.3% - 5%)
                        -*- Detroit, Ml 697 (445 - 946); 3.9% (2.5% - 5.3%)
                        -•- Fresno, CA 110 (70-150); 2% (1.3%-2.7%)
                        -i- Houston, TX 849 (542-1151); 4.6% (2.9%-6.2%)
                        	 Los Angeles, CA 1432 (911 -1948); 2.5%  (1.6%-3.4%)
                            New York, NY 1600 (1019-2174); 3% (1.9%-4.1%)
                        -•- Philadelphia, PA 547 (349-743); 3.8% (2.4%-5.1%)
                        -•- Phoenix, AZ 611  (389 - 831); 2.6% (1.7% - 3.6%)
                        -+- Pittsburgh, PA 366 (233 - 498); 2.6% (1.7% - 3.6%)
                            Salt Lake City, UT 54 (34 - 74); 1.1% (0.7% -1.6%)
                            St. Louis, MO 857 (547-1161); 4.5% (2.9%-6.2%)
                        -•- Tacoma.WA 101 (64-138);  2% (1.3%-2.7%)
*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.
                                                 E-24
September 2009
                  Draft - Do Not Quote or Cite

-------
Figure E-5. Estimated Percent Reductions From the Current Standard to Alternative
Standards in All Cause Mortality Associated with Long-Term Exposure to PMi.s (Exposure
Period: 1999 - 2000): Based on 2006 Air Quality Data*
       re
       •O
       C
       o
       o
       J=
       I
       *-
       o
       3
       •o
-20%
-40%
-60%
-80%
          -100%
       LL.
       g  -120%
       £  -140%
          -160%
          -180%
          -200%
                  2006 air     15/35*
                   quality
                             13/35       12/35

                           Alternative Standard
                                                             13/30
                                                                       12/25
                        -»- Atlanta, GA 706 (451 - 957); 4.5% (2.9% - 6.2%)
                        -m- Baltimore, MD 603 (385-817); 4.3% (2.7%-5.8%)
                        -*- Birmingham, AL 434 (277-589);  4.4% (2.8%-5.9%)
                           Dallas, IX 360 (229 - 490); 2.7% (1.7% - 3.7%)
                        -*- Detroit, Ml 508 (323-690); 2.8% (1.8%-3.9%)
                        -•- Fresno, CA 120 (76-164); 2.1% (1.4%-2.9%)
                        -i- Houston, IX 898 (574-1217); 4.7% (3%-6.3%)
                        	 Los Angeles, CA 1221  (776-1662); 2.1% (1.4%-2.9%)
                           New York, NY  1345 (855-1829);  2.5% (1.6%-3.4%)
                        -»- Philadelphia, PA 536 (341 -727); 3.7% (2.3%-5%)
                        -m- Phoenix, AZ 668 (425-908);  2.8%  (1.8%-3.8%)
                        -*- Pittsburgh, PA 297 (189-404); 2.1% (1.4%-2.9%)
                           Salt Lake City, UT 35 (22-48); 0.7% (0.5%-1%)
                           St. Louis, MO  652 (415-885); 3.4% (2.2%-4.7%)
                        -•- Tacoma, WA 56 (36 - 77); 1.1 % (0.7% -1.5%)
*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.
***The percent reduction for 2006 air quality in Salt Lake City is -274%.
                                                 E-25
September 2009
                                                                     Draft - Do Not Quote or Cite

-------
Figure E-6. Estimated Percent Reductions From the Current Standard to Alternative
Standards in All Cause Mortality Associated with Long-Term Exposure to PMi.s (Exposure
Period: 1999 - 2000): Based on 2007 Air Quality Data*
          100%
           80%
           60%
       1   40%
           20%
       ~    0% -
       3
       o
           -20%
       E   -40%
       e   -60%
       u
       •D
 -80%
-100%
-120%
-140%
-160%
-180%
-200%

                  2007 air
                   quality
                   15/35**      13/35      12/35

                           Alternative Standard
                                                            13/30
                                                                       12/25
                        -»- Atlanta, GA 731 (467-990); 4.6% (2.9%-6.2%)
                        -m- Baltimore, MD  563 (359-764); 4% (2.5%-5.4%)
                        -*- Birmingham, AL 429  (274-581); 4.3% (2.7%-5.8%)
                        -*- Dallas, TX 399 (254-542); 3% (1.9%-4.1%)
                        -*- Detroit, Ml 469 (299-638); 2.7% (1.7%-3.6%)
                        -•- Fresno, CA 144 (92-196); 2.5% (1.6%-3.4%)
                        -i- Houston, TX 880 (562-1193); 4.5% (2.9%-6.1%)
                        	 Los Angeles, CA  1257 (799-1711); 2.2% (1.4%-3%)
                            New York, NY 1370 (871 -1863); 2.6% (1.6%-3.5%)
                        -•- Philadelphia, PA 530 (338-719); 3.6% (2.3%-4.9%)
                        -m- Phoenix, AZ 580  (369-789); 2.3%  (1.5%-3.2%)
                        -*- Pittsburgh, PA  303 (193-413); 2.2% (1.4%-3%)
                            Salt Lake City, UT 67 (43-91); 1.3% (0.8%-1.8%)
                            St. Louis, MO 698 (445-948); 3.7% (2.3%-5%)
                        -•- Tacoma.WA 69 (44-94); 1.3% (0.8%-1.8%)
*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.
                                                 E-26
September 2009
                                                                     Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
247
(204 - 290)
393
(323-461)
214
(177-250)
224
(183-264)
653
(538 - 766)
183
(151 -215)
435
(357-510)
2217
(1824-2596)
2273
(1863-2672)
314
(258 - 369)
346
(283 - 409)
508
(419-593)
48
(39 - 56)
610
(502-715)
91
(74-108)
15/352
221
(181 -259)
363
(298 - 426)
156
(128-183)
224
(183-264)
479
(393 - 564)
66
(54 - 78)
402
(330 - 472)
1048
(855-1238)
1665
(1360-1964)
275
(225 - 324)
346
(283 - 409)
217
(177-256)
17
(14-20)
521
(428-612)
60
(49-71)
13/35
175
(144-206)
296
(242 - 348)
123
(101 -145)
224
(183-264)
416
(340 - 490)
66
(54 - 78)
317
(259 - 373)
1048
(855-1238)
1586
(1295-1871)
246
(201 - 290)
346
(283 - 409)
217
(177-256)
17
(14-20)
417
(341 -491)
60
(49-71)
12/35
152
(124-179)
257
(210-303)
106
(87-125)
199
(163-235)
357
(291 -421)
66
(54 - 78)
273
(223 - 322)
928
(757-1098)
1342
(1095-1586)
211
(172-249)
312
(254 - 368)
217
(177-256)
17
(14-20)
360
(294 - 425)
60
(49-71)
13/30
175
(144-206)
285
(234 - 336)
123
(101 -145)
224
(183-264)
366
(299 - 432)
43
(35-51)
317
(259 - 373)
738
(601 - 874)
1217
(992-1439)
209
(170-246)
302
(246 - 357)
154
(126-183)
9
(7-10)
406
(333 - 479)
40
(32 - 47)
12/25
148
(121 -175)
205
(167-242)
90
(74-107)
199
(163-235)
250
(204 - 296)
20
(17-24)
273
(223 - 322)
420
(342 - 499)
757
(615-897)
140
(114-166)
190
(154-225)
90
(73-107)
0
(0-0)
288
(235 - 340)
19
(16-23)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-27
September 2009
Draft - Do Not Quote or Cite

-------
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 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):
Recent PM2.5
Concentrations
240
(197-282)
345
(283 - 405)
208
(172-243)
173
(142-205)
504
(413-593)
194
(160-227)
459
(377 - 538)
2002
(1645-2349)
1976
(1617-2327)
308
(253 - 362)
378
(309 - 446)
445
(366-521)
40
(33 - 47)
480
(393 - 564)
60
(49-71)
15/352
214
(175-251)
317
(260 - 372)
151
(124-177)
173
(142-205)
353
(288-417)
72
(59 - 85)
424
(348 - 498)
897
(731 -1060)
1407
(1148-1662)
270
(221 -317)
378
(309 - 446)
177
(144-209)
11
(9-13)
401
(328 - 473)
34
(27 - 40)
13/35
168
(138-198)
254
(208 - 299)
118
(97-139)
173
(142-205)
298
(243 - 352)
72
(59 - 85)
335
(274 - 395)
897
(731 - 1060)
1333
(1087-1575)
241
(197-284)
378
(309 - 446)
177
(144-209)
11
(9-13)
310
(253 - 366)
34
(27 - 40)
12/35
145
(119-171)
218
(178-257)
102
(83-120)
151
(124-179)
247
(202 - 293)
72
(59 - 85)
290
(237 - 342)
784
(638 - 928)
1105
(900-1308)
206
(168-243)
341
(278 - 403)
177
(144-209)
11
(9-13)
260
(212-308)
34
(27 - 40)
13/30
168
(138-198)
244
(200 - 288)
118
(97-139)
173
(142-205)
255
(208 - 302)
48
(39 - 57)
335
(274 - 395)
605
(492-717)
988
(804-1170)
204
(166-241)
331
(270-391)
119
(97-141)
3
(3-4)
301
(245 - 355)
17
(14-20)
12/25
142
(116-167)
170
(139-201)
86
(70-102)
151
(124-179)
156
(126-184)
24
(20 - 29)
290
(237 - 342)
306
(249 - 364)
560
(455 - 664)
136
(111 -161)
212
(172-251)
61
(49 - 72)
0
(0-0)
197
(161 -234)
0
(0-0)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-28
September 2009
Draft - Do Not Quote or Cite

-------
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.s Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
248
(204-291)
324
(266-381)
206
(170-241)
191
(156-226)
472
(387 - 556)
218
(180-255)
451
(371 - 530)
2045
(1680-2399)
2008
(1644-2365)
305
(250 - 359)
329
(269 - 389)
450
(370 - 527)
55
(45 - 65)
510
(418-599)
69
(56 - 82)
15/352
221
(181 -260)
296
(243 - 349)
149
(122-175)
191
(156-226)
327
(267 - 386)
86
(70-102)
417
(342 - 489)
923
(752-1091)
1432
(1169-1692)
267
(218-314)
329
(269 - 389)
180
(147-213)
21
(17-25)
429
(351 - 505)
41
(33 - 48)
13/35
174
(143-205)
236
(193-278)
117
(95-138)
191
(156-226)
274
(223 - 324)
86
(70-102)
328
(268 - 386)
923
(752-1091)
1358
(1107-1604)
238
(194-280)
329
(269 - 389)
180
(147-213)
21
(17-25)
335
(273 - 395)
41
(33 - 48)
12/35
151
(123-178)
201
(164-238)
100
(82-118)
168
(137-199)
225
(183-267)
86
(70-102)
282
(231 - 333)
809
(659 - 957)
1127
(918-1334)
203
(166-240)
294
(240 - 348)
180
(147-213)
21
(17-25)
283
(231 - 334)
41
(33 - 48)
13/30
174
(143-205)
227
(185-268)
117
(95-138)
191
(156-226)
233
(190-276)
60
(49 - 72)
328
(268 - 386)
627
(510-743)
1009
(821 -1195)
201
(164-238)
284
(232 - 337)
123
(100-146)
12
(10-14)
325
(265 - 383)
23
(19-27)
12/25
147
(120-173)
155
(126-183)
85
(69-100)
168
(137-199)
137
(111 -162)
34
(28-41)
282
(231 - 333)
324
(263 - 384)
575
(467 - 683)
134
(109-158)
170
(138-202)
64
(52 - 76)
3
(2-3)
218
(177-258)
5
(4-5)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 E-29
September 2009
Draft - Do Not Quote or Cite

-------
Table E-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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
15.8%
(13% -18. 5%)
15.4%
(12. 7% -18.1%)
18.3%
(15.1% -21 .4%)
10.7%
(8. 8% -12. 6%)
15.6%
(12. 8% -18.3%)
16.4%
(13.5% -19.2%)
14.3%
(11. 8% -16.8%)
15.8%
(13% -18. 6%)
12.2%
(10% -14. 4%)
12.6%
(10.3% -14.8%)
7.9%
(6.4% - 9.3%)
18.3%
(15.1% -21 .4%)
9.8%
(8% - 1 1 .5%)
15.5%
(12.7% -18.1%)
9%
(7. 4% -10.7%)
15/352
14.1%
(11. 6% -16.5%)
14.3%
(11. 7% -16.7%)
13.3%
(11% -15. 7%)
10.7%
(8. 8% -12.6%)
1 1 .4%
(9.4% -13.5%)
5.9%
(4.8% - 7%)
13.2%
(10.9% -15.6%)
7.5%
(6.1% -8.8%)
9%
(7.3% -10.6%)
11%
(9% -13%)
7.9%
(6.4% - 9.3%)
7.8%
(6.4% - 9.2%)
3.4%
(2.8% -4.1%)
13.2%
(10.9% -15.5%)
5.9%
(4.8% - 7%)
13/35
1 1 .2%
(9.1% -13.1%)
1 1 .6%
(9.5% -13. 7%)
10.5%
(8.6% -12. 4%)
10.7%
(8.8% -12.6%)
9.9%
(8.1% -11. 7%)
5.9%
(4.8% - 7%)
10.4%
(8.5% -12.3%)
7.5%
(6.1% -8. 8%)
8.5%
(7% -10.1%)
9.8%
(8% - 1 1 .6%)
7.9%
(6.4% - 9.3%)
7.8%
(6.4% - 9.2%)
3.4%
(2. 8% -4.1%)
10.6%
(8.7% -12. 5%)
5.9%
(4.8% - 7%)
12/35
9.7%
(7.9% - 1 1 .4%)
10.1%
(8.3% - 1 1 .9%)
9.1%
(7.4% -10. 7%)
9.5%
(7.8% - 1 1 .3%)
8.5%
(7% - 1 0%)
5.9%
(4.8% - 7%)
9%
(7. 4% -10.6%)
6.6%
(5.4% - 7.8%)
7.2%
(5.9% - 8.5%)
8.4%
(6.9% - 9.9%)
7.1%
(5.8% - 8.4%)
7.8%
(6.4% - 9.2%)
3.4%
(2.8% -4.1%)
9.1%
(7.5% -10. 8%)
5.9%
(4.8% - 7%)
13/30
1 1 .2%
(9.1% -13.1%)
1 1 .2%
(9. 2% -13.2%)
10.5%
(8. 6% -12. 4%)
10.7%
(8.8% -12.6%)
8.7%
(7.1% -10. 3%)
3.9%
(3.2% - 4.6%)
10.4%
(8.5% -12.3%)
5.3%
(4.3% - 6.2%)
6.6%
(5.3% - 7.7%)
8.3%
(6.8% - 9.8%)
6.9%
(5.6% -8.1%)
5.6%
(4.5% - 6.6%)
1 .8%
(1.4% -2.1%)
10.3%
(8. 4% -12.1%)
4%
(3.2% - 4.7%)
12/25
9.5%
(7.7% - 1 1 .2%)
8.1%
(6.6% - 9.5%)
7.7%
(6. 3% -9.1%)
9.5%
(7.8% - 1 1 .3%)
6%
(4.9% -7.1%)
1 .8%
(1 .5% - 2.2%)
9%
(7. 4% -10. 6%)
3%
(2.4% - 3.6%)
4.1%
(3.3% - 4.8%)
5.6%
(4.6% - 6.6%)
4.3%
(3. 5% -5.1%)
3.2%
(2.6% - 3.9%)
0.1%
(0.1% -0.1%)
7.3%
(6% - 8.6%)
1 .9%
(1 .6% - 2.3%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                    E-30
September 2009
                         Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
14.8%
(12. 2% -17.4%)
13.5%
(11.1% -15.9%)
17.6%
(14. 5% -20.5%)
8.1%
(6.6% - 9.6%)
12%
(9. 9% -14.2%)
17.1%
(14.1% -20.1%)
14.6%
(12% -17. 2%)
14.2%
(11. 7% -16.7%)
10.5%
(8. 6% -12.4%)
12.4%
(10.1% -14.5%)
8.3%
(6.7% - 9.7%)
16.1%
(13. 3% -18.9%)
7.9%
(6.4% - 9.3%)
12.1%
(9. 9% -14.2%)
5.8%
(4.7% - 6.9%)
15/352
13.2%
(10.8% -15.5%)
12.4%
(10.2% -14.6%)
12.7%
(10.4% -15%)
8.1%
(6.6% - 9.6%)
8.4%
(6.9% -10%)
6.3%
(5.2% - 7.5%)
13.5%
(11.1% -15.9%)
6.4%
(5.2% - 7.5%)
7.5%
(6.1% -8.9%)
10.8%
(8.8% -12.7%)
8.3%
(6.7% - 9.7%)
6.4%
(5.2% - 7.6%)
2.2%
(1 .8% - 2.6%)
10.1%
(8.3% - 1 1 .9%)
3.3%
(2.7% - 3.9%)
13/35
10.4%
(8.5% -12.3%)
10%
(8.2% - 1 1 .7%)
10%
(8.2% - 1 1 .8%)
8.1%
(6.6% - 9.6%)
7.1%
(5.8% - 8.4%)
6.3%
(5.2% - 7.5%)
10.7%
(8.8% -12.6%)
6.4%
(5.2% - 7.5%)
7.1%
(5.8% - 8.4%)
9.6%
(7.9% - 1 1 .4%)
8.3%
(6.7% - 9.7%)
6.4%
(5.2% - 7.6%)
2.2%
(1 .8% - 2.6%)
7.8%
(6.4% - 9.2%)
3.3%
(2.7% - 3.9%)
12/35
9%
(7. 3% -10.6%)
8.6%
(7% -10.1%)
8.6%
(7% -10.1%)
7.1%
(5.8% - 8.4%)
5.9%
(4.8% - 7%)
6.3%
(5.2% - 7.5%)
9.2%
(7.6% -10. 9%)
5.6%
(4.5% - 6.6%)
5.9%
(4.8% - 7%)
8.3%
(6.7% - 9.7%)
7.5%
(6.1% -8. 8%)
6.4%
(5.2% - 7.6%)
2.2%
(1 .8% - 2.6%)
6.6%
(5.4% - 7.8%)
3.3%
(2.7% - 3.9%)
13/30
10.4%
(8.5% -12.3%)
9.6%
(7.8% - 1 1 .3%)
10%
(8.2% - 1 1 .8%)
8.1%
(6.6% - 9.6%)
6.1%
(5% - 7.2%)
4.3%
(3.5% -5.1%)
10.7%
(8. 8% -12. 6%)
4.3%
(3.5% -5.1%)
5.3%
(4.3% - 6.2%)
8.2%
(6.7% - 9.6%)
7.2%
(5.9% - 8.5%)
4.3%
(3.5% -5.1%)
0.7%
(0.5% - 0.8%)
7.6%
(6.2% - 9%)
1 .6%
(1 .3% - 1 .9%)
12/25
8.8%
(7. 2% -10. 3%)
6.7%
(5.4% - 7.9%)
7.3%
(5.9% - 8.6%)
7.1%
(5.8% - 8.4%)
3.7%
(3% - 4.4%)
2.2%
(1 .8% - 2.6%)
9.2%
(7. 6% -10.9%)
2.2%
(1 .8% - 2.6%)
3%
(2.4% - 3.5%)
5.4%
(4.4% - 6.5%)
4.6%
(3.8% - 5.5%)
2.2%
(1 .8% - 2.6%)
0%
(0% - 0%)
5%
(4.1% -5. 9%)
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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                    E-31
September 2009
                         Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
15%
(12.3% -17.5%)
12.7%
(10. 4% -15%)
17.3%
(14.2% -20. 2%)
8.8%
(7.2% -10.4%)
1 1 .4%
(9.3% -13.4%)
18.9%
(15.6% -22.1%)
14.1%
(11. 6% -16.6%)
14.4%
(11. 9% -16.9%)
10.6%
(8.7% -12.5%)
12.2%
(10% -14. 4%)
7%
(5.7% - 8.3%)
16.4%
(13.5% -19.2%)
10.6%
(8.7% -12.5%)
12.9%
(10.5% -15.1%)
6.6%
(5.4% - 7.8%)
15/352
13.3%
(10.9% -15.6%)
1 1 .7%
(9.6% -13.7%)
12.5%
(10.2% -14.7%)
8.8%
(7.2% -10.4%)
7.9%
(6.4% - 9.3%)
7.5%
(6.1% -8. 8%)
13%
(10. 7% -15.3%)
6.5%
(5.3% - 7.7%)
7.6%
(6.2% - 9%)
10.7%
(8.7% -12. 6%)
7%
(5.7% - 8.3%)
6.6%
(5.3% - 7.8%)
4%
(3.2% - 4.7%)
10.8%
(8.8% -12.7%)
3.9%
(3.2% - 4.6%)
13/35
10.5%
(8.6% -12.4%)
9.3%
(7.6% -10.9%)
9.8%
(8% -11.5%)
8.8%
(7.2% -10.4%)
6.6%
(5.4% - 7.8%)
7.5%
(6.1% -8. 8%)
10.2%
(8. 4% -12.1%)
6.5%
(5.3% - 7.7%)
7.2%
(5.9% - 8.5%)
9.5%
(7.8% - 1 1 .2%)
7%
(5.7% - 8.3%)
6.6%
(5.3% - 7.8%)
4%
(3.2% - 4.7%)
8.4%
(6.9% -10%)
3.9%
(3.2% - 4.6%)
12/35
9.1%
(7.4% -10.7%)
7.9%
(6.5% - 9.4%)
8.4%
(6.8% - 9.9%)
7.8%
(6.3% - 9.2%)
5.4%
(4.4o/0 . g 4o/0)
7.5%
(6.1% -8. 8%)
8.8%
(7. 2% -10. 4%)
5.7%
(4.6% - 6.8%)
6%
(4.9% -7.1%)
8.1%
(6.7% - 9.6%)
6.2%
(5.1% -7. 4%)
6.6%
(5.3% - 7.8%)
4%
(3.2% - 4.7%)
7.1%
(5.8% - 8.4%)
3.9%
(3.2% - 4.6%)
13/30
10.5%
(8.6% -12.4%)
8.9%
(7.3% -10.5%)
9.8%
(8% -11.5%)
8.8%
(7.2% -10.4%)
5.6%
(4.6% - 6.6%)
5.3%
(4.3% - 6.2%)
10.2%
(8. 4% -12.1%)
4.4%
(3.6% - 5.2%)
5.3%
(4.3% - 6.3%)
8.1%
(6.6% - 9.5%)
6%
(4. 9% -7.1%)
4.5%
(3.6% - 5.3%)
2.3%
(1 .8% - 2.7%)
8.2%
(6.7% - 9.7%)
2.2%
(1 .8% - 2.6%)
12/25
8.8%
(7.2% -10.4%)
6.1%
(5% - 7.2%)
7.1%
(5.8% - 8.4%)
7.8%
(6.3% - 9.2%)
3.3%
(2.7% - 3.9%)
3%
(2.4% - 3.5%)
8.8%
(7. 2% -10.4%)
2.3%
(1 .9% - 2.7%)
3%
(2.5% - 3.6%)
5.4%
(4.4% - 6.3%)
3.6%
(2.9% - 4.3%)
2.3%
(1 .9% - 2.8%)
0.5%
(0.4% - 0.6%)
5.5%
(4.5% - 6.5%)
0.4%
(0.4% - 0.5%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                    E-32
September 2009
                         Draft - Do Not Quote or Cite

-------
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 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
-12%
(-12% --12%)
-8%
(.30/0 - -8%)
-37%
(-37% - -38%)
0%
(0% - 0%)
-36%
(-36% - -37%)
-178%
(-175% --181%)
-8%
(-30/0 - -30/0)
-1 1 2%
(-1 1 0% - -1 1 3%)
-37%
(-36% - -37%)
-1 4%
(-14% --14%)
0%
(0% - 0%)
-134%
(-132% --137%)
-186%
(-184% --187%)
-1 7%
(-17% --17%)
-52%
(-52% - -53%)
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%)
13/35
21%
(20% -21%)
19%
(18% -19%)
21%
(21 %-21%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
21%
(21 %-21%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
12/35
31%
(31 % - 32%)
29%
(29% - 29%)
32%
(32% - 32%)
11%
(11% -11%)
26%
(25% - 26%)
0%
(0% - 0%)
32%
(32% - 32%)
11%
(11% -11%)
19%
(19% -20%)
23%
(23% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
31%
(31 %-31%)
0%
(0% - 0%)
13/30
21%
(20% -21%)
21%
(21 % - 22%)
21%
(21 %-21%)
0%
(0% - 0%)
24%
(23% - 24%)
34%
(34% - 34%)
21%
(21 %-21%)
30%
(29% - 30%)
27%
(27% - 27%)
24%
(24% - 24%)
13%
(13% -13%)
29%
(29% - 29%)
48%
(48% - 48%)
22%
(22% - 22%)
33%
(33% - 34%)
12/25
33%
(32% - 33%)
43%
(43% - 44%)
42%
(42% - 42%)
11%
(11% -11%)
48%
(48% - 48%)
69%
(69% - 69%)
32%
(32% - 32%)
60%
(60% - 60%)
55%
(54% - 55%)
49%
(49% - 49%)
45%
(45% - 45%)
58%
(58% - 59%)
98%
(98% - 98%)
45%
(44% - 45%)
68%
(67% - 68%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at'
: 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-33
September 2009
                                                                   Draft - Do Not Quote or Cite

-------
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 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
-12%
(-12% --13%)
-9%
(-9% - -9%)
-38%
(-37% - -39%)
0%
(0% - 0%)
-43%
(-42% - -43%)
-170%
(-167% --173%)
-8%
(.30/0 - -8%)
-123%
(-122% --125%)
-40%
(-40% - -41 %)
-1 4%
(-14% --14%)
0%
(0% - 0%)
-152%
(-149% --154%)
-266%
(-263% - -268%)
-20%
(-1 9% - -20%)
-78%
(-78% - -79%)
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%)
13/35
21%
(21% -21%)
20%
(20% - 20%)
22%
(21 % - 22%)
0%
(0% - 0%)
16%
(15% -16%)
0%
(0% - 0%)
21%
(21% -21%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(32% - 32%)
31%
(31% -31%)
33%
(32% - 33%)
13%
(13% -13%)
30%
(30% - 30%)
0%
(0% - 0%)
32%
(31 % - 32%)
13%
(12% -13%)
21%
(21 % - 22%)
24%
(23% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
13/30
21%
(21% -21%)
23%
(23% - 23%)
22%
(21 % - 22%)
0%
(0% - 0%)
28%
(27% - 28%)
33%
(32% - 33%)
21%
(21% -21%)
33%
(32% - 33%)
30%
(30% - 30%)
24%
(24% - 25%)
12%
(12% -12%)
32%
(32% - 33%)
69%
(69% - 69%)
25%
(25% - 25%)
50%
(50% - 50%)
12/25
34%
(33% - 34%)
46%
(46% - 47%)
43%
(43% - 43%)
13%
(13% -13%)
56%
(56% - 56%)
66%
(66% - 66%)
32%
(31 % - 32%)
66%
(66% - 66%)
60%
(60% - 60%)
50%
(49% - 50%)
44%
(44% - 44%)
66%
(65% - 66%)
100%
(100% -100%)
51%
(51% -51%)
100%
(100% -100%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at'
: 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-34
September 2009
                                                                   Draft - Do Not Quote or Cite

-------
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 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
-12%
(-12% --13%)
-9%
(-9% - -9%)
-38%
(-38% - -39%)
0%
(0% - 0%)
-45%
(-44% - -45%)
-154%
(-151% --157%)
-8%
(.30/0 - -8%)
-122%
(-120% --123%)
-40%
(-40% - -41 %)
-1 4%
(-14% --15%)
0%
(0% - 0%)
-149%
(-147% --152%)
-166%
(-164% --168%)
-1 9%
(-19% --19%)
-69%
(-68% - -69%)
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%)
13/35
21%
(21% -21%)
20%
(20% -21%)
22%
(21 % - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
21%
(21 % - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
12/35
32%
(32% - 32%)
32%
(32% - 32%)
33%
(33% - 33%)
12%
(12% -12%)
31%
(31% -31%)
0%
(0% - 0%)
32%
(32% - 32%)
12%
(12% -12%)
21%
(21% -21%)
24%
(24% - 24%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
0%
(0% - 0%)
34%
(34% - 34%)
0%
(0% - 0%)
13/30
21%
(21% -21%)
24%
(23% - 24%)
22%
(21 % - 22%)
0%
(0% - 0%)
29%
(29% - 29%)
30%
(29% - 30%)
21%
(21 % - 22%)
32%
(32% - 32%)
30%
(29% - 30%)
25%
(24% - 25%)
14%
(14% -14%)
32%
(32% - 32%)
43%
(43% - 43%)
24%
(24% - 24%)
44%
(44% - 44%)
12/25
34%
(33% - 34%)
48%
(47% - 48%)
43%
(43% - 44%)
12%
(12% -12%)
58%
(58% - 58%)
60%
(60% - 60%)
32%
(32% - 32%)
65%
(65% - 65%)
60%
(60% - 60%)
50%
(50% - 50%)
48%
(48% - 49%)
65%
(64% - 65%)
88%
(88% - 88%)
49%
(49% - 49%)
89%
(89% - 89%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at'
: 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-35
September 2009
                                                                   Draft - Do Not Quote or Cite

-------
Figure E-7. Estimated Percent Reductions From the Current Standard to Alternative
Standards in Ischemic Heart Disease Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983): Based on 2005 Air Quality Data*
       S
       OT
       O
       Q

       I
       •s
       1
       01
       O
       s.
 -80%
-100%
-120%
-140%
-160%
-180%
-200%
                  2005 air
                   quality
                   15/35**     13/35       12/35

                            Alternative Standard
13/30
12/25
                        -•- Atlanta, GA 221  (181-259); 14.1% (11.6% -16.5%)
                        -•- Baltimore, MD 363 (298-426);  14.3% (11.7% -16.7%)
                        -*- Birmingham, AL  156 (128-183); 13.3%  (11%-15.7%)
                            Dallas, TX 224 (183-264); 10.7% (8.8%-12.6%)
                        -*- Detroit, Ml 479 (393-564); 11.4% (9.4%-13.5%)
                        -•- Fresno, CA 66 (54 - 78); 5.9% (4.8% - 7%)
                        -i- Houston, TX 402 (330-472); 13.2% (10.9% -15.6%)
                        	 Los Angeles, CA 1048 (855-1238); 7.5% (6.1%-8.8%)
                          - New York, NY  1665 (1360-1964); 9% (7.3%-10.6%)
                        -•- Philadelphia, PA 275 (225-324); 11% (9%-13%)
                        -m- Phoenix, AZ 346 (283 - 409);  7.9% (6.4% - 9.3%)
                        -*- Pittsburgh, PA 217 (177-256);  7.8% (6.4%-9.2%)
                            Salt Lake City, UT 17 (14-20); 3.4% (2.8%-4.1%)
                            St. Louis, MO  521 (428-612); 13.2% (10.9% -15.5%)
                            Tacoma.WA 60 (49-71); 5.9% (4.8%-7%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                                 E-36
September 2009
                                                                     Draft - Do Not Quote or Cite

-------
Figure E-8. Estimated Percent Reductions From the Current Standard to Alternative
Standards in Ischemic Heart Disease Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983):  Based on 2006 Air Quality Data*
          100%
                  2006 air     15/35*
                   quality
   13/35      12/35

Alternative Standard
                                                            13/30
                                                                       12/25
                        -»- Atlanta, GA 214 (175-251); 13.2% (10.8% - 15.5%)
                        -m- Baltimore, MD 317 (260-372); 12.4% (10.2% - 14.6%)
                        -*- Birmingham, AL 151 (124-177);  12.7% (10.4% -15%)
                           Dallas, TX 173 (142-205); 8.1% (6.6% -9.6%)
                        -*- Detroit, Ml 353 (288-417); 8.4% (6.9% -10%)
                        -•- Fresno, CA 72 (59-85); 6.3% (5.2% -7.5%)
                        -i— Houston, TX 424 (348-498);  13.5% (1 1.1% - 15.9%)
                        - Los Angeles, CA 897 (731 -1060); 6.4% (5.2% -7.5%)
                           New York, NY  1407 (1148-1662); 7.5% (6.1% -8.9%)
                        -»- Philadelphia, PA 270 (221 -317); 10.8% (8.8% -12.7%)
                        -m- Phoenix, AZ 378 (309-446);  8.3% (6.7% -9.7%)
                        -*- Pittsburgh, PA 177 (144-209); 6.4% (5.2% -7.6%)
                           Salt Lake City, UT 11  (9-13); 2.2% (1.8% -2.6%)
                           St. Louis, MO  401 (328-473); 10.1% (8.3% -11. 9%)
                           Tacoma.WA 34 (27-40); 3.3% (2.7% -3.9%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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 reduction for 2006 air quality in Salt Lake City is -264%.
                                                 E-37
September 2009
                                          Draft - Do Not Quote or Cite

-------
Figure E-9  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Ischemic Heart Disease Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983): Based on 2007 Air Quality Data*
          100%
          -200%
                  2007 air
                   quality
15/35**      13/35       12/35

        Alternative Standard
13/30
12/25
                       -•- Atlanta, GA 221  (181 -260); 13.3% (10.9% - 15.6%)
                       -m- Baltimore, MD 296 (243-349);  11.7% (9.6% -13.7%)
                       -*- Birmingham, AL  149 (122-175); 12.5% (10.2% - 14.7%)
                           Dallas, TX 191 (156-226); 8.8% (7.2% -10.4%)
                       -*- Detroit, Ml 327 (267 - 386); 7.9% (6.4% - 9.3%)
                       -•- Fresno, CA 86 (70-102); 7.5% (6.1% -8.8%)
                       -i- Houston, TX 417 (342-489); 13% (10.7% - 15.3%)
                       - Los Angeles, CA 923 (752-1091); 6.5% (5.3% -7.7%)
                           New York, NY  1 432 (1 1 69 - 1 692); 7.6% (6.2% - 9%)
                       -»- Philadelphia, PA 267 (218-314); 10.7% (8.7% -12.6%)
                       -m- Phoenix, AZ 329 (269 - 389); 7% (5.7% - 8.3%)
                       -*- Pittsburgh, PA 180 (147-213);  6.6% (5.3% -7.8%)
                           Salt Lake City, UT 21 (17 - 25); 4% (3.2% - 4.7%)
                           St. Louis, MO  429 (351 -505); 10.8% (8.8% -12.7%)
                           Tacoma, WA 41 (33 - 48); 3.9% (3.2% - 4.6%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                                 E-38
September 2009
                                                  Draft - Do Not Quote or Cite

-------
Table E-28.  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 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 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):
Recent PM25
Concentrations
310
(255 - 363)
493
(406 - 577)
268
(221 -312)
283
(231 - 333)
819
(674 - 958)
229
(189-268)
546
(449 - 640)
2778
(2288 - 3247)
2865
(2348 - 3365)
396
(325 - 465)
439
(358-519)
634
(524 - 739)
61
(50-71)
765
(630 - 895)
115
(94-136)
15/352
278
(228 - 325)
456
(375 - 534)
197
(161 -230)
283
(231 - 333)
605
(495-711)
84
(68 - 99)
505
(415-593)
1330
(1083-1572)
2109
(1721 -2487)
347
(284 - 409)
439
(358-519)
275
(224 - 325)
21
(17-25)
656
(538 - 769)
76
(62 - 90)
13/35
221
(181 -260)
373
(305 - 438)
156
(127-183)
283
(231 - 333)
526
(430-619)
84
(68 - 99)
400
(327-471)
1330
(1083-1572)
2010
(1639-2372)
311
(254 - 366)
439
(358-519)
275
(224 - 325)
21
(17-25)
527
(431 -621)
76
(62 - 90)
12/35
192
(157-227)
325
(266 - 383)
135
(110-159)
252
(206 - 297)
452
(369 - 534)
84
(68 - 99)
346
(282 - 408)
1180
(960-1396)
1704
(1388-2015)
267
(218-315)
396
(322 - 468)
275
(224 - 325)
21
(17-25)
456
(372 - 538)
76
(62 - 90)
13/30
221
(181 -260)
360
(295 - 424)
156
(127-183)
283
(231 - 333)
464
(379 - 548)
55
(45 - 66)
400
(327-471)
940
(763-1114)
1546
(1258-1830)
264
(216-312)
384
(312-454)
196
(160-233)
11
(9-13)
514
(420 - 605)
51
(41 -60)
12/25
188
(153-221)
260
(212-307)
115
(94-136)
252
(206 - 297)
318
(258 - 376)
26
(21 -31)
346
(282 - 408)
537
(435 - 639)
965
(783-1146)
178
(145-211)
242
(196-287)
115
(93-137)
1
(0-1)
366
(298 - 432)
25
(20 - 30)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                   E-39
September 2009
                        Draft - Do Not Quote or Cite

-------
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 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 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):
Recent PM25
Concentrations
301
(248 - 353)
434
(356 - 509)
260
(215-303)
220
(179-260)
635
(521 -746)
243
(200-283)
576
(474 - 675)
2515
(2067 - 2947)
2497
(2042 - 2939)
388
(318-456)
479
(391 - 566)
557
(459-651)
51
(41 - 60)
605
(495-710)
76
(62 - 90)
15/352
269
(221 -315)
399
(327 - 468)
190
(156-223)
220
(179-260)
447
(365 - 528)
91
(74-108)
534
(438 - 626)
1140
(927-1349)
1785
(1454-2110)
340
(279-401)
479
(391 - 566)
224
(183-266)
14
(11 -17)
507
(415-597)
43
(35-51)
13/35
213
(174-251)
321
(263 - 378)
150
(122-176)
220
(179-260)
379
(308 - 448)
91
(74-108)
424
(347 - 499)
1140
(927-1349)
1693
(1378-2001)
304
(249 - 359)
479
(391 - 566)
224
(183-266)
14
(11 -17)
393
(321 - 465)
43
(35-51)
12/35
184
(150-217)
276
(225 - 326)
129
(105-152)
192
(157-227)
315
(256 - 373)
91
(74-108)
367
(300-433)
998
(811 -1182)
1406
(1143-1665)
261
(213-308)
433
(353-512)
224
(183-266)
14
(11-17)
331
(269-391)
43
(35-51)
13/30
213
(174-251)
309
(253 - 365)
150
(122-176)
220
(179-260)
325
(264 - 385)
62
(50 - 73)
424
(347 - 499)
771
(626-915)
1258
(1022-1492)
258
(211 -305)
420
(342 - 497)
152
(124-181)
4
(4-5)
382
(311 -451)
21
(17-26)
12/25
180
(146-212)
216
(176-255)
109
(89-129)
192
(157-227)
198
(161 -236)
31
(25 - 37)
367
(300 - 433)
392
(317-466)
715
(579 - 850)
173
(141 -205)
270
(219-320)
78
(63 - 93)
0
(0-0)
251
(204 - 298)
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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                    E-40
September 2009
                         Draft - Do Not Quote or Cite

-------
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 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 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
312
(256 - 365)
408
(334 - 479)
258
(213-301)
243
(198-286)
596
(488-701)
272
(225-316)
567
(466 - 665)
2568
(2111 -3008)
2537
(2075 - 2987)
385
(315-452)
418
(341 - 495)
563
(464 - 658)
70
(57 - 82)
642
(527 - 753)
88
(71 -104)
15/352
278
(228 - 326)
374
(306 - 440)
188
(154-220)
243
(198-286)
415
(338 - 490)
109
(89-129)
524
(430-615)
1173
(954-1388)
1818
(1481 -2148)
337
(276 - 397)
418
(341 -495)
229
(187-271)
27
(22-31)
541
(443 - 637)
52
(42 - 62)
13/35
220
(180-259)
299
(244 - 352)
148
(121 -174)
243
(198-286)
348
(283-412)
109
(89-129)
415
(339 - 488)
1173
(954-1388)
1724
(1404-2038)
301
(246 - 355)
418
(341 -495)
229
(187-271)
27
(22-31)
424
(346-501)
52
(42 - 62)
12/35
191
(156-225)
255
(208 - 302)
127
(104-150)
213
(174-252)
287
(233 - 340)
109
(89-129)
358
(292 - 422)
1029
(836-1219)
1434
(1166-1698)
258
(210-305)
374
(304 - 443)
229
(187-271)
27
(22-31)
359
(292 - 425)
52
(42 - 62)
13/30
220
(180-259)
287
(234 - 339)
148
(121 -174)
243
(198-286)
297
(241 -351)
77
(63-91)
415
(339 - 488)
799
(648 - 948)
1285
(1044-1523)
255
(208-301)
362
(294 - 428)
157
(127-186)
15
(12-18)
412
(336 - 486)
29
(24 - 35)
12/25
186
(152-220)
197
(160-233)
107
(87-127)
213
(174-252)
175
(142-208)
44
(36 - 52)
358
(292 - 422)
414
(335 - 492)
735
(595 - 874)
170
(138-202)
217
(176-258)
82
(66 - 97)
3
(3-4)
277
(225 - 328)
6
(5-7)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                   E-41
September 2009
                        Draft - Do Not Quote or Cite

-------
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.j 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 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):
Recent PM2.5
Concentrations
19.8%
(16.3% -23.1%)
19.4%
(15.9% -22.7%)
22.9%
(18.9% -26.6%)
13.5%
(11.1% -15.9%)
19.5%
(16.1% -22.9%)
20.6%
(16.9% -24%)
18%
(14.8% -21.1%)
19.9%
(16.4% -23.2%)
15.4%
(12.6% -18.1%)
15.8%
(13% -18.6%)
10%
(8.1% -11. 8%)
22.8%
(18.9% -26.6%)
12.4%
(10.1% -14.6%)
19.4%
(16% -22.7%)
1 1 .4%
(9.3% -13.5%)
15/352
17.7%
(14.5% -20.7%)
17.9%
(14.7% -21%)
16.8%
(13.8% -19.7%)
13.5%
(11.1% -15.9%)
14.4%
(11. 8% -17%)
7.5%
(6.1% -8.9%)
16.7%
(13.7% -19.6%)
9.5%
(7.7% - 1 1 .2%)
1 1 .4%
(9.3% -13.4%)
13.9%
(11. 4% -16.3%)
10%
(8.1% -11. 8%)
9.9%
(8.1% -11. 7%)
4.4%
(3.5% - 5.2%)
16.6%
(13.7% -19.5%)
7.6%
(6.1% -9%)
13/35
14.1%
(11. 5% -16.6%)
14.7%
(12% -17.2%)
13.3%
(10.9% -15.7%)
13.5%
(11.1% -15.9%)
12.5%
(10.3% -14.8%)
7.5%
(6.1% -8.9%)
13.2%
(10.8% -15.5%)
9.5%
(7.7% - 1 1 .2%)
10.8%
(8.8% -12.8%)
12.4%
(10.2% -14.6%)
10%
(8.1% -11. 8%)
9.9%
(8.1% -11. 7%)
4.4%
(3.5% - 5.2%)
13.4%
(10.9% -15.7%)
7.6%
(6.1% -9%)
12/35
12.3%
(10% -14.4%)
12.8%
(10.4% -15%)
1 1 .5%
(9.4% - 1 3.6%)
12.1%
(9.9% -14.2%)
10.8%
(8.8% -12.7%)
7.5%
(6.1% -8.9%)
1 1 .4%
(9.3% -13.5%)
8.4%
(6.9% -10%)
9.2%
(7.5% -10.9%)
10.7%
(8.7% -12.6%)
9%
(7.3% - 1 0.6%)
9.9%
(8.1% -11. 7%)
4.4%
(3.5% - 5.2%)
1 1 .6%
(9.4% -13.6%)
7.6%
(6.1% -9%)
13/30
14.1%
(11. 5% -16.6%)
14.2%
(11. 6% -16.6%)
13.3%
(10.9% -15.7%)
13.5%
(11.1% -15.9%)
11.1%
(9% -13.1%)
5%
(4% - 5.9%)
13.2%
(10.8% -15.5%)
6.7%
(5.5% - 8%)
8.3%
(6.8% - 9.9%)
10.6%
(8.6% -12.5%)
8.7%
(7.1% -10.3%)
7.1%
(5.7% - 8.4%)
2.3%
(1 .8% - 2.7%)
13%
(10.7% -15.3%)
5%
(4.1% -6%)
12/25
12%
(9.8% -14.1%)
10.2%
(8.3% -12.1%)
9.8%
(8% - 1 1 .6%)
12.1%
(9.9% -14.2%)
7.6%
(6.2% - 9%)
2.3%
(1 .9% - 2.8%)
1 1 .4%
(9.3% -13.5%)
3.8%
(3.1% -4.6%)
5.2%
(4.2% - 6.2%)
7.1%
(5.8% - 8.4%)
5.5%
(4.5% - 6.5%)
4.2%
(3.4% - 4.9%)
0.1%
(0.1% -0.1%)
9.3%
(7.6% -11%)
2.5%
(2% - 2.9%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-42
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2.5
         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 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):
Recent PM25
Concentrations
18.6%
(15. 3% -21 .8%)
17%
(14% -20%)
22%
(18.1% -25.6%)
10.3%
(8. 4% -12. 2%)
15.2%
(12. 4% -17.8%)
21 .4%
(17. 7% -25%)
18.4%
(15.1% -21 .5%)
17.9%
(14. 7% -20. 9%)
13.3%
(10. 9% -15.7%)
15.6%
(12. 8% -18. 3%)
10.5%
(8. 5% -12. 4%)
20.2%
(16. 6% -23. 6%)
10%
(8.2% - 1 1 .8%)
15.3%
(12. 5% -17.9%)
7.4%
(6% - 8.8%)
15/352
16.6%
(13.6% -19.5%)
15.6%
(12.8% -18.4%)
16%
(13.1% -18.8%)
10.3%
(8. 4% -12.2%)
10.7%
(8. 7% -12.6%)
8.1%
(6.6% - 9.6%)
17%
(14% -20%)
8.1%
(6.6% - 9.6%)
9.5%
(7.8% - 1 1 .3%)
13.6%
(11. 2% -16.1%)
10.5%
(8. 5% -12.4%)
8.1%
(6.6% - 9.6%)
2.8%
(2.2% - 3.3%)
12.8%
(10.5% -15.1%)
4.2%
(3.4% - 5%)
13/35
13.2%
(10.8% -15. 5%)
12.6%
(10. 3% -14. 8%)
12.6%
(10.3% -14. 9%)
10.3%
(8.4% -12. 2%)
9%
(7.4% -10. 7%)
8.1%
(6.6% - 9.6%)
13.5%
(11% -15.9%)
8.1%
(6.6% - 9.6%)
9%
(7.3% -10. 7%)
12.2%
(10% -14.4%)
10.5%
(8.5% -12.4%)
8.1%
(6.6% - 9.6%)
2.8%
(2.2% - 3.3%)
9.9%
(8.1% -11. 7%)
4.2%
(3.4% - 5%)
12/35
1 1 .4%
(9. 3% -13.4%)
10.8%
(8. 8% -12.8%)
10.9%
(8. 9% -12.8%)
9%
(7. 3% -10.7%)
7.5%
(6.1% -8. 9%)
8.1%
(6.6% - 9.6%)
1 1 .7%
(9. 6% -13.8%)
7.1%
(5.8% - 8.4%)
7.5%
(6.1% -8. 9%)
10.5%
(8. 5% -12.4%)
9.5%
(7.7% - 1 1 .2%)
8.1%
(6.6% - 9.6%)
2.8%
(2.2% - 3.3%)
8.3%
(6.8% - 9.9%)
4.2%
(3.4% - 5%)
13/30
13.2%
(10.8% -15. 5%)
12.1%
(9.9% -14.3%)
12.6%
(10.3% -14. 9%)
10.3%
(8.4% -12. 2%)
7.8%
(6.3% - 9.2%)
5.5%
(4.4% - 6.5%)
13.5%
(11% -15.9%)
5.5%
(4.4% - 6.5%)
6.7%
(5.5% - 8%)
10.4%
(8.4% -12.2%)
9.2%
(7.5% -10.9%)
5.5%
(4.5% - 6.5%)
0.9%
(0.7% - 1 %)
9.6%
(7.8% - 1 1 .4%)
2.1%
(1 .7% - 2.5%)
12/25
11.1%
(9.1% -13.1%)
8.5%
(6. 9% -10%)
9.2%
(7. 5% -10. 9%)
9%
(7. 3% -10. 7%)
4.7%
(3.8% - 5.6%)
2.8%
(2.2% - 3.3%)
1 1 .7%
(9. 6% -13.8%)
2.8%
(2.3% - 3.3%)
3.8%
(3.1% -4. 5%)
6.9%
(5.6% - 8.2%)
5.9%
(4.8% - 7%)
2.8%
(2.3% - 3.4%)
0%
(0% - 0%)
6.3%
(5.2% - 7.5%)
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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                    E-43
September 2009
                         Draft - Do Not Quote or Cite

-------
Table E-33.  Estimated Percent of Total Annual Incidence of Ischemic Heart Disease Mortality Associated with Long-Term
         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
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):
Recent PM25
Concentrations
18.8%
(15.4% -22%)
16%
(13. 2% -18.8%)
21 .6%
(17. 8% -25.2%)
1 1 .2%
(9.1% -13. 2%)
14.3%
(11. 7% -16.9%)
23.6%
(19. 5% -27.5%)
17.7%
(14. 6% -20. 8%)
18.1%
(14. 9% -21 .2%)
13.4%
(11% -15. 8%)
15.4%
(12. 6% -18.1%)
8.9%
(7. 2% -10. 5%)
20.5%
(16. 9% -23. 9%)
13.4%
(11% -15. 8%)
16.2%
(13. 3% -19%)
8.4%
(6.8% - 9.9%)
15/352
16.8%
(13.7% -19. 6%)
14.7%
(12% -17. 3%)
15.7%
(12.9% -18.4%)
1 1 .2%
(9.1% -13.2%)
10%
(8.1% -11. 8%)
9.5%
(7.7% - 1 1 .2%)
16.4%
(13.5% -19. 2%)
8.3%
(6.7% - 9.8%)
9.6%
(7.8% - 1 1 .4%)
13.5%
(11% -15. 9%)
8.9%
(7.2% -10.5%)
8.3%
(6.8% - 9.9%)
5.1%
(4.1% -6.1%)
13.7%
(11. 2% -16.1%)
5%
(4.1% -5.9%)
13/35
13.3%
(10. 9% -15.6%)
1 1 .8%
(9.6% -13. 9%)
12.4%
(10.1% -14. 6%)
1 1 .2%
(9.1% -13. 2%)
8.4%
(6.8% - 9.9%)
9.5%
(7.7% - 1 1 .2%)
13%
(10.6% -15. 3%)
8.3%
(6.7% - 9.8%)
9.1%
(7.4% -10. 8%)
12.1%
(9.8% -14.2%)
8.9%
(7. 2% -10. 5%)
8.3%
(6.8% - 9.9%)
5.1%
(4.1% -6.1%)
10.7%
(8.7% -12. 6%)
5%
(4.1% -5.9%)
12/35
1 1 .5%
(9.4% -13.5%)
10%
(8.2% - 1 1 .9%)
10.6%
(8. 7% -12. 5%)
9.8%
(8% - 1 1 .6%)
6.9%
(5.6% - 8.2%)
9.5%
(7.7% - 1 1 .2%)
11.2%
(9.1% -13. 2%)
7.3%
(5.9% - 8.6%)
7.6%
(6.2% - 9%)
10.3%
(8. 4% -12.2%)
7.9%
(6.4% - 9.4%)
8.3%
(6.8% - 9.9%)
5.1%
(4.1% -6.1%)
9.1%
(7. 4% -10. 7%)
5%
(4.1% -5. 9%)
13/30
13.3%
(10. 9% -15. 6%)
1 1 .3%
(9.2% -13. 3%)
12.4%
(10.1% -14. 6%)
1 1 .2%
(9.1% -13.2%)
7.1%
(5.8% - 8.5%)
6.7%
(5.4% - 7.9%)
13%
(10.6% -15.3%)
5.6%
(4.6% - 6.7%)
6.8%
(5. 5% -8.1%)
10.2%
(8.3% -12.1%)
7.7%
(6. 2% -9.1%)
5.7%
(4.6% - 6.8%)
2.9%
(2.3% - 3.4%)
10.4%
(8.5% -12.3%)
2.8%
(2.3% - 3.3%)
12/25
1 1 .2%
(9.1% -13.2%)
7.7%
(6.3% - 9.2%)
9%
(7. 3% -10. 6%)
9.8%
(8% - 1 1 .6%)
4.2%
(3.4% - 5%)
3.8%
(3.1% -4.5%)
1 1 .2%
(9.1% -13. 2%)
2.9%
(2.4% - 3.5%)
3.9%
(3.2% - 4.6%)
6.8%
(5. 5% -8.1%)
4.6%
(3.7% - 5.5%)
3%
(2.4% - 3.5%)
0.6%
(0.5% - 0.8%)
7%
(5.7% - 8.3%)
0.6%
(0.5% - 0.7%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
2009).
                                                                    E-44
September 2009
                         Draft - Do Not Quote or Cite

-------
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 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):
Recent PM2.5
Concentrations
-12%
(-1 1 % - -1 2%)
-8%
(.30/0 - -8%)
-36%
(-35% - -37%)
0%
(0% - 0%)
-35%
(-35% - -36%)
-174%
(-170% --178%)
-8%
(-30/0 - -30/0)
-1 09%
(-107% --111%)
-36%
(-35% - -36%)
-14%
(-14% --14%)
0%
(0% - 0%)
-1 30%
(-127% --134%)
-1 83%
(-181% --185%)
-17%
(-16% --17%)
-51 %
(-51 % - -52%)
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%)
13/35
20%
(20% -21%)
18%
(18% -18%)
21%
(20% -21%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
21%
(21 %-21%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(10% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
20%
(19% -20%)
0%
(0% - 0%)
12/35
31%
(30% - 31 %)
29%
(28% - 29%)
31%
(31 %-32%)
11%
(11% -11%)
25%
(25% - 26%)
0%
(0% - 0%)
32%
(31 %-32%)
11%
(11% -11%)
19%
(19% -19%)
23%
(23% - 23%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
30%
(30% - 31 %)
0%
(0% - 0%)
13/30
20%
(20% - 21 %)
21%
(21 % - 21 %)
21%
(20% - 21 %)
0%
(0% - 0%)
23%
(23% - 24%)
34%
(34% - 34%)
21%
(21 % - 21 %)
29%
(29% - 30%)
27%
(26% - 27%)
24%
(24% - 24%)
13%
(12% -13%)
29%
(28% - 29%)
48%
(48% - 48%)
22%
(21 % - 22%)
33%
(33% - 33%)
12/25
32%
(32% - 33%)
43%
(42% - 43%)
42%
(41 % - 42%)
11%
(11% -11%)
47%
(47% - 48%)
69%
(69% - 69%)
32%
(31 % - 32%)
60%
(59% - 60%)
54%
(54% - 55%)
49%
(48% - 49%)
45%
(45% - 45%)
58%
(58% - 58%)
98%
(98% - 98%)
44%
(44% - 45%)
67%
(67% - 68%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-45
September 2009
Draft - Do Not Quote or Cite

-------
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 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
-12%
(-1 2% - -1 2%)
-9%
(-9% - -9%)
-37%
(-36% - -38%)
0%
(0% - 0%)
-42%
(-41% --43%)
-166%
(-162% --170%)
-8%
(.30/0 - -8%)
-1 21 %
(-11 8% --123%)
-40%
(-39% - -40%)
-14%
(-14% --14%)
0%
(0% - 0%)
-1 48%
(-145% --151%)
-262%
(-260% - -265%)
-19%
(-19% --20%)
-78%
(-77% - -78%)
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%)
13/35
21%
(21% -21%)
19%
(19% -20%)
21%
(21% -21%)
0%
(0% - 0%)
15%
(15% -16%)
0%
(0% - 0%)
21%
(20% -21%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(10% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
22%
(22% - 23%)
0%
(0% - 0%)
12/35
32%
(31% -32%)
31%
(30% - 31 %)
32%
(32% - 32%)
12%
(12% -13%)
30%
(29% - 30%)
0%
(0% - 0%)
31%
(31% -32%)
12%
(12% -13%)
21%
(21% -21%)
23%
(23% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(35% - 35%)
0%
(0% - 0%)
13/30
21%
(21 % - 21 %)
22%
(22% - 23%)
21%
(21 % - 21 %)
0%
(0% - 0%)
27%
(27% - 28%)
32%
(32% - 33%)
21%
(20% - 21 %)
32%
(32% - 33%)
30%
(29% - 30%)
24%
(24% - 24%)
12%
(12% -12%)
32%
(32% - 32%)
69%
(69% - 69%)
25%
(25% - 25%)
50%
(50% - 50%)
12/25
33%
(33% - 34%)
46%
(45% - 46%)
42%
(42% - 43%)
12%
(12% -13%)
56%
(55% - 56%)
66%
(66% - 66%)
31%
(31 % - 32%)
66%
(65% - 66%)
60%
(60% - 60%)
49%
(49% - 50%)
44%
(43% - 44%)
65%
(65% - 65%)
100%
(100% -100%)
50%
(50% -51%)
100%
(100% -100%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-46
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2.5 Concentrations that Just Meet the Current and
Alternative Annual (n) and Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
-12%
(-12% --12%)
-9%
(-9% - -9%)
-37%
(-37% - -38%)
0%
(0% - 0%)
-44%
(-43% - -44%)
-149%
(-145% --153%)
-8%
(.30/0 - -8%)
-1 1 9%
(-11 7% --121%)
-40%
(-39% - -40%)
-1 4%
(-14% --14%)
0%
(0% - 0%)
-146%
(-142% --149%)
-163%
(-161% --166%)
-1 9%
(-18% --19%)
-68%
(-68% - -69%)
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%)
13/35
21%
(20% -21%)
20%
(20% - 20%)
21%
(21 % - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
21%
(21% -21%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
22%
(21 % - 22%)
0%
(0% - 0%)
12/35
31%
(31 % - 32%)
32%
(31 % - 32%)
32%
(32% - 33%)
12%
(12% -12%)
31%
(31% -31%)
0%
(0% - 0%)
32%
(31 % - 32%)
12%
(12% -12%)
21%
(21% -21%)
23%
(23% - 24%)
11%
(1 1 % - 1 1 %)
0%
(0% - 0%)
0%
(0% - 0%)
34%
(33% - 34%)
0%
(0% - 0%)
13/30
21%
(20% -21%)
23%
(23% - 23%)
21%
(21 % - 22%)
0%
(0% - 0%)
28%
(28% - 29%)
29%
(29% - 30%)
21%
(21% -21%)
32%
(32% - 32%)
29%
(29% - 30%)
24%
(24% - 25%)
14%
(13% -14%)
32%
(32% - 32%)
43%
(43% - 43%)
24%
(24% - 24%)
44%
(44% - 44%)
12/25
33%
(33% - 33%)
47%
(47% - 48%)
43%
(42% - 43%)
12%
(12% -12%)
58%
(58% - 58%)
60%
(59% - 60%)
32%
(31 % - 32%)
65%
(65% - 65%)
60%
(59% - 60%)
49%
(49% - 50%)
48%
(48% - 48%)
64%
(64% - 64%)
87%
(87% - 88%)
49%
(48% - 49%)
89%
(89% - 89%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at'
: 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-47
September 2009
                                                                   Draft - Do Not Quote or Cite

-------
Figure E-10.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Ischemic Heart Disease Mortality Associated with Long-Term Exposure to
(Exposure Period: 1999 - 2000): Based on 2005 Air Quality Data*
          100%
                  2005 air
                   quality
15/35**      13/35       12/35

        Alternative Standard
13/30
12/25
                       -»- Atlanta, GA 278 (228-325); 17.7% (14.5% -20.7%)
                       -•- Baltimore, MD 456 (375-534); 17.9% (14.7% -21%)
                       -*- Birmingham, AL 197 (161 -230); 16.8%  (13.8%- 19.7%)
                           Dallas, TX 283 (231 -333);  13.5% (11.1% - 15.9%)
                       -*- Detroit, Ml 605 (495-711);  14.4% (11. 8% -17%)
                       -•- Fresno, CA 84 (68-99); 7.5% (6.1% -8.9%)
                       -t- Houston, TX 505 (415-593); 16.7% (13.7% - 19.6%)
                       - Los Angeles, CA 1330 (1083-1572); 9.5% (7.7% -11. 2%)
                           New York, NY 2109 (1721 -2487); 11.4% (9.3% -13.4%)
                       -•- Philadelphia, PA 347 (284-409); 13.9% (11.4% - 16.3%)
                       -m- Phoenix, AZ 439 (358-519); 10% (8.1% -11. 8%)
                       -*- Pittsburgh, PA 275 (224-325); 9.9% (8.1% -11. 7%)
                           Salt Lake City, UT 21  (17-25); 4.4% (3.5% -5.2%)
                           St. Louis, MO 656 (538-769); 16.6% (13.7% - 19.5%)
                           Tacoma.WA 76 (62-90); 7.6% (6.1% -9%)
*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.
                                                 E-48
September 2009
                                                 Draft - Do Not Quote or Cite

-------
Figure E-ll.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Ischemic Heart Disease Mortality Associated with Long-Term Exposure to
(Exposure Period: 1999 - 2000): Based on 2006 Air Quality Data*
              100%
              80%
              60%
          •5
          re
          •D
          C
          s
           c
           0)
           3
          o
          I
 -40%
           u
           3
          •D
           0)
 -60%
 -80%
-100%
-120%
-140%
-160%
-180%
             -200%
                     2006 air    15/35**
                      quality
                                          13/35
                                                    12/35
                                                             13/30
                                                                       12/25
                                       Alternative Standard
                     -»- Atlanta, GA 269 (221 -315); 16.6% (13.6% -19.5%)
                     -m- Baltimore, MD 399 (327-468); 15.6% (12.8% -18.4%)
                     -*- Birmingham, AL 190  (156-223); 16% (13.1% -18.8%)
                         Dallas, TX 220 (179-260); 10.3% (8.4%-12.2%)
                     -*- Detroit, Ml 447 (365-528); 10.7% (8.7%-12.6%)
                     -•- Fresno, CA 91 (74-108); 8.1% (6.6%-9.6%)
                     -i— Houston, TX 534 (438-626); 17% (14%-20%)
                     	 Los Angeles,  CA  1140 (927-1349);  8.1% (6.6%-9.6%)
                         New York, NY 1785 (1454-2110); 9.5% (7.8% -11.3%)
                     -»- Philadelphia,  PA 340 (279-401); 13.6% (11.2% -16.1%)
                     -m- Phoenix, AZ 479  (391 -566); 10.5%  (8.5%-12.4%)
                     -*- Pittsburgh, PA 224 (183-266); 8.1% (6.6%-9.6%)
                         Salt Lake City, UT 14 (11 -17); 2.8%  (2.2%-3.3%)
                         St. Louis, MO 507 (415-597);  12.8% (10.5% -15.1%)
                         Tacoma.WA  43 (35-51); 4.2% (3.4%-5%)
*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.
***The percent reduction for 2006 air quality in Salt Lake City is -262%.
                                                 E-49
September 2009
                                                                  Draft - Do Not Quote or Cite

-------
Figure E-12  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Ischemic Heart Disease Mortality Associated with Long-Term Exposure to
(Exposure Period: 1999 - 2000): Based on 2007 Air Quality Data*
       I
       OT
       O
       Q

       I
s  -
       01
       O
       8.
           -80%
           100%
           120%
   -140% -
   -160%
   -180%
   -200%
                  2007 air
                   quality
                      15/35**      13/35      12/35

                              Alternative Standard
13/30
12/25
                        -»- Atlanta, GA 278 (228-326); 16.8% (13.7% -19.6%)
                        -m- Baltimore, MD 374 (306-440); 14.7% (12%-17.3%)
                        -*- Birmingham, AL 188 (154-220);  15.7% (12.9% -18.4%)
                           Dallas, TX 243 (198-286); 11.2% (9.1%-13.2%)
                        -*- Detroit, Ml 415 (338-490); 10%  (8.1%-11.8%)
                        -•- Fresno, CA 109 (89-129); 9.5% (7.7%-11.2%)
                        -i- Houston, TX 524 (430-615);  16.4% (13.5% -19.2%)
                        	 Los Angeles, CA 1173 (954-1388); 8.3% (6.7%-9.8%)
                         - New York, NY  1818 (1481 -2148); 9.6% (7.8%-11.4%)
                        -•- Philadelphia, PA 337  (276-397); 13.5% (11%-15.9%)
                        -m- Phoenix, AZ 418 (341 -495);  8.9% (7.2%-10.5%)
                        -*- Pittsburgh, PA 229 (187-271); 8.3% (6.8%-9.9%)
                           Salt Lake City, UT 27  (22-31); 5.1% (4.1%-6.1%)
                           St. Louis, MO  541 (443-637); 13.7% (11.2% -16.1%)
                           Tacoma.WA 52 (42-62); 5% (4.1%-5.9%)
*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.
                                                 E-50
September 2009
                                                                       Draft - Do Not Quote or Cite

-------
Table E-37.  Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
         Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
         on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 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 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
510
(389 - 627)
503
(384-619)
424
(324 - 520)
310
(236 - 383)
708
(541 -871)
235
(180-289)
589
(450 - 726)
2465
(1883-3033)
1908
(1454-2354)
413
(315-510)
401
(305 - 496)
621
(475 - 762)
91
(70-113)
727
(555 - 895)
108
(82-134)
15/352
453
(346 - 558)
463
(353 - 570)
305
(232 - 376)
310
(236 - 383)
514
(391 - 634)
82
(62-102)
543
(414-669)
1140
(866-1412)
1386
(1054-1714)
361
(275 - 445)
401
(305 - 496)
258
(196-319)
32
(24 - 39)
617
(471 -761)
71
(54 - 88)
13/35
357
(272 - 440)
374
(285 - 462)
239
(182-295)
310
(236 - 383)
444
(338 - 548)
82
(62-102)
425
(323 - 525)
1140
(866-1412)
1318
(1002-1631)
321
(244 - 397)
401
(305 - 496)
258
(196-319)
32
(24 - 39)
491
(374 - 606)
71
(54 - 88)
12/35
308
(234-381)
324
(247-401)
205
(156-254)
275
(209 - 340)
380
(288 - 470)
82
(62-102)
365
(278 - 452)
1008
(765-1249)
1112
(845-1377)
274
(209 - 339)
360
(273 - 446)
258
(196-319)
32
(24 - 39)
422
(321 - 522)
71
(54 - 88)
13/30
357
(272 - 440)
361
(275 - 446)
239
(182-295)
310
(236 - 383)
390
(296 - 482)
54
(41 - 67)
425
(323 - 525)
799
(606 - 990)
1007
(764-1247)
271
(206 - 336)
349
(265 - 432)
183
(138-226)
16
(12-20)
478
(364 - 590)
47
(35 - 58)
12/25
301
(229 - 372)
257
(196-318)
174
(132-215)
275
(209 - 340)
264
(200 - 327)
25
(19-31)
365
(278 - 452)
453
(343 - 562)
622
(471 - 772)
181
(137-224)
218
(165-270)
106
(80-132)
1
(1-1)
336
(255-416)
23
(17-28)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-51
September 2009
Draft - Do Not Quote or Cite

-------
Table E-38.  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 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):
Recent PM25
Concentrations
493
(377 - 607)
439
(335-541)
411
(314-505)
239
(181 -295)
541
(412-667)
249
(191 -307)
622
(475 - 767)
2217
(1692-2731)
1651
(1257-2040)
405
(309 - 500)
438
(333 - 542)
540
(413-665)
76
(58 - 94)
566
(432 - 699)
71
(54 - 87)
15/352
437
(333 - 539)
402
(306 - 496)
294
(224 - 362)
239
(181 -295)
375
(285 - 464)
90
(68-111)
574
(438 - 707)
973
(739-1206)
1167
(886-1444)
353
(269 - 436)
438
(333 - 542)
209
(159-259)
20
(16-25)
471
(359 - 582)
39
(30 - 49)
13/35
342
(260 - 422)
320
(244 - 396)
229
(174-283)
239
(181 -295)
316
(240-391)
90
(68-111)
450
(343 - 556)
973
(739-1206)
1104
(838-1367)
314
(239 - 388)
438
(333 - 542)
209
(159-259)
20
(16-25)
362
(275 - 448)
39
(30 - 49)
12/35
294
(223 - 363)
274
(208 - 339)
196
(149-243)
208
(158-257)
261
(198-324)
90
(68-111)
387
(295 - 479)
849
(644-1053)
913
(693-1131)
268
(204-331)
395
(300 - 489)
209
(159-259)
20
(16-25)
303
(230 - 375)
39
(30 - 49)
13/30
342
(260 - 422)
308
(234-381)
229
(174-283)
239
(181 -295)
270
(205 - 335)
60
(46 - 75)
450
(343 - 556)
653
(495-810)
815
(618-1011)
265
(201 - 328)
383
(291 - 474)
141
(107-175)
6
(5-8)
351
(267 - 435)
20
(15-24)
12/25
287
(218-354)
212
(161 -263)
165
(126-205)
208
(158-257)
163
(124-203)
30
(23 - 38)
387
(295 - 479)
329
(249 - 409)
459
(348 - 570)
176
(133-218)
244
(185-302)
71
(54 - 89)
0
(0-0)
229
(174-284)
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).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-52
September 2009
Draft - Do Not Quote or Cite

-------
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 5 Concentrations that Just Meet the Current and Alternative Standards, Based
         on Adjusting 2007 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 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
510
(390 - 628)
411
(313-507)
407
(31 1 - 500)
264
(201 - 326)
506
(385 - 625)
281
(216-346)
611
(466 - 753)
2265
(1729-2791)
1679
(1278-2074)
401
(306 - 495)
381
(289 - 471 )
547
(418-673)
106
(80 - 1 30)
603
(460 - 744)
81
(62-101)
15/352
452
(345 - 558)
375
(286 - 463)
290
(221 - 358)
264
(201 - 326)
347
(264 - 430)
108
(82 - 1 33)
563
(429 - 694)
1002
(761 -1241)
1188
(903-1471)
349
(266 - 431 )
381
(289 - 471 )
214
(162-265)
39
(30 - 48)
505
(384 - 623)
48
(36 - 60)
13/35
354
(270 - 438)
297
(226 - 367)
226
(172-279)
264
(201 - 326)
290
(220 - 359)
108
(82 - 1 33)
440
(335 - 543)
1002
(761 -1241)
1125
(854-1393)
310
(236 - 384)
381
(289 - 471 )
214
(162-265)
39
(30 - 48)
391
(297 - 484)
48
(36 - 60)
12/35
305
(232 - 377)
253
(192-313)
193
(147-239)
231
(176-286)
238
(180-295)
108
(82 - 1 33)
377
(287 - 467)
876
(665 - 1 086)
931
(707 - 1 1 54)
265
(201 - 327)
339
(257 - 420)
214
(162-265)
39
(30 - 48)
330
(250 - 408)
48
(36 - 60)
13/30
354
(270 - 438)
285
(217-353)
226
(172-279)
264
(201 - 326)
246
(187-305)
75
(57 - 93)
440
(335 - 543)
677
(513-840)
832
(631 -1032)
262
(199-324)
328
(249 - 406)
145
(110-180)
22
(17-27)
380
(289 - 470)
27
(20 - 33)
12/25
297
(226 - 367)
193
(147-240)
162
(123-201)
231
(176-286)
144
(109-178)
43
(32 - 53)
377
(287 - 467)
348
(263 - 432)
472
(357 - 586)
173
(131 -214)
195
(148-242)
75
(57 - 93)
5
(4-6)
253
(192-313)
5
(4-7)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-53
September 2009
Draft - Do Not Quote or Cite

-------
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 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 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):
Recent PM2.5
Concentrations
8.7%
(6. 7% -10. 7%)
8.5%
(6. 5% -10. 5%)
10.2%
(7. 8% -12. 5%)
5.9%
(4.5% - 7.2%)
8.6%
(6.6% -10.6%)
9.1%
(7% -11.2%)
7.9%
(6% - 9.7%)
8.8%
(6.7% -10.8%)
6.7%
(5.1% -8.3%)
6.9%
(5.3% - 8.5%)
4.3%
(3.2% - 5.3%)
10.2%
(7.8% -12.5%)
5.3%
(4.1% -6.6%)
8.6%
(6.5% -10.5%)
4.9%
(3.7% -6.1%)
15/352
7.8%
(5.9% - 9.6%)
7.9%
(6% - 9.7%)
7.3%
(5.6% - 9%)
5.9%
(4.5% - 7.2%)
6.3%
(4.8% - 7.7%)
3.2%
(2.4% - 3.9%)
7.3%
(5.6% - 9%)
4.1%
(3.1% -5%)
4.9%
(3.7% - 6%)
6%
(4.6% - 7.4%)
4.3%
(3.2% - 5.3%)
4.2%
(3.2% - 5.2%)
1 .8%
(1 .4% - 2.3%)
7.3%
(5.5% - 9%)
3.2%
(2.4% - 4%)
13/35
6.1%
(4.7% - 7.5%)
6.4%
(4.8% - 7.8%)
5.7%
(4.4% -7.1%)
5.9%
(4.5% - 7.2%)
5.4%
(4.1% -6. 7%)
3.2%
(2.4% - 3.9%)
5.7%
(4.3% - 7%)
4.1%
(3.1% -5%)
4.6%
(3.5% - 5.7%)
5.4%
(4.1% -6.6%)
4.3%
(3.2% - 5.3%)
4.2%
(3.2% - 5.2%)
1 .8%
(1 .4% - 2.3%)
5.8%
(4.4% -7.1%)
3.2%
(2.4% - 4%)
12/35
5.3%
(4% - 6.5%)
5.5%
(4.2% - 6.8%)
4.9%
(3. 8% -6.1%)
5.2%
(4% - 6.4%)
4.6%
(3.5% - 5.7%)
3.2%
(2.4% - 3.9%)
4.9%
(3.7% -6.1%)
3.6%
(2.7% - 4.4%)
3.9%
(3% - 4.8%)
4.6%
(3.5% - 5.7%)
3.8%
(2.9% - 4.7%)
4.2%
(3.2% - 5.2%)
1 .8%
(1 .4% - 2.3%)
5%
(3.8% -6.1%)
3.2%
(2.4% - 4%)
13/30
6.1%
(4.7% - 7.5%)
6.1%
(4.7% - 7.6%)
5.7%
(4.4% -7.1%)
5.9%
(4.5% - 7.2%)
4.7%
(3.6% - 5.9%)
2.1%
(1 .6% - 2.6%)
5.7%
(4.3% - 7%)
2.8%
(2.2% - 3.5%)
3.5%
(2.7% - 4.4%)
4.5%
(3.4% - 5.6%)
3.7%
(2.8% - 4.6%)
3%
(2.3% - 3.7%)
0.9%
(0.7% - 1 .2%)
5.6%
(4.3% - 6.9%)
2.1%
(1 .6% - 2.6%)
12/25
5.1%
(3.9% - 6.4%)
4.4%
(3.3% - 5.4%)
4.2%
(3.2% - 5.2%)
5.2%
(4% - 6.4%)
3.2%
(2.4% - 4%)
1%
(0.7% - 1 .2%)
4.9%
(3. 7% -6.1%)
1 .6%
(1 .2% - 2%)
2.2%
(1 .7% - 2.7%)
3%
(2.3% - 3.7%)
2.3%
(1 .8% - 2.9%)
1 .7%
(1 .3% - 2.2%)
0%
(0%-0.1%)
4%
(3% - 4.9%)
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., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-54
September 2009
Draft - Do Not Quote or Cite

-------
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.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
8.2%
(6.3% -10.1%)
7.4%
(5.7% - 9.2%)
9.8%
(7. 5% -12%)
4.4%
(3.4% - 5.5%)
6.6%
(5% -8.1%)
9.5%
(7. 3% -11. 7%)
8.1%
(6.2% - 9.9%)
7.8%
(6% - 9.7%)
5.8%
(4.4% -7.1%)
6.8%
(5.2% - 8.4%)
4.5%
(3.4% - 5.5%)
8.9%
(6. 8% -11%)
4.3%
(3.3% - 5.3%)
6.6%
(5.1% -8.2%)
3.1%
(2.4% - 3.9%)
15/352
7.3%
(5.5% - 8.9%)
6.8%
(5.2% - 8.4%)
7%
(5.3% - 8.6%)
4.4%
(3.4% - 5.5%)
4.6%
(3.5% - 5.7%)
3.4%
(2.6% - 4.2%)
7.4%
(5.7% - 9.2%)
3.4%
(2.6% - 4.3%)
4.1%
(3.1% -5%)
5.9%
(4.5% -7.3%)
4.5%
(3.4% - 5.5%)
3.5%
(2.6% - 4.3%)
1 .2%
(0.9% -1.4%)
5.5%
(4.2% - 6.8%)
1.8%
(1.3% -2.2%)
13/35
5.7%
(4.3% - 7%)
5.4%
(4.1% -6. 7%)
5.4%
(4.1% -6. 7%)
4.4%
(3.4% - 5.5%)
3.9%
(2.9% - 4.8%)
3.4%
(2.6% - 4.2%)
5.8%
(4.4% - 7.2%)
3.4%
(2.6% - 4.3%)
3.8%
(2.9% - 4.8%)
5.2%
(4% - 6.5%)
4.5%
(3.4% - 5.5%)
3.5%
(2.6% - 4.3%)
1.2%
(0.9% -1.4%)
4.2%
(3.2% - 5.3%)
1.8%
(1.3% -2.2%)
12/35
4.9%
(3.7% - 6%)
4.6%
(3.5% - 5.7%)
4.7%
(3.5% - 5.8%)
3.8%
(2.9% - 4.8%)
3.2%
(2.4% - 4%)
3.4%
(2.6% - 4.2%)
5%
(3.8% - 6.2%)
3%
(2.3% - 3.7%)
3.2%
(2.4% - 3.9%)
4.5%
(3.4% - 5.5%)
4%
(3.1% -5%)
3.5%
(2.6% -4. 3%)
1 .2%
(0.9% -1.4%)
3.5%
(2.7% - 4.4%)
1.8%
(1.3% -2.2%)
13/30
5.7%
(4.3% - 7%)
5.2%
(4% - 6.5%)
5.4%
(4.1% -6. 7%)
4.4%
(3.4% - 5.5%)
3.3%
(2.5% -4.1%)
2.3%
(1.7% -2. 9%)
5.8%
(4.4% - 7.2%)
2.3%
(1.8% -2.9%)
2.8%
(2.2% - 3.5%)
4.4%
(3.4% - 5.5%)
3.9%
(3% - 4.8%)
2.3%
(1.8% -2. 9%)
0.4%
(0.3% - 0.4%)
4.1%
(3.1% -5.1%)
0.9%
(0.7% -1.1%)
12/25
4.8%
(3.6% - 5.9%)
3.6%
(2.7% - 4.5%)
3.9%
(3% -4. 9%)
3.8%
(2.9% - 4.8%)
2%
(1.5% -2.5%)
1.2%
(0.9% -1.4%)
5%
(3.8% - 6.2%)
1.2%
(0.9% -1.4%)
1.6%
(1.2% -2%)
2.9%
(2.2% - 3.6%)
2.5%
(1.9% -3.1%)
1.2%
(0.9% -1.5%)
0%
(0% - 0%)
2.7%
(2% - 3.3%)
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).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-55
September 2009
Draft - Do Not Quote or Cite

-------
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.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
8.3%
(6.3% -10.2%)
7%
(5.3% - 8.6%)
9.6%
(7. 3% -11. 8%)
4.8%
(3.6% - 5.9%)
6.2%
(4.7% - 7.7%)
10.6%
(8.1% -13%)
7.8%
(5.9% - 9.6%)
8%
(6.1% -9.8%)
5.8%
(4.4% - 7.2%)
6.7%
(5.1% -8.3%)
3.8%
(2.9% - 4.7%)
9.1%
(6. 9% -11. 2%)
5.8%
(4.4% - 7.2%)
7.1%
(5.4% - 8.7%)
3.6%
(2.7% - 4.4%)
15/352
7.3%
(5.6% - 9%)
6.4%
(4.9% -7. 9%)
6.8%
(5.2% - 8.4%)
4.8%
(3.6% - 5.9%)
4.3%
(3.2% - 5.3%)
4%
(3.1% -5%)
7.2%
(5.5% - 8.8%)
3.5%
(2.7% - 4.4%)
4.1%
(3.1% -5.1%)
5.8%
(4.4% - 7.2%)
3.8%
(2.9% - 4.7%)
3.5%
(2.7% - 4.4%)
2.1%
(1.6% -2.7%)
5.9%
(4.5% - 7.3%)
2.1%
(1.6% -2.6%)
13/35
5.7%
(4.4% -7.1%)
5%
(3.8% - 6.2%)
5.3%
(4% - 6.6%)
4.8%
(3.6% - 5.9%)
3.6%
(2.7% - 4.4%)
4%
(3.1% -5%)
5.6%
(4.3% - 6.9%)
3.5%
(2.7% - 4.4%)
3.9%
(3% - 4.8%)
5.2%
(3.9% - 6.4%)
3.8%
(2.9% - 4.7%)
3.5%
(2.7% - 4.4%)
2.1%
(1.6% -2. 7%)
4.6%
(3.5% - 5.7%)
2.1%
(1.6% -2.6%)
12/35
4.9%
(3.7% -6.1%)
4.3%
(3.3% - 5.3%)
4.5%
(3.5% - 5.6%)
4.2%
(3.2% - 5.2%)
2.9%
(2.2% - 3.6%)
4%
(3.1% -5%)
4.8%
(3.6% - 5.9%)
3.1%
(2.3% - 3.8%)
3.2%
(2.4% - 4%)
4.4%
(3.4% - 5.5%)
3.4%
(2.6% - 4.2%)
3.5%
(2.7% - 4.4%)
2.1%
(1.6% -2. 7%)
3.9%
(2.9% - 4.8%)
2.1%
(1.6% -2.6%)
13/30
5.7%
(4.4% -7.1%)
4.8%
(3.7% - 6%)
5.3%
(4% - 6.6%)
4.8%
(3.6% - 5.9%)
3%
(2.3% - 3.7%)
2.8%
(2.1% -3. 5%)
5.6%
(4.3% - 6.9%)
2.4%
(1.8% -3%)
2.9%
(2.2% - 3.6%)
4.4%
(3.3% - 5.4%)
3.3%
(2.5% - 4%)
2.4%
(1.8% -3%)
1 .2%
(0.9%- 1.5%)
4.4%
(3.4% - 5.5%)
1 .2%
(0.9% -1.5%)
12/25
4.8%
(3.7% - 5.9%)
3.3%
(2. 5% -4.1%)
3.8%
(2.9% - 4.7%)
4.2%
(3.2% - 5.2%)
1.8%
(1.3% -2.2%)
1 .6%
(1.2% -2%)
4.8%
(3.6% - 5.9%)
1 .2%
(0.9% -1.5%)
1 .6%
(1.2% -2%)
2.9%
(2.2% - 3.6%)
1 .9%
(1.5% -2. 4%)
1 .2%
(0.9% -1.5%)
0.3%
(0.2% - 0.3%)
3%
(2.2% - 3.7%)
0.2%
(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., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-56
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -9%)
-39%
(-39% - -40%)
0%
(0% - 0%)
-38%
(-37% - -38%)
-186%
(-184% --188%)
-9%
(-8% - -9%)
-116%
(-11 5% --11 7%)
-38%
(-37% - -38%)
-15%
(-15% --15%)
0%
(0% - 0%)
-141%
(-139% --142%)
-190%
(-189% --191%)
-18%
(-18% --18%)
-53%
(-53% - -54%)
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%)
13/35
21%
(21% -21%)
19%
(19% -19%)
22%
(21% -22%)
0%
(0% - 0%)
14%
(14% -14%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
20%
(20% -21%)
0%
(0% - 0%)
12/35
32%
(32% - 32%)
30%
(30% - 30%)
33%
(32% - 33%)
11%
(11% -11%)
26%
(26% - 26%)
0%
(0% - 0%)
33%
(33% - 33%)
12%
(12% -12%)
20%
(20% - 20%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
32%
(31% -32%)
0%
(0% - 0%)
13/30
21%
(21% -21%)
22%
(22% - 22%)
22%
(21% -22%)
0%
(0% - 0%)
24%
(24% - 24%)
34%
(34% - 35%)
22%
(22% - 22%)
30%
(30% - 30%)
27%
(27% - 27%)
25%
(25% - 25%)
13%
(13% -13%)
29%
(29% - 29%)
49%
(49% - 49%)
23%
(22% - 23%)
34%
(34% - 34%)
12/25
34%
(33% - 34%)
44%
(44% - 45%)
43%
(43% - 43%)
11%
(11% -11%)
49%
(48% - 49%)
69%
(69% - 69%)
33%
(33% - 33%)
60%
(60% - 60%)
55%
(55% - 55%)
50%
(50% - 50%)
46%
(46% - 46%)
59%
(59% - 59%)
98%
(98% - 98%)
46%
(45% - 46%)
68%
(68% - 68%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-57
September 2009
Draft - Do Not Quote or Cite

-------
Table E-44.  Percent Reduction from the Current Standards:  Estimated Annual Incidence of Cardiopulmonary Disease Mortality
         Associated with Long-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations:
         Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Sta
Long-Term Exposure to PM2.5 Concentr
Alternative Annual (n
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -9%)
-39%
(-38% - -40%)
0%
(0% - 0%)
-43%
(-43% - -44%)
-174%
(-171% --177%)
-8%
(-8% - -8%)
-126%
(-124% --127%)
-41%
(-41% --41%)
-15%
(-14% --15%)
0%
(0% - 0%)
-155%
(-153% --157%)
-268%
(-266% - -270%)
-20%
(-20% - -20%)
-79%
(-79<>/0 . _7go/o)
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%)
ndards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with
ations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and
and Daily (m) Standards (Standard Combination Denoted n/m):
13/35
21%
(21% -22%)
20%
(20% - 20%)
22%
(22% - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
21%
(21% -21%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
31%
(31% -32%)
33%
(33% - 33%)
13%
(13% -13%)
30%
(30% - 30%)
0%
(0% - 0%)
32%
(32% - 32%)
13%
(13% -13%)
22%
(21% -22%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(35% - 36%)
0%
(0% - 0%)
13/30
21%
(21% -22%)
23%
(23% - 23%)
22%
(22% - 22%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
21%
(21% -21%)
33%
(33% - 33%)
30%
(30% - 30%)
25%
(24% - 25%)
12%
(12% -13%)
33%
(32% - 33%)
69%
(69% - 69%)
25%
(25% - 25%)
50%
(50% - 50%)
12/25
34%
(34% - 34%)
47%
(46% - 47%)
43%
(43% - 44%)
13%
(13% -13%)
56%
(56% - 56%)
66%
(66% - 66%)
32%
(32% - 32%)
66%
(66% - 66%)
60%
(60% -61%)
50%
(50% - 50%)
44%
(44% - 44%)
66%
(66% - 66%)
100%
(100% -100%)
51%
(51% -51%)
100%
(100% -100%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-58
September 2009
Draft - Do Not Quote or Cite

-------
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 Sta
Long-Term Exposure to PM2.5 Concentr
Alternative Annual (n
Recent PM2.5
Concentrations
-13%
(-13% --13%)
-10%
(-9% --10%)
-40%
(-40% --41%)
0%
(0% - 0%)
-46%
(-45% - -46%)
-161%
(-159% --164%)
-9%
(-9% - -9%)
-126%
(-125% --127%)
-41%
(-41% --42%)
-15%
(-15% --15%)
0%
(0% - 0%)
-156%
(-154% --157%)
-170%
(-169% --172%)
-20%
(-19% --20%)
-70%
(-70% - -70%)
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%)
ndards: Annual Incidence of Cardiopulmonary Disease Mortality Associated with
ations in a Recent Year and PM2 5 Concentrations that Just Meet the Current and
and Daily (m) Standards (Standard Combination Denoted n/m):
13/35
22%
(22% - 22%)
21%
(21% -21%)
22%
(22% - 22%)
0%
(0% - 0%)
16%
(16% -17%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
22%
(22% - 23%)
0%
(0% - 0%)
12/35
33%
(32% - 33%)
33%
(33% - 33%)
33%
(33% - 34%)
12%
(12% -12%)
32%
(31% -32%)
0%
(0% - 0%)
33%
(33% - 33%)
13%
(13% -13%)
22%
(22% - 22%)
24%
(24% - 24%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(34% - 35%)
0%
(0% - 0%)
13/30
22%
(22% - 22%)
24%
(24% - 24%)
22%
(22% - 22%)
0%
(0% - 0%)
29%
(29% - 29%)
30%
(30% - 30%)
22%
(22% - 22%)
32%
(32% - 33%)
30%
(30% - 30%)
25%
(25% - 25%)
14%
(14% -14%)
32%
(32% - 32%)
44%
(44% - 44%)
25%
(25% - 25%)
44%
(44% - 44%)
12/25
34%
(34% - 35%)
48%
(48% - 49%)
44%
(44% - 44%)
12%
(12% -12%)
59%
(58% - 59%)
60%
(60% -61%)
33%
(33% - 33%)
65%
(65% - 65%)
60%
(60% - 60%)
50%
(50% -51%)
49%
(49% - 49%)
65%
(65% - 65%)
88%
(88% - 88%)
50%
(50% - 50%)
89%
(89% - 89%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-59
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-13.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiopulmonary Disease Mortality Associated with Long-Term Exposure to
PM2.5 (Exposure Period: 1979 - 1983):  Based on 2005 Air Quality Data*
          100%
          -200%
                  2005 air
                   quality
15/35**      13/35      12/35

        Alternative Standard
13/30
12/25
                        -•- Atlanta, GA 453 (346 - 558); 7.8% (5.9% - 9.6%)
                        -•- Baltimore, MD 463 (353-570); 7.9% (6% -9.7%)
                        -*- Birmingham, AL 305 (232-376); 7.3% (5.6% -9%)
                            Dallas, TX 310 (236-383); 5.9% (4.5% -7.2%)
                        -*- Detroit, Ml 514 (391 -634); 6.3% (4.8% -7.7%)
                        -•- Fresno, CA 82 (62-102); 3.2% (2.4% -3.9%)
                        -i- Houston, TX 543 (414-669); 7.3% (5.6% -9%)
                        - Los Angeles, CA  1140 (866-1412); 4.1% (3.1% -5%)
                            New York, NY 1386 (1054-1714); 4.9% (3.7% -6%)
                        -+- Philadelphia, PA 361 (275 - 445); 6% (4.6% - 7.4%)
                        -m- Phoenix, AZ 401  (305 - 496); 4.3% (3.2% - 5.3%)
                        -*- Pittsburgh, PA 258 (196-319); 4.2% (3.2% -5.2%)
                            Salt Lake City, UT 32 (24 - 39); 1 .8% (1 .4% - 2.3%)
                            St. Louis, MO 617 (471 -761); 7.3% (5.5% -9%)
                        —- Tacoma, WA  71 (54 - 88); 3.2% (2.4% - 4%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                                E-60
September 2009
                                                 Draft - Do Not Quote or Cite

-------
Figure E-14.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiopulmonary Disease Mortality Associated with Long-Term Exposure to
PM2.5 (Exposure Period:  1979 - 1983): Based on 2006 Air Quality Data*
       re
       •D
       C
       s
       v>
       •£
       I
       o
       I
       I
       
-------
Figure E-15.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiopulmonary Disease Mortality Associated with Long-Term Exposure to
PM2.5 (Exposure Period: 1979 - 1983): Based on 2007 Air Quality Data*
          100%
          -200%
                  2007 air
                   quality
15/35**      13/35       12/35

        Alternative Standard
13/30
12/25
                        -•- Atlanta, GA 452 (345 - 558); 7.3% (5.6% - 9%)
                        -m- Baltimore, MD 375 (286 - 463); 6.4% (4.9% - 7.9%)
                        -*- Birmingham, AL 290 (221 - 358); 6.8%  (5.2% - 8.4%)
                            Dallas, TX 264 (201 - 326); 4.8% (3.6% - 5.9%)
                        -*- Detroit, Ml 347 (264 - 430); 4.3% (3.2% - 5.3%)
                        -•- Fresno, CA 108 (82-133); 4% (3.1% -5%)
                        -i- Houston, TX 563 (429 - 694); 7.2% (5.5% - 8.8%)
                        - Los Angeles, CA 1002 (761 -1241); 3.5% (2.7% -4.4%)
                          - New York, NY 1188 (903-1471); 4.1%  (3.1% -5.1%)
                        -•- Philadelphia, PA 349 (266-431); 5.8% (4.4% -7.2%)
                        -m- Phoenix, AZ 381 (289-471); 3.8% (2.9% -4.7%)
                        -*- Pittsburgh, PA 214 (162-265); 3.5% (2.7% -4.4%)
                            Salt Lake City, UT 39 (30-48); 2.1% (1.6% -2.7%)
                            St. Louis, MO  505 (384 - 623); 5.9% (4.5% - 7.3%)
                            Tacoma.WA 48 (36-60); 2.1% (1.6%  -2.6%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                                 E-62
September 2009
                                                  Draft - Do Not Quote or Cite

-------
Table E-46.  Estimated Annual Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to Ambient PM2 5
         Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
         on Adjusting 2005 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2.5 from 1999 - 20001
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Cardiopulmonary Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent
Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
719
(566 - 868)
710
(558 - 857)
596
(470-717)
440
(345 - 534)
999
(786-1205)
331
(261 - 399)
833
(655-1007)
3477
(2737-4196)
2704
(2121 -3274)
586
(459 - 709)
571
(446 - 695)
873
(689-1051)
130
(102- 158)
1026
(807- 1238)
154
(121 -187)
15/352
640
(503 - 774)
654
(514-790)
431
(339 - 522)
440
(345 - 534)
729
(571 -883)
117
(92-143)
768
(603 - 929)
1626
(1270-1977)
1972
(1542-2395)
512
(401 -620)
571
(446 - 695)
368
(287 - 447)
45
(35 - 55)
874
(686- 1057)
101
(79-123)
13/35
506
(397-614)
531
(416-643)
339
(266-411)
440
(345 - 534)
631
(494 - 765)
117
(92- 143)
604
(473 - 732)
1626
(1270-1977)
1878
(1468-2281)
457
(357 - 554)
571
(446 - 695)
368
(287 - 447)
45
(35 - 55)
697
(546 - 845)
101
(79-123)
12/35
438
(343 - 532)
461
(361 - 559)
292
(228 - 355)
391
(306 - 475)
541
(423 - 657)
117
(92- 143)
520
(406-631)
1439
(1123-1752)
1586
(1239-1930)
391
(305 - 475)
514
(401 -625)
368
(287 - 447)
45
(35-55)
601
(470 - 729)
101
(79-123)
13/30
506
(397-614)
512
(401 -621)
339
(266-411)
440
(345 - 534)
555
(434 - 674)
77
(60 - 94)
604
(473 - 732)
1142
(890- 1392)
1437
(1121 -1749)
387
(302 - 470)
498
(389 - 606)
261
(203-318)
23
(18-28)
679
(532 - 823)
67
(52 - 82)
12/25
428
(335-519)
367
(287 - 446)
248
(194-301)
391
(306 - 475)
377
(294 - 459)
36
(28 - 44)
520
(406-631)
649
(505 - 792)
891
(694-1087)
259
(202-315)
312
(243-381)
152
(118-186)
1
(1-1)
479
(374 - 583)
33
(25 - 40)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-63
September 2009
Draft - Do Not Quote or Cite

-------
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 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
697
(548-841)
621
(488-751)
578
(456 - 696)
340
(266-413)
767
(601 - 929)
351
(277 - 423)
879
(691 -1062)
3134
(2463 - 3788)
2346
(1837-2845)
574
(450-695)
624
(488-758)
762
(600-919)
108
(85-132)
803
(630 - 972)
101
(79-123)
15/352
619
(486 - 748)
569
(447 - 689)
416
(326 - 503)
340
(266-413)
535
(418-650)
128
(100-156)
812
(637 - 982)
1390
(1084-1692)
1663
(1299-2023)
501
(393 - 608)
624
(488 - 758)
299
(233 - 364)
29
(23 - 36)
670
(524-813)
56
(44 - 69)
13/35
486
(380 - 589)
455
(356 - 552)
325
(255 - 395)
340
(266-413)
451
(352 - 548)
128
(100-156)
639
(501 -775)
1390
(1084-1692)
1575
(1230-1916)
447
(349 - 542)
624
(488 - 758)
299
(233 - 364)
29
(23 - 36)
516
(403 - 628)
56
(44 - 69)
12/35
418
(327 - 508)
390
(305 - 474)
279
(218-339)
297
(232-361)
373
(291 -455)
128
(100-156)
551
(431 -669)
1214
(946-1479)
1304
(1017-1588)
382
(298 - 464)
563
(440 - 685)
299
(233 - 364)
29
(23 - 36)
433
(338 - 526)
56
(44 - 69)
13/30
486
(380 - 589)
438
(343 - 532)
325
(255 - 395)
340
(266-413)
386
(301 - 470)
86
(67- 105)
639
(501 - 775)
935
(728-1140)
1165
(908-1420)
378
(295-459)
546
(426 - 664)
202
(157-246)
9
(7-11)
500
(391 -609)
28
(22 - 34)
12/25
408
(319-496)
303
(237 - 369)
236
(184-287)
297
(232-361)
234
(182-286)
43
(34 - 53)
551
(431 - 669)
473
(367 - 577)
658
(512-804)
251
(196-306)
349
(271 - 425)
103
(80-125)
0
(0-0)
327
(255 - 399)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-64
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
721
(567 - 870)
582
(457 - 705)
573
(451 -690)
376
(294 - 456)
718
(563 - 870)
395
(312-476)
864
(679- 1044)
3202
(2517-3869)
2385
(1868-2891)
568
(446 - 688)
543
(424-661)
771
(607 - 930)
150
(118-182)
854
(670-1034)
116
(91 -142)
15/352
640
(503 - 774)
533
(418-645)
411
(322 - 498)
376
(294 - 456)
495
(387 - 602)
154
(120-187)
797
(625 - 964)
1431
(1116-1741)
1694
(1323-2060)
496
(388-601)
543
(424-661)
305
(238 - 372)
56
(44 - 68)
717
(561 - 869)
69
(53 - 84)
13/35
503
(394-610)
423
(331 -513)
321
(251 -389)
376
(294 - 456)
414
(323 - 504)
154
(120-187)
625
(489 - 758)
1431
(1116-1741)
1605
(1253-1952)
441
(345 - 536)
543
(424-661)
305
(238 - 372)
56
(44 - 68)
557
(436 - 677)
69
(53 - 84)
12/35
434
(339 - 527)
360
(281 - 438)
275
(215-334)
330
(258-401)
340
(265-414)
154
(120- 187)
537
(420 - 652)
1252
(976-1525)
1331
(1038-1620)
377
(295-458)
485
(378 - 590)
305
(238 - 372)
56
(44 - 68)
471
(367 - 572)
69
(53 - 84)
13/30
503
(394-610)
406
(317-493)
321
(251 - 389)
376
(294 - 456)
352
(274 - 429)
108
(84- 131)
625
(489 - 758)
969
(755-1181)
1190
(928-1450)
373
(291 - 453)
469
(365-571)
208
(162-253)
32
(25 - 39)
541
(423 - 657)
38
(30 - 47)
12/25
423
(331 -514)
276
(216-337)
232
(181 -282)
330
(258-401)
206
(160-252)
61
(48 - 75)
537
(420 - 652)
499
(388-610)
677
(526 - 826)
247
(193-301)
279
(217-341)
108
(84-132)
7
(5-8)
361
(282 - 440)
8
(6-9)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-65
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2 5
Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
12.3%
(9.7% -14.9%)
12.1%
(9. 5% -14. 6%)
14.3%
(11. 3% -17. 3%)
8.3%
(6.5% -10.1%)
12.2%
(9. 6% -14. 7%)
12.8%
(10.1% -15.5%)
11.2%
(8.8% -13.5%)
12.4%
(9.7% -14.9%)
9.5%
(7.5% -11. 5%)
9.8%
(7.7% - 1 1 .8%)
6.1%
(4.7% - 7.4%)
14.3%
(11. 3% -17. 2%)
7.6%
(5.9% - 9.2%)
12.1%
(9.5% -14.6%)
7%
(5.5% - 8.5%)
15/352
11%
(8.6% -13.2%)
11.1%
(8.7% -13. 4%)
10.4%
(8.2% -12.6%)
8.3%
(6.5% -10.1%)
8.9%
(7% -10. 8%)
4.5%
(3.5% - 5.5%)
10.3%
(8.1% -12.5%)
5.8%
(4. 5% -7%)
6.9%
(5.4% - 8.4%)
8.5%
(6.7% -10. 3%)
6.1%
(4.7% - 7.4%)
6%
(4.7% - 7.3%)
2.6%
(2% - 3.2%)
10.3%
(8.1% -12.4%)
4.6%
(3.6% - 5.6%)
13/35
8.7%
(6.8% -10.5%)
9%
(7.1% -10.9%)
8.2%
(6.4% - 9.9%)
8.3%
(6.5% -10.1%)
7.7%
(6% - 9.3%)
4.5%
(3.5% - 5.5%)
8.1%
(6.3% - 9.8%)
5.8%
(4.5% - 7%)
6.6%
(5.2% - 8%)
7.6%
(6% - 9.2%)
6.1%
(4.7% - 7.4%)
6%
(4.7% - 7.3%)
2.6%
(2% - 3.2%)
8.2%
(6.4% -10%)
4.6%
(3.6% - 5.6%)
12/35
7.5%
(5.9% -9.1%)
7.8%
(6.1% -9. 5%)
7%
(5.5% - 8.5%)
7.4%
(5.8% - 9%)
6.6%
(5.1% -8%)
4.5%
(3.5% - 5.5%)
7%
(5.5% - 8.5%)
5.1%
(4% - 6.2%)
5.6%
(4.4% - 6.8%)
6.5%
(5.1% -7. 9%)
5.5%
(4.3% -6. 6%)
6%
(4.7% -7. 3%)
2.6%
(2% - 3.2%)
7.1%
(5.5% - 8.6%)
4.6%
(3.6% - 5.6%)
13/30
8.7%
(6.8% -10.5%)
8.7%
(6. 8% -10. 5%)
8.2%
(6.4% - 9.9%)
8.3%
(6.5% -10.1%)
6.8%
(5.3% - 8.2%)
3%
(2.3% - 3.6%)
8.1%
(6.3% - 9.8%)
4.1%
(3.2% - 5%)
5.1%
(3.9% - 6.2%)
6.5%
(5% - 7.8%)
5.3%
(4.1% -6. 4%)
4.3%
(3.3% - 5.2%)
1.4%
(1.1% -1.7%)
8%
(6.3% - 9.7%)
3%
(2.4% - 3.7%)
12/25
7.3%
(5.7% - 8.9%)
6.2%
(4.9% - 7.6%)
6%
(4.7% - 7.3%)
7.4%
(5.8% - 9%)
4.6%
(3.6% - 5.6%)
1 .4%
(1.1% -1.7%)
7%
(5.5% - 8.5%)
2.3%
(1.8% -2.8%)
3.1%
(2.4% - 3.8%)
4.3%
(3.4% - 5.3%)
3.3%
(2.6% - 4%)
2.5%
(1.9% -3%)
0.1%
(0%-0.1%)
5.6%
(4.4% - 6.9%)
1.5%
(1.1% -1.8%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-66
September 2009
Draft - Do Not Quote or Cite

-------
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 Exposure to PM2 5
Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
11.6%
(9.1% -14%)
10.5%
(8.3% -12. 7%)
13.8%
(10.8% -16.6%)
6.3%
(4.9% - 7.6%)
9.4%
(7. 3% -11. 3%)
13.4%
(10.6% -16.2%)
11.4%
(9% -13.8%)
11.1%
(8.7% -13.4%)
8.2%
(6.4% - 9.9%)
9.6%
(7.5% -11. 6%)
6.4%
(5% -7. 8%)
12.6%
(9. 9% -15. 2%)
6.1%
(4.8% - 7.4%)
9.4%
(7.4% - 1 1 .4%)
4.5%
(3.5% - 5.5%)
15/352
10.3%
(8.1% -12.4%)
9.7%
(7.6% - 1 1 .7%)
9.9%
(7.8% -12%)
6.3%
(4.9% -7.6%)
6.5%
(5.1% -7. 9%)
4.9%
(3.8% - 6%)
10.5%
(8.3% -12.7%)
4.9%
(3.8% - 6%)
5.8%
(4.5% - 7%)
8.4%
(6.6% -10. 2%)
6.4%
(5% - 7.8%)
4.9%
(3. 8% -6%)
1 .7%
(1.3% -2%)
7.8%
(6.1% -9.5%)
2.5%
(2% -3.1%)
13/35
8.1%
(6.3% - 9.8%)
7.7%
(6% - 9.4%)
7.7%
(6.1% -9.4%)
6.3%
(4.9% -7.6%)
5.5%
(4.3% - 6.7%)
4.9%
(3.8% - 6%)
8.3%
(6.5% -10.1%)
4.9%
(3.8% - 6%)
5.5%
(4.3% -6.7%)
7.5%
(5. 8% -9.1%)
6.4%
(5% - 7.8%)
4.9%
(3.8% - 6%)
1.7%
(1.3% -2%)
6.1%
(4.7% - 7.4%)
2.5%
(2% -3.1%)
12/35
6.9%
(5.4% - 8.4%)
6.6%
(5.2% - 8%)
6.6%
(5.2% -8.1%)
5.5%
(4.3% - 6.7%)
4.6%
(3.6% - 5.5%)
4.9%
(3.8% - 6%)
7.2%
(5.6% - 8.7%)
4.3%
(3.3% - 5.2%)
4.5%
(3.5% - 5.5%)
6.4%
(5% - 7.8%)
5.8%
(4.5% - 7%)
4.9%
(3.8% - 6%)
1.7%
(1.3% -2%)
5.1%
(4% - 6.2%)
2.5%
(2% -3.1%)
13/30
8.1%
(6.3% - 9.8%)
7.4%
(5.8% - 9%)
7.7%
(6.1% -9.4%)
6.3%
(4.9% - 7.6%)
4.7%
(3.7% - 5.7%)
3.3%
(2.6% - 4%)
8.3%
(6.5% -10.1%)
3.3%
(2.6% - 4%)
4.1%
(3.2% - 4.9%)
6.3%
(4.9% - 7.7%)
5.6%
(4.4% - 6.8%)
3.3%
(2. 6% -4.1%)
0.5%
(0.4% - 0.6%)
5.9%
(4.6% -7.1%)
1.2%
(1%-1.5%)
12/25
6.8%
(5.3% - 8.2%)
5.1%
(4% - 6.3%)
5.6%
(4.4% - 6.8%)
5.5%
(4.3% - 6.7%)
2.9%
(2.2% - 3.5%)
1 .7%
(1.3% -2%)
7.2%
(5.6% - 8.7%)
1 .7%
(1.3% -2%)
2.3%
(1.8% -2. 8%)
4.2%
(3. 3% -5.1%)
3.6%
(2.8% -4. 3%)
1 .7%
(1.3% -2.1%)
0%
(0% - 0%)
3.8%
(3% - 4.7%)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-67
September 2009
Draft - Do Not Quote or Cite

-------
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 - 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 PM2 5
Concentrations in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
11.7%
(9.2% -14.1%)
9.9%
(7. 8% -12%)
13.5%
(10.6% -16.3%)
6.8%
(5.3% - 8.3%)
8.8%
(6. 9% -10. 7%)
14.8%
(11. 7% -17.8%)
11%
(8.6% -13.3%)
11.2%
(8.8% -13.6%)
8.2%
(6.5% -10%)
9.5%
(7. 4% -11. 5%)
5.4%
(4.2% - 6.6%)
12.8%
(10.1% -15.4%)
8.2%
(6.5% -10%)
10%
(7.8% -12.1%)
5.1%
(4% - 6.2%)
15/352
10.4%
(8.1% -12.5%)
9.1%
(7.1% -11%)
9.7%
(7.6% - 1 1 .7%)
6.8%
(5.3% - 8.3%)
6.1%
(4.7% - 7.4%)
5.8%
(4.5% - 7%)
10.1%
(8% -12.3%)
5%
(3.9% -6.1%)
5.9%
(4.6% -7.1%)
8.3%
(6. 5% -10%)
5.4%
(4.2% - 6.6%)
5.1%
(4% - 6.2%)
3.1%
(2.4% - 3.7%)
8.4%
(6.6% -10.2%)
3%
(2.3% - 3.7%)
13/35
8.1%
(6.4% - 9.9%)
7.2%
(5.6% - 8.7%)
7.6%
(5.9% - 9.2%)
6.8%
(5.3% - 8.3%)
5.1%
(4% - 6.2%)
5.8%
(4.5% - 7%)
7.9%
(6.2% - 9.6%)
5%
(3.9% -6.1%)
5.5%
(4.3% - 6.8%)
7.4%
(5.8% - 9%)
5.4%
(4.2% - 6.6%)
5.1%
(4% - 6.2%)
3.1%
(2.4% - 3.7%)
6.5%
(5.1% -7.9%)
3%
(2.3% - 3.7%)
12/35
7%
(5.5% - 8.5%)
6.1%
(4.8% - 7.4%)
6.5%
(5.1% -7.9%)
6%
(4.7% - 7.3%)
4.2%
(3. 3% -5.1%)
5.8%
(4.5% - 7%)
6.8%
(5. 3% -8. 3%)
4.4%
(3.4% - 5.4%)
4.6%
(3.6% - 5.6%)
6.3%
(4.9% - 7.7%)
4.8%
(3.8% - 5.9%)
5.1%
(4% - 6.2%)
3.1%
(2.4% - 3.7%)
5.5%
(4.3% - 6.7%)
3%
(2.3% - 3.7%)
13/30
8.1%
(6.4% - 9.9%)
6.9%
(5.4% - 8.4%)
7.6%
(5.9% - 9.2%)
6.8%
(5.3% - 8.3%)
4.3%
(3.4% - 5.3%)
4%
(3.2% - 4.9%)
7.9%
(6.2% - 9.6%)
3.4%
(2.7% - 4.2%)
4.1%
(3.2% - 5%)
6.2%
(4.9% - 7.6%)
4.6%
(3.6% - 5.7%)
3.4%
(2.7% - 4.2%)
1.7%
(1.3% -2.1%)
6.3%
(4.9% - 7.7%)
1.7%
(1.3% -2.1%)
12/25
6.8%
(5.4% - 8.3%)
4.7%
(3.7% - 5.7%)
5.5%
(4.3% -6.6%)
6%
(4.7% -7.3%)
2.5%
(2% -3.1%)
2.3%
(1.8% -2.8%)
6.8%
(5.3% - 8.3%)
1.8%
(1.4% -2.1%)
2.3%
(1.8% -2.9%)
4.1%
(3.2% - 5%)
2.8%
(2.2% - 3.4%)
1.8%
(1.4% -2. 2%)
0.4%
(0.3% - 0.5%)
4.2%
(3.3% - 5.2%)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-68
September 2009
Draft - Do Not Quote or Cite

-------
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 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
-12%
(-12% --12%)
-9%
(-80/0 . -go/0)
-38%
(-38% - -39%)
0%
(0% - 0%)
-37%
(-36% - -38%)
-182%
(-179% --185%)
-8%
(-80/0 - -go/0)
-114%
(-11 2% --11 6%)
-37%
(-37% - -38%)
-14%
(-14% --15%)
0%
(0% - 0%)
-137%
(-135% --140%)
-188%
(-186% --189%)
-17%
(-17% --18%)
-53%
(-52% - -53%)
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%)
13/35
21%
(21% -21%)
19%
(19% -19%)
21%
(21% -22%)
0%
(0% - 0%)
13%
(13% -14%)
0%
(0% - 0%)
21%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
20%
(20% - 20%)
0%
(0% - 0%)
12/35
32%
(31% -32%)
30%
(29% - 30%)
32%
(32% - 33%)
11%
(11% -11%)
26%
(26% - 26%)
0%
(0% - 0%)
32%
(32% - 33%)
11%
(11% -12%)
20%
(19% -20%)
24%
(23% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
31%
(31% -32%)
0%
(0% - 0%)
13/30
21%
(21% -21%)
22%
(21% -22%)
21%
(21% -22%)
0%
(0% - 0%)
24%
(24% - 24%)
34%
(34% - 34%)
21%
(21% -22%)
30%
(30% - 30%)
27%
(27% - 27%)
24%
(24% - 25%)
13%
(13% -13%)
29%
(29% - 29%)
48%
(48% - 49%)
22%
(22% - 23%)
34%
(33% - 34%)
12/25
33%
(33% - 33%)
44%
(44% - 44%)
42%
(42% - 43%)
11%
(11% -11%)
48%
(48% - 49%)
69%
(69% - 69%)
32%
(32% - 33%)
60%
(60% - 60%)
55%
(55% - 55%)
49%
(49% - 50%)
45%
(45% - 46%)
59%
(58% - 59%)
98%
(98% - 98%)
45%
(45% - 45%)
68%
(68% - 68%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-69
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -9%)
-39%
(-38% - -40%)
0%
(0% - 0%)
-43%
(-43% - -44%)
-174%
(-171% --177%)
-8%
(-8% - -8%)
-126%
(-124% --127%)
-41%
(-41% --41%)
-15%
(-14% --15%)
0%
(0% - 0%)
-155%
(-153% --157%)
-268%
(-266% - -270%)
-20%
(-20% - -20%)
-79%
(-79<>/0 . _7go/o)
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%)
13/35
21%
(21% -22%)
20%
(20% - 20%)
22%
(22% - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
21%
(21% -21%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
31%
(31% -32%)
33%
(33% - 33%)
13%
(13% -13%)
30%
(30% - 30%)
0%
(0% - 0%)
32%
(32% - 32%)
13%
(13% -13%)
22%
(21% -22%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(35% - 36%)
0%
(0% - 0%)
13/30
21%
(21% -22%)
23%
(23% - 23%)
22%
(22% - 22%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
21%
(21% -21%)
33%
(33% - 33%)
30%
(30% - 30%)
25%
(24% - 25%)
12%
(12% -13%)
33%
(32% - 33%)
69%
(69% - 69%)
25%
(25% - 25%)
50%
(50% - 50%)
12/25
34%
(34% - 34%)
47%
(46% - 47%)
43%
(43% - 44%)
13%
(13% -13%)
56%
(56% - 56%)
66%
(66% - 66%)
32%
(32% - 32%)
66%
(66% - 66%)
60%
(60% -61%)
50%
(50% - 50%)
44%
(44% - 44%)
66%
(66% - 66%)
100%
(100% -100%)
51%
(51% -51%)
100%
(100% -100%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-70
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -9%)
-39%
(-39% - -40%)
0%
(0% - 0%)
-45%
(-45% - -46%)
-157%
(-155% --160%)
-8%
(-80/0 . -go/0)
-124%
(-122% --125%)
-41%
(-40% --41%)
-15%
(-14% --15%)
0%
(0% - 0%)
-152%
(-150% --155%)
-168%
(-167% --170%)
-19%
(-19% --19%)
-69%
(-69% - -70%)
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%)
13/35
21%
(21% -22%)
21%
(20% -21%)
22%
(22% - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
22%
(22% - 22%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
32%
(32% - 33%)
33%
(33% - 33%)
12%
(12% -12%)
31%
(31% -31%)
0%
(0% - 0%)
33%
(32% - 33%)
12%
(12% -13%)
21%
(21% -22%)
24%
(24% - 24%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
34%
(34% - 35%)
0%
(0% - 0%)
13/30
21%
(21% -22%)
24%
(24% - 24%)
22%
(22% - 22%)
0%
(0% - 0%)
29%
(29% - 29%)
30%
(30% - 30%)
22%
(21% -22%)
32%
(32% - 32%)
30%
(30% - 30%)
25%
(25% - 25%)
14%
(14% -14%)
32%
(32% - 32%)
44%
(43% - 44%)
25%
(24% - 25%)
44%
(44% - 44%)
12/25
34%
(34% - 34%)
48%
(48% - 48%)
44%
(43% - 44%)
12%
(12% -12%)
58%
(58% - 59%)
60%
(60% - 60%)
33%
(32% - 33%)
65%
(65% - 65%)
60%
(60% - 60%)
50%
(50% - 50%)
49%
(48% - 49%)
65%
(65% - 65%)
88%
(88% - 88%)
50%
(49% - 50%)
89%
(89% - 89%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-71
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-16.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiopulmonary Disease Mortality Associated with Long-Term Exposure to
PM2.5 (Exposure Period: 1999 - 2000):  Based on 2005 Air Quality Data*
          100%
          -200%
                  2005 air
                   quality
15/35**      13/35      12/35

        Alternative Standard
13/30
12/25
                        -.-Atlanta, GA 640 (503-774);  11% (8.6% -13.2%)
                        -•- Baltimore, MD 654 (514-790); 11.1% (8.7% -13.4%)
                        -»- Birmingham, AL 431 (339-522); 10.4%  (8.2% -12.6%)
                            Dallas, TX 440 (345-534); 8.3% (6.5% -10.1%)
                        -*- Detroit, Ml 729 (571 -883); 8.9% (7% -10.8%)
                        -•- Fresno, CA 117 (92-143); 4.5% (3.5% -5.5%)
                        -i- Houston, TX 768 (603-929); 10.3% (8.1% -12.5%)
                        - Los Angeles, CA 1626 (1270-1977); 5.8% (4.5% -7%)
                          - New York, NY 1972 (1542-2395); 6.9%  (5.4% -8.4%)
                        -•- Philadelphia, PA 512 (401 -620); 8.5% (6.7% -10.3%)
                        -m- Phoenix, AZ 571  (446-695); 6.1% (4.7% -7.4%)
                        -*- Pittsburgh, PA 368 (287 - 447); 6% (4.7% - 7.3%)
                            Salt Lake City, UT 45 (35 - 55); 2.6% (2% - 3.2%)
                            St. Louis, MO 874 (686-1057); 10.3% (8.1% -12.4%)
                        -^- Tacoma.WA 101 (79-123);  4.6%  (3.6% -5.6%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                                 E-72
September 2009
                                                 Draft - Do Not Quote or Cite

-------
Figure E-17.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiopulmonary Disease Mortality Associated with Long-Term Exposure to
PM2.5 (Exposure Period:  1999 - 2000): Based on 2006 Air Quality Data*
          100%
           80%
       re
       •D
       C
       s
       v>
       •H
       I
       o
       E
       I
       
-------
Figure E-18  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiopulmonary Disease Mortality Associated with Long-Term Exposure to
PM2.5 (Exposure Period: 1999 - 2000):  Based on 2007 Air Quality Data*
       S
       OT
          100%
           80%
           60%
           40%
       O
       Q

       I
 -40%
       s  ,
       01
       o
       8.
 -60%
 -80%
 100%
-120%
-140%
-160%
-180%
-200%
                  2007 air
                   quality
                   15/35**      13/35       12/35

                           Alternative Standard
13/30
12/25
                        -.-Atlanta, GA 640 (503-774); 10.4% (8.1%-12.5%)
                        -•- Baltimore, MD 533 (418-645); 9.1% (7.1%-11%)
                        -*- Birmingham, AL 411 (322-498);  9.7% (7.6%-11.7%)
                            Dallas, TX 376 (294-456); 6.8% (5.3%-8.3%)
                        -*- Detroit, Ml 495 (387-602); 6.1% (4.7%-7.4%)
                        -•- Fresno, CA 154 (120-187); 5.8% (4.5%-7%)
                        -i- Houston, TX 797 (625-964); 10.1% (8%-12.3%)
                        	 Los Angeles, CA 1431  (1116-1741); 5%  (3.9%-6.1%)
                          - New York, NY  1694 (1323-2060); 5.9% (4.6%-7.1%)
                        -•- Philadelphia, PA 496 (388-601); 8.3% (6.5%-10%)
                        -m- Phoenix, AZ 543 (424-661); 5.4%  (4.2%-6.6%)
                        -*- Pittsburgh, PA 305 (238-372); 5.1% (4%-6.2%)
                            Salt Lake City, UT 56 (44-68); 3.1% (2.4%-3.7%)
                            St. Louis, MO  717 (561 -869); 8.4% (6.6%-10.2%)
                        -^- Tacoma, WA 69 (53 - 84); 3% (2.3% - 3.7%)
*Based on Krewski et al. (2009), exposure period from 1979 - 1983. 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.
                                                E-74
September 2009
                                                                    Draft - Do Not Quote or Cite

-------
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 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
76
(29-121)
80
(31 - 127)
64
(24-101)
48
(18-77)
106
(40-168)
26
(10-41)
90
(34-144)
259
(99-411)
180
(68 - 287)
66
(25-105)
58
(22 - 93)
94
(36-148)
9
(4-15)
111
(42-176)
18
(7 - 29)
15/352
68
(26-108)
74
(28-117)
46
(17-73)
48
(18-77)
77
(29-123)
9
(3-14)
83
(32-132)
120
(45-193)
131
(49-210)
57
(22 - 92)
58
(22 - 93)
39
(15-63)
3
(1-5)
94
(36-150)
12
(4-19)
13/35
53
(20 - 85)
60
(23 - 95)
36
(14-58)
48
(18-77)
67
(25-107)
9
(3-14)
65
(25-104)
120
(45-193)
124
(47 - 200)
51
(19-82)
58
(22 - 93)
39
(15-63)
3
(1-5)
75
(28-120)
12
(4-19)
12/35
46
(17-74)
52
(20 - 83)
31
(12-50)
43
(16-68)
57
(21 -91)
9
(3-14)
56
(21 - 90)
106
(40-171)
105
(39-169)
44
(16-70)
52
(19-83)
39
(15-63)
3
(1-5)
64
(24-103)
12
(4-19)
13/30
53
(20 - 85)
58
(22 - 92)
36
(14-58)
48
(18-77)
59
(22 - 94)
6
(2-10)
65
(25-104)
84
(31 -136)
95
(36-153)
43
(16-69)
50
(19-81)
28
(10-45)
2
(1-3)
73
(28-117)
8
(3-13)
12/25
45
(17-72)
41
(15-66)
26
(10-42)
43
(16-68)
40
(15-64)
3
(1-4)
56
(21 - 90)
48
(18-77)
59
(22 - 95)
29
(11 -46)
31
(12-51)
16
(6 - 26)
0
(0-0)
51
(19-83)
4
(1-6)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 E-75
September 2009
Draft - Do Not Quote or Cite

-------
Table E-56.  Estimated Annual Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to Ambient PM2 5
         Concentrations in a Recent Year (2006) and PM2 5 Concentrations that Just Meet the Current and Alternative Standards, Based
         on Adjusting 2006 PM2 5 Concentrations: Estimates Based on Krewski et al. (2009), Using Ambient PM2 5 from 1979 -19831
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Incidence of Lung Cancer Mortality Associated with Long-Term Exposure to PM2.5 Concentrations in a Recent Year
and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards (Standard
Combination Denoted n/m):
Recent PM2.5
Concentrations
74
(28-117)
70
(27-111)
62
(24 - 98)
37
(14-59)
81
(31 - 130)
27
(10-43)
95
(36-152)
233
(89-371)
156
(59 - 249)
64
(24-103)
63
(24-101)
82
(31 - 130)
8
(3-13)
87
(33-138)
12
(4-19)
15/352
65
(25-104)
64
(24-102)
44
(17-71)
37
(14-59)
56
(21 - 90)
10
(4-16)
88
(33-140)
102
(38-165)
110
(41 -177)
56
(21 - 90)
63
(24-101)
32
(12-51)
2
(1-3)
72
(27-115)
7
(2-11)
13/35
51
(19-82)
51
(19-82)
34
(13-55)
37
(14-59)
47
(18-76)
10
(4-16)
69
(26-110)
102
(38-165)
104
(39-168)
50
(19-80)
63
(24-101)
32
(12-51)
2
(1-3)
55
(21 - 89)
7
(2-11)
12/35
44
(17-71)
44
(16-70)
30
(11-47)
32
(12-52)
39
(15-63)
10
(4-16)
59
(22 - 95)
89
(33-144)
86
(32-139)
43
(16-68)
57
(21 -91)
32
(12-51)
2
(1-3)
46
(17-75)
7
(2-11)
13/30
51
(19-82)
49
(19-79)
34
(13-55)
37
(14-59)
41
(15-65)
7
(2-11)
69
(26-110)
69
(26-111)
77
(29-124)
42
(16-68)
55
(21 - 89)
21
(8 - 34)
1
(0-1)
54
(20 - 86)
3
(1-5)
12/25
43
(16-69)
34
(13-55)
25
(9 - 40)
32
(12-52)
25
(9 - 40)
3
(1-5)
59
(22 - 95)
35
(13-56)
43
(16-70)
28
(10-45)
35
(13-57)
11
(4-18)
0
(0-0)
35
(13-56)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 E-76
September 2009
Draft - Do Not Quote or Cite

-------
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.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
76
(29-121)
66
(25-105)
61
(23 - 97)
41
(15-66)
76
(29-121)
31
(12-48)
94
(36-149)
238
(91 - 379)
158
(60 - 254)
64
(24-102)
55
(21 - 88)
83
(32-131)
11
(4-17)
92
(35-147)
14
(5 - 22)
15/352
68
(26-108)
60
(23 - 96)
44
(17-70)
41
(15-66)
52
(20 - 84)
12
(4-19)
86
(33-137)
105
(39-170)
112
(42-180)
56
(21 - 89)
55
(21 - 88)
32
(12-52)
4
(2-7)
77
(29-123)
8
(3-13)
13/35
53
(20 - 85)
47
(18-76)
34
(13-54)
41
(15-66)
44
(16-70)
12
(4-19)
67
(25-108)
105
(39-170)
106
(40-171)
49
(19-79)
55
(21 - 88)
32
(12-52)
4
(2-7)
60
(22 - 96)
8
(3-13)
12/35
46
(17-73)
40
(15-65)
29
(11-47)
36
(13-58)
36
(13-58)
12
(4-19)
58
(22 - 93)
92
(34-149)
88
(33-142)
42
(16-68)
49
(18-79)
32
(12-52)
4
(2-7)
50
(19-81)
8
(3-13)
13/30
53
(20 - 85)
46
(17-73)
34
(13-54)
41
(15-66)
37
(14-60)
8
(3-13)
67
(25-108)
71
(27-115)
79
(29-127)
42
(16-67)
47
(18-76)
22
(8 - 35)
2
(1-4)
58
(22 - 93)
5
(2-7)
12/25
45
(17-71)
31
(12-50)
24
(9 - 39)
36
(13-58)
22
(8 - 35)
5
(2-8)
58
(22 - 93)
37
(14-59)
45
(17-72)
27
(10-44)
28
(10-45)
11
(4-18)
0
(0-1)
39
(14-62)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 E-77
September 2009
Draft - Do Not Quote or Cite

-------
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 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 Lung Cancer Mortality Associated with Long-Term Exposure to PM
in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual
Standards (Standard Combination Denoted n/m):
Recent PM2 5
Concentrations
8.6%
(3.3% -13.6%)
8.4%
(3.2% -13.3%)
10%
(3.8% -15.8%)
5.7%
(2.2% - 9.2%)
8.5%
(3. 2% -13. 4%)
8.9%
(3.4% -14.1%)
7.8%
(2.9% -12.3%)
8.6%
(3.3% -13.6%)
6.6%
(2. 5% -10. 5%)
6.8%
(2. 6% -10. 8%)
4.2%
(1.6% -6.7%)
10%
(3.8% -15.7%)
5.2%
(2% - 8.4%)
8.4%
(3. 2% -13. 3%)
4.8%
(1.8% -7.7%)
15/352
7.6%
(2.9% -12.1%)
7.7%
(2.9% -12.2%)
7.2%
(2.7% - 1 1 .4%)
5.7%
(2.2% - 9.2%)
6.1%
(2.3% - 9.8%)
3.1%
(1.2% -5%)
7.1%
(2.7% - 1 1 .4%)
4%
(1.5% -6.4%)
4.8%
(1.8% -7. 7%)
5.9%
(2.2% - 9.4%)
4.2%
(1.6% -6.7%)
4.1%
(1.6% -6.7%)
1.8%
(0.7% - 2.9%)
7.1%
(2.7% -11. 3%)
3.1%
(1.2% -5.1%)
13/35
6%
(2.3% - 9.6%)
6.2%
(2.4% -10%)
5.6%
(2.1% -9%)
5.7%
(2.2% - 9.2%)
5.3%
(2% - 8.5%)
3.1%
(1.2% -5%)
5.6%
(2.1% -8.9%)
4%
(1.5% -6.4%)
4.5%
(1.7% -7. 3%)
5.2%
(2% - 8.4%)
4.2%
(1.6% -6.7%)
4.1%
(1.6% -6.7%)
1.8%
(0.7% -2. 9%)
5.7%
(2.1% -9.1%)
3.1%
(1.2% -5.1%)
12/35
5.2%
(1.9% -8.3%)
5.4%
(2% - 8.6%)
4.8%
(1.8% -7.8%)
5.1%
(1.9% -8. 2%)
4.5%
(1.7% -7. 3%)
3.1%
(1.2% -5%)
4.8%
(1.8% -7.7%)
3.5%
(1.3% -5.7%)
3.8%
(1.4% -6. 2%)
4.5%
(1.7% -7. 2%)
3.7%
(1.4% -6%)
4.1%
(1.6% -6.7%)
1.8%
(0.7% -2. 9%)
4.9%
(1.8% -7. 8%)
3.1%
(1.2% -5.1%)
13/30
6%
(2.3% - 9.6%)
6%
(2.3% - 9.6%)
5.6%
(2.1% -9%)
5.7%
(2.2% - 9.2%)
4.7%
(1.8% -7. 5%)
2%
(0.8% - 3.3%)
5.6%
(2.1% -8.9%)
2.8%
(1%-4.5%)
3.5%
(1.3% -5. 6%)
4.4%
(1.7% -7.1%)
3.6%
(1.4% -5.9%)
2.9%
(1.1% -4.7%)
0.9%
(0.3% -1.5%)
5.5%
(2.1% -8. 8%)
2.1%
(0.8% - 3.4%)
2.5 Concentrations
n) and Daily (m)
12/25
5%
(1.9% -8.1%)
4.3%
(1.6% -6.9%)
4.1%
(1.5% -6.6%)
5.1%
(1.9% -8.2%)
3.2%
(1.2% -5.1%)
1%
(0.4% -1.6%)
4.8%
(1.8% -7.7%)
1 .6%
(0.6% - 2.6%)
2.1%
(0.8% - 3.5%)
3%
(1.1% -4. 8%)
2.3%
(0.8% - 3.7%)
1 .7%
(0.6% - 2.8%)
0%
(0%-0.1%)
3.9%
(1.5% -6.2%)
1%
(0.4% - 1 .6%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3and a daily standard set at 35 ug/m3.
                                                                   E-78
September 2009
Draft - Do Not Quote or Cite

-------
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 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 Lung Cancer Mortality Associated with Long-Term Exposure to PM
in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual
Standards (Standard Combination Denoted n/m):
Recent PM2 5
Concentrations
8%
(3.1% -12.7%)
7.3%
(2.8% -11. 6%)
9.6%
(3.7% -15.1%)
4.3%
(1.6% -6.9%)
6.5%
(2.4% -10. 3%)
9.3%
(3.6% -14.8%)
7.9%
(3% -12.6%)
7.7%
(2.9% -12.2%)
5.6%
(2.1% -9%)
6.6%
(2.5% -10. 6%)
4.4%
(1.7% -7.1%)
8.7%
(3.3% -13.8%)
4.2%
(1.6% -6.7%)
6.5%
(2.5% -10. 4%)
3.1%
(1.2% -5%)
15/352
7.1%
(2.7% - 1 1 .3%)
6.7%
(2.5% -10.6%)
6.8%
(2.6% -10.9%)
4.3%
(1.6% -6.9%)
4.5%
(1.7% -7. 2%)
3.4%
(1.3% -5.4%)
7.3%
(2.8% -11. 6%)
3.4%
(1.3% -5.4%)
4%
(1.5% -6.4%)
5.8%
(2.2% - 9.3%)
4.4%
(1.7% -7.1%)
3.4%
(1.3% -5. 5%)
1.1%
(0.4% -1.8%)
5.4%
(2% - 8.7%)
1 .7%
(0.6% - 2.8%)
13/35
5.6%
(2.1% -8.9%)
5.3%
(2% - 8.5%)
5.3%
(2% - 8.5%)
4.3%
(1.6% -6.9%)
3.8%
(1.4% -6.1%)
3.4%
(1.3% -5.4%)
5.7%
(2.2% - 9.2%)
3.4%
(1.3% -5.4%)
3.8%
(1.4% -6.1%)
5.1%
(1.9% -8.2%)
4.4%
(1.7% -7.1%)
3.4%
(1.3% -5. 5%)
1.1%
(0.4% -1.8%)
4.2%
(1.6% -6. 7%)
1.7%
(0.6% - 2.8%)
12/35
4.8%
(1.8% -7.7%)
4.6%
(1.7% -7.3%)
4.6%
(1.7% -7.3%)
3.8%
(1.4% -6.1%)
3.1%
(1.2% -5%)
3.4%
(1.3% -5.4%)
4.9%
(1.9% -7. 9%)
2.9%
(1.1% -4. 8%)
3.1%
(1.2% -5%)
4.4%
(1.6% -7.1%)
4%
(1.5% -6.4%)
3.4%
(1.3% -5.5%)
1.1%
(0.4% -1.8%)
3.5%
(1.3% -5.6%)
1.7%
(0.6% - 2.8%)
13/30
5.6%
(2.1% -8.9%)
5.1%
(1.9% -8.2%)
5.3%
(2% - 8.5%)
4.3%
(1.6% -6. 9%)
3.2%
(1.2% -5. 2%)
2.3%
(0.8% - 3.6%)
5.7%
(2.2% - 9.2%)
2.3%
(0.8% - 3.7%)
2.8%
(1%-4.5%)
4.3%
(1.6% -7%)
3.8%
(1.4% -6.2%)
2.3%
(0.9% - 3.7%)
0.4%
(0.1% -0.6%)
4%
(1.5% -6. 5%)
0.9%
(0.3% -1.4%)
2.5 Concentrations
n) and Daily (m)
12/25
4.7%
(1.8% -7.5%)
3.5%
(1.3% -5.7%)
3.9%
(1.4% -6.2%)
3.8%
(1.4% -6.1%)
2%
(0.7% - 3.2%)
1.1%
(0.4% -1.8%)
4.9%
(1.9% -7.9%)
1.1%
(0.4% -1.9%)
1 .6%
(0.6% - 2.5%)
2.9%
(1.1% -4. 6%)
2.4%
(0.9% - 3.9%)
1 .2%
(0.4% -1.9%)
0%
(0% - 0%)
2.6%
(1%-4.2%)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-79
September 2009
Draft - Do Not Quote or Cite

-------
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 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 Lung Cancer Mortality Associated with Long-Term Exposure to PM
in a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual
Standards (Standard Combination Denoted n/m):
Recent PM2 5
Concentrations
8.1%
(3.1% -12.9%)
6.9%
(2.6% -10.9%)
9.4%
(3.6% -14.9%)
4.7%
(1.8% -7. 5%)
6.1%
(2.3% - 9.7%)
10.4%
(4% -16. 3%)
7.6%
(2.9% -12.1%)
7.8%
(3% -12.4%)
5.7%
(2.1% -9.1%)
6.6%
(2. 5% -10. 5%)
3.7%
(1.4% -6%)
8.9%
(3.4% -14.1%)
5.7%
(2.1% -9.1%)
6.9%
(2.6% -11%)
3.5%
(1.3% -5.6%)
15/352
7.2%
(2.7% - 1 1 .4%)
6.3%
(2.4% -10%)
6.7%
(2.5% -10.7%)
4.7%
(1.8% -7. 5%)
4.2%
(1.6% -6. 7%)
4%
(1.5% -6.4%)
7%
(2.7% - 1 1 .2%)
3.5%
(1.3% -5.6%)
4%
(1.5% -6. 5%)
5.7%
(2. 2% -9.1%)
3.7%
(1.4% -6%)
3.5%
(1.3% -5.6%)
2.1%
(0.8% - 3.4%)
5.8%
(2.2% - 9.3%)
2.1%
(0.8% - 3.3%)
13/35
5.6%
(2.1% -9%)
4.9%
(1.9% -7. 9%)
5.2%
(2% - 8.4%)
4.7%
(1.8% -7. 5%)
3.5%
(1.3% -5. 6%)
4%
(1.5% -6.4%)
5.5%
(2.1% -8.8%)
3.5%
(1.3% -5.6%)
3.8%
(1.4% -6.1%)
5.1%
(1.9% -8.1%)
3.7%
(1.4% -6%)
3.5%
(1.3% -5.6%)
2.1%
(0.8% - 3.4%)
4.5%
(1.7% -7. 2%)
2.1%
(0.8% - 3.3%)
12/35
4.8%
(1.8% -7.8%)
4.2%
(1.6% -6.8%)
4.5%
(1 .7% - 7.2%)
4.1%
(1.5% -6.6%)
2.9%
(1.1% -4.6%)
4%
(1.5% -6.4%)
4.7%
(1.8% -7.5%)
3%
(1.1% -4.9%)
3.2%
(1.2% -5.1%)
4.3%
(1.6% -7%)
3.3%
(1.2% -5.3%)
3.5%
(1.3% -5.6%)
2.1%
(0.8% - 3.4%)
3.8%
(1.4% -6.1%)
2.1%
(0.8% - 3.3%)
13/30
5.6%
(2.1% -9%)
4.8%
(1.8% -7.6%)
5.2%
(2% - 8.4%)
4.7%
(1.8% -7. 5%)
3%
(1.1% -4. 8%)
2.8%
(1%-4.5%)
5.5%
(2.1% -8.8%)
2.3%
(0.9% - 3.8%)
2.8%
(1.1% -4. 6%)
4.3%
(1.6% -6. 9%)
3.2%
(1.2% -5.1%)
2.4%
(0.9% - 3.8%)
1 .2%
(0.4% -1.9%)
4.4%
(1.6% -7%)
1.1%
(0.4% -1.9%)
2.5 Concentrations
n) and Daily (m)
12/25
4.7%
(1.8% -7.6%)
3.2%
(1.2% -5.2%)
3.8%
(1.4% -6%)
4.1%
(1.5% -6.6%)
1 .7%
(0.6% - 2.8%)
1 .6%
(0.6% - 2.5%)
4.7%
(1.8% -7. 5%)
1 .2%
(0.4% -1.9%)
1 .6%
(0.6% - 2.6%)
2.8%
(1.1% -4.6%)
1.9%
(0.7% -3.1%)
1 .2%
(0.5% - 2%)
0.3%
(0.1% -0.4%)
2.9%
(1.1% -4.7%)
0.2%
(0.1% -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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                   E-80
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-80/0 . -go/0)
-39%
(-38% - -40%)
0%
(0% - 0%)
-38%
(-37% - -39%)
-186%
(-181% --192%)
-9%
(-80/0 - -go/0)
-116%
(-11 3% --120%)
-38%
(-37% - -39%)
-15%
(-14% --15%)
0%
(0% - 0%)
-141%
(-136% --146%)
-190%
(-187% --193%)
-18%
(-17% --18%)
-53%
(-52% - -54%)
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%)
13/35
21%
(21% -22%)
19%
(19% -19%)
22%
(21% -22%)
0%
(0% - 0%)
14%
(13% -14%)
0%
(0% - 0%)
22%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
20%
(20% -21%)
0%
(0% - 0%)
12/35
32%
(31% -33%)
30%
(29% - 30%)
33%
(32% - 33%)
11%
(11% -11%)
26%
(26% - 27%)
0%
(0% - 0%)
33%
(32% - 33%)
12%
(11% -12%)
20%
(20% - 20%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
32%
(31% -32%)
0%
(0% - 0%)
13/30
21%
(21% -22%)
22%
(22% - 22%)
22%
(21% -22%)
0%
(0% - 0%)
24%
(24% - 24%)
34%
(34% - 35%)
22%
(21% -22%)
30%
(30% - 30%)
27%
(27% - 28%)
25%
(24% - 25%)
13%
(13% -13%)
29%
(29% - 30%)
49%
(48% - 49%)
23%
(22% - 23%)
34%
(34% - 34%)
12/25
34%
(33% - 34%)
44%
(44% - 45%)
43%
(42% - 44%)
11%
(11% -11%)
49%
(48% - 49%)
69%
(69% - 70%)
33%
(32% - 33%)
60%
(60% -61%)
55%
(55% - 55%)
50%
(49% - 50%)
46%
(45% - 46%)
59%
(59% - 59%)
98%
(98% - 98%)
46%
(45% - 46%)
68%
(68% - 68%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-81
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-13% --13%)
-9%
(-9% - -9%)
-40%
(-39% --41%)
0%
(0% - 0%)
-44%
(-43% - -45%)
-178%
(-172% --183%)
-9%
(-80/0 . -go/0)
-128%
(-125% --131%)
-42%
(-41% --42%)
-15%
(-14% --15%)
0%
(0% - 0%)
-158%
(-154% --163%)
-271%
(-267% - -274%)
-20%
(-20% --21%)
-80%
(-79% - -80%)
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%)
13/35
22%
(21% -22%)
20%
(20% -21%)
22%
(22% - 22%)
0%
(0% - 0%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
33%
(32% - 33%)
32%
(31% -32%)
33%
(33% - 34%)
13%
(13% -13%)
30%
(30% -31%)
0%
(0% - 0%)
32%
(32% - 33%)
13%
(13% -13%)
22%
(22% - 22%)
24%
(24% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
36%
(35% - 36%)
0%
(0% - 0%)
13/30
22%
(21% -22%)
23%
(23% - 24%)
22%
(22% - 22%)
0%
(0% - 0%)
28%
(28% - 28%)
33%
(33% - 33%)
22%
(21% -22%)
33%
(33% - 33%)
30%
(30% - 30%)
25%
(25% - 25%)
13%
(12% -13%)
33%
(33% - 33%)
69%
(69% - 69%)
26%
(25% - 26%)
50%
(50% - 50%)
12/25
34%
(34% - 35%)
47%
(47% - 48%)
44%
(43% - 44%)
13%
(13% -13%)
56%
(56% - 57%)
66%
(66% - 67%)
32%
(32% - 33%)
66%
(66% - 66%)
61%
(60% -61%)
50%
(50% -51%)
44%
(44% - 45%)
66%
(66% - 66%)
100%
(100% -100%)
51%
(51% -52%)
100%
(100% -100%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-82
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-10%
(-9% --10%)
-40%
(-39% - -42%)
0%
(0% - 0%)
-46%
(-45% - -47%)
-162%
(-156% --167%)
-9%
(-80/0 . -go/0)
-126%
(-123% --129%)
-41%
(-41% --42%)
-15%
(-15% --15%)
0%
(0% - 0%)
-156%
(-151% --160%)
-170%
(-167% --174%)
-20%
(-19% --20%)
-70%
(-69% --71%)
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%)
13/35
22%
(21% -22%)
21%
(21% -21%)
22%
(22% - 23%)
0%
(0% - 0%)
16%
(16% -17%)
0%
(0% - 0%)
22%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
22%
(22% - 23%)
0%
(0% - 0%)
12/35
33%
(32% - 33%)
33%
(32% - 33%)
34%
(33% - 34%)
12%
(12% -13%)
32%
(31% -32%)
0%
(0% - 0%)
33%
(32% - 33%)
13%
(12% -13%)
22%
(21% -22%)
24%
(24% - 25%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(34% - 35%)
0%
(0% - 0%)
13/30
22%
(21% -22%)
24%
(24% - 24%)
22%
(22% - 23%)
0%
(0% - 0%)
29%
(29% - 29%)
30%
(30% - 30%)
22%
(21% -22%)
32%
(32% - 33%)
30%
(30% - 30%)
25%
(25% - 25%)
14%
(14% -14%)
32%
(32% - 32%)
44%
(43% - 44%)
25%
(24% - 25%)
44%
(44% - 44%)
12/25
34%
(34% - 35%)
48%
(48% - 49%)
44%
(44% - 45%)
12%
(12% -13%)
59%
(58% - 59%)
60%
(60% -61%)
33%
(32% - 33%)
65%
(65% - 66%)
60%
(60% -61%)
51%
(50% -51%)
49%
(49% - 49%)
65%
(65% - 65%)
88%
(88% - 88%)
50%
(49% - 50%)
89%
(89% - 89%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-83
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-19.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Lung Cancer Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983):  Based on 2005 Air Quality Data*
       ^   -20%
       o
       p
       I
       I
       1
       a:
       01
       o
       8.
 -40%
 -60%
 -80%
-100%
          -120%
-140% -
-160%
          -180%
          -200%
           ^   /
   Ji
TIL
                  2005 air
                   quality
                   15/35**      13/35       12/35

                           Alternative Standard
                                       13/30
12/25
                          -»-Atlanta, GA 68 (26-108); 7.6% (2.9%-12.1%)
                          -m- Baltimore, MD 74 (28-117); 7.7% (2.9%-12.2%)
                          -*- Birmingham, AL 46 (17-73); 7.2%  (2.7%-11.4%)
                              Dallas, TX 48 (18-77); 5.7% (2.2%-9.2%)
                          -*- Detroit, Ml 77 (29-123); 6.1% (2.3%-9.8%)
                          -•- Fresno, CA 9 (3-14); 3.1% (1.2%-5%)
                          -i- Houston, TX 83 (32-132); 7.1% (2.7%-11.4%)
                          	 Los Angeles, CA 120 (45-193);  4% (1.5%-6.4%)
                            - New York, NY  131 (49 - 210); 4.8%  (1.8% - 7.7%)
                          -•- Philadelphia, PA 57 (22-92); 5.9% (2.2%-9.4%)
                          -m- Phoenix, AZ 58  (22 - 93); 4.2%  (1.6% - 6.7%)
                          -*- Pittsburgh, PA 39 (15-63);  4.1% (1.6%-6.7%)
                              Salt Lake City, UT 3 (1 - 5);  1.8% (0.7% - 2.9%)
                              St. Louis, MO  94 (36-150); 7.1% (2.7%-11.3%)
                          -•- Tacoma.WA 12 (4-19); 3.1% (1.2%-5.1%)
*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.
                                                 E-84
September 2009
                                                                    Draft - Do Not Quote or Cite

-------
Figure E-20.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Lung Cancer Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983):  Based on 2006 Air Quality Data*
          100%
           80%
       •5
       re
       •O
       v>
       o
       I
       c
       O
       'o
       3
       •o
       
-------
Figure E-21  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Lung Cancer Mortality Associated with Long-Term Exposure to
(Exposure Period: 1979 - 1983): Based on 2007 Air Quality Data*
       I
       OT
       O
       Q

       I
s  -
       01
       O
       8.
           -80%
          100%
   -120%
   -140% -
   -160%
   -180%
   -200%
                  2007 air
                   quality
                      15/35**      13/35      12/35

                              Alternative Standard
13/30
12/25
                         -»- Atlanta, GA 68 (26-108); 7.2% (2.7%-11.4%)
                         -m- Baltimore, MD 60 (23-96); 6.3%  (2.4%-10%)
                         -*- Birmingham, AL 44  (17-70); 6.7% (2.5%-10.7%)
                             Dallas, TX 41 (15-66); 4.7% (1.8%-7.5%)
                         -*- Detroit, Ml 52 (20 - 84); 4.2% (1.6% - 6.7%)
                         -•- Fresno, CA 12 (4-19); 4% (1.5%-6.4%)
                         -i- Houston, TX 86 (33-137); 7% (2.7%-11.2%)
                         	 Los Angeles, CA 105 (39-170); 3.5% (1.3%-5.6%)
                             New York, NY 112 (42-180); 4%  (1.5% - 6.5%)
                         -•- Philadelphia, PA 56 (21 -89); 5.7% (2.2%-9.1%)
                         -m- Phoenix, AZ 55  (21  - 88); 3.7% (1.4% - 6%)
                         -*- Pittsburgh, PA 32 (12-52); 3.5%  (1.3%-5.6%)
                             Salt Lake City, UT 4 (2-7); 2.1%  (0.8%-3.4%)
                             St. Louis, MO 77 (29-123); 5.8% (2.2%-9.3%)
                         -•- Tacoma.WA 8 (3-13); 2.1% (0.8%-3.3%)
*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.
                                                 E-86
September 2009
                                                                       Draft - Do Not Quote or Cite

-------
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 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
110
(49-166)
116
(51 -175)
92
(41 -138)
69
(31 -107)
153
(68 - 232)
37
(16-56)
130
(58-198)
373
(166-565)
260
(115-398)
95
(42-145)
84
(37-129)
135
(60 - 203)
14
(6-21)
160
(71 - 242)
27
(12-41)
15/352
98
(43-149)
106
(47-162)
66
(29-101)
69
(31 -107)
112
(49-171)
13
(6 - 20)
120
(53-183)
174
(76 - 269)
190
(83 - 292)
83
(37-127)
84
(37-129)
57
(25 - 88)
5
(2-7)
136
(60 - 207)
17
(8 - 27)
13/35
77
(34-118)
86
(38-132)
52
(23 - 80)
69
(31 -107)
97
(42-148)
13
(6 - 20)
94
(42-145)
174
(76 - 269)
181
(79 - 279)
74
(33-114)
84
(37-129)
57
(25 - 88)
5
(2-7)
109
(48-167)
17
(8 - 27)
12/35
67
(29-103)
75
(33-115)
45
(20 - 69)
62
(27 - 95)
83
(36-128)
13
(6 - 20)
81
(36-125)
154
(67 - 239)
153
(67 - 236)
63
(28 - 98)
75
(33-117)
57
(25 - 88)
5
(2-7)
94
(41 - 144)
17
(8 - 27)
13/30
77
(34-118)
83
(37-128)
52
(23 - 80)
69
(31 - 107)
85
(37-131)
9
(4-13)
94
(42-145)
123
(53-190)
138
(60-214)
63
(27 - 97)
73
(32-113)
40
(18-63)
2
(1-4)
106
(47-162)
12
(5-18)
12/25
65
(29-100)
60
(26 - 92)
38
(17-59)
62
(27 - 95)
58
(25 - 90)
4
(2-6)
81
(36-125)
70
(30-109)
86
(37-134)
42
(18-65)
46
(20-71)
24
(10-37)
0
(0-0)
75
(33-115)
6
(2-9)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35
ug/m3.
                                                                  E-87
September 2009
                                Draft - Do Not Quote or Cite

-------
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.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
106
(47-162)
101
(45-154)
89
(40-134)
54
(23 - 83)
117
(52-179)
39
(18-59)
138
(61 - 209)
336
(149-511)
226
(99 - 346)
93
(41 - 142)
92
(40-141)
118
(52-178)
11
(5-18)
125
(55-191)
17
(8 - 27)
15/352
95
(42-144)
93
(41 -141)
64
(28 - 98)
54
(23 - 83)
82
(36-126)
14
(6 - 22)
127
(56-193)
149
(65-231)
160
(70 - 247)
81
(36-125)
92
(40-141)
46
(20-71)
3
(1-5)
104
(46-160)
10
(4-15)
13/35
74
(33-114)
74
(33-114)
50
(22 - 77)
54
(23 - 83)
69
(30-107)
14
(6 - 22)
100
(44-153)
149
(65-231)
152
(66 - 234)
72
(32-111)
92
(40-141)
46
(20-71)
3
(1-5)
80
(35-124)
10
(4-15)
12/35
64
(28 - 98)
64
(28 - 98)
43
(19-66)
47
(20 - 72)
57
(25 - 89)
14
(6 - 22)
86
(38-133)
130
(57 - 202)
126
(55-195)
62
(27 - 95)
83
(36-128)
46
(20-71)
3
(1-5)
67
(29-104)
10
(4-15)
13/30
74
(33-114)
71
(31 -110)
50
(22 - 77)
54
(23 - 83)
59
(26 - 92)
10
(4-15)
100
(44-153)
100
(43-156)
112
(49-174)
61
(27 - 94)
80
(35-124)
31
(13-48)
1
(0-2)
78
(34-120)
5
(2-8)
12/25
62
(27 - 96)
49
(22 - 76)
36
(16-56)
47
(20 - 72)
36
(16-56)
5
(2-8)
86
(38-133)
51
(22 - 79)
63
(27 - 99)
41
(18-63)
51
(22 - 79)
16
(7 - 25)
0
(0-0)
51
(22 - 79)
0
(0-0)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35
ug/m3.
                                                                  E-88
September 2009
                                Draft - Do Not Quote or Cite

-------
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 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
110
(49-167)
95
(42-145)
88
(39-133)
59
(26-91)
110
(48-168)
44
(20 - 66)
135
(60 - 205)
343
(153-522)
230
(101 -352)
92
(41 -141)
80
(35-123)
119
(53-180)
16
(7 - 24)
133
(59 - 203)
20
(9-31)
15/352
98
(43-149)
87
(38-133)
63
(28 - 96)
59
(26-91)
76
(33-117)
17
(7-26)
125
(55-190)
153
(67 - 238)
163
(71 - 252)
80
(35-123)
80
(35-123)
47
(21 - 73)
6
(3-9)
112
(49-171)
12
(5-18)
13/35
77
(34-118)
69
(30-106)
49
(22 - 76)
59
(26-91)
63
(28 - 98)
17
(7-26)
98
(43-150)
153
(67 - 238)
155
(67 - 239)
72
(31 -110)
80
(35-123)
47
(21 - 73)
6
(3-9)
87
(38-134)
12
(5-18)
12/35
66
(29-102)
59
(26 - 90)
42
(18-65)
52
(23 - 80)
52
(23-81)
17
(7-26)
84
(37-129)
134
(58 - 208)
128
(56-199)
61
(27 - 94)
71
(31 -110)
47
(21 - 73)
6
(3-9)
73
(32-113)
12
(5-18)
13/30
77
(34-118)
66
(29-102)
49
(22 - 76)
59
(26-91)
54
(23 - 84)
12
(5-19)
98
(43-150)
104
(45-162)
115
(50-178)
61
(26 - 93)
69
(30-106)
32
(14-50)
3
(1-5)
84
(37-130)
7
(3-10)
12/25
65
(28-100)
45
(20 - 70)
36
(16-55)
52
(23 - 80)
32
(14-49)
7
(3-11)
84
(37-129)
54
(23 - 84)
65
(28-102)
40
(17-62)
41
(18-64)
17
(7-26)
1
(0-1)
56
(24 - 87)
1
(1-2)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-89
September 2009
Draft - Do Not Quote or Cite

-------
Table E-67.  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 PM2.5 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 Lung Cancer 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):
Recent PM2 5
Concentrations
12.3%
(5.5% -18.6%)
12.1%
(5. 4% -18. 3%)
14.3%
(6.4% - 21 .6%)
8.3%
(3.7% -12.7%)
12.2%
(5. 4% -18. 4%)
12.8%
(5. 7% -19. 4%)
1 1 .2%
(5% -17%)
12.4%
(5.5% -18.7%)
9.5%
(4.2% -14.5%)
9.8%
(4.3% -14.9%)
6.1%
(2.7% - 9.4%)
14.3%
(6.4% -21. 5%)
7.6%
(3. 3% -11. 6%)
12.1%
(5. 4% -18.3%)
7%
(3.1% -10.8%)
15/352
11%
(4.9% -16.7%)
11.1%
(4. 9% -16. 9%)
10.4%
(4.6% -15.8%)
8.3%
(3.7% -12.7%)
8.9%
(3. 9% -13. 6%)
4.5%
(2% - 7%)
10.3%
(4.6% -15.7%)
5.8%
(2.5% - 8.9%)
6.9%
(3% -10.7%)
8.5%
(3.8% -13.1%)
6.1%
(2.7% - 9.4%)
6%
(2.6% - 9.3%)
2.6%
(1.1% -4.1%)
10.3%
(4. 6% -15. 7%)
4.6%
(2% -7.1%)
13/35
8.7%
(3.8% -13.3%)
9%
(4% -13. 8%)
8.2%
(3.6% -12.5%)
8.3%
(3.7% -12.7%)
7.7%
(3.4% - 1 1 .8%)
4.5%
(2% - 7%)
8.1%
(3.6% -12.4%)
5.8%
(2.5% - 8.9%)
6.6%
(2.9% -10.2%)
7.6%
(3.3% - 1 1 .7%)
6.1%
(2.7% - 9.4%)
6%
(2.6% - 9.3%)
2.6%
(1.1% -4.1%)
8.2%
(3. 6% -12. 6%)
4.6%
(2% -7.1%)
12/35
7.5%
(3.3% -11. 5%)
7.8%
(3.4% -12%)
7%
(3.1% -10.8%)
7.4%
(3.2% - 1 1 .4%)
6.6%
(2. 9% -10.1%)
4.5%
(2% - 7%)
7%
(3.1% -10.7%)
5.1%
(2.2% - 7.9%)
5.6%
(2.4% - 8.6%)
6.5%
(2. 9% -10%)
5.5%
(2.4% - 8.4%)
6%
(2.6% - 9.3%)
2.6%
(1.1% -4.1%)
7.1%
(3.1% -10. 9%)
4.6%
(2% -7.1%)
13/30
8.7%
(3.8% -13.3%)
8.7%
(3. 8% -13. 3%)
8.2%
(3.6% -12.5%)
8.3%
(3.7% -12.7%)
6.8%
(3% -10. 4%)
3%
(1.3% -4.6%)
8.1%
(3.6% -12.4%)
4.1%
(1.8% -6.3%)
5.1%
(2.2% - 7.8%)
6.5%
(2.8% - 9.9%)
5.3%
(2.3% - 8.2%)
4.3%
(1.9% -6.6%)
1.4%
(0.6% -2.1%)
8%
(3. 5% -12.3%)
3%
(1.3% -4.7%)
12/25
7.3%
(3.2% -11. 3%)
6.2%
(2.7% - 9.6%)
6%
(2.6% - 9.2%)
7.4%
(3.2% - 1 1 .4%)
4.6%
(2% -7.1%)
1.4%
(0.6% - 2.2%)
7%
(3.1% -10.7%)
2.3%
(1%-3.6%)
3.1%
(1.4% -4.9%)
4.3%
(1.9% -6.7%)
3.3%
(1.4% -5.2%)
2.5%
(1.1% -3.9%)
0.1%
(0%-0.1%)
5.6%
(2.5% - 8.7%)
1.5%
(0.6% - 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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-90
September 2009
Draft - Do Not Quote or Cite

-------
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 - 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.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
11.6%
(5.1% -17.6%)
10.5%
(4.7% -16%)
13.8%
(6.2% - 20.7%)
6.3%
(2.7% - 9.7%)
9.4%
(4.1% -14. 3%)
13.4%
(6% - 20.2%)
11.4%
(5.1% -17.3%)
11.1%
(4.9% -16.8%)
8.2%
(3.6% -12.5%)
9.6%
(4.2% -14.7%)
6.4%
(2.8% - 9.8%)
12.6%
(5.6% -19%)
6.1%
(2.7% - 9.4%)
9.4%
(4.2% -14. 4%)
4.5%
(2% - 7%)
15/352
10.3%
(4.5% -15.7%)
9.7%
(4. 3% -14. 7%)
9.9%
(4.4% -15.1%)
6.3%
(2.7% - 9.7%)
6.5%
(2. 9% -10%)
4.9%
(2.1% -7. 6%)
10.5%
(4.7% -16%)
4.9%
(2.1% -7.6%)
5.8%
(2.5% - 8.9%)
8.4%
(3.7% -12.8%)
6.4%
(2.8% - 9.8%)
4.9%
(2.1% -7.6%)
1 .7%
(0.7% - 2.6%)
7.8%
(3.4% -12%)
2.5%
(1.1% -3. 9%)
13/35
8.1%
(3.5% -12.4%)
7.7%
(3.4% -11. 9%)
7.7%
(3.4% -11. 9%)
6.3%
(2.7% - 9.7%)
5.5%
(2.4% - 8.5%)
4.9%
(2.1% -7. 6%)
8.3%
(3.6% -12.7%)
4.9%
(2.1% -7.6%)
5.5%
(2.4% - 8.5%)
7.5%
(3.3% -11. 5%)
6.4%
(2.8% - 9.8%)
4.9%
(2.1% -7.6%)
1 .7%
(0.7% - 2.6%)
6.1%
(2.6% - 9.3%)
2.5%
(1.1% -3. 9%)
12/35
6.9%
(3% -10.7%)
6.6%
(2. 9% -10. 2%)
6.6%
(2.9% -10.2%)
5.5%
(2.4% - 8.5%)
4.6%
(2% -7.1%)
4.9%
(2.1% -7.6%)
7.2%
(3.1% -11%)
4.3%
(1.9% -6.7%)
4.5%
(2% - 7%)
6.4%
(2.8% - 9.8%)
5.8%
(2.5% - 8.9%)
4.9%
(2.1% -7.6%)
1.7%
(0.7% - 2.6%)
5.1%
(2.2% - 7.8%)
2.5%
(1.1% -3. 9%)
13/30
8.1%
(3.5% -12.4%)
7.4%
(3. 3% -11. 4%)
7.7%
(3.4% -11. 9%)
6.3%
(2.7% - 9.7%)
4.7%
(2% - 7.3%)
3.3%
(1.4% -5.1%)
8.3%
(3.6% -12.7%)
3.3%
(1.4% -5.1%)
4.1%
(1.8% -6.3%)
6.3%
(2.8% - 9.7%)
5.6%
(2.4% - 8.6%)
3.3%
(1.4% -5.2%)
0.5%
(0.2% - 0.8%)
5.9%
(2.6% -9.1%)
1 .2%
(0.5% -2%)
12/25
6.8%
(3% -10.4%)
5.1%
(2.2% - 8%)
5.6%
(2.4% - 8.7%)
5.5%
(2.4% - 8.5%)
2.9%
(1.2% -4. 4%)
1 .7%
(0.7% - 2.6%)
7.2%
(3.1% -11%)
1 .7%
(0.7% - 2.6%)
2.3%
(1%-3.6%)
4.2%
(1.8% -6. 5%)
3.6%
(1.5% -5.5%)
1 .7%
(0.7% - 2.6%)
0%
(0% - 0%)
3.8%
(1.7% -6%)
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).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-91
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2.5 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.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
1 1 .7%
(5.2% -17.7%)
9.9%
(4.4% -15.1%)
13.5%
(6% - 20.4%)
6.8%
(3% -10.5%)
8.8%
(3.9% -13. 5%)
14.8%
(6.7% -22. 3%)
11%
(4.9% -16.7%)
1 1 .2%
(5% -17.1%)
8.2%
(3.6% -12.6%)
9.5%
(4.2% -14.5%)
5.4%
(2.4% - 8.3%)
12.8%
(5.7% -19.3%)
8.2%
(3.6% -12. 6%)
10%
(4.4% -15. 2%)
5.1%
(2.2% - 7.9%)
15/352
10.4%
(4.6% -15.8%)
9.1%
(4% -13. 8%)
9.7%
(4.3% -14.8%)
6.8%
(3% -10.5%)
6.1%
(2.7% - 9.4%)
5.8%
(2.5% - 8.9%)
10.1%
(4.5% -15.4%)
5%
(2.2% - 7.8%)
5.9%
(2.6% - 9%)
8.3%
(3.6% -12.7%)
5.4%
(2.4% - 8.3%)
5.1%
(2.2% - 7.8%)
3.1%
(1.3% -4. 8%)
8.4%
(3. 7% -12. 8%)
3%
(1.3% -4. 7%)
13/35
8.1%
(3.6% -12.5%)
7.2%
(3. 2% -11.1%)
7.6%
(3.3% -11. 6%)
6.8%
(3% -10.5%)
5.1%
(2.2% - 7.9%)
5.8%
(2.5% - 8.9%)
7.9%
(3.5% -12.2%)
5%
(2.2% - 7.8%)
5.5%
(2.4% - 8.6%)
7.4%
(3.2% - 1 1 .3%)
5.4%
(2.4% - 8.3%)
5.1%
(2.2% - 7.8%)
3.1%
(1.3% -4. 8%)
6.5%
(2. 9% -10%)
3%
(1.3% -4. 7%)
12/35
7%
(3.1% -10. 8%)
6.1%
(2.7% - 9.4%)
6.5%
(2.8% -10%)
6%
(2.6% - 9.2%)
4.2%
(1.8% -6. 5%)
5.8%
(2.5% - 8.9%)
6.8%
(3% -10. 5%)
4.4%
(1.9% -6.8%)
4.6%
(2% -7.1%)
6.3%
(2.8% - 9.7%)
4.8%
(2.1% -7.4%)
5.1%
(2.2% - 7.8%)
3.1%
(1.3% -4. 8%)
5.5%
(2.4% - 8.5%)
3%
(1.3% -4. 7%)
13/30
8.1%
(3.6% -12.5%)
6.9%
(3% -10.6%)
7.6%
(3.3% -11. 6%)
6.8%
(3% -10. 5%)
4.3%
(1.9% -6. 7%)
4%
(1.8% -6. 3%)
7.9%
(3.5% -12.2%)
3.4%
(1.5% -5. 3%)
4.1%
(1.8% -6.4%)
6.2%
(2.7% - 9.6%)
4.6%
(2% - 7.2%)
3.4%
(1.5% -5. 3%)
1.7%
(0.7% - 2.7%)
6.3%
(2.8% - 9.8%)
1 .7%
(0.7% - 2.6%)
12/25
6.8%
(3% -10.5%)
4.7%
(2% - 7.3%)
5.5%
(2.4% - 8.4%)
6%
(2.6% - 9.2%)
2.5%
(1.1% -3. 9%)
2.3%
(1%-3.6%)
6.8%
(3% -10.5%)
1.8%
(0.8% -2.7%)
2.3%
(1%-3.7%)
4.1%
(1.8% -6.4%)
2.8%
(1.2% -4.3%)
1.8%
(0.8% - 2.8%)
0.4%
(0.2% - 0.6%)
4.2%
(1.8% -6. 6%)
0.3%
(0.1% -0.5%)
1Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-92
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-12%
(-12% --13%)
-9%
(-80/0 . -go/0)
-38%
(-37% - -40%)
0%
(0% - 0%)
-37%
(-36% - -38%)
-182%
(-175% --189%)
-8%
(-80/0 - -go/0)
-114%
(-110% --11 8%)
-37%
(-36% - -38%)
-14%
(-14% --15%)
0%
(0% - 0%)
-137%
(-131% --144%)
-188%
(-184% --192%)
-17%
(-17% --18%)
-53%
(-52% - -54%)
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%)
13/35
21%
(20% -21%)
19%
(18% -19%)
21%
(21% -22%)
0%
(0% - 0%)
13%
(13% -14%)
0%
(0% - 0%)
21%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
20%
(20% -21%)
0%
(0% - 0%)
12/35
32%
(31% -32%)
30%
(29% - 30%)
32%
(32% - 33%)
11%
(11% -11%)
26%
(25% - 26%)
0%
(0% - 0%)
32%
(32% - 33%)
11%
(11% -12%)
20%
(19% -20%)
24%
(23% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
31%
(31% -32%)
0%
(0% - 0%)
13/30
21%
(20% -21%)
22%
(21% -22%)
21%
(21% -22%)
0%
(0% - 0%)
24%
(23% - 24%)
34%
(34% - 35%)
21%
(21% -22%)
30%
(29% - 30%)
27%
(27% - 28%)
24%
(24% - 25%)
13%
(13% -13%)
29%
(29% - 29%)
48%
(48% - 49%)
22%
(22% - 23%)
34%
(33% - 34%)
12/25
33%
(32% - 34%)
44%
(43% - 45%)
42%
(42% - 43%)
11%
(11% -11%)
48%
(48% - 49%)
69%
(69% - 69%)
32%
(32% - 33%)
60%
(60% - 60%)
55%
(54% - 55%)
49%
(49% - 50%)
45%
(45% - 46%)
59%
(58% - 59%)
98%
(98% - 98%)
45%
(440/0 - 450/0)
68%
(67% - 68%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-93
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-9% - -9%)
-39%
(-37% --41%)
0%
(0% - 0%)
-43%
(-42% - -45%)
-174%
(-167% --181%)
-8%
(-80/0 . -go/0)
-126%
(-121% --130%)
-41%
(-40% - -42%)
-15%
(-14% --15%)
0%
(0% - 0%)
-155%
(-149% --161%)
-268%
(-263% - -273%)
-20%
(-19% --20%)
-79%
(-78% - -80%)
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%)
13/35
21%
(21% -22%)
20%
(20% - 20%)
22%
(21% -22%)
0%
(0% - 0%)
16%
(15% -16%)
0%
(0% - 0%)
21%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
23%
(23% - 23%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
31%
(31% -32%)
33%
(32% - 34%)
13%
(13% -13%)
30%
(30% -31%)
0%
(0% - 0%)
32%
(31% -33%)
13%
(12% -13%)
22%
(21% -22%)
24%
(23% - 24%)
10%
(10% -10%)
0%
(0% - 0%)
0%
(0% - 0%)
35%
(35% - 36%)
0%
(0% - 0%)
13/30
21%
(21% -22%)
23%
(23% - 24%)
22%
(21% -22%)
0%
(0% - 0%)
28%
(27% - 28%)
33%
(32% - 33%)
21%
(21% -22%)
33%
(32% - 33%)
30%
(30% - 30%)
25%
(24% - 25%)
12%
(12% -13%)
33%
(32% - 33%)
69%
(69% - 69%)
25%
(25% - 26%)
50%
(50% - 50%)
12/25
34%
(33% - 35%)
47%
(46% - 47%)
43%
(43% - 44%)
13%
(13% -13%)
56%
(56% - 57%)
66%
(66% - 66%)
32%
(31% -33%)
66%
(66% - 66%)
60%
(60% -61%)
50%
(49% -51%)
44%
(44% - 45%)
66%
(65% - 66%)
100%
(100% -100%)
51%
(51% -52%)
100%
(100% -100%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-94
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-13%
(-12% --13%)
-9%
(-9% --10%)
-39%
(-38% --41%)
0%
(0% - 0%)
-45%
(-44% - -46%)
-157%
(-150% --165%)
-8%
(-80/0 - -go/0)
-124%
(-120% --128%)
-41%
(-40% - -42%)
-15%
(-14% --15%)
0%
(0% - 0%)
-152%
(-147% --158%)
-168%
(-164% --172%)
-19%
(-19% --20%)
-69%
(-68% - -70%)
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%)
13/35
21%
(21% -22%)
21%
(20% -21%)
22%
(21% -22%)
0%
(0% - 0%)
16%
(16% -17%)
0%
(0% - 0%)
22%
(21% -22%)
0%
(0% - 0%)
5%
(5% - 5%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
0%
(0% - 0%)
22%
(22% - 23%)
0%
(0% - 0%)
12/35
32%
(32% - 33%)
32%
(32% - 33%)
33%
(32% - 34%)
12%
(12% -12%)
31%
(31% -32%)
0%
(0% - 0%)
33%
(32% - 33%)
12%
(12% -13%)
21%
(21% -22%)
24%
(24% - 24%)
11%
(11% -11%)
0%
(0% - 0%)
0%
(0% - 0%)
34%
(34% - 35%)
0%
(0% - 0%)
13/30
21%
(21% -22%)
24%
(23% - 24%)
22%
(21% -22%)
0%
(0% - 0%)
29%
(29% - 29%)
30%
(29% - 30%)
22%
(21% -22%)
32%
(32% - 33%)
30%
(29% - 30%)
25%
(24% - 25%)
14%
(14% -14%)
32%
(32% - 32%)
44%
(43% - 44%)
25%
(24% - 25%)
44%
(44% - 44%)
12/25
34%
(33% - 35%)
48%
(47% - 49%)
44%
(43% - 44%)
12%
(12% -12%)
58%
(58% - 59%)
60%
(60% -61%)
33%
(32% - 33%)
65%
(65% - 65%)
60%
(60% - 60%)
50%
(50% -51%)
49%
(48% - 49%)
65%
(64% - 65%)
88%
(88% - 88%)
50%
(49% - 50%)
89%
(89% - 89%)
1 Based on follow-up through 2000, using models with 44 individual and 7 ecological covariates (see Table 33 in Krewski et al., 2009).
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-95
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-22.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Lung Cancer Mortality Associated with Long-Term Exposure to
(Exposure Period: 1999 - 2000):  Based on 2005 Air Quality Data*
          100%
           80%
       I
       OT
       ^   -20%
       O
       p
       I
       •s
       1
       a:
       01
       o
       s.
 -40% -
 -60% -
 -80% -
-100% -
-120% -
-140%
          -160%
          -180%
          -200%
          ^    I
Ji
                  2005 air
                   quality
                   15/35**      13/35       12/35

                           Alternative Standard
                                    13/30
12/25
                         -»-Atlanta, GA 98 (43-149);  11% (4.9%-16.7%)
                         -m- Baltimore, MD 106 (47-162); 11.1%  (4.9%-16.9%)
                         -*- Birmingham, AL  66 (29-101); 10.4%  (4.6%-15.8%)
                             Dallas, TX 69 (31 -107); 8.3% (3.7%-12.7%)
                         -*- Detroit, Ml 112 (49-171);  8.9% (3.9%-13.6%)
                         -•- Fresno, CA 13 (6-20); 4.5% (2%-7%)
                         -i- Houston, TX 120 (53-183); 10.3% (4.6%-15.7%)
                         	 Los Angeles, CA 174 (76-269); 5.8%  (2.5%-8.9%)
                           - New York, NY 190 (83-292); 6.9% (3%-10.7%)
                         -•- Philadelphia, PA 83 (37-127); 8.5% (3.8%-13.1%)
                         -m- Phoenix, AZ 84  (37-129); 6.1% (2.7%-9.4%)
                         -*- Pittsburgh, PA 57 (25 - 88); 6% (2.6% - 9.3%)
                             Salt Lake City, UT 5 (2-7); 2.6% (1.1%-4.1%)
                             St. Louis, MO 136 (60-207); 10.3% (4.6%-15.7%)
                             Tacoma.WA 17 (8-27); 4.6% (2%-7.1%)
*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.
                                                E-96
   September 2009
                                                                       Draft - Do Not Quote or Cite

-------
Figure E-23.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Lung Cancer Mortality Associated with Long-Term Exposure to
(Exposure Period: 1999 - 2000):  Based on 2006 Air Quality Data*
          100%
           80%
           60%
       •5
       re
       •O
       v>
       o
       I
       c
       O
       'o
       3
       •D
       
-------
 Figure E-24. Estimated Percent Reductions From the Current Standard to Alternative
Standards in Lung Cancer Mortality Associated with Long-Term Exposure to
(Exposure Period: 1999 - 2000): Based on 2007 Air Quality Data*
          100%
           80%
       I
       OT
       ^   -20%
       O
       p
       I
       s  -
 -40% -
 -60% -
 -80% -
 100% -
       01
       o
       8.
-120% -
-140%
          -160%
          -180%
          -200%
                  2007 air
                   quality
                   15/35**      13/35       12/35

                           Alternative Standard
13/30
12/25
                         -»-Atlanta, GA 98 (43-149); 10.4% (4.6%-15.8%)
                         -•- Baltimore, MD 87 (38-133); 9.1%  (4%-13.8%)
                         -*- Birmingham, AL 63  (28-96); 9.7% (4.3%-14.8%)
                           - Dallas, TX 59 (26-91); 6.8% (3%-10.5%)
                         -*- Detroit, Ml 76 (33-117); 6.1% (2.7%-9.4%)
                         -•- Fresno, CA 17 (7-26); 5.8% (2.5%-8.9%)
                         -i- Houston, TX 125 (55-190); 10.1% (4.5%-15.4%)
                         	 Los Angeles,  CA 153 (67-238); 5% (2.2%-7.8%)
                             New York, NY 163 (71 -252); 5.9% (2.6%-9%)
                         -+- Philadelphia, PA 80 (35-123); 8.3% (3.6%-12.7%)
                         -m- Phoenix, AZ 80  (35-123); 5.4% (2.4%-8.3%)
                         -*- Pittsburgh, PA 47 (21 -73); 5.1% (2.2%-7.8%)
                             Salt Lake City, UT 6 (3 - 9); 3.1 % (1.3% - 4.8%)
                             St. Louis, MO 112 (49-171); 8.4% (3.7%-12.8%)
                             Tacoma, WA  12 (5-18); 3% (1.3% - 4.7%)
*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.
                                                E-98
   September 2009
                                                                       Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
191
(36 - 344)
271
(110-429)
49
(-77-173)
151
(36 - 264)
174
(-31 - 375)
80
(11 -147)
245
(50 - 438)
134
(-192-458)
858
(504-1209)
229
(85 - 372)
240
(40 - 438)
233
(69 - 395)
52
(11 -93)
281
(81 - 478)
59
(10-107)
15/352
175
(33-316)
256
(104-406)
38
(-60-135)
151
(36 - 264)
141
(-25 - 305)
44
(6 - 82)
232
(47-414)
85
(-121 -289)
714
(419-1007)
211
(78 - 342)
240
(40 - 438)
135
(40 - 230)
33
(7 - 59)
252
(73 - 429)
48
(8 - 87)
13/35
149
(28 - 269)
224
(91 - 355)
33
(-51 -115)
151
(36 - 264)
129
(-23 - 280)
44
(6 - 82)
197
(40 - 352)
85
(-121 -289)
696
(408-981)
197
(73-321)
240
(40-438)
135
(40 - 230)
33
(7 - 59)
219
(63 - 373)
48
(8 - 87)
12/35
136
(26 - 245)
206
(84 - 327)
30
(-46-105)
140
(34 - 245)
119
(-21 - 257)
44
(6 - 82)
180
(37-321)
80
(-114-273)
639
(375 - 902)
181
(67 - 295)
228
(38-417)
135
(40 - 230)
33
(7 - 59)
201
(58 - 342)
48
(8 - 87)
13/30
149
(28 - 269)
219
(89 - 348)
33
(-51 -115)
151
(36 - 264)
121
(-21 - 260)
38
(5 - 70)
197
(40 - 352)
72
(-103-247)
611
(358-861)
180
(67 - 293)
225
(37-411)
116
(34-197)
28
(6-51)
215
(62 - 367)
41
(7 - 74)
12/25
134
(25-241)
182
(74 - 289)
27
(-42 - 96)
140
(34 - 245)
100
(-18-216)
31
(4 - 58)
180
(37-321)
60
(-86 - 205)
507
(297-716)
150
(55 - 244)
187
(31 - 342)
96
(28-163)
23
(5 - 42)
179
(51 - 305)
34
(6 - 62)
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.  Numbers are rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-99
September 2009
Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
188
(36 - 338)
248
(101 -393)
48
(-75-169)
129
(31 - 226)
145
(-25-313)
84
(12-155)
258
(53 - 460)
124
(-178-423)
786
(461 -1107)
226
(83 - 366)
255
(42 - 466)
210
(62 - 357)
48
(10-86)
238
(69 - 406)
48
(8 - 88)
15/352
172
(33-310)
234
(95 - 372)
38
(-59-132)
129
(31 - 226)
117
(-21 - 254)
46
(7 - 86)
243
(50 - 434)
78
(-112-267)
654
(384 - 922)
208
(77 - 337)
255
(42 - 466)
122
(36 - 208)
30
(6 - 54)
214
(62 - 364)
39
(7-71)
13/35
146
(28 - 264)
205
(83 - 325)
32
(-50-113)
129
(31 - 226)
108
(-19-233)
46
(7 - 86)
207
(42 - 369)
78
(-112-267)
637
(374 - 898)
194
(72-315)
255
(42 - 466)
122
(36 - 208)
30
(6 - 54)
185
(53-316)
39
(7-71)
12/35
134
(25-241)
188
(76 - 299)
29
(-45-103)
120
(29-210)
99
(-17-214)
46
(7 - 86)
188
(38 - 337)
74
(-106-252)
585
(343 - 826)
179
(66 - 290)
243
(40 - 443)
122
(36 - 208)
30
(6 - 54)
170
(49 - 290)
39
(7-71)
13/30
146
(28 - 264)
201
(81 -318)
32
(-50-113)
129
(31 -226)
100
(-18-217)
40
(6 - 73)
207
(42 - 369)
67
(-96 - 228)
559
(328 - 789)
178
(65 - 288)
239
(40 - 436)
104
(31 -177)
26
(5 - 46)
183
(53-311)
34
(6-61)
12/25
132
(25 - 237)
167
(68 - 265)
26
(-41 - 94)
120
(29-210)
83
(-15-180)
33
(5-61)
188
(38 - 337)
55
(-79-189)
464
(272 - 655)
147
(54 - 240)
199
(33 - 363)
87
(25-147)
21
(4 - 38)
151
(44 - 258)
28
(5-51)
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.  Numbers are rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-100
September 2009
Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
193
(37 - 347)
238
(97 - 377)
48
(-74-168)
137
(33 - 240)
138
(-24 - 298)
92
(13-168)
254
(52 - 454)
125
(-180-427)
793
(465-1117)
224
(83 - 363)
242
(40-441)
212
(62 - 359)
58
(12-102)
248
(71 -421)
52
(9 - 94)
15/352
177
(34-319)
225
(91 - 357)
37
(-58-131)
137
(33 - 240)
112
(-20 - 242)
51
(7 - 94)
240
(49 - 429)
79
(-113-270)
659
(387 - 930)
206
(76 - 334)
242
(40-441)
123
(36 - 209)
36
(7 - 65)
222
(64 - 378)
42
(7 - 76)
13/35
151
(29-271)
197
(80-312)
32
(-49-112)
137
(33 - 240)
102
(-18-222)
51
(7 - 94)
204
(42 - 365)
79
(-113-270)
642
(377 - 906)
193
(71 -313)
242
(40-441)
123
(36 - 209)
36
(7 - 65)
193
(56 - 328)
42
(7 - 76)
12/35
137
(26 - 248)
181
(73 - 287)
29
(-45-102)
127
(31 - 223)
94
(-16-204)
51
(7 - 94)
186
(38 - 333)
74
(-107-255)
590
(346 - 833)
177
(65 - 287)
230
(38-419)
123
(36 - 209)
36
(7 - 65)
177
(51 -301)
42
(7 - 76)
13/30
151
(29-271)
192
(78 - 306)
32
(-49-112)
137
(33-240)
95
(-17-207)
43
(6 - 80)
204
(42 - 365)
67
(-97-231)
564
(331 - 795)
176
(65 - 286)
226
(37-413)
105
(31 -179)
31
(6 - 56)
190
(55 - 323)
36
(6 - 65)
12/25
135
(26 - 244)
160
(65 - 254)
26
(-41 - 93)
127
(31 - 223)
79
(-14-171)
36
(5 - 67)
186
(38 - 333)
56
(-80-191)
468
(274-661)
146
(54 - 238)
188
(31 - 344)
87
(26-148)
26
(5 - 46)
157
(45 - 268)
30
(5 - 54)
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.  Numbers are rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-101
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
1 .3%
(0.3% - 2.4%)
2%
(0.8% - 3.2%)
0.5%
(-0.8% - 1 .8%)
1 .2%
(0.3% -2.1%)
1%
(-0.2% - 2.2%)
1 .5%
(0.2% - 2.7%)
1 .4%
(0.3% - 2.5%)
0.2%
(-0.3% - 0.8%)
1 .7%
(1%-2.3%)
1 .6%
(0.6% - 2.6%)
1.1%
(0.2% - 2%)
1 .7%
(0.5% - 2.9%)
1.1%
(0.2% - 2%)
1 .5%
(0.4% - 2.6%)
1 .2%
(0.2% - 2.2%)
15/352
1 .2%
(0.2% - 2.2%)
1 .9%
(0.8% - 3%)
0.4%
(-0.6% - 1 .4%)
1 .2%
(0.3% -2.1%)
0.8%
(-0.1% -1.8%)
0.8%
(0.1% -1.5%)
1 .3%
(0.3% - 2.3%)
0.2%
(-0.2% - 0.5%)
1 .4%
(0.8% - 2%)
1 .5%
(0.6% - 2.4%)
1.1%
(0.2% - 2%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.3%)
1 .4%
(0.4% - 2.4%)
1%
(0.2% - 1 .8%)
13/35
1%
(0.2% - 1 .9%)
1 .7%
(0.7% - 2.6%)
0.3%
(-0.5% - 1 .2%)
1 .2%
(0.3% -2.1%)
0.7%
(-0.1% -1.6%)
0.8%
(0.1% -1.5%)
1.1%
(0.2% - 2%)
0.2%
(-0.2% - 0.5%)
1 .3%
(0.8% - 1 .9%)
1 .4%
(0.5% - 2.3%)
1.1%
(0.2% - 2%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.3%)
1 .2%
(0.3% - 2%)
1%
(0.2% - 1 .8%)
12/35
0.9%
(0.2% - 1 .7%)
1 .5%
(0.6% - 2.4%)
0.3%
(-0.5% -1.1%)
1.1%
(0.3% - 2%)
0.7%
(-0.1% -1.5%)
0.8%
(0.1% -1.5%)
1%
(0.2% - 1 .8%)
0.1%
(-0.2% - 0.5%)
1 .2%
(0.7% - 1 .7%)
1 .3%
(0.5% -2.1%)
1%
(0.2% - 1 .9%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.3%)
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 .2%)
1 .2%
(0.3% -2.1%)
0.7%
(-0.1% -1.5%)
0.7%
(0.1% -1.3%)
1.1%
(0.2% - 2%)
0.1%
(-0.2% - 0.4%)
1 .2%
(0.7% - 1 .7%)
1 .3%
(0.5% -2.1%)
1%
(0.2% - 1 .9%)
0.9%
(0.3% - 1 .5%)
0.6%
(0.1% -1.1%)
1 .2%
(0.3% - 2%)
0.8%
(0.1% -1.5%)
12/25
0.9%
(0.2% - 1 .7%)
1 .3%
(0.5% -2.1%)
0.3%
(-0.4% - 1 %)
1.1%
(0.3% - 2%)
0.6%
(-0.1% -1.3%)
0.6%
(0.1% -1.1%)
1%
(0.2% - 1 .8%)
0.1%
(-0.2% - 0.4%)
1%
(0.6% - 1 .4%)
1.1%
(0.4% - 1 .7%)
0.9%
(0.1% -1.6%)
0.7%
(0.2% - 1 .2%)
0.5%
(0.1% -0.9%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.3%)
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. Percents are rounded to the nearest hundredth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-102
September 2009
Draft - Do Not Quote or Cite

-------
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 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
1 .3%
(0.2% - 2.3%)
1 .8%
(0.7% - 2.9%)
0.5%
(-0.8% - 1 .8%)
1%
(0.2% - 1 .8%)
0.8%
(-0.1% -1.8%)
1 .5%
(0.2% - 2.8%)
1 .4%
(0.3% - 2.5%)
0.2%
(-0.3% - 0.8%)
1 .5%
(0.9% -2.1%)
1 .6%
(0.6% - 2.6%)
1.1%
(0.2% -2.1%)
1 .6%
(0.5% - 2.7%)
1%
(0.2% - 1 .8%)
1 .3%
(0.4% - 2.2%)
1%
(0.2% - 1 .8%)
15/352
1 .2%
(0.2% -2.1%)
1 .7%
(0.7% - 2.7%)
0.4%
(-0.6% - 1 .4%)
1%
(0.2% - 1 .8%)
0.7%
(-0.1% -1.5%)
0.8%
(0.1% -1.6%)
1 .3%
(0.3% - 2.3%)
0.1%
(-0.2% - 0.5%)
1 .3%
(0.7% - 1 .8%)
1 .5%
(0.5% - 2.4%)
1.1%
(0.2% -2.1%)
0.9%
(0.3% - 1 .5%)
0.6%
(0.1% -1.1%)
1 .2%
(0.3% - 2%)
0.8%
(0.1% -1.4%)
13/35
1%
(0.2% - 1 .8%)
1 .5%
(0.6% - 2.4%)
0.3%
(-0.5% - 1 .2%)
1%
(0.2% - 1 .8%)
0.6%
(-0.1% -1.4%)
0.8%
(0.1% -1.6%)
1.1%
(0.2% - 2%)
0.1%
(-0.2% - 0.5%)
1 .2%
(0.7% - 1 .7%)
1 .4%
(0.5% - 2.2%)
1.1%
(0.2% -2.1%)
0.9%
(0.3% - 1 .5%)
0.6%
(0.1% -1.1%)
1%
(0.3% - 1 .7%)
0.8%
(0.1% -1.4%)
12/35
0.9%
(0.2% - 1 .6%)
1 .4%
(0.6% - 2.2%)
0.3%
(-0.5% -1.1%)
0.9%
(0.2% - 1 .7%)
0.6%
(-0.1% -1.2%)
0.8%
(0.1% -1.6%)
1%
(0.2% - 1 .8%)
0.1%
(-0.2% - 0.5%)
1.1%
(0.7% - 1 .6%)
1 .3%
(0.5% -2.1%)
1.1%
(0.2% - 2%)
0.9%
(0.3% - 1 .5%)
0.6%
(0.1% -1.1%)
0.9%
(0.3% - 1 .6%)
0.8%
(0.1% -1.4%)
13/30
1%
(0.2% - 1 .8%)
1 .5%
(0.6% - 2.3%)
0.3%
(-0.5% - 1 .2%)
1%
(0.2% - 1 .8%)
0.6%
(-0.1% -1.3%)
0.7%
(0.1% -1.3%)
1.1%
(0.2% - 2%)
0.1%
(-0.2% - 0.4%)
1.1%
(0.6% - 1 .5%)
1 .3%
(0.5% -2.1%)
1.1%
(0.2% - 1 .9%)
0.8%
(0.2% - 1 .3%)
0.5%
(0.1% -1%)
1%
(0.3% - 1 .7%)
0.7%
(0.1% -1.2%)
12/25
0.9%
(0.2% - 1 .6%)
1 .2%
(0.5% - 1 .9%)
0.3%
(-0.4% -1%)
0.9%
(0.2% - 1 .7%)
0.5%
(-0.1% -1.1%)
0.6%
(0.1% -1.1%)
1%
(0.2% - 1 .8%)
0.1%
(-0.1% -0.3%)
0.9%
(0.5% - 1 .3%)
1%
(0.4% - 1 .7%)
0.9%
(0.1% -1.6%)
0.6%
(0.2% -1.1%)
0.4%
(0.1% -0.8%)
0.8%
(0.2% - 1 .4%)
0.6%
(0.1% -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. Percents are rounded to the nearest hundredth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-103
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
1 .3%
(0.2% - 2.3%)
1 .7%
(0.7% - 2.8%)
0.5%
(-0.8% - 1 .7%)
1.1%
(0.3% - 1 .9%)
0.8%
(-0.1% -1.8%)
1 .6%
(0.2% - 3%)
1 .4%
(0.3% - 2.4%)
0.2%
(-0.3% - 0.8%)
1 .5%
(0.9% - 2.2%)
1 .6%
(0.6% - 2.6%)
1%
(0.2% - 1 .9%)
1 .6%
(0.5% - 2.7%)
1 .2%
(0.2% -2.1%)
1 .4%
(0.4% - 2.3%)
1%
(0.2% - 1 .9%)
15/352
1 .2%
(0.2% -2.1%)
1 .7%
(0.7% - 2.6%)
0.4%
(-0.6% - 1 .4%)
1.1%
(0.3% - 1 .9%)
0.7%
(-0.1% -1.4%)
0.9%
(0.1% -1.7%)
1 .3%
(0.3% - 2.3%)
0.1%
(-0.2% - 0.5%)
1 .3%
(0.7% - 1 .8%)
1 .5%
(0.5% - 2.4%)
1%
(0.2% - 1 .9%)
0.9%
(0.3% - 1 .6%)
0.7%
(0.2% - 1 .3%)
1 .2%
(0.4% -2.1%)
0.8%
(0.1% -1.5%)
13/35
1%
(0.2% - 1 .8%)
1 .4%
(0.6% - 2.3%)
0.3%
(-0.5% - 1 .2%)
1.1%
(0.3% - 1 .9%)
0.6%
(-0.1% -1.3%)
0.9%
(0.1% -1.7%)
1.1%
(0.2% - 1 .9%)
0.1%
(-0.2% - 0.5%)
1 .2%
(0.7% - 1 .7%)
1 .4%
(0.5% - 2.2%)
1%
(0.2% - 1 .9%)
0.9%
(0.3% - 1 .6%)
0.7%
(0.2% - 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%
(-0.5% -1.1%)
1%
(0.2% - 1 .7%)
0.6%
(-0.1% -1.2%)
0.9%
(0.1% -1.7%)
1%
(0.2% - 1 .8%)
0.1%
(-0.2% - 0.5%)
1.1%
(0.7% - 1 .6%)
1 .3%
(0.5% -2.1%)
1%
(0.2% - 1 .8%)
0.9%
(0.3% - 1 .6%)
0.7%
(0.2% - 1 .3%)
1%
(0.3% - 1 .7%)
0.8%
(0.1% -1.5%)
13/30
1%
(0.2% - 1 .8%)
1 .4%
(0.6% - 2.2%)
0.3%
(-0.5% - 1 .2%)
1.1%
(0.3% - 1 .9%)
0.6%
(-0.1% -1.2%)
0.8%
(0.1% -1.4%)
1.1%
(0.2% - 1 .9%)
0.1%
(-0.2% - 0.4%)
1.1%
(0.6% - 1 .5%)
1 .3%
(0.5% - 2%)
1%
(0.2% - 1 .8%)
0.8%
(0.2% - 1 .3%)
0.6%
(0.1% -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%)
0.3%
(-0.4% -1%)
1%
(0.2% - 1 .7%)
0.5%
(-0.1% -1%)
0.6%
(0.1% -1.2%)
1%
(0.2% - 1 .8%)
0.1%
(-0.1% -0.3%)
0.9%
(0.5% - 1 .3%)
1%
(0.4% - 1 .7%)
0.8%
(0.1% -1.5%)
0.7%
(0.2% -1.1%)
0.5%
(0.1% -0.9%)
0.9%
(0.2% - 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.  Percents are rounded to the nearest hundredth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-104
September 2009
Draft - Do Not Quote or Cite

-------
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):
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%)
-72%
(-72% - -73%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -23%)
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%)
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%)
0%
(0% - 0%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(1 9% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(1 6% - 1 6%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(1 0% - 1 0%)
14%
(1 4% - 1 4%)
5%
(5% - 5%)
0%
(0% - 0%)
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%)
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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                        E-105
September 2009
Draft - Do Not Quote or Cite

-------
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):
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%)
-72%
(-72% - -73%)
-58%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -23%)
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%)
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%)
0%
(0% - 0%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
20%
(20% - 20%)
23%
(22% - 23%)
7%
(7% - 7%)
16%
(16% -16%)
0%
(0% - 0%)
22%
(22% - 23%)
6%
(6% - 6%)
10%
(10% -10%)
14%
(140/0- 140/0)
5%
(5% - 5%)
0%
(0% - 0%)
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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                        E-106
September 2009
Draft - Do Not Quote or Cite

-------
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 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):
Recent PM25
Concentrations
-9%
(-9% - -9%)
-6%
(-50/0 - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-79% - -82%)
-6%
(-60/0 - -60/o)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-72% - -73%)
-58%
(-58% - -59%)
-1 2%
(-12% --12%)
-23%
(-23% - -23%)
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%)
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%)
0%
(0% - 0%)
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%)
0%
(0% - 0%)
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%
(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% - 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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                        E-107
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-25.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5:
Based on 2005 Air Quality Data*
          30%

          20%

          10%

           0%

          -10%

          -20%

          -30%

          -40%

          -50%

          -60%

          -70%

          -80%

          -90%
3
                 2005 air
                  quality
   15/35**      13/35       12/35

           Alternative Standard
13/30
12/25
                        -•- Atlanta, GA 175 (33-316); 1.2% (0.2%-2.2%)
                        -m- Baltimore, MD 256 (104-406); 1.9% (0.8%-3%)
                        -*- Birmingham, AL 38 (-60-135); 0.4% (-0.6%-1.4%)
                            Dallas, TX 151 (36 - 264); 1.2% (0.3% - 2.1 %)
                        -*- Detroit, Ml 141 (-25-305); 0.8%  (-0.1%-1.8%)
                        -•- Fresno, CA 44 (6-82); 0.8% (0.1%-1.5%)
                        -i- Houston, TX 232 (47-414); 1.3% (0.3%-2.3%)
                        — Los Angeles, CA 85 (-121 -289); 0.2%  (-0.2%-0.5%)
                            New York, NY 714 (419-1007);  1.4% (0.8%-2%)
                        -•- Philadelphia, PA 211 (78 - 342); 1.5% (0.6% - 2.4%)
                        -•- Phoenix, AZ 240 (40-438); 1.1% (0.2%-2%)
                        -*- Pittsburgh, PA 135 (40-230); 1% (0.3%-1.7%)
                            Salt Lake City, UT  33 (7 - 59); 0.7% (0.1 % -1.3%)
                            St. Louis, MO 252 (73 - 429); 1.4% (0.4% - 2.4%)
                        -•- Tacoma, WA 48 (8-87);  1% (0.2%-1.8%)
*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.
                                                E-108
   September 2009
                                                        Draft - Do Not Quote or Cite

-------
Figure E-26.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5:
Based on 2006 Air Quality Data*
          30%
          20%
=  -20%
E
8  -30%
I
          -40%
       S  -I
       01
       o
       8.
          50%
   -60%
          -70%
          -80%
          -90%
                           1
                         1
                     UL
                 2006 air
                  quality
                     15/35**      13/35       12/35

                              Alternative Standard
13/30
12/25
                        -•- Atlanta, GA 172 (33-310); 1.2% (0.2%-2.1%)
                        -m- Baltimore, MD 234 (95-372);  1.7% (0.7%-2.7%)
                        -*- Birmingham, AL 38 (-59-132); 0.4% (-0.6%-1.4%)
                            Dallas, TX 129 (31 -226); 1% (0.2%-1.8%)
                        -*- Detroit, Ml 117 (-21 -254); 0.7%  (-0.1%-1.5%)
                        -•- Fresno, CA 46 (7-86); 0.8% (0.1%-1.6%)
                        -i- Houston, TX 243 (50 - 434); 1.3% (0.3% - 2.3%)
                        	 Los Angeles, CA 78 (-112-267); 0.1% (-0.2%-0.5%)
                            New York, NY 654 (384-922); 1.3%  (0.7%-1.8%)
                        -•- Philadelphia, PA 208 (77 - 337);  1.5% (0.5% - 2.4%)
                        -m- Phoenix, AZ 255 (42-466); 1.1% (0.2%-2.1%)
                        -*- Pittsburgh, PA 122 (36-208);  0.9% (0.3%-1.5%)
                            Salt Lake City, UT 30 (6 - 54); 0.6% (0.1 % -1.1 %)
                            St. Louis, MO 214 (62-364); 1.2% (0.3%-2%)
                        -•- Tacoma.WA 39 (7-71); 0.8% (0.1%-1.4%)
*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.
                                                 E-109
   September 2009
                                                                           Draft - Do Not Quote or Cite

-------
Figure E-27.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Non-Accidental Mortality Associated with Short-Term Exposure to PM2.5:
Based on 2007 Air Quality Data*
          10%

           o%
          -10%

          -20%
          -30%

          -40%
          -50%
          -60%

          -70%
          -80%

          -90%
  3
J
                 2007 air
                  quality
    15/35**      13/35      12/35

             Alternative Standard
13/30
12/25
                        -•- Atlanta, GA 177 (34-319); 1.2% (0.2%-2.1%)
                        -m- Baltimore, MD 225 (91 -357);  1.7% (0.7%-2.6%)
                        -*- Birmingham, AL 37 (-58-131); 0.4% (-0.6%-1.4%)
                            Dallas, TX 137 (33-240); 1.1% (0.3%-1.9%)
                        -*- Detroit, Ml 112 (-20-242); 0.7%  (-0.1%-1.4%)
                        -•- Fresno, CA 51  (7-94); 0.9% (0.1%-1.7%)
                        -i- Houston, TX 240 (49 - 429); 1.3% (0.3% - 2.3%)
                        	 Los Angeles, CA 79 (-113-270); 0.1%  (-0.2%-0.5%)
                            New York, NY 659 (387-930); 1.3% (0.7%-1.8%)
                        -•- Philadelphia, PA 206  (76 - 334); 1.5% (0.5% - 2.4%)
                        -m- Phoenix, AZ 242 (40-441); 1% (0.2%-1.9%)
                        -*- Pittsburgh, PA 123 (36-209);  0.9% (0.3%-1.6%)
                            Salt Lake City, UT 36  (7 - 65); 0.7% (0.2% -1.3%)
                            St. Louis, MO 222 (64-378); 1.2% (0.4%-2.1%)
                        -•- Tacoma.WA 42 (7-76); 0.8% (0.1%-1.5%)
*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.
                                                 E-110
   September 2009
                                                          Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
34
(-35-103)
74
(-5-151)
-1
(-61 - 58)
31
(-21 - 82)
86
(-11 -180)
21
(-14-55)
54
(-36-142)
-52
(-231 -126)
552
(323 - 779)
92
(24-159)
83
(-4-169)
70
(-14-151)
14
(-3 - 30)
132
(29 - 232)
15
(-8 - 38)
15/352
32
(-32 - 94)
70
(-5-142)
-1
(-48 - 45)
31
(-21 - 82)
70
(-9-147)
11
(-8 - 30)
51
(-34-134)
-33
(-146-79)
460
(268 - 649)
85
(22-147)
83
(-4-169)
40
(-8 - 88)
9
(-2-19)
118
(26 - 208)
12
(-7-31)
13/35
27
(-28 - 80)
61
(-4-125)
-1
(-41 - 38)
31
(-21 - 82)
64
(-8-135)
11
(-8 - 30)
43
(-29-114)
-33
(-146-79)
448
(261 - 633)
80
(21 -137)
83
(-4-169)
40
(-8 - 88)
9
(-2-19)
103
(23-181)
12
(-7-31)
12/35
25
(-25 - 73)
56
(-4-115)
-1
(-37 - 35)
29
(-19-76)
59
(-7-124)
11
(-8 - 30)
39
(-27-104)
-31
(-137-75)
412
(240 - 582)
73
(19-126)
79
(-3-160)
40
(-8 - 88)
9
(-2-19)
94
(21 - 167)
12
(-7-31)
13/30
27
(-28 - 80)
60
(-4-122)
-1
(-41 - 38)
31
(-21 - 82)
60
(-7-126)
10
(-7 - 26)
43
(-29-114)
-28
(-124-68)
394
(229 - 556)
73
(19-126)
78
(-3-158)
35
(-7 - 75)
7
(-2-16)
101
(23-179)
11
(-6 - 27)
12/25
24
(-25 - 72)
50
(-3-102)
-1
(-34 - 32)
29
(-19-76)
49
(-6-104)
8
(-6 - 22)
39
(-27-104)
-23
(-103-56)
327
(190-463)
61
(16-105)
65
(-3-131)
29
(-6 - 63)
6
(-1-14)
84
(19-148)
9
(-5 - 22)
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.  Numbers are rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-lll
September 2009
Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
34
(-35-101)
68
(-5-138)
-1
(-60 - 56)
27
(-18-71)
72
(-9-151)
22
(-15-57)
56
(-38-149)
-48
(-214-116)
506
(295-713)
91
(24-157)
88
(-4-179)
63
(-13-136)
13
(-3 - 28)
112
(25-198)
13
(-7-31)
15/352
31
(-32 - 93)
64
(-4-130)
-1
(-47 - 44)
27
(-18-71)
58
(-7-122)
12
(-8 - 32)
53
(-36-140)
-30
(-134-73)
421
(246 - 595)
84
(22-144)
88
(-4-179)
36
(-7 - 80)
8
(-2-18)
101
(22-177)
10
(-5 - 26)
13/35
26
(-27 - 79)
56
(-4-114)
-1
(-40 - 38)
27
(-18-71)
53
(-7-112)
12
(-8 - 32)
45
(-31 -119)
-30
(-134-73)
410
(239 - 580)
78
(20-135)
88
(-4-179)
36
(-7 - 80)
8
(-2-18)
87
(19-154)
10
(-5 - 26)
12/35
24
(-25 - 72)
51
(-3-105)
-1
(-36 - 34)
25
(-16-66)
49
(-6 - 1 03)
12
(-8 - 32)
41
(-28 - 1 09)
-28
(-127-69)
377
(220 - 533)
72
(19-124)
84
(-4-170)
36
(-7 - 80)
8
(-2-18)
80
(18-142)
10
(-5 - 26)
13/30
26
(-27 - 79)
55
(-4-112)
-1
(-40 - 38)
27
(-18-71)
50
(-6-105)
10
(-7 - 27)
45
(-31 -119)
-26
(-115-63)
360
(210-509)
72
(19-124)
83
(-4-168)
31
(-6 - 68)
7
(-2-15)
86
(19-152)
9
(-5 - 22)
12/25
24
(-24-71)
45
(-3 - 93)
-1
(-33-31)
25
(-16-66)
41
(-5 - 87)
9
(-6 - 23)
41
(-28-109)
-21
(-95 - 52)
299
(174-424)
60
(15-103)
69
(-3-140)
26
(-5 - 56)
6
(-1-12)
71
(16-126)
7
(-4-18)
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.  Numbers are rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-112
September 2009
Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m) Standards
(Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
35
(-36 - 1 04)
65
(-4-132)
-1
(-59 - 56)
29
(-19-75)
68
(-9-143)
24
(-17-62)
56
(-38-147)
-48
(-216-117)
510
(298 - 720)
90
(23-155)
84
(-4-170)
63
(-13-137)
15
(-4 - 33)
116
(26 - 205)
14
(-7 - 34)
15/352
32
(-33 - 95)
61
(-4-125)
-1
(-46 - 44)
29
(-19-75)
55
(-7-117)
13
(-9 - 35)
52
(-36-139)
-30
(-136-74)
425
(248 - 600)
83
(22-143)
84
(-4-170)
37
(-7 - 80)
10
(-2-21)
104
(23-184)
11
(-6 - 27)
13/35
27
(-28-81)
54
(-4-110)
-1
(-39 - 37)
29
(-19-75)
51
(-6 - 1 07)
13
(-9 - 35)
45
(-30-118)
-30
(-136-74)
414
(241 - 585)
78
(20-134)
84
(-4-170)
37
(-7 - 80)
10
(-2-21)
91
(20-160)
11
(-6 - 27)
12/35
25
(-25 - 74)
49
(-3-101)
-1
(-36 - 34)
26
(-17-70)
47
(-6 - 98)
13
(-9 - 35)
41
(-28 - 1 08)
-29
(-128-70)
381
(222 - 538)
71
(18-123)
80
(-3-161)
37
(-7 - 80)
10
(-2-21)
83
(18-147)
11
(-6 - 27)
13/30
27
(-28-81)
52
(-4-107)
-1
(-39 - 37)
29
(-19-75)
47
(-6-100)
11
(-8 - 30)
45
(-30-118)
-26
(-116-63)
364
(212-514)
71
(18-123)
78
(-3-159)
31
(-6 - 69)
8
(-2-18)
89
(20-157)
9
(-5 - 23)
12/25
24
(-25 - 73)
44
(-3 - 89)
-1
(-33-31)
26
(-17-70)
39
(-5 - 83)
9
(-6 - 25)
41
(-28-108)
-22
(-96 - 52)
302
(176-427)
59
(15-102)
65
(-3 - 1 32)
26
(-5 - 57)
7
(-2-15)
74
(16-131)
8
(-4-19)
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.  Numbers are rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-113
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM25
Concentrations
0.9%
(-1 % - 2.8%)
1 .9%
(-0.1% -3. 9%)
0%
(-2. 3% -2.1%)
0.9%
(-0.6% - 2.4%)
1 .4%
(-0.2% - 3%)
1 .3%
(-0.9% - 3.3%)
1.1%
(-0.7% - 2.9%)
-0.3%
(-1 .2% - 0.7%)
2.5%
(1 .4% - 3.5%)
2.3%
(0.6% - 4%)
1 .4%
(-0.1% -2. 9%)
1 .7%
(-0.3% - 3.7%)
1 .3%
(-0.3% - 2.8%)
2.3%
(0.5% -4.1%)
1.1%
(-0.6% - 2.7%)
15/352
0.9%
(-0.9% - 2.5%)
1 .8%
(-0.1% -3.7%)
0%
(-1 .8% - 1 .7%)
0.9%
(-0.6% - 2.4%)
1 .2%
(-0.1% -2. 4%)
0.7%
(-0.5% - 1 .8%)
1%
(-0.7% - 2.7%)
-0.2%
(-0.8% - 0.4%)
2%
(1 .2% - 2.9%)
2.1%
(0.5% - 3.6%)
1 .4%
(-0.1% -2.9%)
1%
(-0.2% - 2.2%)
0.8%
(-0.2% - 1 .8%)
2.1%
(0.5% - 3.7%)
0.9%
(-0.5% - 2.2%)
13/35
0.7%
(-0.7% - 2.2%)
1 .6%
(-0.1% -3.2%)
0%
(-1 .5% - 1 .4%)
0.9%
(-0.6% - 2.4%)
1.1%
(-0.1% -2.2%)
0.7%
(-0.5% - 1 .8%)
0.9%
(-0.6% - 2.3%)
-0.2%
(-0.8% - 0.4%)
2%
(1 .2% - 2.8%)
2%
(0.5% - 3.4%)
1 .4%
(-0.1% -2.9%)
1%
(-0.2% - 2.2%)
0.8%
(-0.2% - 1 .8%)
1 .8%
(0.4% - 3.2%)
0.9%
(-0.5% - 2.2%)
12/35
0.7%
(-0.7% - 2%)
1 .4%
(-0.1% -2. 9%)
0%
(-1 .4% - 1 .3%)
0.9%
(-0.6% - 2.2%)
1%
(-0.1% -2.1%)
0.7%
(-0.5% - 1 .8%)
0.8%
(-0.5% -2.1%)
-0.2%
(-0.7% - 0.4%)
1 .8%
(1.1% -2.6%)
1 .8%
(0.5% -3.1%)
1 .3%
(-0.1% -2.7%)
1%
(-0.2% - 2.2%)
0.8%
(-0.2% - 1 .8%)
1 .7%
(0.4% - 2.9%)
0.9%
(-0.5% - 2.2%)
13/30
0.7%
(-0.7% - 2.2%)
1 .5%
(-0.1% -3.1%)
0%
(-1 .5% - 1 .4%)
0.9%
(-0.6% - 2.4%)
1%
(-0.1% -2.1%)
0.6%
(-0.4% - 1 .6%)
0.9%
(-0.6% - 2.3%)
-0.1%
(-0.7% - 0.4%)
1 .8%
(1 % - 2.5%)
1 .8%
(0.5% -3.1%)
1 .3%
(-0.1% -2. 7%)
0.8%
(-0.2% - 1 .8%)
0.7%
(-0.2% - 1 .5%)
1 .8%
(0.4% -3.1%)
0.7%
(-0.4% - 1 .8%)
12/25
0.7%
(-0.7% - 1 .9%)
1 .3%
(-0.1% -2. 6%)
0%
(-1 .2% - 1 .2%)
0.9%
(-0.6% - 2.2%)
0.8%
(-0.1% -1.7%)
0.5%
(-0.3% - 1 .3%)
0.8%
(-0.5% -2.1%)
-0.1%
(-0.5% - 0.3%)
1 .5%
(0.8% -2.1%)
1 .5%
(0.4% - 2.6%)
1.1%
(0% - 2.2%)
0.7%
(-0.1% -1.5%)
0.6%
(-0.1% -1.2%)
1 .5%
(0.3% - 2.6%)
0.6%
(-0.3% - 1 .5%)
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.  Percents are rounded to the nearest tenth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-114
September 2009
Draft - Do Not Quote or Cite

-------
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 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):
Recent PM2 5
Concentrations
0.9%
(-0.9% - 2.6%)
1 .7%
(-0.1% -3.5%)
0%
(-2.2% -2.1%)
0.8%
(-0.5% - 2%)
1 .2%
(-0.2% - 2.5%)
1 .3%
(-0.9% - 3.4%)
1.1%
(-0.8% - 2.9%)
-0.3%
(-1.1% -0.6%)
2.2%
(1 .3% - 3.2%)
2.3%
(0.6% - 3.9%)
1 .5%
(-0.1% -2. 9%)
1 .6%
(-0.3% - 3.4%)
1.1%
(-0.3% - 2.5%)
2%
(0.4% - 3.5%)
0.9%
(-0.5% -2.1%)
15/352
0.8%
(-0.8% - 2.4%)
1 .6%
(-0.1% -3. 3%)
0%
(-1 .7% - 1 .6%)
0.8%
(-0.5% - 2%)
1%
(-0.1% -2.1%)
0.7%
(-0.5% - 1 .9%)
1.1%
(-0.7% - 2.8%)
-0.2%
(-0.7% - 0.4%)
1 .9%
(1.1% -2. 6%)
2.1%
(0.5% - 3.6%)
1 .5%
(-0.1% -2. 9%)
0.9%
(-0.2% - 2%)
0.7%
(-0.2% - 1 .6%)
1 .8%
(0.4% -3.1%)
0.7%
(-0.4% - 1 .7%)
13/35
0.7%
(-0.7% -2.1%)
1 .4%
(-0.1% -2.9%)
0%
(-1 .5% - 1 .4%)
0.8%
(-0.5% - 2%)
0.9%
(-0.1% -1.9%)
0.7%
(-0.5% - 1 .9%)
0.9%
(-0.6% - 2.4%)
-0.2%
(-0.7% - 0.4%)
1 .8%
(1.1% -2.6%)
2%
(0.5% - 3.4%)
1 .5%
(-0.1% -2.9%)
0.9%
(-0.2% - 2%)
0.7%
(-0.2% - 1 .6%)
1 .5%
(0.3% - 2.7%)
0.7%
(-0.4% - 1 .7%)
12/35
0.6%
(-0.6% - 1 .9%)
1 .3%
(-0.1% -2.7%)
0%
(-1 .3% - 1 .3%)
0.7%
(-0.5% - 1 .9%)
0.8%
(-0.1% -1.7%)
0.7%
(-0.5% - 1 .9%)
0.8%
(-0.6% - 2.2%)
-0.2%
(-0.7% - 0.4%)
1 .7%
(1 % - 2.4%)
1 .8%
(0.5% -3.1%)
1 .4%
(-0.1% -2.8%)
0.9%
(-0.2% - 2%)
0.7%
(-0.2% - 1 .6%)
1 .4%
(0.3% - 2.5%)
0.7%
(-0.4% - 1 .7%)
13/30
0.7%
(-0.7% -2.1%)
1 .4%
(-0.1% -2.9%)
0%
(-1 .5% - 1 .4%)
0.8%
(-0.5% - 2%)
0.8%
(-0.1% -1.8%)
0.6%
(-0.4% - 1 .6%)
0.9%
(-0.6% - 2.4%)
-0.1%
(-0.6% - 0.3%)
1 .6%
(0.9% - 2.3%)
1 .8%
(0.5% -3.1%)
1 .4%
(-0.1% -2.8%)
0.8%
(-0.2% - 1 .7%)
0.6%
(-0.1% -1.3%)
1 .5%
(0.3% - 2.7%)
0.6%
(-0.3% - 1 .5%)
12/25
0.6%
(-0.6% - 1 .9%)
1 .2%
(-0.1% -2.4%)
0%
(-1.2% -1.1%)
0.7%
(-0.5% - 1 .9%)
0.7%
(-0.1% -1.5%)
0.5%
(-0.4% - 1 .4%)
0.8%
(-0.6% - 2.2%)
-0.1%
(-0.5% - 0.3%)
1 .3%
(0.8% - 1 .9%)
1 .5%
(0.4% - 2.6%)
1.1%
(0% - 2.3%)
0.6%
(-0.1% -1.4%)
0.5%
(-0.1% -1.1%)
1 .3%
(0.3% - 2.2%)
0.5%
(-0.3% - 1 .2%)
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.  Percents are rounded to the nearest tenth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-115
September 2009
Draft - Do Not Quote or Cite

-------
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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and
Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
0.9%
(-0.9% - 2.7%)
1 .7%
(-0.1% -3.4%)
0%
(-2.2% - 2%)
0.8%
(-0.5% -2.1%)
1 .2%
(-0.1% -2.4%)
1 .4%
(-1%-3.7%)
1.1%
(-0.7% - 2.9%)
-0.3%
(-1.1% -0.6%)
2.3%
(1 .3% - 3.2%)
2.3%
(0.6% - 3.9%)
1 .3%
(-0.1% -2.7%)
1 .6%
(-0.3% - 3.4%)
1 .3%
(-0.3% - 2.9%)
2%
(0.5% - 3.6%)
0.9%
(-0.5% - 2.3%)
15/352
0.8%
(-0.8% - 2.4%)
1 .6%
(-0.1% -3.2%)
0%
(-1 .7% - 1 .6%)
0.8%
(-0.5% -2.1%)
0.9%
(-0.1% -2%)
0.8%
(-0.5% -2.1%)
1%
(-0.7% - 2.7%)
-0.2%
(-0.7% - 0.4%)
1 .9%
(1.1% -2.7%)
2.1%
(0.5% - 3.6%)
1 .3%
(-0.1% -2.7%)
0.9%
(-0.2% - 2%)
0.8%
(-0.2% - 1 .8%)
1 .8%
(0.4% - 3.2%)
0.7%
(-0.4% - 1 .8%)
13/35
0.7%
(-0.7% -2.1%)
1 .4%
(-0.1% -2. 8%)
0%
(-1 .4% - 1 .4%)
0.8%
(-0.5% -2.1%)
0.9%
(-0.1% -1.8%)
0.8%
(-0.5% -2.1%)
0.9%
(-0.6% - 2.3%)
-0.2%
(-0.7% - 0.4%)
1 .8%
(1.1% -2. 6%)
1 .9%
(0.5% - 3.4%)
1 .3%
(-0.1% -2. 7%)
0.9%
(-0.2% - 2%)
0.8%
(-0.2% - 1 .8%)
1 .6%
(0.4% - 2.8%)
0.7%
(-0.4% - 1 .8%)
12/35
0.6%
(-0.6% - 1 .9%)
1 .3%
(-0.1% -2.6%)
0%
(-1 .3% - 1 .2%)
0.8%
(-0.5% - 2%)
0.8%
(-0.1% -1.7%)
0.8%
(-0.5% -2.1%)
0.8%
(-0.5% -2.1%)
-0.2%
(-0.7% - 0.4%)
1 .7%
(1 % - 2.4%)
1 .8%
(0.5% -3.1%)
1 .3%
(-0.1% -2. 6%)
0.9%
(-0.2% - 2%)
0.8%
(-0.2% - 1 .8%)
1 .5%
(0.3% - 2.6%)
0.7%
(-0.4% - 1 .8%)
13/30
0.7%
(-0.7% -2.1%)
1 .3%
(-0.1% -2.8%)
0%
(-1 .4% - 1 .4%)
0.8%
(-0.5% -2.1%)
0.8%
(-0.1% -1.7%)
0.7%
(-0.5% - 1 .8%)
0.9%
(-0.6% - 2.3%)
-0.1%
(-0.6% - 0.3%)
1 .6%
(0.9% - 2.3%)
1 .8%
(0.5% -3.1%)
1 .3%
(-0.1% -2.5%)
0.8%
(-0.2% - 1 .7%)
0.7%
(-0.2% - 1 .6%)
1 .6%
(0.3% - 2.8%)
0.6%
(-0.3% - 1 .6%)
12/25
0.6%
(-0.6% - 1 .9%)
1.1%
(-0.1% -2.3%)
0%
(-1.2% -1.1%)
0.8%
(-0.5% - 2%)
0.7%
(-0.1% -1.4%)
0.5%
(-0.4% - 1 .5%)
0.8%
(-0.5% -2.1%)
-0.1%
(-0.5% - 0.3%)
1 .3%
(0.8% - 1 .9%)
1 .5%
(0.4% - 2.6%)
1%
(0%-2.1%)
0.6%
(-0.1% -1.4%)
0.6%
(-0.1% -1.3%)
1 .3%
(0.3% - 2.3%)
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.  Percents are rounded to the nearest tenth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-116
September 2009
Draft - Do Not Quote or Cite

-------
Table E-88. 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 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 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):
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-79% - -83%)
-6%
(-6% - -6%)
-59%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-71% --73%)
-58%
(-57% - -59%)
-12%
(-11% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(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%)
0%
(0% - 0%)
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%)
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.  Percents are rounded to whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-117
September 2009
Draft - Do Not Quote or Cite

-------
Table E-89. Percent Reduction from the Current Standards: Estimated Annual Incidence of Cardiovascular Mortality Associated with
          Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Cardiovascular Mortality Associated with Short-Term
Exposure to PM2 5 Concentrations in a Recent Year and 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
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -83%)
-6%
(-6% - -6%)
-59%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-71% --73%)
-58%
(-58% - -59%)
-12%
(-11% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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%)
0%
(0% - 0%)
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%)
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%)
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.  Percents are rounded to whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-118
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-79% - -83%)
-6%
(-6% - -6%)
-59%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-71% --73%)
-58%
(-57% - -59%)
-12%
(-11% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(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%)
0%
(0% - 0%)
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%
(14% -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%)
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.  Percents are rounded to whole number.
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
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-28.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiovascular Mortality Associated with Short-Term Exposure to
Based on 2005 Air Quality Data*
          30%

          20%

          10%

           0%

          -10%

          -20%

          -30%

          -40%

          -50%

          -60%

          -70%

          -80%

          -90%
                  2005 air
                  quality
15/35"      13/35       12/35

         Alternative Standard
13/30
12/25
                         -•- Atlanta, GA 32 (-32 - 94); 0.9% (-0.9% - 2.5%)
                         -m- Baltimore, MD 70 (-5-142); 1.8% (-0.1%-3.7%)
                         -*- Birmingham, AL -1 (-48-45); 0% (-1.8%-1.7%)
                            Dallas, TX 31  (-21 - 82); 0.9% (-0.6% - 2.4%)
                         -*- Detroit, Ml 70 (-9-147); 1.2% (-0.1%-2.4%)
                         -•- Fresno, CA 11 (-8 - 30); 0.7% (-0.5% -1.8%)
                         -i- Houston, TX 51  (-34-134); 1% (-0.7%-2.7%)
                         — Los Angeles, CA -33 (-146-79); -0.2% (-0.8%-0.4%)
                           - New York, NY  460 (268-649); 2% (1.2%-2.9%)
                         -»- Philadelphia, PA 85 (22-147);  2.1%  (0.5%-3.6%)
                         -•- Phoenix, AZ 83  (-4-169);  1.4% (-0.1%-2.9%)
                         -*- Pittsburgh, PA 40 (-8-88); 1% (-0.2%-2.2%)
                            Salt Lake City, UT 9 (-2 -19); 0.8% (-0.2% -1.8%)
                            St. Louis, MO  118 (26-208); 2.1% (0.5%-3.7%)
                         -•- Tacoma.WA 12 (-7-31); 0.9% (-0.5%-2.2%)
*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.
                                                 E-120
   September 2009
                                                      Draft - Do Not Quote or Cite

-------
Figure E-29. Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiovascular Mortality Associated with Short-Term Exposure to
Based on 2006 Air Quality Data*
           10%
           o%
          -10%
          -20%
          -30%

          -40%
          -50%
          -60%
          -70%

          -80%
          -90%
/
                  2006 air     15/35**
                  quality
             13/35
12/35
13/30
12/25
                                     Alternative Standard
                         -»- Atlanta, GA 31 (-32 - 93); 0.8% (-0.8% - 2.4%)
                         -•- Baltimore, MD 64 (-4-130);  1.6% (-0.1%-3.3%)
                         -*- Birmingham, AL -1 (-47-44); 0% (-1.7%-1.6%)
                            Dallas, TX 27 (-18-71); 0.8% (-0.5%-2%)
                         -*- Detroit, Ml 58 (-7-122); 1%  (-0.1%-2.1%)
                         -•- Fresno, CA 12 (-8-32); 0.7% (-0.5%-1.9%)
                         -i- Houston, TX 53 (-36-140); 1.1% (-0.7%-2.8%)
                         	 Los Angeles, CA -30 (-134-73); -0.2% (-0.7%-0.4%)
                            New York, NY 421  (246-595); 1.9% (1.1%-2.6%)
                         -•- Philadelphia, PA 84 (22-144); 2.1%  (0.5%-3.6%)
                         -m- Phoenix, AZ 88  (-4-179);  1.5%  (-0.1%-2.9%)
                         -*- Pittsburgh, PA 36 (-7 - 80); 0.9% (-0.2% - 2%)
                            Salt Lake City, UT 8 (-2 -18); 0.7% (-0.2% -1.6%)
                            St. Louis, MO 101  (22-177); 1.8% (0.4%-3.1%)
                         -•- Tacoma.WA 10 (-5-26); 0.7% (-0.4%-1.7%)
*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.
                                                 E-121
   September 2009
                                                        Draft - Do Not Quote or Cite

-------
Figure E-30. Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiovascular Mortality Associated with Short-Term Exposure to
Based on 2007 Air Quality Data*
          30%
          -90%
                  2007 air
                  quality
15/35**      13/35      12/35

         Alternative Standard
13/30
12/25
                         -»- Atlanta, GA 32 (-33 - 95); 0.8% (-0.8% - 2.4%)
                         -•- Baltimore, MD 61 (-4-125);  1.6% (-0.1%-3.2%)
                         -*- Birmingham, AL -1 (-46-44); 0% (-1.7%-1.6%)
                             Dallas, TX 29 (-19-75); 0.8% (-0.5%-2.1%)
                         -*- Detroit, Ml 55 (-7-117); 0.9% (-0.1%-2%)
                         -•- Fresno, CA 13 (-9-35); 0.8% (-0.5%-2.1%)
                         -i- Houston, TX 52 (-36-139); 1% (-0.7%-2.7%)
                         	 Los Angeles, CA -30 (-136-74); -0.2% (-0.7%-0.4%)
                             New York, NY 425 (248-600); 1.9% (1.1%-2.7%)
                         -•- Philadelphia, PA 83 (22-143); 2.1%  (0.5%-3.6%)
                         -m- Phoenix, AZ 84  (-4-170);  1.3%  (-0.1%-2.7%)
                         -*- Pittsburgh, PA 37 (-7 - 80); 0.9% (-0.2% - 2%)
                             Salt Lake City, UT 10 (-2-21); 0.8% (-0.2%-1.8%)
                             St. Louis, MO 104 (23-184); 1.8% (0.4%-3.2%)
                         -•- Tacoma, WA 11 (-6 - 27); 0.7% (-0.4% -1.8%)
*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.
                                                 E-122
   September 2009
                                                      Draft - Do Not Quote or Cite

-------
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 s 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 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):
Recent PM2.5
Concentrations
21
(-9 - 50)
38
(7 - 67)
13
(-1 1 - 36)
11
(-10-31)
34
(2 - 64)
16
(1 - 30)
38
(6 - 70)
94
(9-177)
117
(41 -190)
24
(-2 - 50)
47
(4 - 89)
29
(-3 - 60)
9
(1-17)
34
(-9 - 75)
9
(0-18)
15/352
19
(-8 - 46)
35
(7 - 63)
10
(-8 - 29)
11
(-10-31)
27
(1 - 52)
9
(0-17)
36
(6 - 66)
59
(6-112)
97
(34-159)
23
(-2 - 46)
47
(4 - 89)
17
(-2-35)
6
(1-11)
30
(-8 - 67)
7
(0-15)
13/35
17
(-7 - 39)
31
(6 - 56)
9
(-7 - 24)
11
(-10-31)
25
(1 - 48)
9
(0-17)
31
(5 - 56)
59
(6-112)
95
(33-155)
21
(-2 - 44)
47
(4 - 89)
17
(-2 - 35)
6
(1-11)
26
(-7 - 59)
7
(0-15)
12/35
15
(-6 - 36)
29
(5-51)
8
(-6 - 22)
10
(-9 - 29)
23
(1-44)
9
(0-17)
28
(5-51)
56
(5-106)
87
(31 -142)
19
(-2 - 40)
45
(4 - 85)
17
(-2 - 35)
6
(1-11)
24
(-6 - 54)
7
(0-15)
13/30
17
(-7 - 39)
30
(6-54)
9
(-7 - 24)
11
(-10-31)
23
(1 - 45)
7
(0-14)
31
(5 - 56)
51
(5 - 96)
83
(29 - 1 36)
19
(-2 - 40)
44
(3 - 84)
15
(-1-30)
5
(1-9)
26
(-7 - 58)
6
(0-13)
12/25
15
(-6 - 35)
25
(5-45)
7
(-6 - 20)
10
(-9 - 29)
19
(1 - 37)
6
(0-12)
28
(5-51)
42
(4 - 80)
69
(24 - 1 1 3)
16
(-1 - 33)
37
(3 - 70)
12
(-1-25)
4
(1-8)
21
(-6 - 48)
5
(0-10)
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.
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
September 2009
Draft - Do Not Quote or Cite

-------
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 s 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 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):
Recent PM2.5
Concentrations
21
(-8 - 49)
34
(6-61)
13
(-1 1 - 36)
9
(-9 - 27)
28
(1 - 54)
16
(1 - 32)
40
(6 - 73)
87
(8 - 1 64)
107
(38-174)
24
(-2 - 50)
50
(4 - 95)
27
(-3 - 55)
8
(1-15)
29
(-8 - 64)
7
(0-15)
15/352
19
(-8 - 45)
32
(6 - 58)
10
(-8 - 28)
9
(-9 - 27)
23
(1 - 44)
9
(0-18)
38
(6 - 69)
55
(5 - 1 04)
89
(31 -145)
22
(-2 - 46)
50
(4 - 95)
15
(-2 - 32)
5
(1-10)
26
(-7 - 57)
6
(0-12)
13/35
16
(-7 - 39)
28
(5-51)
9
(-7 - 24)
9
(-9 - 27)
21
(1-40)
9
(0-18)
32
(5 - 59)
55
(5-104)
87
(31 -142)
21
(-2 - 43)
50
(4 - 95)
15
(-2 - 32)
5
(1-10)
22
(-6 - 50)
6
(0-12)
12/35
15
(-6 - 35)
26
(5-47)
8
(-6 - 22)
9
(-8 - 25)
19
(1-37)
9
(0-18)
30
(5 - 54)
52
(5 - 98)
80
(28 - 1 30)
19
(-2 - 39)
47
(4 - 90)
15
(-2 - 32)
5
(1-10)
20
(-5 - 46)
6
(0-12)
13/30
16
(-7 - 39)
28
(5 - 50)
9
(-7 - 24)
9
(-9 - 27)
19
(1-38)
8
(0-15)
32
(5 - 59)
47
(5 - 89)
76
(27 - 1 25)
19
(-2 - 39)
47
(4 - 89)
13
(-1-27)
4
(1-8)
22
(-6 - 49)
5
(0-10)
12/25
15
(-6 - 35)
23
(4-42)
7
(-6 - 20)
9
(-8 - 25)
16
(1-31)
6
(0-13)
30
(5 - 54)
39
(4 - 74)
63
(22-104)
16
(-1 - 33)
39
(3 - 74)
11
(-1 - 23)
4
(0-7)
18
(-5-41)
4
(0-9)
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.
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
September 2009
Draft - Do Not Quote or Cite

-------
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 s 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 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):
Recent PM2.5
Concentrations
21
(-9-51)
33
(6 - 59)
13
(-10-35)
10
(-9 - 29)
27
(1-51)
18
(1 - 34)
40
(6 - 72)
88
(9 - 1 65)
108
(38-176)
24
(-2 - 49)
47
(4 - 90)
27
(-3 - 55)
10
(1-18)
30
(-8 - 66)
8
(0-16)
15/352
20
(-8 - 47)
31
(6 - 56)
10
(-8 - 28)
10
(-9 - 29)
22
(1 - 42)
10
(0-19)
38
(6 - 68)
56
(5 - 1 05)
90
(32-147)
22
(-2 - 45)
47
(4 - 90)
16
(-2 - 32)
6
(1-12)
27
(-7 - 59)
6
(0-13)
13/35
17
(-7 - 40)
27
(5-49)
9
(-7 - 24)
10
(-9 - 29)
20
(1-38)
10
(0-19)
32
(5 - 58)
56
(5-105)
87
(31 -143)
21
(-2 - 42)
47
(4 - 90)
16
(-2 - 32)
6
(1-12)
23
(-6 - 52)
6
(0-13)
12/35
15
(-6 - 36)
25
(5-45)
8
(-6 - 22)
9
(-8 - 27)
18
(1-35)
10
(0-19)
29
(5 - 53)
52
(5 - 99)
80
(28-132)
19
(-2 - 39)
45
(4 - 85)
16
(-2 - 32)
6
(1-12)
21
(-6 - 48)
6
(0-13)
13/30
17
(-7 - 40)
27
(5-48)
9
(-7 - 24)
10
(-9 - 29)
19
(1-36)
8
(0-16)
32
(5 - 58)
47
(5 - 90)
77
(27-126)
19
(-2 - 39)
44
(4 - 84)
13
(-1 - 28)
5
(1-10)
23
(-6-51)
5
(0-11)
12/25
15
(-6 - 36)
22
(4 - 40)
7
(-6 - 20)
9
(-8 - 27)
15
(1-30)
7
(0-14)
29
(5 - 53)
39
(4 - 75)
64
(22-105)
16
(-1 - 32)
37
(3 - 70)
11
(-1 - 23)
4
(1-8)
19
(-5 - 42)
5
(0-9)
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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-125
September 2009
Draft - Do Not Quote or Cite

-------
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,
a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n
Standards (Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
1 .7%
(-0.7% - 4%)
3.1%
(0.6% - 5.5%)
1 .5%
(-1 .2% - 4.2%)
1%
(-0.9% - 2.9%)
2.5%
(0.1% -4. 8%)
2.7%
(0.1% -5.1%)
2.7%
(0.4% - 5%)
1 .7%
(0.2% - 3.2%)
2.7%
(1%-4.4%)
2%
(-0.2% - 4.2%)
1 .9%
(0.2% - 3.7%)
2.5%
(-0.2% -5.1%)
2.1%
(0.3% - 3.8%)
1 .9%
(-0.5% - 4.3%)
1 .8%
(0% - 3.6%)
15/352
1 .6%
(-0.6% - 3.7%)
2.9%
(0.5% - 5.3%)
1 .2%
(-1 % - 3.3%)
1%
(-0.9% - 2.9%)
2%
(0.1% -3. 9%)
1 .5%
(0% - 2.9%)
2.6%
(0.4% - 4.7%)
1.1%
(0.1% -2%)
2.3%
(0.8% - 3.7%)
1 .9%
(-0.2% - 3.8%)
1 .9%
(0.2% - 3.7%)
1 .4%
(-0.1% -3%)
1 .3%
(0.2% - 2.4%)
1 .7%
(-0.5% - 3.9%)
1 .5%
(0% - 3%)
13/35
1 .3%
(-0.5% - 3.2%)
2.6%
(0.5% - 4.6%)
1%
(-0.8% - 2.8%)
1%
(-0.9% - 2.9%)
1 .9%
(0.1% -3. 6%)
1 .5%
(0% - 2.9%)
2.2%
(0.4% - 4%)
1.1%
(0.1% -2%)
2.2%
(0.8% - 3.6%)
1 .7%
(-0.2% - 3.6%)
1 .9%
(0.2% - 3.7%)
1 .4%
(-0.1% -3%)
1 .3%
(0.2% - 2.4%)
1 .5%
(-0.4% - 3.4%)
1 .5%
(0% - 3%)
12/35
1 .2%
(-0.5% - 2.9%)
2.4%
(0.4% - 4.2%)
0.9%
(-0.7% - 2.6%)
0.9%
(-0.8% - 2.7%)
1 .7%
(0.1% -3. 3%)
1 .5%
(0% - 2.9%)
2%
(0.3% - 3.7%)
1%
(0.1% -1.9%)
2%
(0.7% - 3.3%)
1 .6%
(-0.1% -3.3%)
1 .8%
(0.1% -3.5%)
1 .4%
(-0.1% -3%)
1 .3%
(0.2% - 2.4%)
1 .4%
(-0.4% -3.1%)
1 .5%
(0% - 3%)
13/30
1 .3%
(-0.5% - 3.2%)
2.5%
(0.5% - 4.5%)
1%
(-0.8% - 2.8%)
1%
(-0.9% - 2.9%)
1 .7%
(0.1% -3. 3%)
1 .3%
(0% - 2.5%)
2.2%
(0.4% - 4%)
0.9%
(0.1% -1.7%)
1 .9%
(0.7% - 3.2%)
1 .6%
(-0.1% -3.3%)
1 .8%
(0.1% -3. 4%)
1 .2%
(-0.1% -2. 6%)
1.1%
(0.1% -2.1%)
1 .5%
(-0.4% - 3.3%)
1 .3%
(0% - 2.5%)
, Concentrations in
and Daily (m)
12/25
1 .2%
(-0.5% - 2.8%)
2.1%
(0.4% - 3.8%)
0.8%
(-0.7% - 2.3%)
0.9%
(-0.8% - 2.7%)
1 .4%
(0.1% -2. 8%)
1%
(0% - 2%)
2%
(0.3% - 3.7%)
0.8%
(0.1% -1.4%)
1 .6%
(0.6% - 2.6%)
1 .3%
(-0.1% -2.7%)
1 .5%
(0.1% -2. 9%)
1%
(-0.1% -2.1%)
0.9%
(0.1% -1.7%)
1 .2%
(-0.3% - 2.8%)
1.1%
(0%-2.1%)
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. Percents are rounded to the nearest hundredth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-126
September 2009
Draft - Do Not Quote or Cite

-------
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,
a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n
Standards (Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
1 .6%
(-0.7% - 3.9%)
2.8%
(0.5% -5.1%)
1 .5%
(-1 .2% - 4%)
0.8%
(-0.8% - 2.4%)
2.1%
(0.1% -4%)
2.8%
(0.1% -5. 3%)
2.8%
(0.4% -5.1%)
1 .6%
(0.2% - 2.9%)
2.5%
(0.9% - 4%)
2%
(-0.2% -4.1%)
2%
(0.2% - 3.8%)
2.2%
(-0.2% - 4.6%)
1 .8%
(0.2% - 3.4%)
1 .6%
(-0.4% - 3.7%)
1 .5%
(0% - 2.9%)
15/352
1 .5%
(-0.6% - 3.5%)
2.7%
(0.5% - 4.8%)
1 .2%
(-0.9% - 3.2%)
0.8%
(-0.8% - 2.4%)
1 .7%
(0.1% -3. 3%)
1 .5%
(0% - 3%)
2.6%
(0.4% - 4.8%)
1%
(0.1% -1.9%)
2.1%
(0.7% - 3.4%)
1 .8%
(-0.2% - 3.8%)
2%
(0.2% - 3.8%)
1 .3%
(-0.1% -2. 7%)
1 .2%
(0.1% -2.2%)
1 .5%
(-0.4% - 3.3%)
1 .2%
(0% - 2.4%)
13/35
1 .3%
(-0.5% - 3%)
2.3%
(0.4% - 4.2%)
1%
(-0.8% - 2.7%)
0.8%
(-0.8% - 2.4%)
1 .6%
(0.1% -3%)
1 .5%
(0% - 3%)
2.2%
(0.4% -4.1%)
1%
(0.1% -1.9%)
2%
(0.7% - 3.3%)
1 .7%
(-0.2% - 3.5%)
2%
(0.2% - 3.8%)
1 .3%
(-0.1% -2. 7%)
1 .2%
(0.1% -2.2%)
1 .3%
(-0.3% - 2.9%)
1 .2%
(0% - 2.4%)
12/35
1 .2%
(-0.5% - 2.8%)
2.2%
(0.4% - 3.9%)
0.9%
(-0.7% - 2.5%)
0.8%
(-0.7% - 2.2%)
1 .4%
(0.1% -2. 8%)
1 .5%
(0% - 3%)
2%
(0.3% - 3.7%)
0.9%
(0.1% -1.8%)
1 .8%
(0.6% - 3%)
1 .6%
(-0.1% -3. 3%)
1 .9%
(0.1% -3.6%)
1 .3%
(-0.1% -2. 7%)
1 .2%
(0.1% -2.2%)
1 .2%
(-0.3% - 2.6%)
1 .2%
(0% - 2.4%)
13/30
1 .3%
(-0.5% - 3%)
2.3%
(0.4% -4.1%)
1%
(-0.8% - 2.7%)
0.8%
(-0.8% - 2.4%)
1 .5%
(0.1% -2. 8%)
1 .3%
(0% - 2.5%)
2.2%
(0.4% -4.1%)
0.8%
(0.1% -1.6%)
1 .8%
(0.6% - 2.9%)
1 .6%
(-0.1% -3.2%)
1 .9%
(0.1% -3.5%)
1.1%
(-0.1% -2. 3%)
1%
(0.1% -1.9%)
1 .3%
(-0.3% - 2.8%)
1%
(0% - 2%)
, Concentrations in
and Daily (m)
12/25
1.1%
(-0.5% - 2.7%)
1 .9%
(0.4% - 3.4%)
0.8%
(-0.7% - 2.3%)
0.8%
(-0.7% - 2.2%)
1 .2%
(0.1% -2. 3%)
1.1%
(0%-2.1%)
2%
(0.3% - 3.7%)
0.7%
(0.1% -1.3%)
1 .5%
(0.5% - 2.4%)
1 .3%
(-0.1% -2. 7%)
1 .5%
(0.1% -2.9%)
0.9%
(-0.1% -1.9%)
0.8%
(0.1% -1.5%)
1%
(-0.3% - 2.3%)
0.8%
(0% - 1 .7%)
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. Percents are rounded to the nearest hundredth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-127
September 2009
Draft - Do Not Quote or Cite

-------
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,
a Recent Year and PM25 Concentrations that Just Meet the Current and Alternative Annual (n
Standards (Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
1 .6%
(-0.7% - 3.9%)
2.7%
(0.5% - 4.9%)
1 .5%
(-1 .2% - 4%)
0.9%
(-0.8% - 2.5%)
2%
(0.1% -3. 9%)
2.9%
(0.1% -5. 6%)
2.7%
(0.4% - 4.9%)
1 .6%
(0.2% - 3%)
2.5%
(0.9% -4.1%)
2%
(-0.2% -4.1%)
1 .8%
(0.1% -3.5%)
2.3%
(-0.2% - 4.7%)
2.2%
(0.3% - 4%)
1 .7%
(-0.4% - 3.8%)
1 .6%
(0%-3.1%)
15/352
1 .5%
(-0.6% - 3.6%)
2.6%
(0.5% - 4.6%)
1.1%
(-0.9% -3.1%)
0.9%
(-0.8% - 2.5%)
1 .6%
(0.1% -3. 2%)
1 .6%
(0.1% -3. 2%)
2.6%
(0.4% - 4.7%)
1%
(0.1% -1.9%)
2.1%
(0.7% - 3.4%)
1 .8%
(-0.2% - 3.8%)
1 .8%
(0.1% -3.5%)
1 .3%
(-0.1% -2. 7%)
1 .4%
(0.2% - 2.5%)
1 .5%
(-0.4% - 3.4%)
1 .3%
(0% - 2.5%)
13/35
1 .3%
(-0.5% - 3%)
2.3%
(0.4% - 4%)
1%
(-0.8% - 2.7%)
0.9%
(-0.8% - 2.5%)
1 .5%
(0.1% -2. 9%)
1 .6%
(0.1% -3. 2%)
2.2%
(0.3% - 4%)
1%
(0.1% -1.9%)
2%
(0.7% - 3.3%)
1 .7%
(-0.2% - 3.5%)
1 .8%
(0.1% -3.5%)
1 .3%
(-0.1% -2. 7%)
1 .4%
(0.2% - 2.5%)
1 .3%
(-0.3% - 3%)
1 .3%
(0% - 2.5%)
12/35
1 .2%
(-0.5% - 2.8%)
2.1%
(0.4% - 3.7%)
0.9%
(-0.7% - 2.4%)
0.8%
(-0.7% - 2.4%)
1 .4%
(0.1% -2. 7%)
1 .6%
(0.1% -3. 2%)
2%
(0.3% - 3.6%)
0.9%
(0.1% -1.8%)
1 .9%
(0.7% - 3%)
1 .6%
(-0.1% -3. 2%)
1 .7%
(0.1% -3.3%)
1 .3%
(-0.1% -2. 7%)
1 .4%
(0.2% - 2.5%)
1 .2%
(-0.3% - 2.7%)
1 .3%
(0% - 2.5%)
13/30
1 .3%
(-0.5% - 3%)
2.2%
(0.4% - 4%)
1%
(-0.8% - 2.7%)
0.9%
(-0.8% - 2.5%)
1 .4%
(0.1% -2. 7%)
1 .4%
(0% - 2.7%)
2.2%
(0.3% - 4%)
0.8%
(0.1% -1.6%)
1 .8%
(0.6% - 2.9%)
1 .6%
(-0.1% -3.2%)
1 .7%
(0.1% -3.2%)
1.1%
(-0.1% -2. 4%)
1 .2%
(0.1% -2. 2%)
1 .3%
(-0.3% - 2.9%)
1.1%
(0% - 2.2%)
, Concentrations in
and Daily (m)
12/25
1.1%
(-0.5% - 2.7%)
1 .8%
(0.3% - 3.3%)
0.8%
(-0.6% - 2.2%)
0.8%
(-0.7% - 2.4%)
1 .2%
(0.1% -2. 3%)
1 .2%
(0% - 2.3%)
2%
(0.3% - 3.6%)
0.7%
(0.1% -1.3%)
1 .5%
(0.5% - 2.4%)
1 .3%
(-0.1% -2. 7%)
1 .4%
(0.1% -2.7%)
0.9%
(-0.1% -2%)
1%
(0.1% -1.8%)
1.1%
(-0.3% - 2.4%)
0.9%
(0% - 1 .8%)
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. Percents are rounded to the nearest hundredth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                       E-128
September 2009
Draft - Do Not Quote or Cite

-------
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 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
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-80%
(-78% - -82%)
-6%
(-6% - -6%)
-58%
(-57% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-71% --73%)
-58%
(-57% - -59%)
-12%
(-11% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
0%
(0% - 0%)
13%
(13% -13%)
0%
(0% - 0%)
12/35
22%
(22% - 23%)
19%
(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%)
0%
(0% - 0%)
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%
(28% - 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%)
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. Percents are rounded to whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-129
September 2009
Draft - Do Not Quote or Cite

-------
Table E-98. Percent Reduction from the Current Standards:  Estimated Annual Incidence of Respiratory Mortality Associated with
          Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2006 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Mortality Associated with Short-Term
Exposure to PM2.5 Concentrations in a Recent Year and 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
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-80%
(-79% - -82%)
-6%
(-6% - -6%)
-58%
(-57% - -58%)
-20%
(-20% - -20%)
-9%
(-8% - -9%)
0%
(0% - 0%)
-72%
(-71% --73%)
-58%
(-57% - -59%)
-12%
(-11% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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% -10%)
14%
(14%- 14%)
5%
(5% - 5%)
0%
(0% - 0%)
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%
(28% - 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%)
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. Percents are rounded to whole number.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-130
September 2009
Draft - Do Not Quote or Cite

-------
Table E-99. Percent Reduction from the Current Standards:  Estimated Annual Incidence of Respiratory Mortality Associated with
          Short-Term Exposure to Ambient PM2 5 Concentrations, Based on Adjusting 2007 PM2 5 Concentrations1
Risk Assessment
Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, CA
Houston, TX
Los Angeles, CA
New York, NY
Philadelphia, PA
Phoenix, AZ
Pittsburgh, PA
Salt Lake City, UT
St. Louis, MO
Tacoma, WA
Percent Reduction from the Current Standards: Annual Incidence of Respiratory Mortality Associated with Short-Term
Exposure to PM2.5 Concentrations in a Recent Year and 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
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-27% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-79%
(-77% - -82%)
-6%
(-6% - -6%)
-58%
(-57% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-71% --73%)
-58%
(-56% - -59%)
-12%
(-11% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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%)
0%
(0% - 0%)
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%)
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. Percents are rounded to whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-131
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-31.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Respiratory Mortality Associated with Short-Term Exposure to PM2.5:  Based
on 2005 Air Quality Data*
          -80%
                  2005 air     15/35**
                  quality
13/35
12/35
13/30
12/25
                                     Alternative Standard
                          -•-Atlanta, GA 19 (-8-46);  (-0.6%-3.7%) 2.9%
                          -•- Baltimore, MD 35 (7-63); (0.5%-5.3%)  1.2%
                          -*- Birmingham, AL 10 (-8-29); (-1%-3.3%) 1%
                              Dallas, TX  11  (-10 - 31);  (-0.9% - 2.9%) 2%
                          -*- Detroit, Ml 27  (1 -52); (0.1%-3.9%) 1.5%
                          -•- Fresno, CA 9  (0-17); (0%-2.9%) 2.6%
                          -i- Houston, TX 36 (6-66);  (0.4%-4.7%) 1.1%
                          	 Los Angeles, CA 59  (6-112); (0.1%-2%) 2.3%
                              New York, NY 97 (34-159); (0.8%-3.7%) 1.9%
                          -•- Philadelphia, PA 23  (-2 - 46);  (-0.2% - 3.8%) 1.9%
                          -m- Phoenix, AZ 47 (4 - 89);  (0.2% - 3.7%) 1.4%
                          -*- Pittsburgh, PA 17 (-2-35); (-0.1%-3%)  1.3%
                              Salt Lake City, UT 6  (1-11); (0.2% - 2.4%) 1.7%
                              St. Louis, MO 30 (-8 - 67); (-0.5% - 3.9%) 1.5%
                              Tacoma.WA 7 (0-15); (0%-3%)
*Based on Bell et al. (2008). 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.
                                                 E-132
   September 2009
                                           Draft - Do Not Quote or Cite

-------
Figure E-32. Estimated Percent Reductions From the Current Standard to Alternative
Standards in Respiratory Mortality Associated with Short-Term Exposure to PM2.5: Based
on 2006 Air Quality Data*
          -80%
                  2006 air     15/35**
                  quality
13/35
12/35
13/30
12/25
                                     Alternative Standard
                          -•-Atlanta, GA  19 (-8-45); (-0.6%-3.5%) 2.7%
                          -m- Baltimore, MD 32 (6 - 58); (0.5% - 4.8%)  1.2%
                          -*- Birmingham, AL 10 (-8-28); (-0.9%-3.2%) 0.8%
                              Dallas, TX 9 (-9 - 27); (-0.8% - 2.4%) 1.7%
                          -*- Detroit, Ml 23 (1  -44); (0.1%-3.3%) 1.5%
                          -•- Fresno, CA  9 (0-18); (0%-3%) 2.6%
                          -i- Houston, TX 38 (6 - 69); (0.4% - 4.8%) 1 %
                          	 Los Angeles, CA 55  (5-104); (0.1%-1.9%) 2.1%
                              New York, NY 89 (31 -145);  (0.7%-3.4%) 1.8%
                          -•- Philadelphia, PA  22  (-2-46); (-0.2%-3.8%) 2%
                          -m- Phoenix, AZ 50 (4 - 95); (0.2% - 3.8%) 1.3%
                          -*- Pittsburgh, PA 15 (-2-32); (-0.1%-2.7%) 1.2%
                              Salt Lake City, UT 5  (1 -10); (0.1%-2.2%)  1.5%
                              St. Louis, MO 26 (-7 - 57); (-0.4% - 3.3%) 1.2%
                          -•- Tacoma.WA 6 (0-12);  (0%-2.4%)
*Based on Bell et al. (2008). 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.
                                                 E-133
   September 2009
                                           Draft - Do Not Quote or Cite

-------
Figure E-33.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Respiratory Mortality Associated with Short-Term Exposure to PM2.5:  Based
on 2007 Air Quality Data*
          30%
          20%
       S
       OT
          -10%
       o
       Q

       I

-30%
          -40%
          -50%
          -60%
          -70%
          -80%
                  2007 air     15/35**
                  quality
                              13/35
12/35
13/30
12/25
                                     Alternative Standard
                          -•- Atlanta, GA 20 (-8 - 47);  (-0.6% - 3.6%) 2.6%
                          -m- Baltimore, MD 31  (6-56); (0.5%-4.6%) 1.1%
                          -*- Birmingham, AL 10 (-8-28); (-0.9%-3.1%) 0.9%
                              Dallas, TX 10 (-9-29); (-0.8%-2.5%) 1.6%
                          -*- Detroit, Ml 22 (1 -42); (0.1%-3.2%) 1.6%
                          -•- Fresno, CA 10 (0-19); (0.1%-3.2%) 2.6%
                          -i- Houston, TX 38 (6 - 68);  (0.4% - 4.7%) 1 %
                          	 Los Angeles,  CA 56 (5-105); (0.1%-1.9%) 2.1%
                              New York, NY 90 (32-147); (0.7%-3.4%)  1.8%
                          -•- Philadelphia,  PA 22 (-2-45);  (-0.2%-3.8%) 1.8%
                          -m- Phoenix, AZ 47 (4-90);  (0.1%-3.5%) 1.3%
                          -*- Pittsburgh, PA 16  (-2-32); (-0.1%-2.7%)  1.4%
                              Salt Lake City, UT 6 (1 -12); (0.2%-2.5%) 1.5%
                              St. Louis, MO 27 (-7 - 59); (-0.4% - 3.4%) 1.3%
                          -•- Tacoma.WA  6 (0-13); (0%-2.5%)
*Based on Bell et al. (2008). 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.
                                                 E-134
   September 2009
                                                                         Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
43
(-28-114)
262
(192-331)
23
(-15-62)
30
(-19-78)
331
(243-418)
38
(0 - 76)
65
(-42- 171)
434
(5 - 858)
870
(639-1100)
228
(167-288)
107
(1 -212)
231
(170-291)
14
(0 - 28)
223
(164-282)
26
(-65-113)
15/352
40
(-26-105)
247
(182-313)
18
(-12-48)
30
(-19-78)
269
(198-340)
21
(0 - 42)
61
(-40- 161)
274
(3 - 543)
724
(532-915)
210
(154-265)
107
(1 -212)
134
(98-169)
9
(0-17)
200
(147-253)
21
(-52 - 92)
13/35
34
(-22 - 89)
216
(159-273)
16
(-10-41)
30
(-19-78)
247
(181 -312)
21
(0 - 42)
52
(-34-137)
274
(3 - 543)
705
(518-891)
196
(144-248)
107
(1 -212)
134
(98-169)
9
(0-17)
174
(128-220)
21
(-52 - 92)
12/35
31
(-20-81)
199
(146-251)
14
(-9 - 37)
27
(-18-73)
227
(166-286)
21
(0 - 42)
47
(-31 -125)
259
(3-513)
648
(476-819)
180
(132-228)
102
(1 -201)
134
(98-169)
9
(0-17)
159
(117-202)
21
(-52 - 92)
13/30
34
(-22 - 89)
212
(155-267)
16
(-10-41)
30
(-19-78)
230
(169-290)
18
(0 - 36)
52
(-34-137)
234
(3 - 464)
619
(454 - 783)
179
(132-227)
100
(1 -198)
114
(84-145)
7
(0-15)
171
(126-216)
18
(-44 - 79)
12/25
30
(-20 - 80)
176
(129-222)
13
(-8 - 34)
27
(-18-73)
191
(140-241)
15
(0-30)
47
(-31 -125)
194
(2 - 385)
514
(377 - 650)
149
(109-188)
83
(1 -165)
95
(70-120)
6
(0-12)
142
(104-179)
15
(-37 - 66)
 '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.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-135
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
43
(-28-113)
237
(174-300)
23
(-15-60)
25
(-17-67)
276
(203 - 349)
40
(0-79)
68
(-45- 180)
408
(5 - 807)
801
(589-1013)
222
(163-281)
113
(1 - 225)
207
(152-261)
13
(0 - 25)
190
(140-241)
22
(-53 - 93)
15/352
39
(-26-103)
224
(165-284)
18
(-12-47)
25
(-17-67)
224
(165-284)
22
(0 - 44)
64
(-42- 170)
258
(3-511)
666
(489 - 843)
204
(150-258)
113
(1 - 225)
120
(88-152)
8
(0-16)
171
(125-216)
18
(-43 - 76)
13/35
33
(-22 - 88)
196
(144-248)
15
(-10-40)
25
(-17-67)
206
(151 -260)
22
(0 - 44)
54
(-36-144)
258
(3-511)
649
(477-821)
191
(140-242)
113
(1 - 225)
120
(88-152)
8
(0-16)
148
(109-187)
18
(-43 - 76)
12/35
30
(-20 - 80)
180
(132-228)
14
(-9 - 36)
24
(-15-63)
189
(139-239)
22
(0 - 44)
50
(-33-131)
243
(3 - 482)
597
(438 - 755)
176
(129-222)
108
(1 -213)
120
(88-152)
8
(0-16)
136
(100-172)
18
(-43 - 76)
13/30
33
(-22 - 88)
192
(141 -243)
15
(-10-40)
25
(-17-67)
192
(141 -242)
19
(0 - 38)
54
(-36-144)
220
(3 - 437)
570
(418-721)
175
(128-221)
106
(1 -210)
102
(75-130)
7
(0-14)
146
(107-184)
15
(-36 - 65)
12/25
30
(-20 - 79)
159
(117-201)
12
(-8 - 33)
24
(-15-63)
159
(117-201)
16
(0-31)
50
(-33-131)
183
(2 - 362)
473
(347 - 598)
145
(106-183)
88
(1-175)
85
(62-107)
6
(0-11)
121
(89- 153)
12
(-30 - 54)
 '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.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-136
September 2009
Draft - Do Not Quote or Cite

-------
 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.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 Ambient
PM2.5
Concentrations
44
(-29-117)
227
(167-287)
22
(-15-59)
27
(-18-72)
265
(195-335)
44
(1 - 87)
68
(-45-180)
420
(5-831)
813
(597-1028)
218
(160-276)
108
(1-215)
207
(152-261)
16
(0-31)
196
(144-248)
23
(-57-100)
15/352
41
(-27-108)
215
(158-271)
17
(-1 1 - 46)
27
(-18-72)
215
(158-272)
24
(0 - 48)
64
(-42-169)
265
(3 - 526)
676
(496 - 855)
201
(148-254)
108
(1-215)
120
(88-152)
10
(0-20)
176
(129-222)
19
(-46 - 82)
13/35
35
(-23 - 92)
188
(138-237)
15
(-10-39)
27
(-18-72)
197
(145-249)
24
(0 - 48)
54
(-36-144)
265
(3 - 526)
659
(484 - 833)
188
(138-238)
108
(1-215)
120
(88-152)
10
(0-20)
152
(112-193)
19
(-46 - 82)
12/35
32
(-21 -84)
173
(127-218)
14
(-9 - 36)
25
(-17-67)
181
(133-229)
24
(0 - 48)
50
(-33- 131)
250
(3 - 496)
605
(444 - 765)
173
(127-218)
103
(1 -204)
120
(88-152)
10
(0-20)
140
(103-177)
19
(-46 - 82)
13/30
35
(-23 - 92)
184
(135-232)
15
(-10-39)
27
(-18-72)
184
(135-232)
21
(0-41)
54
(-36- 144)
227
(3 - 449)
578
(424-731)
172
(126-217)
101
(1 -201)
102
(75-130)
8
(0-17)
150
(110-190)
16
(-39 - 70)
12/25
31
(-20 - 82)
152
(112-193)
12
(-8 - 32)
25
(-17-67)
152
(112-193)
17
(0 - 34)
50
(-33-131)
188
(2 - 373)
480
(352 - 607)
143
(105-180)
84
(1-167)
85
(62-108)
7
(0-14)
124
(91 -157)
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.
 2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-137
September 2009
Draft - Do Not Quote or Cite

-------
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 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
0.41%
(-0.27% -1.09%)
1.59%
(1.17% -2. 01%)
0.47%
(-0.31%- 1.23%)
0.31%
(-0.2% -0.83%)
1.59%
(1.17% -2. 01%)
0.82%
(0.01% -1.61%)
0.38%
(-0.25% -1.01%)
0.8%
(0.01% -1.58%)
1.36%
(1 %-1.72%)
1.38%
(1.02% -1.75%)
0.53%
(0.01% -1.05%)
1 .79%
(1.32% -2.26%)
0.57%
(0.01% -1.1 3%)
1.58%
(1.1 6% -2%)
0.76%
(-1.86% -3. 26%)
15/352
0.38%
(-0.25% -1%)
1.51%
(1.11% -1.9%)
0.37%
(-0.24% - 0.97%)
0.31%
(-0.2% - 0.83%)
1 .29%
(0.95% -1.63%)
0.45%
(0.01% -0.89%)
0.36%
(-0.24% - 0.95%)
0.5%
(0.01% -1%)
1.13%
(0.83% -1.43%)
1 .27%
(0.94% -1.61%)
0.53%
(0.01% -1.05%)
1.04%
(0.76% -1.31%)
0.36%
(0% - 0.72%)
1 .42%
(1.04% -1.8%)
0.62%
(-1.5% -2. 65%)
13/35
0.32%
(-0.21% -0.85%)
1.32%
(0.97%- 1.66%)
0.31%
(-0.2% - 0.82%)
0.31%
(-0.2% - 0.83%)
1.18%
(0.87% -1.5%)
0.45%
(0.01% -0.89%)
0.31%
(-0.2% -0.81%)
0.5%
(0.01% -1%)
1.1%
(0.81% -1.39%)
1.19%
(0.88% -1.51%)
0.53%
(0.01% -1.05%)
1.04%
(0.76% -1.31%)
0.36%
(0% - 0.72%)
1 .23%
(0.91% -1.56%)
0.62%
(-1.5% -2. 65%)
12/35
0.29%
(-0.1 9% -0.77%)
1.21%
(0.89% -1.53%)
0.28%
(-0.1 9% -0.75%)
0.29%
(-0.1 9% -0.77%)
1.09%
(0.8% -1.37%)
0.45%
(0.01% -0.89%)
0.28%
(-0.1 8% -0.74%)
0.48%
(0.01% -0.94%)
1.01%
(0.74% - 1 .28%)
1.09%
(0.8% -1.38%)
0.5%
(0.01% -0.99%)
1.04%
(0.76% -1.31%)
0.36%
(0% - 0.72%)
1 . 1 3%
(0.83% -1.43%)
0.62%
(-1.5% -2. 65%)
13/30
0.32%
(-0.21% -0.85%)
1 .29%
(0.95% -1.63%)
0.31%
(-0.2% - 0.82%)
0.31%
(-0.2% - 0.83%)
1.1%
(0.81%- 1.39%)
0.38%
(0% - 0.76%)
0.31%
(-0.2% -0.81%)
0.43%
(0.01% -0.85%)
0.97%
(0.71% -1.22%)
1.09%
(0.8% -1.38%)
0.49%
(0.01% -0.98%)
0.89%
(0.65% -1.1 2%)
0.31%
(0%-0.61%)
1.21%
(0.89% -1.53%)
0.53%
(-1.28% -2. 27%)
12/25
0.29%
(-0.1 9% -0.76%)
1.07%
(0.79% -1.35%)
0.26%
(-0.1 7% -0.68%)
0.29%
(-0.1 9% -0.77%)
0.91%
(0.67% -1.1 6%)
0.32%
(0% - 0.63%)
0.28%
(-0.1 8% -0.74%)
0.36%
(0%-0.71%)
0.8%
(0.59% -1.02%)
0.9%
(0.66% -1.1 4%)
0.41%
(0%-0.81%)
0.74%
(0.54% - 0.93%)
0.26%
(0%-0.51%)
1.01%
(0.74% -1.27%)
0.44%
(-1.06% -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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-138
September 2009
Draft - Do Not Quote or Cite

-------
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 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
0.39%
(-0.26% -1.04%)
1 .45%
(1.07%- 1.84%)
0.45%
(-0.3%- 1.19%)
0.26%
(-0.1 7% -0.7%)
1.33%
(0.98% -1.69%)
0.85%
(0.01% -1.67%)
0.39%
(-0.25% -1.03%)
0.74%
(0.01% -1.46%)
1 .24%
(0.91% -1.57%)
1.37%
(1.01% -1.73%)
0.54%
(0.01% -1.07%)
1 .63%
(1.2% -2.06%)
0.51%
(0.01% -1.01%)
1.36%
(1%-1.71%)
0.61%
(-1 .49% - 2.63%)
15/352
0.36%
(-0.24% - 0.95%)
1.37%
(1.01% -1.74%)
0.35%
(-0.23% - 0.93%)
0.26%
(-0.1 7% -0.7%)
1.08%
(0.79% -1.37%)
0.47%
(0.01% -0.92%)
0.37%
(-0.24% - 0.97%)
0.47%
(0.01% -0.93%)
1.03%
(0.76% -1.3%)
1 .26%
(0.92% -1.59%)
0.54%
(0.01% -1.07%)
0.94%
(0.69% -1.1 9%)
0.32%
(0% - 0.64%)
1.21%
(0.89% -1.54%)
0.5%
(-1.2% -2. 14%)
13/35
0.31%
(-0.2% -0.81%)
1.2%
(0.88%- 1.52%)
0.3%
(-0.2% - 0.79%)
0.26%
(-0.1 7% -0.7%)
0.99%
(0.73%- 1.26%)
0.47%
(0.01% -0.92%)
0.31%
(-0.2% - 0.82%)
0.47%
(0.01% -0.93%)
1%
(0.74% -1.27%)
1.18%
(0.87% -1.49%)
0.54%
(0.01% -1.07%)
0.94%
(0.69% -1.1 9%)
0.32%
(0% - 0.64%)
1.05%
(0.77% -1.33%)
0.5%
(-1.2% -2. 14%)
12/35
0.28%
(-0.1 8% -0.74%)
1.1%
(0.81% -1.4%)
0.27%
(-0.1 8% -0.72%)
0.24%
(-0.1 6% -0.65%)
0.91%
(0.67% -1.1 5%)
0.47%
(0.01% -0.92%)
0.28%
(-0.1 9% -0.75%)
0.44%
(0.01% -0.87%)
0.92%
(0.68% -1.1 7%)
1.08%
(0.79% -1.37%)
0.51%
(0.01% -1.02%)
0.94%
(0.69% -1.1 9%)
0.32%
(0% - 0.64%)
0.97%
(0.71% -1.22%)
0.5%
(-1.2% -2. 14%)
13/30
0.31%
(-0.2% -0.81%)
1.17%
(0.86%- 1.49%)
0.3%
(-0.2% - 0.79%)
0.26%
(-0.1 7% -0.7%)
0.92%
(0.68%- 1.17%)
0.4%
(0% - 0.79%)
0.31%
(-0.2% - 0.82%)
0.4%
(0% - 0.79%)
0.88%
(0.65% -1.1 2%)
1.08%
(0.79% -1.36%)
0.51%
(0.01% -1.01%)
0.81%
(0.59% -1.02%)
0.27%
(0%-0.54%)
1.04%
(0.76% -1.31%)
0.42%
(-1.02% -1.83%)
12/25
0.27%
(-0.1 8% -0.73%)
0.98%
(0.72% -1.23%)
0.25%
(-0.1 6% -0.66%)
0.24%
(-0.1 6% -0.65%)
0.77%
(0.56% - 0.97%)
0.33%
(0% - 0.65%)
0.28%
(-0.1 9% -0.75%)
0.33%
(0% - 0.66%)
0.73%
(0.54% -0.93%)
0.89%
(0.66% -1.1 3%)
0.42%
(0.01% -0.84%)
0.67%
(0.49% - 0.85%)
0.23%
(0% - 0.45%)
0.86%
(0.63% -1.09%)
0.35%
(-0.85% -1.52%)
 '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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-139
September 2009
Draft - Do Not Quote or Cite

-------
Table E-105. Estimated Percent of Total Annual Incidence of Hospital Admissions for Cardiovascular Illness Associated with
          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 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
0.4%
(-0.26% -1.05%)
1.4%
(1.03%- 1.76%)
0.44%
(-0.29% -1.1 7%)
0.28%
(-0.1 8% -0.73%)
1 .29%
(0.95%- 1.63%)
0.91%
(0.01%- 1.79%)
0.38%
(-0.25% -1%)
0.75%
(0.01% -1.48%)
1 .25%
(0.92% -1.58%)
1.36%
(1 %-1.72%)
0.5%
(0.01% -0.99%)
1 .65%
(1.21% -2.08%)
0.6%
(0.01% -1.1 9%)
1 .4%
(1.03% -1.77%)
0.65%
(-1.58% -2. 77%)
15/352
0.36%
(-0.24% - 0.96%)
1.32%
(0.97% -1.67%)
0.35%
(-0.23% - 0.92%)
0.28%
(-0.1 8% -0.73%)
1.04%
(0.77% -1.32%)
0.5%
(0.01% -0.99%)
0.36%
(-0.23% - 0.94%)
0.47%
(0.01% -0.93%)
1.04%
(0.76% -1.31%)
1 .25%
(0.92% -1.58%)
0.5%
(0.01% -0.99%)
0.96%
(0.7% -1.21%)
0.38%
(0% - 0.75%)
1 .25%
(0.92% -1.58%)
0.52%
(-1 .28% - 2.26%)
13/35
0.31%
(-0.2% - 0.82%)
1 . 1 5%
(0.85% -1.46%)
0.3%
(-0.1 9% -0.78%)
0.28%
(-0.1 8% -0.73%)
0.96%
(0.7%- 1.21%)
0.5%
(0.01% -0.99%)
0.3%
(-0.2% - 0.8%)
0.47%
(0.01% -0.93%)
1.01%
(0.74% -1.28%)
1.17%
(0.86% -1.48%)
0.5%
(0.01% -0.99%)
0.96%
(0.7% -1.21%)
0.38%
(0% - 0.75%)
1.09%
(0.8% -1.37%)
0.52%
(-1.28% -2.26%)
12/35
0.28%
(-0.1 8% -0.75%)
1.06%
(0.78% -1.34%)
0.27%
(-0.1 8% -0.71%)
0.26%
(-0.1 7% -0.68%)
0.88%
(0.65% -1.11%)
0.5%
(0.01% -0.99%)
0.28%
(-0.1 8% -0.73%)
0.45%
(0.01% -0.88%)
0.93%
(0.68% -1.1 7%)
1.08%
(0.79% -1.36%)
0.47%
(0.01% -0.94%)
0.96%
(0.7% -1.21%)
0.38%
(0% - 0.75%)
1%
(0.73% - 1 .26%)
0.52%
(-1 .28% - 2.26%)
13/30
0.31%
(-0.2% - 0.82%)
1.13%
(0.83% -1.43%)
0.3%
(-0.1 9% -0.78%)
0.28%
(-0.1 8% -0.73%)
0.89%
(0.65%- 1.13%)
0.43%
(0.01% -0.85%)
0.3%
(-0.2% - 0.8%)
0.4%
(0% - 0.8%)
0.89%
(0.65% -1.1 2%)
1.07%
(0.79% -1.35%)
0.47%
(0.01% -0.93%)
0.82%
(0.6% -1.03%)
0.32%
(0% - 0.64%)
1.07%
(0.79% -1.35%)
0.45%
(-1.09% -1.93%)
12/25
0.28%
(-0.1 8% -0.73%)
0.94%
(0.69% -1.1 9%)
0.24%
(-0.1 6% -0.65%)
0.26%
(-0.1 7% -0.68%)
0.74%
(0.54% -0.94%)
0.36%
(0% - 0.7%)
0.28%
(-0.1 8% -0.73%)
0.33%
(0% - 0.66%)
0.74%
(0.54% - 0.93%)
0.89%
(0.65% -1.1 2%)
0.39%
(0% - 0.77%)
0.68%
(0.5% - 0.86%)
0.27%
(0% - 0.53%)
0.89%
(0.65% -1.1 2%)
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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-140
September 2009
Draft - Do Not Quote or Cite

-------
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 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):
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-81% --82%)
-6%
(-6% - -6%)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-72% - -73%)
-59%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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%)
10%
(10% -10%)
14%
(14% -14%)
5%
(5% - 5%)
0%
(0% - 0%)
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%
(15% -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%
(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%)
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.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.

                                                                    E-141
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-81% --82%)
-6%
(-6% - -6%)
-58%
(-58% - -59%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-72%
(-72% - -73%)
-59%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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%)
10%
(10% -10%)
14%
(14% -14%)
5%
(5% - 5%)
0%
(0% - 0%)
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%
(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%)
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.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                    E-142
September 2009
Draft - Do Not Quote or Cite

-------
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 Sta
Short-Term Exposure to PM2.5 Concenti
Alternative Annual (n
Recent PM2.5
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%)
-72%
(-72% - -73%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -24%)
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%)
idards: Annual Incidence of Cardiovascular Hospital Admissions Associated with
ations in a Recent Year and PM25 Concentrations that Just Meet the Current and
and Daily (m) Standards (Standard Combination Denoted n/m):
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%)
0%
(0% - 0%)
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%)
10%
(10% -10%)
14%
(14% -14%)
5%
(5% - 5%)
0%
(0% - 0%)
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%
(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%)
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.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.

                                                                    E-143
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-34.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiovascular Hospital Admissions Associated with Short-Term Exposure to
PM2.5: Based on 2005 Air Quality Data*
       ra
       •c
       c
       S
       V)
       -  -10%
       o
       Q

       I
          -20%
-30%

-40%
       S  -i
          50%
       fi  -60%
          -70%
          -80%
          -90%
7
                 2005 air
                  quality
                   15/35**      13/35       12/35

                           Alternative Standard
                                     13/30
12/25
                       -•-Atlanta, GA 40 (-26-105); 0.38% (-0.25%-1%)
                       -•- Baltimore, MD 247  (182-313); 1.51% (1.11%-1.9%)
                       -*- Birmingham, AL 18 (-12-48); 0.37% (-0.24% - 0.97%)
                         - Dallas, TX 30 (-19-78); 0.31% (-0.2% - 0.83%)
                       -*- Detroit, Ml 269 (198-340); 1.29% (0.95% -1.63%)
                       -+- Fresno, CA 21 (0-42); 0.45% (0.01%-0.89%)
                       -i- Houston, TX 61 (-40-161); 0.36% (-0.24% - 0.95%)
                       — Los Angeles,  CA 274 (3 - 543); 0.5% (0.01 % -1 %)
                           New York, NY 724 (532-915); 1.13% (0.83% -1.43%)
                       -•- Philadelphia, PA  210 (154-265); 1.27%  (0.94% -1.61 %)
                       -•- Phoenix, AZ 107 (1 -212); 0.53% (0.01 % -1.05%)
                       -+- Pittsburgh, PA 134  (98-169); 1.04% (0.76% -1.31 %)
                           Salt Lake City, UT 9 (0 -17); 0.36% (0% - 0.72%)
                           St. Louis, MO 200 (147-253); 1.42% (1.04%-1.8%)
                           Tacoma, WA  21  (-52 - 92); 0.62% (-1.5% - 2.65%)
*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.
                                                 E-144
   September 2009
                                                                        Draft - Do Not Quote or Cite

-------
Figure E-35.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiovascular Hospital Admissions Associated with Short-Term Exposure to
PM2.5: Based on 2006 Air Quality Data*
          30%
          20%
       S
       OT
       -  -10%
=  -20%
E
8  -30%
I
       a:
          -40%
          -50%
          -60%
          -70%
          -80%
          -90%
                        JL
                 2006 air
                  quality
                     15/35**      13/35      12/35

                              Alternative Standard
13/30
12/25
                       -•-Atlanta, GA 39 (-26-103); 0.36% (-0.24% - 0.95%)
                       -•- Baltimore, MD 224 (165-284); 1.37% (1.01% -1.74%)
                       -*- Birmingham, AL 18  (-12-47); 0.35% (-0.23% - 0.93%)
                         - Dallas, TX 25 (-17-67); 0.26% (-0.17%-0.7%)
                       -*- Detroit, Ml 224 (165-284); 1.08% (0.79% -1.37%)
                       -•- Fresno, CA 22 (0-44); 0.47% (0.01%-0.92%)
                       -i- Houston, TX 64 (-42-170); 0.37% (-0.24% - 0.97%)
                       	 Los Angeles,  CA 258 (3-511); 0.47% (0.01%-0.93%)
                           New York, NY 666 (489 - 843); 1.03% (0.76% -1.3%)
                       -»- Philadelphia, PA 204 (150-258);  1.26% (0.92% -1.59%)
                       -m- Phoenix, AZ 113 (1 -225); 0.54% (0.01 % -1.07%)
                       -*- Pittsburgh, PA 120 (88-152); 0.94% (0.69% -1.19%)
                           Salt Lake City, UT 8 (0 -16); 0.32% (0% - 0.64%)
                           St. Louis, MO 171 (125-216); 1.21% (0.89% -1.54%)
                           Tacoma.WA  18  (-43-76); 0.5%  (-1.2%-2.14%)
*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.
                                                E-145
   September 2009
                                                                           Draft - Do Not Quote or Cite

-------
Figure E-36.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Cardiovascular Hospital Admissions Associated with Short-Term Exposure to
PM2.5: Based on 2007 Air Quality Data*
       S
       OT
       -  -10%
       O
       Q

       I
          -20%
-30%

-40%
       S  -i
       o:
       01
       o
       8.
          50%
-60%
          -70%
          -80%
          -90%
7
                  2007 air
                  quality
                   15/35**      13/35       12/35

                           Alternative Standard
                                     13/30
12/25
                       -•-Atlanta, GA 41 (-27-108); 0.36% (-0.24% - 0.96%)
                       -•- Baltimore, MD 215 (158-271); 1.32% (0.97% -1.67%)
                       -*- Birmingham, AL 17 (-11 -46); 0.35% (-0.23% - 0.92%)
                           Dallas, TX 27 (-18-72);  0.28% (-0.18%-0.73%)
                       -*- Detroit, Ml 215 (158-272); 1.04% (0.77% -1.32%)
                       -•- Fresno, CA 24 (0-48); 0.5% (0.01%-0.99%)
                       -i- Houston, TX 64 (-42-169); 0.36% (-0.23% - 0.94%)
                       	 Los Angeles,  CA 265  (3 - 526); 0.47% (0.01 % - 0.93%)
                           New York, NY 676 (496 - 855); 1.04% (0.76% -1.31 %)
                       -»- Philadelphia, PA  201 (148-254); 1.25% (0.92% -1.58%)
                       -m- Phoenix, AZ 108 (1 -215); 0.5% (0.01 %-0.99%)
                       -*- Pittsburgh, PA 120 (88-152); 0.96% (0.7%-1.21%)
                           Salt Lake City, UT 10 (0-20); 0.38% (0%-0.75%)
                           St. Louis, MO 176 (129-222); 1.25% (0.92% -1.58%)
                       -^- Tacoma.WA  19  (-46-82); 0.52% (-1.28% - 2.26%)
*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.
                                                 E-146
   September 2009
                                                                        Draft - Do Not Quote or Cite

-------
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 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
19
(-23 - 60)
21
(-12-54)
10
(-13-32)
14
(-18-46)
30
(-17-76)
26
(6 - 45)
29
(-36 - 93)
279
(65 - 490)
72
(-42-185)
18
(-11-47)
61
(14-106)
19
(-11 -49)
10
(2-18)
27
(-16-69)
2
(-34 - 37)
15/352
17
(-21 - 55)
20
(-12-51)
8
(-10-25)
14
(-18-46)
24
(-14-62)
14
(3 - 25)
27
(-34 - 88)
177
(41 -311)
60
(-35-154)
17
(-10-43)
61
(14-106)
11
(-6 - 28)
6
(1-11)
24
(-14-62)
2
(-27 - 30)
13/35
14
(-18-47)
17
(-10-45)
7
(-8-21)
14
(-18-46)
22
(-13-57)
14
(3-25)
23
(-29 - 75)
177
(41 -311)
58
(-34-150)
16
(-9-41)
61
(14-106)
11
(-6 - 28)
6
(1-11)
21
(-12-54)
2
(-27 - 30)
12/35
13
(-17-43)
16
(-9-41)
6
(-8 - 20)
13
(-17-43)
20
(-12-52)
14
(3 - 25)
21
(-27 - 68)
167
(39 - 293)
53
(-31 -138)
14
(-8 - 37)
57
(13-101)
11
(-6 - 28)
6
(1-11)
19
(-11 -49)
2
(-27 - 30)
13/30
14
(-18-47)
17
(-10-44)
7
(-8-21)
14
(-18-46)
20
(-12-53)
12
(3-21)
23
(-29 - 75)
151
(35 - 266)
51
(-30-132)
14
(-8 - 37)
57
(13-100)
9
(-5 - 24)
5
(1-10)
21
(-12-53)
2
(-23 - 26)
12/25
13
(-16-42)
14
(-8 - 36)
6
(-7-18)
13
(-17-43)
17
(-10-44)
10
(2-18)
21
(-27 - 68)
125
(29-221)
42
(-25-109)
12
(-7-31)
47
(11 -83)
8
(-5 - 20)
5
(1-8)
17
(-10-44)
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.
2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-147
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
18
(-23 - 59)
19
(-11 -49)
10
(-12-31)
12
(-15-40)
25
(-15-64)
27
(6 - 47)
30
(-38 - 98)
263
(61 - 462)
66
(-39-171)
18
(-10-46)
64
(15-113)
17
(-10-43)
9
(2-16)
23
(-13-59)
2
(-28 - 30)
15/352
17
(-21 - 54)
18
(-11 -46)
8
(-10-24)
12
(-15-40)
20
(-12-52)
15
(3 - 26)
29
(-36 - 93)
166
(39 - 293)
55
(-32-142)
16
(-10-42)
64
(15-113)
10
(-6 - 25)
6
(1-10)
21
(-12-53)
2
(-22 - 25)
13/35
14
(-18-46)
16
(-9-41)
6
(-8-21)
12
(-15-40)
18
(-11 -47)
15
(3 - 26)
24
(-31 -79)
166
(39 - 293)
54
(-32-138)
15
(-9 - 39)
64
(15-113)
10
(-6 - 25)
6
(1-10)
18
(-10-46)
2
(-22 - 25)
12/35
13
(-16-42)
14
(-9 - 37)
6
(-7-19)
11
(-14-37)
17
(-10-43)
15
(3 - 26)
22
(-28 - 72)
157
(37 - 276)
49
(-29-127)
14
(-8 - 36)
61
(14-107)
10
(-6 - 25)
6
(1-10)
16
(-10-42)
2
(-22 - 25)
13/30
14
(-18-46)
15
(-9 - 40)
6
(-8-21)
12
(-15-40)
17
(-10-44)
13
(3 - 22)
24
(-31 - 79)
142
(33 - 250)
47
(-28-121)
14
(-8 - 36)
60
(14-106)
8
(-5 - 22)
5
(1-9)
18
(-10-45)
1
(-19-21)
12/25
13
(-16-41)
13
(-8 - 33)
5
(-7-17)
11
(-14-37)
14
(-8 - 36)
11
(2-19)
22
(-28 - 72)
118
(27 - 208)
39
(-23-101)
12
(-7 - 30)
50
(12-88)
7
(-4-18)
4
(1-7)
15
(-9 - 37)
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.
2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-148
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
19
(-24-61)
18
(-11-47)
10
(-12-31)
13
(-16-42)
24
(-14-61)
29
(7-51)
30
(-38 - 98)
270
(63 - 475)
67
(-40-173)
18
(-10-45)
61
(14-108)
17
(-10-43)
11
(3 - 20)
24
(-14-61)
2
(-30 - 33)
15/352
17
(-22 - 56)
17
(-10-44)
7
(-9 - 24)
13
(-16-42)
19
(-11 -49)
16
(4 - 29)
29
(-36 - 93)
171
(40-301)
56
(-33-144)
16
(-9 - 42)
61
(14-108)
10
(-6 - 25)
7
(2-13)
21
(-12-54)
2
(-24 - 27)
13/35
15
(-19-48)
15
(-9 - 39)
6
(-8-21)
13
(-16-42)
18
(-10-45)
16
(4 - 29)
24
(-31 -79)
171
(40-301)
54
(-32-140)
15
(-9 - 39)
61
(14-108)
10
(-6 - 25)
7
(2-13)
18
(-11-47)
2
(-24 - 27)
12/35
14
(-17-44)
14
(-8 - 36)
6
(-7-19)
12
(-15-39)
16
(-9 - 42)
16
(4 - 29)
22
(-28 - 72)
161
(38 - 284)
50
(-29-129)
14
(-8 - 36)
58
(14-103)
10
(-6 - 25)
7
(2-13)
17
(-10-43)
2
(-24 - 27)
13/30
15
(-19-48)
15
(-9 - 38)
6
(-8-21)
13
(-16-42)
16
(-10-42)
14
(3 - 24)
24
(-31 - 79)
146
(34 - 257)
48
(-28-123)
14
(-8 - 35)
57
(13-101)
8
(-5 - 22)
6
(1-11)
18
(-11 -46)
2
(-21 - 23)
12/25
13
(-17-43)
12
(-7-31)
5
(-7-17)
12
(-15-39)
14
(-8 - 35)
12
(3 - 20)
22
(-28 - 72)
121
(28-214)
40
(-23-102)
11
(-7 - 29)
48
(11-84)
7
(-4-18)
5
(1-9)
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.
2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-149
September 2009
Draft - Do Not Quote or Cite

-------
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 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
0.5%
(-0.63% - 1 .6%)
0.42%
(-0.25% -1.08%)
0.56%
(-0.71%- 1.82%)
0.38%
(-0.47% - 1 .22%)
0.42%
(-0.25% -1.07%)
1 .44%
(0.34% -2. 51%)
0.46%
(-0.58% -1.49%)
1.41%
(0.33% - 2.47%)
0.36%
(-0.21% -0.92%)
0.36%
(-0.21% -0.93%)
0.93%
(0.22% - 1 .64%)
0.47%
(-0.28% -1.21%)
1.01%
(0.24% - 1 .77%)
0.42%
(-0.25% -1.07%)
0.2%
(-2.72% - 2.97%)
15/352
0.46%
(-0.57% -1.47%)
0.4%
(-0.23% -1.02%)
0.44%
(-0.56% -1.43%)
0.38%
(-0.47% -1.22%)
0.34%
(-0.2% - 0.87%)
0.79%
(0.1 9% -1.4%)
0.44%
(-0.55%- 1.41%)
0.89%
(0.21% -1.57%)
0.3%
(-0.1 7% -0.76%)
0.33%
(-0.2% - 0.86%)
0.93%
(0.22% - 1 .64%)
0.27%
(-0.1 6% -0.7%)
0.64%
(0.15%- 1.12%)
0.37%
(-0.22% - 0.96%)
0.16%
(-2. 19% -2.41%)
13/35
0.39%
(-0.49% - 1 .25%)
0.35%
(-0.2% - 0.89%)
0.37%
(-0.47% -1.21%)
0.38%
(-0.47% - 1 .22%)
0.31%
(-0.1 8% -0.8%)
0.79%
(0.1 9% -1.4%)
0.37%
(-0.47% -1.2%)
0.89%
(0.21% -1.57%)
0.29%
(-0.1 7% -0.74%)
0.31%
(-0.1 8% -0.8%)
0.93%
(0.22% - 1 .64%)
0.27%
(-0.1 6% -0.7%)
0.64%
(0.1 5% -1.1 2%)
0.32%
(-0.1 9% -0.83%)
0.16%
(-2. 19% -2.41%)
12/35
0.35%
(-0.44%- 1.14%)
0.32%
(-0.1 9% -0.82%)
0.34%
(-0.43%- 1.11%)
0.35%
(-0.44% -1.1 3%)
0.28%
(-0.1 7% -0.73%)
0.79%
(0.19%- 1.4%)
0.34%
(-0.42% -1.09%)
0.84%
(0.2% - 1 .48%)
0.27%
(-0.1 6% -0.68%)
0.29%
(-0.1 7% -0.74%)
0.89%
(0.21% -1.56%)
0.27%
(-0.1 6% -0.7%)
0.64%
(0.1 5% -1.1 2%)
0.3%
(-0.1 7% -0.76%)
0.16%
(-2. 19% -2.41%)
13/30
0.39%
(-0.49% - 1 .25%)
0.34%
(-0.2% - 0.87%)
0.37%
(-0.47% -1.21%)
0.38%
(-0.47% -1.22%)
0.29%
(-0.1 7% -0.74%)
0.68%
(0.1 6% -1.1 9%)
0.37%
(-0.47% - 1 .2%)
0.76%
(0.18%- 1.34%)
0.25%
(-0.1 5% -0.65%)
0.29%
(-0.1 7% -0.74%)
0.87%
(0.2% -1.54%)
0.23%
(-0.1 4% -0.6%)
0.54%
(0.1 3% -0.96%)
0.32%
(-0.1 9% -0.82%)
0.14%
(-1.87% -2.07%)
12/25
0.35%
(-0.44% -1.1 2%)
0.28%
(-0.1 6% -0.72%)
0.31%
(-0.39% -1.01%)
0.35%
(-0.44% -1.1 3%)
0.24%
(-0.1 4% -0.62%)
0.56%
(0.1 3% -0.99%)
0.34%
(-0.42% -1.09%)
0.63%
(0.1 5% -1.11%)
0.21%
(-0.1 2% -0.54%)
0.24%
(-0.1 4% -0.61%)
0.73%
(0.17%- 1.28%)
0.19%
(-0.11% -0.5%)
0.45%
(0.11% -0.79%)
0.26%
(-0.1 6% -0.68%)
0.11%
(-1.54% -1.72%)
 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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-150
September 2009
Draft - Do Not Quote or Cite

-------
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.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
0.47%
(-0.6% -1.53%)
0.38%
(-0.22% - 0.98%)
0.54%
(-0.69% - 1 .76%)
0.32%
(-0.4% -1.03%)
0.35%
(-0.21% -0.9%)
1 .49%
(0.35% -2.61%)
0.47%
(-0.59% -1.51%)
1.3%
(0.3% - 2.29%)
0.32%
(-0.1 9% -0.84%)
0.36%
(-0.21% -0.92%)
0.96%
(0.22% - 1 .68%)
0.43%
(-0.25%- 1.1%)
0.9%
(0.21%- 1.58%)
0.36%
(-0.21% -0.92%)
0.16%
(-2. 18% -2.39%)
15/352
0.43%
(-0.55% -1.4%)
0.36%
(-0.21% -0.93%)
0.43%
(-0.54% -1.38%)
0.32%
(-0.4%- 1.03%)
0.28%
(-0.1 7% -0.73%)
0.82%
(0.1 9% -1.45%)
0.44%
(-0.56% -1.43%)
0.82%
(0.1 9% -1.45%)
0.27%
(-0.1 6% -0.7%)
0.33%
(-0.1 9% -0.85%)
0.96%
(0.22% - 1 .68%)
0.25%
(-0.1 5% -0.64%)
0.57%
(0.13%-1%)
0.32%
(-0.1 9% -0.82%)
0.13%
(-1.76% -1.94%)
13/35
0.37%
(-0.46% -1.1 9%)
0.31%
(-0.1 9% -0.81%)
0.36%
(-0.46% -1.1 7%)
0.32%
(-0.4% -1.03%)
0.26%
(-0.1 5% -0.67%)
0.82%
(0.1 9% -1.45%)
0.38%
(-0.47% - 1 .22%)
0.82%
(0.1 9% -1.45%)
0.26%
(-0.1 5% -0.68%)
0.31%
(-0.1 8% -0.8%)
0.96%
(0.22% - 1 .68%)
0.25%
(-0.1 5% -0.64%)
0.57%
(0.13%-1%)
0.28%
(-0.1 6% -0.71%)
0.13%
(-1.76% -1.94%)
12/35
0.34%
(-0.42% -1.09%)
0.29%
(-0.1 7% -0.74%)
0.33%
(-0.42% -1.07%)
0.29%
(-0.37% - 0.95%)
0.24%
(-0.1 4% -0.62%)
0.82%
(0.1 9% -1.45%)
0.34%
(-0.43%- 1.11%)
0.78%
(0.1 8% -1.37%)
0.24%
(-0.1 4% -0.62%)
0.28%
(-0.1 7% -0.73%)
0.91%
(0.21% -1.6%)
0.25%
(-0.1 5% -0.64%)
0.57%
(0.13%-1%)
0.25%
(-0.1 5% -0.65%)
0.13%
(-1.76% -1.94%)
13/30
0.37%
(-0.46% -1.1 9%)
0.31%
(-0.1 8% -0.79%)
0.36%
(-0.46% -1.1 7%)
0.32%
(-0.4% -1.03%)
0.24%
(-0.1 4% -0.62%)
0.7%
(0.1 6% -1.24%)
0.38%
(-0.47% - 1 .22%)
0.7%
(0.1 6% -1.24%)
0.23%
(-0.14% -0.59%)
0.28%
(-0.1 7% -0.73%)
0.9%
(0.21% -1.58%)
0.21%
(-0.1 2% -0.54%)
0.48%
(0.11% -0.85%)
0.27%
(-0.1 6% -0.7%)
0.11%
(-1 .49% - 1 .66%)
12/25
0.33%
(-0.42% -1.07%)
0.26%
(-0.1 5% -0.66%)
0.3%
(-0.38% - 0.97%)
0.29%
(-0.37% - 0.95%)
0.2%
(-0.1 2% -0.52%)
0.58%
(0.1 4% -1.03%)
0.34%
(-0.43%- 1.11%)
0.58%
(0.1 4% -1.03%)
0.19%
(-0.11% -0.49%)
0.23%
(-0.1 4% -0.6%)
0.74%
(0.17% -1.31%)
0.17%
(-0.1% -0.45%)
0.4%
(0.09% -0.71%)
0.22%
(-0.1 3% -0.58%)
0.09%
(-1.23% -1.38%)
1 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.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-151
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
0.48%
(-0.6% -1.54%)
0.37%
(-0.22% - 0.94%)
0.54%
(-0.68% - 1 .73%)
0.33%
(-0.42% -1.08%)
0.34%
(-0.2% - 0.87%)
1 .6%
(0.38% - 2.79%)
0.46%
(-0.57% -1.47%)
1.32%
(0.31% -2. 31%)
0.33%
(-0.1 9% -0.84%)
0.36%
(-0.21% -0.92%)
0.88%
(0.21% -1.55%)
0.43%
(-0.26% -1.11%)
1.06%
(0.25% -1.86%)
0.37%
(-0.22% - 0.94%)
0.17%
(-2.32% - 2.52%)
15/352
0.44%
(-0.55% -1.42%)
0.35%
(-0.2% - 0.89%)
0.42%
(-0.53% -1.36%)
0.33%
(-0.42% -1.08%)
0.27%
(-0.1 6% -0.7%)
0.89%
(0.21% -1.55%)
0.43%
(-0.54% -1.39%)
0.83%
(0.1 9% -1.47%)
0.27%
(-0.1 6% -0.7%)
0.33%
(-0.1 9% -0.84%)
0.88%
(0.21% -1.55%)
0.25%
(-0.1 5% -0.64%)
0.67%
(0.1 6% -1.1 8%)
0.33%
(-0.1 9% -0.85%)
0.14%
(-1.87% -2. 05%)
13/35
0.37%
(-0.47% -1.21%)
0.3%
(-0.1 8% -0.78%)
0.36%
(-0.45% -1.1 5%)
0.33%
(-0.42% -1.08%)
0.25%
(-0.1 5% -0.65%)
0.89%
(0.21% -1.55%)
0.37%
(-0.46% -1.1 8%)
0.83%
(0.1 9% -1.47%)
0.26%
(-0.1 6% -0.68%)
0.31%
(-0.1 8% -0.79%)
0.88%
(0.21% -1.55%)
0.25%
(-0.1 5% -0.64%)
0.67%
(0.1 6% -1.1 8%)
0.28%
(-0.1 7% -0.73%)
0.14%
(-1.87% -2. 05%)
12/35
0.34%
(-0.43%- 1.1%)
0.28%
(-0.1 6% -0.72%)
0.32%
(-0.41% -1.05%)
0.31%
(-0.39%- 1%)
0.23%
(-0.1 4% -0.59%)
0.89%
(0.21% -1.55%)
0.33%
(-0.42%- 1.08%)
0.79%
(0.18%- 1.38%)
0.24%
(-0.1 4% -0.63%)
0.28%
(-0.1 7% -0.73%)
0.84%
(0.2% -1.47%)
0.25%
(-0.1 5% -0.64%)
0.67%
(0.1 6% -1.1 8%)
0.26%
(-0.1 5% -0.67%)
0.14%
(-1.87% -2. 05%)
13/30
0.37%
(-0.47% -1.21%)
0.3%
(-0.1 7% -0.76%)
0.36%
(-0.45% -1.1 5%)
0.33%
(-0.42% -1.08%)
0.23%
(-0.1 4% -0.6%)
0.76%
(0.1 8% -1.33%)
0.37%
(-0.46%- 1.18%)
0.71%
(0.17%- 1.25%)
0.23%
(-0.1 4% -0.6%)
0.28%
(-0.1 6% -0.72%)
0.83%
(0.1 9% -1.45%)
0.21%
(-0.1 3% -0.55%)
0.57%
(0.13%- 1.01%)
0.28%
(-0.1 6% -0.72%)
0.12%
(-1.59% -1.76%)
12/25
0.33%
(-0.42% -1.08%)
0.25%
(-0.1 4% -0.63%)
0.3%
(-0.37% - 0.96%)
0.31%
(-0.39%-1%)
0.19%
(-0.11% -0.5%)
0.63%
(0.1 5% -1.1%)
0.33%
(-0.42% -1.08%)
0.59%
(0.14%- 1.04%)
0.19%
(-0.11% -0.5%)
0.23%
(-0.1 4% -0.6%)
0.69%
(0.1 6% -1.21%)
0.18%
(-0.1% -0.46%)
0.47%
(0.11% -0.83%)
0.23%
(-0.1 4% -0.6%)
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.
 2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                     E-152
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -82%)
-6%
(-6% - -6%)
-58%
(-58% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-73%
(-73% - -73%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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%
(10% -11%)
14%
(14% -14%)
5%
(5% - 5%)
0%
(0% - 0%)
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%)
1 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. Percents are rounded to the nearest whole number.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-153
September 2009
Draft - Do Not Quote or Cite

-------
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):
Recent PM2.5
Concentrations
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -82%)
-6%
(-6% - -6%)
-58%
(-58% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-73%
(-73% - -73%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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%)
0%
(0% - 0%)
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%)
1 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. Percents are rounded to the nearest whole number.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-154
September 2009
Draft - Do Not Quote or Cite

-------
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 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
-9%
(-9% - -9%)
-6%
(-6% - -6%)
-28%
(-28% - -28%)
0%
(0% - 0%)
-23%
(-23% - -23%)
-81%
(-80% - -82%)
-6%
(-6% - -6%)
-58%
(-58% - -58%)
-20%
(-20% - -20%)
-9%
(-9% - -9%)
0%
(0% - 0%)
-73%
(-73% - -73%)
-58%
(-58% - -59%)
-12%
(-12% --12%)
-23%
(-23% - -24%)
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%)
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%)
0%
(0% - 0%)
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%)
0%
(0% - 0%)
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%)
1 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. Percents are rounded to the nearest whole number.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                      E-155
September 2009
Draft - Do Not Quote or Cite

-------
Figure E-37.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Respiratory Hospital Admissions Associated with Short-Term Exposure to
PM2.5: Based on 2005 Air Quality Data*
           10%
       S
       OT
       -  -10%
       =  -20%
       E
       8  -30%
       I
       a:
          -40%
          -50%
          -60%
          -70%
          -80%
          -90%
                          3
                  2005 air
                  quality
15/35**      13/35       12/35

         Alternative Standard
13/30
12/25
                        -»- Atlanta, GA 17 (-21 -55); 0.46% (-0.57% -1.47%)
                        -•- Baltimore, MD 20 (-12-51); 0.4% (-0.23% -1.02%)
                        -*- Birmingham, AL 8 (-10-25); 0.44%  (-0.56% -1.43%)
                           Dallas, TX 14 (-18-46); 0.38% (-0.47% -1.22%)
                        -*- Detroit, Ml 24 (-14-62); 0.34% (-0.2% - 0.87%)
                        -•- Fresno, CA 14 (3-25); 0.79% (0.19%-1.4%)
                        -i- Houston, TX 27 (-34 - 88); 0.44%  (-0.55% -1.41 %)
                        	 Los Angeles, CA 177 (41 -311); 0.89%  (0.21 % -1.57%)
                           New York, NY 60 (-35-154); 0.3% (-0.17%-0.76%)
                        -•- Philadelphia, PA 17 (-10-43); 0.33% (-0.2% - 0.86%)
                        -m- Phoenix, AZ 61  (14-106); 0.93% (0.22% -1.64%)
                        -*- Pittsburgh, PA 11 (-6-28); 0.27% (-0.16%-0.7%)
                           Salt Lake City, UT 6 (1-11);  0.64%  (0.15% -1.12%)
                           St. Louis, MO  24 (-14-62); 0.37% (-0.22% - 0.96%)
                           Tacoma.WA 2 (-27-30); 0.16% (-2.19%-2.41%)
*Based on Bell et al. (2008). 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.
                                                 E-156
   September 2009
                                                      Draft - Do Not Quote or Cite

-------
Figure E-38  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Respiratory Hospital Admissions Associated with Short-Term Exposure to
PM2.5: Based on 2006 Air Quality Data*
          30%
          20%
       S
       OT
       -  -10%
=  -20%
E
8  -30%
I
       a:
          -40%
          -50%
          -60%
          -70%
          -80%
          -90%
                            V
                          /I,
             I
                  2006 air
                  quality
                      15/35**      13/35       12/35

                              Alternative Standard
13/30
12/25
                       -»- Atlanta, GA 17 (-21 -54); 0.43% (-0.55% -1.4%)
                       -m- Baltimore, MD 18 (-11 -46); 0.36% (-0.21 %-0.93%)
                       -*- Birmingham, AL 8 (-10-24); 0.43% (-0.54% -1.38%)
                          - Dallas, TX 12 (-15-40); 0.32% (-0.4% -1.03%)
                       -*- Detroit, Ml 20 (-12-52); 0.28% (-0.17%-0.73%)
                       -•- Fresno, CA 15 (3-26); 0.82% (0.19% -1.45%)
                       -i- Houston, TX 29 (-36 - 93);  0.44%  (-0.56% -1.43%)
                       	 Los Angeles, CA 166 (39-293); 0.82%  (0.19% -1.45%)
                           New York, NY 55 (-32-142); 0.27% (-0.16%-0.7%)
                       -»- Philadelphia, PA 16 (-10-42); 0.33% (-0.19%-0.85%)
                       -m- Phoenix, AZ 64 (15-113); 0.96% (0.22% -1.68%)
                       -*- Pittsburgh, PA 10 (-6-25); 0.25% (-0.15%-0.64%)
                           Salt Lake City, UT 6 (1 -10); 0.57% (0.13% -1%)
                           St. Louis, MO  21 (-12-53); 0.32% (-0.19%-0.82%)
                           Tacoma.WA 2 (-22-25); 0.13% (-1.76% -1.94%)
*Based on Bell et al. (2008). 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.
                                                 E-157
   September 2009
                                                                           Draft - Do Not Quote or Cite

-------
Figure E-39.  Estimated Percent Reductions From the Current Standard to Alternative
Standards in Respiratory Hospital Admissions Associated with Short-Term Exposure to
PM2.5: Based on 2007 Air Quality Data*
           10%
       S
       OT
       -  -10%
       =  -20%
       E
       8  -30%
       I
       a:
          -40%
          -50%
          -60%
          -70%
          -80%
          -90%
                          3
                     /
                  2007 air
                  quality
15/35**      13/35       12/35

         Alternative Standard
13/30
12/25
                        -»- Atlanta, GA 17 (-22-56); 0.44% (-0.55% -1.42%)
                        -•- Baltimore, MD 17 (-10-44); 0.35% (-0.2% - 0.89%)
                        -*- Birmingham, AL  7 (-9-24); 0.42% (-0.53% -1.36%)
                           Dallas, TX 13 (-16-42); 0.33% (-0.42% -1.08%)
                        -*- Detroit, Ml 19 (-11 -49); 0.27% (-0.16%-0.7%)
                        -•- Fresno, CA 16 (4-29); 0.89% (0.21 % -1.55%)
                        -i- Houston, TX 29  (-36 - 93);  0.43% (-0.54% -1.39%)
                        	 Los Angeles, CA 171 (40-301); 0.83%  (0.19% -1.47%)
                           New York, NY 56 (-33-144); 0.27% (-0.16%-0.7%)
                        -•- Philadelphia, PA 16 (-9-42); 0.33% (-0.19%-0.84%)
                        -m- Phoenix, AZ 61  (14-108); 0.88% (0.21 % -1.55%)
                        -*- Pittsburgh, PA 10 (-6-25); 0.25% (-0.15%-0.64%)
                           Salt Lake City, UT 7 (2 -13); 0.67% (0.16% -1.18%)
                           St. Louis, MO  21 (-12-54); 0.33% (-0.19%-0.85%)
                           Tacoma.WA 2 (-24-27); 0.14% (-1.87% - 2.05%)
*Based on Bell et al. (2008). 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.
                                                 E-158
   September 2009
                                                      Draft - Do Not Quote or Cite

-------
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
Tolbert et al. (2007)
Tolbert et al. (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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily
(m) Standards (Standard Combination Denoted n/m):
Recent PM2.5
Concentrations
215
(-302 - 722)
808
(-811 -2400)
1162
(743-1567)
15/352
197
(-277 - 663)
741
(-743 - 2203)
971
(620-1314)
13/35
167
(-235 - 564)
630
(-631 -1875)
947
(604-1281)
12/35
153
(-214-514)
574
(-575-1710)
872
(556-1181)
13/30
167
(-235 - 564)
630
(-631 -1875)
834
(531 -1130)
12/25
150
(-211 -507)
566
(-566-1685)
695
(442 - 943)
 Numbers rounded to the nearest whole number.
2The current primary PM2.5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 E-159
September 2009
Draft - Do Not Quote or Cite

-------
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
Tolbert et al. (2007)
Tolbert et al. (2007)
Itoetal. (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 PM25 Concentrations that Just Meet the Current and Alternative Annual (n) and Daily (m)
Standards (Standard Combination Denoted n/m):
Recent PM25
Concentrations
212
(-298-713)
798
(-800 - 2373)
991
(633- 1340)
15/352
195
(-274 - 655)
732
(-734-2178)
828
(528-1122)
13/35
165
(-232 - 557)
622
(-623-1853)
807
(515-1094)
12/35
151
(-212-508)
567
(-568-1690)
743
(473-1008)
13/30
165
(-232 - 557)
622
(-623 - 1 853)
711
(452 - 964)
12/25
149
(-208 - 501 )
559
(-559 - 1 665)
592
(376 - 804)
1 Numbers rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-160
September 2009
Draft - Do Not Quote or Cite

-------
Table E-120. Estimated 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
Tolbert et al. (2007)
Tolbert et al. (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.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
(-310-739)
827
(-830 - 2457)
1114
(712-1504)
15/352
202
(-284 - 678)
759
(-761 - 2256)
931
(594-1260)
13/35
171
(-241 - 577)
645
(-646-1920)
907
(579-1229)
12/35
156
(-219-527)
588
(-589-1751)
836
(532-1133)
13/30
171
(-241 - 577)
645
(-646-1920)
799
(509-1083)
12/25
154
(-216-519)
580
(-580-1726)
665
(423 - 904)
1 Numbers rounded to the nearest whole number.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 E-161
September 2009
Draft - Do Not Quote or Cite

-------
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
Tolbert et al. (2007)
Tolbert et al. (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% - 2.2%)
0.6%
(-0.6% -1.9%)
6.5%
(4.2% - 8.8%)
15/352
0.6%
(-0.8% -2%)
0.6%
(-0.6% -1.8%)
5.5%
(3.5% - 7.4%)
13/35
0.5%
(-0.7% - 1 .7%)
0.5%
(-0.5% -1.5%)
5.3%
(3.4% - 7.2%)
12/35
0.5%
(-0.6% -1.5%)
0.5%
(-0.5% -1.4%)
4.9%
(3.1% -6. 7%)
13/30
0.5%
(-0.7% - 1 .7%)
0.5%
(-0.5% -1.5%)
4.7%
(3% - 6.4%)
12/25
0.5%
(-0.6% -1.5%)
0.5%
(-0.5% -1.3%)
3.9%
(2.5% - 5.3%)
1Percents rounded to the nearest tenth.
2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 E-162
September 2009
Draft - Do Not Quote or Cite

-------
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)
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.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% -2.1%)
0.6%
(-0.6% -1.8%)
5.6%
(3.6% - 7.5%)
15/352
0.6%
(-0.8% -1.9%)
0.6%
(-0.6% - 1 .7%)
4.7%
(3% - 6.3%)
13/35
0.5%
(-0.7% - 1 .6%)
0.5%
(-0.5% -1.4%)
4.5%
(2.9% -6.1%)
12/35
0.4%
(-0.6% -1.5%)
0.4%
(-0.4% -1.3%)
4.2%
(2.7% - 5.7%)
13/30
0.5%
(-0.7% -1.6%)
0.5%
(-0.5% -1.4%)
4%
(2.5% - 5.4%)
12/25
0.4%
(-0.6% - 1 .5%)
0.4%
(-0.4% -1.3%)
3.3%
(2.1% -4.5%)
1Percents rounded to the nearest tenth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-163
September 2009
Draft - Do Not Quote or Cite

-------
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.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% -2.1%)
0.6%
(-0.6% -1.8%)
6.2%
(4% - 8.4%)
15/352
0.6%
(-0.8% -1.9%)
0.6%
(-0.6% - 1 .7%)
5.2%
(3.3% - 7%)
13/35
0.5%
(-0.7% - 1 .6%)
0.5%
(-0.5% -1.4%)
5.1%
(3.2% - 6.9%)
12/35
0.4%
(-0.6% - 1 .5%)
0.4%
(-0.4% - 1 .3%)
4.7%
(3% - 6.3%)
13/30
0.5%
(-0.7% -1.6%)
0.5%
(-0.5% -1.4%)
4.5%
(2.8% -6.1%)
12/25
0.4%
(-0.6% -1.5%)
0.4%
(-0.4% -1.3%)
3.7%
(2.4% - 5%)
1Percents rounded to the nearest tenth.
2The current primary PM25 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                  E-164
September 2009
Draft - Do Not Quote or Cite

-------
            APPENDIX F: SENSITIVITY ANALYSIS RESULTS
                               F-l
September 2009                                          Draft - Do Not Quote or Cite

-------
 1                                 Appendix F. Sensitivity Analysis Results
 2
 3           This Appendix provides detailed results of the single- and multi-factor sensitivity
 4    analyses completed as part of this risk analysis. For additional detail on the sensitivity analysis
 5    results completed for this analysis, as well as the types of results generated, see section 4.3.
 6           We have identified an error in the approach used to simulate ambient PM2.5 levels for the
 7    Pittsburgh study area for the scenarios involving just meeting the current and alternative sets of
 8    standards.  Sensitivity analyses involving the Pittsburgh study area (focusing on the impact of
 9    using the alternative hybrid rollback approach - see section 3.5.4) involved the current and
10    alternative  sets of standard levels and consequently, these sensitivity analyses results are
11    impacted. However, there was  insufficient time after identifying this error to either generate
12    corrected risk estimates or remove the erroneous risk estimates from the summary tables
13    presented in this Appendix.  We will correct this error and release updated results for the
14    Pittsburgh study area as  soon as is practicable and will include the corrected results in the next
15    version of this document. Note, that this error does not impact sensitivity analysis results for any
16    of the other urban study  areas.
                                                 F-2
      September 2009                                                            Draft - Do Not Quote or Cite

-------
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
PM2.5 Concentrations Using:2
Standard Fixed
Effects Log-Linear
(Cox Proportional
Hazard) Model3
Random Effects Log-
Linear Model 4
Random Effects Log-
Log Model 5
Percent Difference 6
Fixed Effects vs.
Random Effects Log-
Linear Models
Random Effects Log-
Linear vs. Log-Log
Models
Los Angeles, CA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
1432
(911 -1948)
1626
(1270-1977)
1330
(1083-1572)
174
(76 - 269)
1767
(824 - 2694)
7
1486
(903 - 2044)
—
3526
(2179-4841)
2694
(1794-3563)
2652
(1878-3376)
322
(168-467)
23%
—
12%
—
146%
66%
99%
85%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
547
(349 - 743)
512
(401 - 620)
347
(284 - 409)
83
(37-127)
674
(316- 1023)
—
387
(238 - 527)
—
1201
(745-1641)
757
(507 - 996)
614
(439 - 774)
137
(72-196)
23%
—
12%
—
120%
48%
77%
65%
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 the study (5.8 ug/m3).

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 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 et al. (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.
September 2009
Draft - Do Not Quote or Cite

-------
 Table F-2. 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 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 Model 4
Random Effects Log-
Log Model 5
Percent Difference 6
Fixed Effects vs.
Random Effects Log-
Linear Models
Random Effects Log-
Linear vs. Log-Log
Models
Los Angeles, CA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
1221
(776 - 1 662)
1390
(1 084 - 1 692)
1140
(927 - 1 349)
149
(65-231)
1507
(702 - 2300)
7
1275
(772 - 1 760)
—
3130
(1931 -4303)
2396
(1592-3176)
2374
(1 675 - 3034)
286
(149-417)
23%
—
12%
—
1 56%
72%
1 08%
92%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
536
(341 - 727)
501
(393 - 608)
340
(279 - 401 )
81
(36 - 1 25)
660
(309 - 1 002)
—
380
(233 - 51 7)
—
1182
(733-1617)
746
(500 - 982)
606
(433 - 764)
134
(71 -194)
23%
—
12%
—
1 21 %
49%
78%
65%
1The current primary PM25 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 the study (5.8 ug/m3).

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 9. Autocorrelation at MSA and ZCA levels; MSA & DIFF, in Krewski et al. (2009) - exposure period from 1999 - 2000.
5Estimates based on Table 11, "MSA and DIFF" rows, in Krewski et al. (2009) - exposure period from 1999-2000.
6Calculated 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
September 2009
Draft - Do Not Quote or Cite

-------
Table F-3. 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 2007 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 Model 4
Random Effects Log-
Log Model 5
Percent Difference 6
Fixed Effects vs.
Random Effects Log-
Linear Models
Random Effects Log-
Linear vs. Log-Log
Models
Los Angeles, CA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
1257
(799-1711)
1431
(1116-1741)
1173
(954-1388)
153
(67 - 238)
1552
(723 - 2368)
7
1312
(795-1810)
—
3206
(1979-4407)
2454
(1631 -3251)
2429
(1715-3102)
293
(153-427)
23%
—
12%
—
155%
71%
107%
92%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Ischemic Heart Disease Mortality
Lung Cancer Mortality
530
(338-719)
496
(388-601)
337
(276 - 397)
80
(35-123)
653
(305-991)
—
376
(230-512)
—
1174
(728-1605)
741
(496 - 975)
602
(430 - 759)
134
(70-192)
23%
—
12%
—
122%
49%
79%
68%
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 the study (5.8 ug/m3).

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).
"Estimates 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 et al. (2009) - exposure period from 1999 - 2000.
6Calculated 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
September 2009
Draft - Do Not Quote or Cite

-------
Table F-4. Sensitivity Analysis: Impact of Limiting Estimated Annual Incidence of All-Cause Mortality
          Associated with Long-Term Exposure to PM2.S Concentrations that Just Meet the Current

          Standards to the Lowest Measured Level in the Study vs. to PRB, Based on Adjusting 2005
          PM2 5 Concentrations '
Risk Assessment Location
Atlanta, GA
Baltimore, MD
Birmingham, AL
Dallas, TX
Detroit, Ml
Fresno, 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 Measured
Down to:
Lowest Measured Level in
Study (5.8 ug/m3)
732
(468 - 992)
696
(444 - 942)
451
(288-611)
469
(299 - 637)
697
(445 - 946)
110
(70 - 1 50)
849
(542-1151)
1432
(911 -1948)
1600
(1019-2174)
547
(349 - 743)
611
(389 - 831 )
366
(233 - 498)
54
(34 - 74)
857
(547-1161)
101
(64 - 1 38)
Estimated PRB
1053
(676-1421)
1067
(685-1437)
660
(423 - 891 )
746
(477-1009)
1160
(744-1566)
258
(1 65 - 350)
1247
(800-1683)
2934
(1876-3972)
3012
(1928-4073)
935
(599-1262)
1247
(797-1687)
732
(468 - 991 )
177
(113-240)
1342
(861 -1810)
232
(148-314)
Percent Difference3
44%
53%
46%
59%
66%
135%
47%
105%
88%
71%
104%
100%
228%
57%
130%
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.
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 estimated down to PRB - mortality estimated down to LML)/(mortality estimated down to LML).
                                                      F-6
September 2009
Draft - Do Not Quote or Cite

-------
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.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
Incidence of Ischemic Heart Disease Mortality
Associated with Long-Term Exposure to PM25
Concentrations Measured Down to:
Lowest Measured Level in
Study (5.8 ug/m3)
269
(221 -315)
399
(327 - 468)
190
(156-223)
220
(179-260)
447
(365 - 528)
91
(74 - 1 08)
534
(438 - 626)
1140
(927 - 1 349)
1785
(1454-2110)
340
(279 - 401 )
479
(391 - 566)
224
(183-266)
14
(11 -17)
507
(415-597)
43
(35 - 51 )
Estimated PRB
383
(317-446)
622
(515-723)
274
(227-319)
382
(314-447)
827
(681 - 967)
198
(1 62 - 232)
754
(624 - 877)
2459
(2020 - 2882)
3546
(2918-4149)
564
(466 - 657)
915
(753-1069)
482
(396 - 565)
63
(51 - 74)
858
(708-1001)
140
(114- 165)
Percent Difference3
42%
56%
44%
74%
85%
118%
41%
116%
99%
66%
91%
115%
350%
69%
226%
 '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.
 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 (IHD mortality estimated down to PRB - IHD mortality estimated down to LML)/(IHD mortality estimated down to LML).
                                                      F-7
September 2009
Draft - Do Not Quote or Cite

-------
 Table F-6. Sensitivity Analysis: Impact of Limiting Estimated Annual Incidence of All-Cause Mortality
          Associated with Long-Term Exposure to PM2.s Concentrations that Just Meet the Current

          Standards to the Lowest Measured Level in the Study vs. to PRB, 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
Incidence of All Cause Mortality Associated with Long-
Term Exposure to PM2 5 Concentrations Measured
Down to:
Lowest Measured Level in
Study (5.8 ug/m3)
731
(467 - 990)
563
(359 - 764)
429
(274 - 581 )
399
(254 - 542)
469
(299 - 638)
144
(92- 196)
880
(562-1193)
1257
(799-1711)
1370
(871 - 1 863)
530
(338-719)
580
(369 - 789)
303
(193-413)
67
(43 - 91 )
698
(445 - 948)
69
(44 - 94)
Estimated PRB
1071
(687 - 1 446)
937
(601 - 1 265)
643
(41 2 - 869)
689
(440 - 932)
934
(598 - 1 265)
296
(189-401)
1300
(834-1755)
2783
(1 778 - 3770)
2813
(1 799 - 3807)
917
(588 - 1 239)
1262
(807 - 1 709)
667
(426 - 904)
197
(126-268)
1191
(763 - 1 609)
205
(131 -278)
Percent Difference3
47%
66%
50%
73%
99%
106%
48%
121%
105%
73%
118%
120%
194%
71%
197%
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.
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 estimated down to PRB - mortality estimated down to LML)/(mortality estimated down to LML).
                                                      F-8
September 2009
Draft - Do Not Quote or Cite

-------
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 s Concentrations1
Health End point
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
1432
(91 1 - 1 948)
1626
(1270-1977)
174
(76 - 269)
1072
(-1096-3160)
718
(-726 - 2089)
-162
(-604 - 227)
-25%
-56%
-193%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
547
(349 - 743)
512
(401 - 620)
83
(37 - 1 27)
410
(-423-1198)
228
(-233 - 655)
-79
(-306-107)
-25%
-55%
-195%
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).
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].
Calculated as (Krewski et al. (2000) estimate - Krewski et al. (2009) estimate)/(Krewski et al.  (2009) estimate).
                                                        F-9
September 2009
Draft - Do Not Quote or Cite

-------
Table F-8. 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 2006 PM2 s Concentrations1
Health End point
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
1221
(776-1662)
1390
(1084- 1692)
149
(65-231)
914
(-931 - 2700)
612
(-616-1787)
-137
(-507-194)
-25%
-56%
-192%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
536
(341 - 727)
501
(393 - 608)
81
(36-125)
402
(-414-1173)
223
(-228-641)
-77
(-299-105)
-25%
-55%
-195%
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).
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].
Calculated as (Krewski et al. (2000) estimate - Krewski et al. (2009) estimate)/(Krewski et al. (2009) estimate).
                                                       F-10
September 2009
Draft - Do Not Quote or Cite

-------
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 s Concentrations1
Health End point
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
1257
(799-1711)
1431
(1116-1741)
153
(67 - 238)
941
(-959 - 2780)
630
(-635-1840)
-141
(-523 - 200)
-25%
-56%
-192%
Philadelphia, PA
All Cause Mortality
Cardiopulmonary Mortality
Lung Cancer Mortality
530
(338-719)
496
(388-601)
80
(35-123)
397
(-410-1160)
221
(-226 - 635)
-76
(-295-104)
-25%
-55%
-195%
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).
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].
Calculated as (Krewski et al. (2000) estimate - Krewski et al. (2009) estimate)/(Krewski et al. (2009) estimate).
                                                       F-ll
September 2009
Draft - Do Not Quote or Cite

-------
              Table F-10.  Sensitivity Analysis:  Estimated Annual Incidence of All Cause Mortality Associated with Long-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
Pittsburgh, PA
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
Proportional
Hybrid
Percent Difference
Incidence of All Cause Mortality Associated with Long -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):
15/352
696
(444 - 942)
687
(439-931)
-1%
451
(288-611)
452
(289-613)
0%
697
(445 - 946)
738
(471 - 1001)
6%
1432
(911 -1948)
1775
(1131 -2412)
24%
1600
(1019-2174)
1650
(1051 -2242)
3%
366
(233 - 498)
518
(330 - 703)
42%
857
(547-1161)
930
(594-1260)
9%
13/35
561
(358-761)
586
(374 - 795)
4%
352
(224 - 478)
353
(225 - 480)
0%
601
(383-817)
716
(457 - 972)
19%
1432
(911 -1948)
1698
(1081 -2308)
19%
1521
(969 - 2068)
1518
(966 - 2063)
0%
366
(233 - 498)
487
(310-661)
33%
679
(433 - 922)
732
(467 - 993)
8%
12/35
485
(309 - 659)
508
(324 - 690)
5%
302
(193-411)
304
(193-412)
1%
514
(327 - 698)
620
(395-841)
21%
1265
(804-1722)
1435
(913-1952)
13%
1282
(815-1744)
1278
(813-1739)
0%
366
(233 - 498)
417
(266 - 567)
14%
583
(371 - 792)
631
(402 - 858)
8%
13/30
541
(345 - 734)
534
(340 - 724)
-1%
352
(224 - 478)
353
(225 - 480)
0%
528
(336-717)
563
(358 - 765)
7%
1001
(636-1363)
1296
(824-1764)
29%
1159
(737-1578)
1202
(765-1637)
4%
259
(164-352)
389
(248 - 530)
50%
661
(421 - 897)
732
(467 - 993)
11%
12/25
384
(245 - 522)
378
(241 -514)
-2%
256
(163-348)
268
(170-364)
5%
356
(226 - 485)
386
(245 - 525)
8%
566
(359 - 772)
812
(516-1107)
43%
714
(454 - 974)
751
(477-1023)
5%
150
(95 - 205)
259
(165-353)
73%
463
(295 - 630)
529
(337 - 720)
14%
              'bstimates based on Krewski et al. (^uua), exposure period trom laaa - ^UUU, using models witn 44 individual and I ecological covanates (see lable 33 in Krewski et al.,
              2009).
              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-12
September 2009
Draft - Do Not Quote or Cite

-------
              Table F-ll.  Sensitivity Analysis:  Estimated Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2 5
                        Concentrations that Just Meet the Current and Alternative Standards, Based on Adjusting 2006 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
Pittsburgh, PA
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
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
603
(385-817)
594
(379 - 806)
-1%
434
(277 - 589)
434
(277 - 589)
0%
508
(323 - 690)
541
(344 - 735)
6%
1221
(776-1662)
1540
(980 - 2094)
26%
1345
(855-1829)
1389
(884-1889)
3%
297
(189-404)
430
(274 - 584)
45%
652
(415-885)
714
(455 - 969)
10%
13/35
479
(305-651)
501
(320-681)
5%
337
(215-458)
337
(215-458)
0%
427
(271 -581)
523
(333-710)
22%
1221
(776-1662)
1468
(934-1996)
20%
1272
(809-1731)
1267
(806-1724)
0%
297
(189-404)
402
(256 - 546)
35%
500
(318-680)
543
(346 - 739)
9%
12/35
409
(261 - 557)
430
(274 - 584)
5%
289
(184-392)
289
(184-392)
0%
353
(224 - 480)
441
(280 - 600)
25%
1064
(676-1450)
1222
(777-1664)
15%
1050
(668-1430)
1045
(664-1424)
0%
297
(189-404)
339
(215-461)
14%
417
(265 - 568)
458
(291 - 623)
10%
13/30
460
(293 - 626)
453
(289-616)
-2%
337
(215-458)
337
(215-458)
0%
364
(232 - 496)
393
(250 - 535)
8%
817
(519-1114)
1092
(694-1487)
34%
937
(595-1277)
975
(620-1329)
4%
199
(126-272)
314
(200 - 427)
58%
484
(308 - 659)
543
(346 - 739)
12%
12/25
317
(201 -431)
311
(198-423)
-2%
243
(155-331)
253
(161 -345)
4%
220
(140-300)
244
(155-332)
11%
411
(261 -561)
640
(406 - 873)
56%
526
(334-718)
558
(354 - 762)
6%
101
(64-138)
197
(125-268)
95%
315
(200 - 429)
370
(235 - 504)
17%
              'bstimates based on Krewski et al. (^uua), exposure period trom laaa - ^UUU, using models witn 44 individual and I ecological covanates (see lable 33 in Krewski et al.,
              2009).
              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-13
September 2009
Draft - Do Not Quote or Cite

-------
              Table F-12.  Sensitivity Analysis:  Estimated Annual Incidence of All Cause Mortality Associated with Long-Term Exposure to PM2 5
                         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
Pittsburgh, PA
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
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
563
(359 - 764)
555
(354 - 555)
-1%
429
(274-581)
429
(274 - 429)
0%
469
(299 - 638)
502
(319-502)
7%
1257
(799-1711)
1582
(1007-1582)
26%
1370
(871 - 1863)
1415
(900-1415)
3%
303
(193-413)
437
(278 - 437)
44%
698
(445 - 948)
766
(489 - 766)
10%
13/35
444
(283 - 604)
465
(296 - 465)
5%
332
(212-452)
333
(212-333)
0%
392
(249 - 533)
484
(308 - 484)
23%
1257
(799-1711)
1508
(960-1508)
20%
1296
(824 - 1 764)
1291
(821 -1291)
0%
303
(193-413)
408
(260 - 408)
35%
540
(344 - 735)
589
(375 - 589)
9%
12/35
377
(240-513)
397
(252 - 397)
5%
284
(181 -386)
284
(181 -284)
0%
321
(204 - 437)
406
(258 - 406)
26%
1098
(698-1496)
1259
(800-1259)
15%
1072
(681 -1459)
1067
(678-1067)
0%
303
(193-413)
345
(219-345)
14%
455
(289-619)
500
(318-500)
10%
13/30
426
(271 - 579)
419
(267-419)
-2%
332
(212-452)
333
(212-333)
0%
332
(211 -452)
360
(229 - 360)
8%
847
(538-1155)
1126
(716-1126)
33%
957
(608-1304)
996
(633 - 996)
4%
205
(130-280)
320
(203 - 320)
56%
524
(334-713)
589
(375 - 589)
12%
12/25
288
(183-393)
282
(179-282)
-2%
239
(152-325)
249
(159-249)
4%
193
(123-264)
217
(138-217)
12%
434
(275 - 593)
668
(424 - 668)
54%
541
(343 - 739)
574
(364 - 574)
6%
106
(67-145)
202
(128-202)
91%
348
(221 - 474)
408
(260 - 408)
17%
              'bstimates based on Krewski et al. (^uua), exposure period trom laaa -^UUU, using models witn 44 individual and I ecological covanates (see  I able 33 in Krewski et al.,
              2009).
              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-14
September 2009
Draft - Do Not Quote or Cite

-------
  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 PM25 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, 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
53
(8 - 97)
67
(9-124)
21
(-5 - 47)
30
(-5 - 64)
-6
(-81 - 67)
0
(-34 - 34)
49
(-5-102)
17
(-86-120)
264
(97 - 428)
90
(19-159)
—
39
(-4 - 80)
17
(-2 - 35)
36
(-36-106)
1
(-53 - 53)
Spring
52
(3 - 99)
47
(1 - 93)
28
(-3 - 59)
28
(-8 - 64)
76
(19-132)
15
(-1 - 32)
64
(5-123)
66
(-35-167)
142
(1 -281)
29
(-33 - 89)
—
56
(10-101)
7
(-2-16)
70
(13-126)
9
(-7 - 25)
Summer
43
(-15-101)
59
(-7-123)
18
(-23 - 59)
41
(-3 - 85)
53
(-32-136)
3
(-14-21)
56
(-14-125)
-109
(-271 - 50)
122
(-49 - 289)
33
(-47-110)
—
36
(-19-91)
7
(-6-19)
67
(-6-138)
4
(-9-17)
Fall
33
(-17-83)
49
(7-91)
11
(-24 - 45)
44
(6 - 82)
30
(-28 - 86)
11
(-12-34)
59
(-10-127)
-2
(-96 - 90)
189
(82 - 295)
63
(16-110)
—
20
(-23 - 62)
9
(-4-21)
70
(12-126)
14
(-10-37)
Sum of Four
Seasons
181
4
222
78
143
153
29
228
-28
717
215
—
151
40
243
28
All Year
175
(33-316)
256
(104-406)
38
(-60-135)
151
(36 - 264)
141
(-25 - 305)
44
(6 - 82)
232
(47-414)
85
(-121 -289)
714
(419-1007)
211
(78 - 342)
240
(40 - 438)
135
(40 - 230)
33
(7 - 59)
252
(73 - 429)
48
(8 - 87)
Percent
Difference 3
3%
-13%
105%
-5%
9%
-34%
-2%
-133%
0%
2%
—
12%
21%
-4%
-42%
 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.

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
September 2009
Draft - Do Not Quote or Cite

-------
  Table F-14. Sensitivity Analysis: Impact of Using Season-Specific Concentration-Response Functions vs. an Annual Concentration-Response
            Function to Estimate the Incidence of Non-Accidental Mortality Associated with Short-Term Exposure to PM2 5 Concentrations
            that Just Meet the Current Standards, Based on Adjusting 2006 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 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
47
(7 - 86)
55
(7-102)
22
(-5 - 49)
23
(-4 - 50)
-4
(-59 - 50)
1
(-37 - 37)
49
(-5-103)
18
(-88-123)
230
(84 - 374)
78
(16-139)
—
34
(-3-71)
13
(-2 - 28)
27
(-27 - 79)
1
(-36 - 36)
Spring
54
(3-105)
42
(1 - 82)
31
(-3 - 64)
29
(-9 - 66)
75
(19- 130)
15
(-1 - 30)
82
(6-156)
57
(-30-144)
126
(1 - 250)
27
(-31 - 84)
—
52
(9-94)
7
(-2-16)
65
(12-118)
10
(-8 - 26)
Summer
49
(-17-113)
56
(-6-118)
19
(-24-61)
36
(-3 - 75)
42
(-25-107)
4
(-15-23)
61
(-15-135)
-102
(-252 - 47)
113
(-45 - 268)
33
(-47-111)
—
35
(-18-88)
7
(-7-21)
57
(-5-118)
4
(-10-19)
Fall
28
(-14-69)
49
(7-91)
9
(-19-36)
33
(4-61)
26
(-24 - 76)
11
(-12-34)
51
(-8-109)
-2
(-83 - 79)
187
(81 - 292)
72
(18-125)
—
16
(-19-50)
8
(-3-19)
60
(10- 110)
12
(-8-31)
Sum of Four
Seasons
178
4
202
81
121
139
31
243
-29
656
210
—
137
35
209
27
All Year
172
(33-310)
234
(95 - 372)
38
(-59-132)
129
(31 - 226)
117
(-21 - 254)
46
(7-86)
243
(50 - 434)
78
(-112-267)
654
(384 - 922)
208
(77 - 337)
255
(42 - 466)
122
(36 - 208)
30
(6 - 54)
214
(62 - 364)
39
(7-71)
Percent
Difference 3
3%
-14%
113%
-6%
19%
-33%
0%
-137%
0%
1%
—
12%
17%
-2%
-31%
 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.
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.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
                                                                                 F-16
September 2009
Draft - Do Not Quote or Cite

-------
  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 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 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
47
(7 - 86)
55
(7-102)
23
(-5-50)
28
(-5-60)
-5
(-76 - 63)
1
(-84 - 80)
58
(-6-121)
23
(-115-160)
269
(98 - 436)
87
(18-155)
:::
54
(-5-112)
29
(-3 - 60)
31
(-32 - 93)
1
(-47 - 46)
Spring
70
(4-134)
44
(1-87)
50
(-6- 104)
25
(-8 - 57)
71
(18-124)
26
(-2 - 53)
73
(6-140)
117
(-62 - 294)
161
(1 -319)
31
(-35 - 95)
:::
89
(16-160)
13
(-4-29)
80
(15-144)
12
(-9 - 32)
Summer
53
(-18-122)
57
(-6-118)
22
(-28 - 70)
38
(-3 - 79)
43
(-26-109)
6
(-24 - 34)
65
(-16-146)
-148
(-369 - 68)
133
(-54-316)
34
(-49-115)
:::
59
(-31 -146)
12
(-10-33)
64
(-6-132)
4
(-9-16)
Fall
31
(-16-77)
50
(7-92)
11
(-23 - 43)
39
(5 - 72)
39
(-37- 113)
23
(-24 - 68)
55
(-9-118)
-3
(-145-136)
234
(102-365)
76
(19-132)
:::
32
(-37 - 99)
14
(-6 - 34)
69
(12-124)
20
(-14-52)
Sum of Four
Seasons
201
4
206
106
130
148
56
251
-11
797
228
:::
234
68
244
37
All Year
177
(34-319)
225
(91 - 357)
37
(-58-131)
137
(33 - 240)
112
(-20 - 242)
51
(7-94)
240
(49 - 429)
79
(-113-270)
659
(387 - 930)
206
(76 - 334)
242
(40-441)
123
(36 - 209)
36
(7 - 65)
222
(64 - 378)
42
(7 - 76)
Percent
Difference 3
14%
-8%
186%
-5%
32%
10%
5%
-114%
21%
11%
—
90%
89%
10%
-12%
 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.
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.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
                                                                                  F-17
September 2009
Draft - Do Not Quote or Cite

-------
  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 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 5
Pittsburgh, PA
Salt Lake City, UT 5
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
14
(-20 - 46)
16
(-31 - 60)
5
(-18-27)
10
(-18-36)
-1
(-47 - 42)
-2
(-13-8)
9
(-37 - 53)
-7
(-55 - 40)
141
(33 - 246)
27
(-5 - 58)
—
13
(-9 - 34)
—
-3
(-66 - 57)
0
(-12-13)
Spring
9
(-32 - 47)
10
(-32 - 49)
1
(-27 - 28)
11
(-19-40)
25
(-7 - 56)
1
(-3 - 5)
2
(-49 - 50)
3
(-45 - 50)
116
(26 - 204)
16
(-15-46)
—
26
(2-48)
—
44
(-2 - 88)
0
(-3 - 4)
Summer
9
(-32 - 47)
11
(-44 - 63)
0
(-32 - 29)
12
(-24 - 47)
28
(-21 - 74)
0
(-4 - 4)
30
(-24-81)
-46
(-111 - 18)
143
(27 - 256)
26
(-12-63)
—
11
(-19-39)
—
38
(-18-91)
0
(-2 - 3)
Fall
-2
(-37-31)
32
(-2 - 64)
-18
(-46-10)
-2
(-32 - 27)
32
(0 - 63)
3
(-4 - 9)
8
(-44 - 58)
0
(-46 - 46)
92
(21 -160)
27
(4 - 49)
—
4
(-17-24)
—
41
(-4 - 84)
2
(-4 - 7)
Sum of Four
Seasons
30
4
69
-12
31
84
2
49
-50
492
96
—
54
—
120
2
All Year
32
(-32 - 94)
70
(-5-142)
-1
(-48 - 45)
31
(-21 - 82)
70
(-9-147)
11
(-8 - 30)
51
(-34-134)
-33
(-146-79)
460
(268 - 649)
85
(22-147)
83
(-4-169)
40
(-8 - 88)
9
(-2-19)
118
(26 - 208)
12
(-7-31)
Percent
Difference 3
-6%
-1%
1100%
0%
20%
-82%
-4%
52%
7%
13%
—
35%
—
2%
-83%
 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.
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.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
                                                                                 F-18
September 2009
Draft - Do Not Quote or Cite

-------
    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 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
12
(-18-41)
13
(-25 - 50)
5
(-19-28)
8
(-14-28)
-1
(-34-31)
-2
(-14-9)
9
(-37 - 53)
-8
(-57-41)
123
(29-215)
23
(-5-51)
—
11
(-8 - 30)
—
-2
(-49 - 43)
0
(-9 - 9)
Spring
9
(-33 - 50)
9
(-28 - 43)
1
(-29-31)
11
(-20-41)
25
(-7 - 55)
1
(-3 - 5)
2
(-63 - 64)
2
(-39 - 43)
103
(23-182)
15
(-14-43)
—
24
(2 - 45)
—
41
(-2 - 83)
0
(-3 - 4)
Summer
10
(-36 - 53)
11
(-42 - 60)
0
(-33 - 30)
11
(-21-41)
22
(-16-58)
0
(-4 - 5)
32
(-26 - 87)
-43
(-103-17)
133
(25 - 237)
27
(-13-63)
—
10
(-18-38)
—
33
(-15-79)
0
(-2 - 3)
Fall
-2
(-31 - 26)
32
(-2 - 64)
-14
(-36 - 8)
-1
(-23 - 20)
28
(0 - 56)
3
(-4 - 9)
7
(-38 - 50)
0
(-40 - 40)
90
(21 -158)
31
(5 - 56)
—
3
(-14-20)
—
36
(-3 - 73)
1
(-3 - 6)
Sum of Four
Seasons
29
4
65
-8
29
74
2
50
-49
449
96
—
48
—
108
1
All Year
31
(-32 - 93)
64
(-4-130)
-1
(-47 - 44)
27
(-18-71)
58
(-7-122)
12
(-8 - 32)
53
(-36- 140)
-30
(-134-73)
421
(246 - 595)
84
(22-144)
88
(-4-179)
36
(-7-80)
8
(-2-18)
101
(22-177)
10
(-5 - 26)
Percent
Difference 3
-6%
2%
700%
7%
28%
-83%
-6%
63%
7%
14%
—
33%
—
7%
-90%
   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.
   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
September 2009
Draft - Do Not Quote or Cite

-------
  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 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 5
Pittsburgh, PA
Salt Lake City, UT 5
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
(-16-38)
12
(-24 - 47)
4
(-15-23)
9
(-16-34)
-1
(-36 - 33)
-3
(-17-11)
10
(-42 - 59)
-6
(-47 - 34)
120
(28 - 209)
24
(-5 - 52)
:::
10
(-7 - 28)
—
-2
(-52 - 45)
0
(-9 - 9)
Spring
11
(-40 - 58)
9
(-28 - 44)
2
(-38 - 39)
10
(-17-36)
19
(-5 - 43)
1
(-3 - 5)
2
(-53 - 54)
3
(-50 - 55)
109
(24-193)
16
(-14-45)
:::
24
(2 - 45)
—
45
(-2-91)
0
(-3 - 4)
Summer
10
(-35 - 52)
10
(-40 - 57)
0
(-30 - 27)
11
(-22 - 44)
18
(-13-48)
0
(-3 - 4)
33
(-26 - 89)
-39
(-94-15)
130
(25 - 233)
25
(-12-60)
:::
10
(-18-37)
—
33
(-15-79)
0
(-2 - 2)
Fall
-2
(-31 - 27)
31
(-2 - 62)
-13
(-34 - 7)
-2
(-28 - 24)
34
(0 - 67)
3
(-4-10)
7
(-39-51)
0
(-44 - 43)
94
(22-165)
30
(4 - 54)
:::
4
(-16-23)
—
36
(-3 - 74)
2
(-5 - 8)
Sum of Four
Seasons
30
4
62
-7
28
70
1
52
-42
453
95
:::
48
—
112
2
All Year
32
(-33 - 95)
61
(-4-125)
-1
(-46 - 44)
29
(-19-75)
55
(-7-117)
13
(-9-35)
52
(-36-139)
-30
(-136-74)
425
(248 - 600)
83
(22-143)
84
(-4-170)
37
(-7 - 80)
10
(-2-21)
104
(23-184)
11
(-6 - 27)
Percent
Difference 3
-6%
2%
600%
-3%
27%
-92%
0%
40%
7%
14%
—
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.
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.
5 Season-specific coefficient estimates were not available from Zanobetti and Schwartz (2009) for this location.
                                                                                 F-20
September 2009
Draft - Do Not Quote or Cite

-------
  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
(-6-13)
5
(-6-16)
1
(-4 - 6)
1
(-6 - 9)
5
(-6-16)
-1
(-1 1 - 9)
6
(-4-16)
27
(-3 - 57)
48
(18-78)
10
(-1-21)
27
(-28 - 77)
4
(-3-10)
4
(-1-10)
1
(-14-16)
0
(-15-13)
Spring
1
(-8-11)
6
(-4-15)
2
(-4-8)
3
(-4-9)
9
(-1-18)
4
(-1 - 9)
5
(-4-14)
28
(-2 - 56)
16
(-5 - 37)
7
(-1-15)
30
(-8 - 66)
6
(-1-13)
2
(-2 - 6)
7
(-6-19)
2
(-2 - 6)
Summer
3
(-7-13)
6
(-6-17)
-1
(-9 - 7)
2
(-5 - 9)
10
(0-19)
1
(-2 - 4)
4
(-6-14)
-16
(-61 - 27)
20
(-9 - 47)
7
(-3-16)
21
(-3 - 45)
6
(-2-14)
-2
(-7 - 3)
4
(-10-17)
1
(-2 - 3)
Fall
3
(-5-11)
3
(-4-10)
4
(-2-10)
1
(-5 - 7)
8
(0-16)
1
(-6 - 8)
5
(-7-16)
0
(-24 - 23)
20
(1 - 39)
5
(-2-11)
41
(14-67)
6
(0-12)
-1
(-6 - 4)
6
(-5-18)
1
(-4 - 6)
Sum of Four
Seasons
11
4
20
6
7
32
5
20
39
104
29
119
22
3
18
4
All Year
19
(-8-46)
35
(7 - 63)
10
(-8 - 29)
11
(-10-31)
27
(1 - 52)
9
(0-17)
36
(6 - 66)
59
(6-112)
97
(34-159)
23
(-2 - 46)
47
(4-89)
17
(-2-35)
6
(1-11)
30
(-8-67)
7
(0-15)
Percent
Difference 3
-42%
-43%
-40%
-36%
19%
-44%
-44%
-34%
7%
26%
153%
29%
-50%
-40%
-43%
 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.

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.
                                                                               F-21
September 2009
Draft - Do Not Quote or Cite

-------
  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 PM25 Concentrations
that Just Meet the Current Standards
Winter
3
(-5-11)
4
(-5-13)
1
(-4 - 6)
1
(-5 - 7)
4
(-5-12)
-1
(-12-10)
6
(-4-16)
28
(-3 - 59)
42
(16-68)
9
(-1-18)
31
(-33 - 89)
3
(-3 - 9)
4
(-1 - 8)
1
(-11-12)
0
(-10-9)
Spring
1
(-9-11)
5
(-4-13)
2
(-5 - 9)
3
(-4-10)
8
(-1-17)
4
(-1 - 9)
6
(-6-18)
24
(-2 - 49)
14
(-5 - 33)
7
(-1 -14)
30
(-8 - 66)
6
(-1-12)
2
(-2 - 6)
6
(-6-18)
2
(-2 - 6)
Summer
4
(-8-15)
5
(-6-16)
-1
(-9 - 7)
2
(-5 - 8)
8
(0-15)
1
(-2 - 5)
5
(-7-15)
-15
(-57 - 26)
18
(-8 - 44)
7
(-3-16)
22
(-3 - 46)
6
(-2-14)
-2
(-7 - 3)
3
(-9-14)
1
(-2 - 4)
Fall
3
(-4-9)
3
(-4-10)
3
(-2 - 8)
1
(-4-5)
7
(0-14)
1
(-6 - 8)
4
(-6-14)
0
(-21 - 20)
20
(1 - 38)
5
(-2-12)
41
(14-67)
5
(0-10)
-1
(-5 - 3)
5
(-5-15)
1
(-3 - 5)
Sum of Four
Seasons
11
4
17
5
7
27
5
21
37
94
28
124
20
3
15
4
All Year
19
(-8-45)
32
(6 - 58)
10
(-8 - 28)
9
(-9-27)
23
(1 - 44)
9
(0-18)
38
(6 - 69)
55
(5-104)
89
(31 -145)
22
(-2 - 46)
50
(4-95)
15
(-2 - 32)
5
(1-10)
26
(-7-57)
6
(0-12)
Percent
Difference 3
-42%
-47%
-50%
-22%
17%
-44%
-45%
-33%
6%
27%
148%
33%
-40%
-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.

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.
                                                                               F-22
September 2009
Draft - Do Not Quote or Cite

-------
  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 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
3
(-5-10)
4
(-5-12)
1
(-4 - 5)
1
(-6 - 8)
4
(-5-12)
-1
(-14-12)
6
(-5-18)
23
(-2 - 48)
41
(15-66)
9
(-1-18)
24
(-25 - 69)
3
(-2 - 8)
5
(-2-11)
1
(-11 -13)
0
(-11 -10)
Spring
2
(-10-13)
5
(-4-14)
3
(-6-11)
3
(-3 - 9)
7
(0-13)
4
(-1 - 8)
5
(-5-15)
31
(-2 - 63)
15
(-5 - 35)
7
(-1-15)
29
(-8 - 63)
6
(-1-12)
2
(-3 - 7)
7
(-6-19)
2
(-2 - 6)
Summer
4
(-8-15)
5
(-5-15)
-1
(-8 - 7)
2
(-5 - 8)
7
(0-13)
1
(-2 - 4)
5
(-7-15)
-14
(-52 - 23)
18
(-8 - 43)
7
(-3-15)
25
(-4-51)
6
(-2-13)
-2
(-7 - 3)
3
(-9-14)
1
(-1 - 3)
Fall
3
(-4-9)
3
(-4-10)
3
(-2 - 7)
1
(-5 - 6)
9
(0-17)
2
(-6 - 9)
4
(-6-14)
0
(-23 - 21 )
21
(1 - 40)
5
(-2-12)
45
(15-73)
5
(0-11)
-1
(-6 - 4)
6
(-5-15)
1
(-5 - 7)
Sum of Four
Seasons
12
4
17
6
7
27
6
20
40
95
28
123
20
4
17
4
All Year
20
(-8 - 47)
31
(6 - 56)
10
(-8 - 28)
10
(-9 - 29)
22
(1 - 42)
10
(0-19)
38
(6 - 68)
56
(5-105)
90
(32-147)
22
(-2 - 45)
47
(4 - 90)
16
(-2 - 32)
6
(1 -12)
27
(-7 - 59)
6
(0-13)
Percent
Difference 3
-40%
-45%
-40%
-30%
23%
-40%
-47%
-29%
6%
27%
162%
25%
-33%
-37%
-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.

 2The current primary PM2 5 standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
 3 Calculated as (sum-of-4-seasons mortality - all-year mortality)/(all-year mortality).
 4 It was not possible to calculate the 2.5th and 97.5th percentile estimates of the sum of the season-specific incidences because the variance-covariance matrix for the
 season-specific coefficient estimators was not available.
                                                                               F-23
September 2009
Draft - Do Not Quote or Cite

-------
  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 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
29
(-2 - 59)
131
(92-171)
12
(-1 - 24)
24
(-2 - 49)
149
(104-193)
14
(-5 - 32)
48
(-3 - 98)
106
(-36 - 247)
370
(258-481)
114
(79-148)
57
(-19-132)
52
(36 - 68)
6
(-2-14)
100
(70-130)
12
(-65 - 82)
Spring
24
(-8 - 55)
46
(16-76)
11
(-4 - 25)
18
(-6 - 42)
54
(18-88)
11
(-5 - 27)
36
(-13-85)
196
(-99-481)
144
(49 - 238)
39
(13-64)
82
(-41 - 202)
27
(9 - 44)
4
(-2-11)
41
(14-67)
0
(-61 - 55)
Summer
-29
(-70-11)
39
(6 - 72)
-14
(-34-5)
-20
(-49 - 8)
40
(6 - 73)
-7
(-29-15)
-39
(-94-15)
-152
(-647 - 322)
117
(17-216)
34
(5 - 63)
-47
(-1 98 - 99)
23
(3 - 43)
-4
(-17-8)
30
(4-55)
-5
(-56 - 42)
Fall
6
(-27 - 39)
47
(22 - 72)
3
(-13-19)
4
(-19-27)
50
(24-77)
3
(-11-17)
10
(-44 - 64)
44
(-147-232)
133
(63 - 203)
38
(18-58)
14
(-47 - 75)
31
(14-47)
1
(-4-6)
41
(19-63)
-5
(-53 - 40)
Sum of Four
Seasons
30
4
263
12
26
293
21
55
194
764
225
106
133
7
212
2
All Year
40
(-26-105)
247
(182-313)
18
(-12-48)
30
(-19-78)
269
(198-340)
21
(0-42)
61
(-40-161)
274
(3 - 543)
724
(532-915)
210
(154-265)
107
(1 -212)
134
(98 - 1 69)
9
(0-17)
200
(147-253)
21
(-52 - 92)
Percent
Difference 3
-25%
6%
-33%
-13%
9%
0%
-10%
-29%
6%
7%
-1%
-1%
-22%
6%
-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.

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
September 2009
Draft - Do Not Quote or Cite

-------
  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
26
(-2 - 53)
107
(75 - 1 39)
12
(-1 - 24)
19
(-1 - 39)
110
(77 - 1 44)
15
(-5 - 35)
48
(-3-100)
111
(-38 - 258)
325
(227 - 422)
98
(69 - 1 28)
66
(-22-153)
46
(32 - 60)
5
(-2-11)
76
(53 - 98)
9
(-45 - 57)
Spring
25
(-9 - 58)
41
(14-67)
12
(-4 - 27)
19
(-7 - 44)
53
(18-87)
10
(-5 - 26)
46
(-17-108)
172
(-86 - 424)
129
(44-213)
36
(12-60)
81
(-41 -199)
25
(8-41)
5
(-2-12)
38
(13-63)
0
(-64 - 59)
Summer
-32
(-78-13)
37
(6 - 69)
-14
(-34-6)
-18
(-43 - 7)
31
(5 - 57)
-7
(-32-16)
-42
(-102-16)
-144
(-612-306)
109
(16-201)
34
(5 - 63)
-48
(-205-103)
22
(3-41)
-4
(-19-9)
26
(4-48)
-6
(-61 - 46)
Fall
5
(-23 - 33)
47
(22 - 72)
2
(-10- 15)
3
(-14-21)
45
(21 - 68)
3
(-11-17)
9
(-38 - 55)
39
(-129-205)
132
(62 - 202)
43
(20 - 65)
14
(-47 - 75)
25
(12-38)
1
(-3 - 5)
36
(17-55)
-4
(-43 - 33)
Sum of Four
Seasons
24
4
232
12
23
239
21
61
178
695
211
113
118
7
176
-1
All Year
39
(-26-103)
224
(165-284)
18
(-12-47)
25
(-17-67)
224
(165-284)
22
(0 - 44)
64
(-42-170)
258
(3-511)
666
(489 - 843)
204
(150-258)
113
(1 - 225)
120
(88-152)
8
(0-16)
171
(125-216)
18
(-43 - 76)
Percent
Difference 3
-38%
4%
-33%
-8%
7%
-5%
-5%
-31%
4%
3%
0%
-2%
-13%
3%
-1 06%
 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.

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 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
September 2009
Draft - Do Not Quote or Cite

-------
  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 5 Concentrations that Just Meet the Current Standards
Winter
24
(-2 - 49)
101
(70-131)
10
(-1 - 20)
23
(-2 - 47)
115
(81 - 1 50)
18
(-6 - 42)
55
(-4-112)
93
(-32-217)
318
(222-413)
100
(70 - 1 30)
51
(-17-118)
42
(29 - 54)
7
(-2-15)
79
(55 - 1 03)
9
(-48 - 59)
Spring
30
(-1 1 - 69)
41
(14-67)
15
(-5 - 35)
16
(-6 - 38)
41
(14-68)
10
(-5 - 25)
40
(-14-93)
226
(-114-556)
137
(47 - 227)
37
(13-62)
78
(-39-192)
24
(8 - 40)
5
(-3-13)
41
(14-69)
0
(-64 - 59)
Summer
-33
(-79-13)
35
(5 - 65)
-13
(-31 - 5)
-19
(-46 - 7)
26
(4 - 48)
-6
(-27 - 1 3)
-43
(-105-17)
-135
(-571 - 286)
108
(16-199)
32
(5 - 59)
-54
(-228-114)
21
(3 - 39)
-4
(-19-10)
26
(4 - 47)
-4
(-43 - 33)
Fall
5
(-23 - 34)
45
(21 - 69)
2
(-10- 14)
4
(-17-25)
54
(26 - 83)
4
(-12-19)
9
(-39 - 57)
44
(-144-228)
139
(65-212)
41
(1 9 - 63)
16
(-52 - 82)
28
(1 3 - 43)
1
(-4-6)
36
(17-55)
-6
(-63 - 47)
Sum of Four
Seasons
26
4
222
14
24
236
26
61
228
702
210
91
115
9
182
-1
All Year
41
(-27-108)
215
(158-271)
17
(-1 1 - 46)
27
(-18-72)
215
(1 58 - 272)
24
(0 - 48)
64
(-42-169)
265
(3 - 526)
676
(496 - 855)
201
(1 48 - 254)
108
(1-215)
120
(88-152)
10
(0 - 20)
176
(1 29 - 222)
19
(-46 - 82)
Percent
Difference 3
-37%
3%
-18%
-11%
10%
8%
-5%
-14%
4%
4%
-16%
-4%
-10%
3%
-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.

 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-26
September 2009
Draft - Do Not Quote or Cite

-------
  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 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 Respiratory Illness Associated with Short-Term Exposure to
PM2.5 Concentrations that Just Meet the Current Standards
Winter
6
(-20-31)
18
(-5-41)
2
(-8-12)
4
(-17-25)
23
(-6-51)
10
(-1-21)
8
(-30 - 45)
73
(-6-151)
54
(-15-122)
16
(-4 - 36)
34
(-3-70)
7
(-2-16)
4
(0-9)
23
(-6-51)
0
(-50 - 43)
Spring
9
(-10-27)
1
(-14- 15)
4
(-5-13)
7
(-8-21)
1
(-18-20)
3
(-6-11)
13
(-14-40)
46
(-98-184)
2
(-46 - 50)
1
(-12-13)
17
(-36 - 68)
0
(-9-9)
1
(-3-5)
1
(-18-19)
4
(-27 - 32)
Summer
-6
(-25-12)
14
(0 - 28)
-3
(-12-6)
-5
(-20-10)
17
(0 - 35)
4
(-4-12)
-9
(-37-18)
90
(-103-273)
44
(-1 - 88)
12
(0 - 24)
23
(-26 - 70)
8
(0-16)
3
(-3-8)
15
(0-29)
1
(-19-19)
Fall
1
(-13-15)
2
(-12-15)
1
(-6 - 7)
1
(-13-15)
2
(-12-16)
3
(-5-11)
3
(-29 - 35)
44
(-64-149)
4
(-31 - 39)
1
(-9-12)
13
(-18-43)
1
(-8-10)
1
(-2 - 4)
2
(-15- 19)
-2
(-25-18)
Sum of Four
Seasons
10
4
35
4
7
43
20
15
253
104
30
87
16
9
41
3
All Year
17
(-21 - 55)
20
(-12-51)
8
(-10-25)
14
(-18-46)
24
(-14-62)
14
(3-25)
27
(-34 - 88)
177
(41 -311)
60
(-35-154)
17
(-10-43)
61
(14-106)
11
(-6 - 28)
6
(1-11)
24
(-14-62)
2
(-27 - 30)
Percent
Difference 3
-41%
75%
-50%
-50%
79%
43%
-44%
43%
73%
76%
43%
45%
50%
71%
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.
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-27
September 2009
Draft - Do Not Quote or Cite

-------
  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 PM2.5
            Concentrations that Just Meet the Current Standards, Based on Adjusting 2006 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 Hospital Admissions for Respiratory Illness Associated with Short-Term Exposure to
PM2.s Concentrations that Just Meet the Current Standards
Winter
5
(-18-27)
15
(-4 - 34)
2
(-8-13)
4
(-13-20)
17
(-5-38)
11
(-1 - 23)
8
(-30 - 45)
77
(-7-158)
48
(-13- 107)
14
(-4-31)
40
(-3 - 82)
6
(-2-14)
3
(0-7)
17
(-5-39)
0
(-35 - 29)
Spring
9
(-1 1 - 29)
1
(-12-13)
4
(-5-14)
7
(-8 - 22)
1
(-18-19)
3
(-5-10)
17
(-19-51)
40
(-86-162)
2
(-41 - 45)
0
(-11-12)
17
(-36 - 67)
0
(-8 - 9)
1
(-3-5)
1
(-17-18)
4
(-29 - 34)
Summer
-7
(-29-14)
13
(0 - 26)
-3
(-13-6)
-4
(-18-9)
14
(0 - 27)
4
(-5-13)
-10
(-40 - 20)
86
(-98-261)
41
(-1 - 82)
12
(0 - 24)
24
(-27 - 73)
7
(0-15)
3
(-3-9)
13
(0 - 25)
1
(-21 - 20)
Fall
1
(-11-13)
2
(-12-15)
0
(-5 - 6)
1
(-10-11)
2
(-11-14)
3
(-5-11)
3
(-25 - 30)
39
(-56-132)
4
(-30 - 39)
2
(-10-13)
13
(-18-43)
1
(-6 - 8)
1
(-2 - 4)
2
(-13-16)
-1
(-20-15)
Sum of Four
Seasons
8
4
31
3
8
34
21
18
242
95
28
94
14
8
33
4
All Year
17
(-21 - 54)
18
(-1 1 - 46)
8
(-10-24)
12
(-15-40)
20
(-12-52)
15
(3 - 26)
29
(-36 - 93)
166
(39 - 293)
55
(-32-142)
16
(-10-42)
64
(15-113)
10
(-6 - 25)
6
(1-10)
21
(-12-53)
2
(-22 - 25)
Percent
Difference 3
-53%
72%
-63%
-33%
70%
40%
-38%
46%
73%
75%
47%
40%
33%
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.

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-28
September 2009
Draft - Do Not Quote or Cite

-------
  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 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
5
(-17-25)
14
(-4-32)
2
(-7-10)
4
(-16-24)
18
(-5 - 40)
13
(-1 - 28)
9
(-34-51)
64
(-6-133)
46
(-12-105)
14
(-4-31)
31
(-3 - 63)
6
(-2-13)
5
(0-10)
18
(-5 - 40)
0
(-37-31)
Spring
11
(-13-34)
1
(-12- 13)
6
(-6-17)
6
(-7-20)
1
(-14-15)
2
(-5-10)
14
(-16-44)
53
(-113-213)
2
(-44 - 48)
1
(-11 - 12)
16
(-34 - 65)
0
(-8-9)
1
(-3 - 6)
1
(-18-20)
4
(-29 - 34)
Summer
-7
(-29- 14)
12
(0-25)
-3
(-11-6)
-5
(-19-9)
11
(0 - 23)
4
(-4-11)
-10
(-41 - 20)
80
(-92 - 245)
41
(-1 -81)
11
(0 - 23)
27
(-30-81)
7
(0-14)
3
(-3 - 9)
13
(0 - 25)
1
(-15-15)
Fall
1
(-11-13)
2
(-11-14)
0
(-4 - 5)
1
(-11-14)
2
(-13-17)
4
(-6-13)
3
(-26-31)
44
(-63-146)
5
(-32 - 40)
1
(-10-13)
14
(-20 - 47)
1
(-7 - 9)
1
(-2 - 5)
2
(-13-16)
-2
(-29 - 22)
Sum of Four
Seasons
10
4
29
5
6
32
23
16
241
94
27
88
14
10
34
3
All Year
17
(-22 - 56)
17
(-10-44)
7
(-9 - 24)
13
(-16-42)
19
(-1 1 - 49)
16
(4-29)
29
(-36 - 93)
171
(40-301)
56
(-33- 144)
16
(-9 - 42)
61
(14-108)
10
(-6 - 25)
7
(2-13)
21
(-12-54)
2
(-24 - 27)
Percent
Difference 3
-41%
71%
-29%
-54%
68%
44%
-45%
41%
68%
69%
44%
40%
43%
62%
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.
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-29
September 2009
Draft - Do Not Quote or Cite

-------
Table F-28. Sensitivity Analysis: Impact of Using an Annual Concentration-Response Function vs. a Seasonal Function (for April - August)
         Applied Only to that Period to Estimate the Incidence of Emergency Room Visits for Asthma Associated with Short-Term
         Exposure to Concentrations in a Recent Year (2005) and PM2 5 Concentrations that Just Meet the Current and Alternative
         Standards in New York City, Based on Adjusting 2005 PM2 5 Concentrations1
Concentration-Response (C-R) Function and
Period to Which Applied:
Annual C-R Function Applied to the Whole Year
Seasonal C-R Function for April - August Applied
Only to that Period:
Incidence of ER Visits Associated with Short-Term Exposure to PM2.5 Concentrations in a
Recent Year and PM2.5 Concentrations that Just Meet the Current and Alternative Annual (n)
and Daily (m) Standards (Standard Combination Denoted n/m):
Recent PM2.5
Concentration
1162
(743-1567)
737
(485 - 977)
15/352
971
(620- 1314)
620
(406 - 824)
13/35
947
(604- 1281)
605
(396 - 805)
12/35
872
(556-1181)
558
(364 - 744)
13/30
834
(531 -1130)
534
(349-713)
12/25
695
(442 - 943)
447
(291 - 598)
1Based on Ito et al. (2007). New York City in this study consisted only of Manhattan.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 F-30
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2.5
Concentration
991
(633-1340)
627
(411 -832)
15/352
828
(528-1122)
526
(344 - 702)
13/35
807
(515-1094)
513
(335 - 685)
12/35
743
(473-1008)
474
(309-632)
13/30
711
(452 - 964)
453
(295 - 606)
12/25
592
(376 - 804)
379
(246 - 508)
1Based on Ito et al. (2007). New York City in this study consisted only of Manhattan.
2The current primary PM2s standards include an annual standard set at 15 ug/m3 and a daily standard set at 35 ug/m3.
                                                                 F-31
September 2009
Draft - Do Not Quote or Cite

-------
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 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
1114
(712-1504)
684
(449 - 908)
15/352
931
(594-1260)
575
(376 - 766)
13/35
907
(579-1229)
561
(366 - 747)
12/35
836
(532-1133)
517
(337-691)
13/30
799
(509-1083)
495
(323-661)
12/25
665
(423 - 904)
414
(269 - 554)
1Based on Ito et al. (2007). New York City in this study consisted only of Manhattan.
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
September 2009
Draft - Do Not Quote or Cite

-------
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 s 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 Minimum 2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Non-Accidental Mortality Associated with Short-Term Exposure to PM 2.5 — 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
286
(-36 - 606)
312
(0 - 622)
201
(-100-501)
-80
(-388 - 226)
-48
(-342 - 244)
-299
(-616- 16)
Max. positive est. =
312
Min. positive est. =
201
Percent diff. =
55%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2.s — Impact of Changing the Type of 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-linear, GLM, 100 df
Non-Accidental Mortality Associated with Short-Term Exposure to 1
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-linear, GLM, 100 df
Oday
Oday
0 day
0 day
none
none
none
none
286
(-36 - 606)
212
(-181 -601)
170
(-120-457)
159
(-227 - 542)
Max. positive est. =
286
Min. positive est. =
159
Percent diff. =
80%
'/M 2.5 - Impact of Changing the Type of Model, with a 1-Day Lai
1 day
1 day
1 day
1 day
none
none
none
none
312
(0 - 622)
291
(-90 - 669)
53
(-245 - 349)
-5
(-530-512)
Max. positive est. =
312
Min. positive est. =
53
Percent diff. =
489%
85
(-121 -289)
236%
267%
136%
-194%
-156%
-452%
r
85
(-121 -289)
236%
149%
100%
87%
r
85
(-121 -289)
267%
242%
-38%
-106%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2.s — Impact of 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-linear, GLM, 100 df
1 day
1 day
1 day
CO
CO
CO
-283
(-703- 133)
-176
(-561 - 206)
-176
(-627 - 270)

85
(-121 -289)
-433%
-307%
-307%
September 2009
                                                                        F-33
Draft - Do Not Quote or Cite

-------
Table F-31 cont'd.  Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
           Exposure to PM2 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
PM2.5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum 2
Cardiovascular Mortality Associated with Short- Term Exposure to PM 2.s - Impact of Changing the Type of Model, with a 0-Day Lag
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
Oday
none
none
none
178
(17-336)
174
(25 - 322)
174
(-4 - 350)
Max. positive est. =
178
Min. positive est. =
174
Percent diff. =
2%
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
1
-33
(-146-79)
109%
105%
105%
Cardiovascular Mortality Associated with Short- Term Exposure to PM 2.s - Impact of Changing the Type of Model, with a 1-Day Lag
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
1 day
1 day
1 day
none
none
none
185
(27-341)
144
(-6 - 292)
124
(-58 - 304)
Max. positive est. =
185
Min. positive est. =
124
Percent diff. =
49%
-33
(-146-79)
118%
69%
46%
Cardiovascular Mortality Associated with Short-Term Exposure to PM2.s - 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), 100 df
log-linear, GLM, 100 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
1 day
1 day
Respiratory Mortality Associated with Short-Term Exposure to PM2.s - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
Oday
Oday
Oday
Respiratory Mortality Associated with Short-Term Exposure to PM2S - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term 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
CO
CO
CO
CO
318
(135-499)
336
(120-548)
163
(-23 - 347)
163
(-63 - 386)
pact of Changing the Type of Model
none
none
none
-15
(-83-51)
-39
(-106-26)
-33
(-114-44)
pact of Changing the Type of Model
none
none
none
10
(-58 - 76)
23
(-44 - 88)
5
(-78 - 86)
Max. positive est. =
336
Min. positive est. =
163
Percent diff. =
106%
-33
(-146-79)
274%
295%
92%
92%
, with a 0-Day Lag

5
—
—
—
, with a 1-Day Lag
Max. positive est. =
23
Min. positive est. =
5
Percent diff. =
360%

—
—
—
September 2009
                                                                       F-34
Draft - Do Not Quote or Cite

-------
Table F-31 cont'd.  Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
           Exposure to PM2 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
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 PM2.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
824
(474-1170)
606
(264 - 946)
658
(235-1077)
Max. positive est. =
824
Min. positive est. =
606
Percent diff. =
36%
274
(3 - 543)
869%
613%
674%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM2.s - Impact of Changing the Type of Model, with a 1-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GLM, 100 df
log-linear, GLM, 100 df
log-linear, GLM, 100df
Oday
Oday
Oday
CO
CO
CO
725
(361 - 1087)
591
(243 - 935)
627
(201 - 1048)
Max. positive est. =
725
Min. positive est. =
591
Percent diff. =
23%
274
(3 - 543)
753%
595%
638%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM2.s - 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, GLM, 100df
log-linear, GLM, 100df
Oday
Oday
Oday
Oday
CO
CO
CO
CO
205
(-233 - 638)
304
(-216-818)
126
(-342 - 589)
142
(-396 - 672)
Max. positive est. =
304
Min. positive est. =
126
Percent diff. =
141%
274
(3 - 543)
141%
258%
48%
67%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - Impact of Changing the Type of Model, with a 0-Day Lag
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GLM, 100 df
log-linear, GLM, 100df
log-linear, GLM, 100df
Oday
Oday
Oday
CO
CO
CO
348
(143-551)
288
(108-466)
311
(86 - 533)
Max. positive est. =
348
Min. positive est. =
288
Percent diff. =
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
249
(47 - 448)
157
(-23 - 336)
161
(-57 - 377)
Max. positive est. =
249
Min. positive est. =
157
Percent diff. =
59%

—
—
—
September 2009
                                                                       F-35
Draft - Do Not Quote or Cite

-------
Table F-31 cont'd.  Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
             Exposure to PM2 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
PM2.5 Above Policy
Relevant
Background
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - Impact of Changing the 7j
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
2 day
2 day
2 day
none
none
none
385
(172-595)
238
(45 - 430)
216
(-25 - 452)
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
rpe of Model, with a 2-Day Lag
Max. positive est. =
385
Min. positive est. =
216
Percent diff. =
78%

—
—
—
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - 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), 100 df
log-linear, GAM (stringent), 100 df
log-linear, GAM (stringent), 100 df
log-linear, GAM (stringent), 100 df
Oday
1 day
2 day
3 day
NO2
NO2
NO2
NO2
88
(-192-364)
-8
(-342-319)
74
(-217-359)
-231
(-510-43)
Max. positive est. =
88
Min. positive est. =
74
Percent diff. =
19%

—
—
—
—
1The current primary PM25 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 Moolgavkar (2000a, 2000b, and
2000c)].

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).
4Calculated as (Moolgavkar (2003) estimate - core analysis estimate)/(core analysis estimate).
5Because "respiratory illness" was much more broadly defined in both Zanobetti and Schwartz (2009) and Bell et al. (2008) than in Moolgavkar (2003), a comparison between the Moolgavkar (2003) estimates
and the corresponding core analysis estimates is not shown.
                                                                                  F-36
September 2009
Draft - Do Not Quote or Cite

-------
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 s 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 Minimum 2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Non-Accidental Mortality Associated with Short-Term Exposure to PM 2.5 — 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
265
(-34-561)
289
(0 - 576)
186
(-93 - 464)
-74
(-358 - 209)
-44
(-316-226)
-276
(-569- 15)
Max. positive est. =
289
Min. positive est. =
186
Percent diff. =
55%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2.s — Impact of Changing the Type of 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-linear, GLM, 100 df
Non-Accidental Mortality Associated with Short-Term Exposure to 1
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-linear, GLM, 100 df
Oday
Oday
0 day
0 day
none
none
none
none
265
(-34-561)
196
(-167-556)
157
(-1 1 1 - 423)
147
(-210-501)
Max. positive est. =
265
Min. positive est. =
147
Percent diff. =
80%
'/M 2.5 - Impact of Changing the Type of Model, with a 1-Day Lai
1 day
1 day
1 day
1 day
none
none
none
none
289
(0 - 576)
269
(-83-619)
49
(-227 - 323)
-5
(-489 - 474)
Max. positive est. =
289
Min. positive est. =
49
Percent diff. =
490%
78
(-112-267)
240%
271 %
138%
-195%
-156%
-454%
r
78
(-112-267)
240%
151%
101%
88%
r
78
(-112-267)
271 %
245%
-37%
-106%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2.s — Impact of 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-linear, GLM, 100 df
1 day
1 day
1 day
CO
CO
CO
-261
(-649- 123)
-162
(-518- 191)
-162
(-579 - 250)

78
(-112-267)
-435%
-308%
-308%
September 2009
                                                                        F-37
Draft - Do Not Quote or Cite

-------
Table F-32 cont'd.  Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
           Exposure to PM2 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
PM2.5 Above Policy
Relevant
Background
Range of Positive
Estimates and
Percent Difference
Between Maximum
and Minimum 2
Cardiovascular Mortality Associated with Short- Term Exposure to PM 2.s - Impact of Changing the Type of Model, with a 0-Day Lag
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
Oday
none
none
none
165
(16-312)
161
(23 - 298)
161
(-3 - 324)
Max. positive est. =
165
Min. positive est. =
161
Percent diff. =
2%
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
1
-30
(-134-73)
112%
106%
106%
Cardiovascular Mortality Associated with Short- Term Exposure to PM 2.s - Impact of Changing the Type of Model, with a 1-Day Lag
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
1 day
1 day
1 day
none
none
none
171
(25-315)
133
(-6-271)
115
(-53-281)
Max. positive est. =
171
Min. positive est. =
115
Percent diff. =
49%
-30
(-134-73)
119%
71%
47%
Cardiovascular Mortality Associated with Short-Term Exposure to PM2.s - 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), 100 df
log-linear, GLM, 100 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
1 day
1 day
Respiratory Mortality Associated with Short-Term Exposure to PM2.s - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
Oday
Oday
Oday
Respiratory Mortality Associated with Short-Term Exposure to PM2S - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term 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
CO
CO
CO
CO
294
(125-462)
311
(111 -508)
151
(-21 - 322)
151
(-58 - 358)
pact of Changing the Type of Model
none
none
none
-14
(-77 - 47)
-36
(-98 - 24)
-31
(-105-41)
pact of Changing the Type of Model
none
none
none
10
(-54-71)
22
(-40 - 82)
5
(-72 - 79)
Max. positive est. =
311
Min. positive est. =
151
Percent diff. =
106%
-30
(-134-73)
277%
299%
94%
94%
, with a 0-Day Lag

5
—
—
—
, with a 1-Day Lag
Max. positive est. =
22
Min. positive est. =
5
Percent diff. =
340%

—
—
—
September 2009
                                                                       F-38
Draft - Do Not Quote or Cite

-------
Table F-32 cont'd.  Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
           Exposure to PM2 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
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 PM2.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 diff. =
36%
258
(3-511)
894%
631%
694%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM2.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
683
(339-1024)
556
(229-881)
590
(189-987)
Max. positive est. =
683
Min. positive est. =
556
Percent diff. =
23%
258
(3-511)
776%
613%
656%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM2.s - Impact of a Copollutant Model
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
1 day
1 day
CO
CO
CO
CO
193
(-219-600)
286
(-203 - 770)
119
(-322 - 555)
133
(-372 - 633)
Max. positive est. =
286
Min. positive est. =
119
Percent diff. =
140%
258
(3-511)
147%
267%
53%
71%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - 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
322
(132-510)
267
(100-432)
288
(79 - 494)
Max. positive est. =
322
Min. positive est. =
267
Percent diff. =
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
230
(43-415)
146
(-21 -311)
149
(-53 - 349)
Max. positive est. =
230
Min. positive est. =
146
Percent diff. =
58%

—
—
—
September 2009
                                                                       F-39
Draft - Do Not Quote or Cite

-------
Table F-32 cont'd.  Sensitivity Analysis: Estimated Annual Incidence and Percent of Total Incidence of Mortality in Los Angeles, CA Associated with Short-Term
             Exposure to PM2 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
PM2.5 Above Policy
Relevant
Background
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - Impact of Changing the 7j
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
2 day
2 day
2 day
none
none
none
357
(160-552)
221
(41 - 398)
200
(-23-419)
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
rpe of Model, with a 2-Day Lag
Max. positive est. =
357
Min. positive est. =
200
Percent diff. =
79%

—
—
—
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - 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), 100 df
log-linear, GAM (stringent), 100 df
log-linear, GAM (stringent), 100 df
log-linear, GAM (stringent), 100 df
Oday
1 day
2 day
3 day
NO2
NO2
NO2
NO2
82
(-178-337)
-8
(-316-295)
68
(-200 - 332)
-214
(-471 - 40)
Max. positive est. =
82
Min. positive est. =
68
Percent diff. =
21%

—
—
—
—
1The current primary PM25 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 Moolgavkar (2000a, 2000b, and

                                                                                                                                             . The core analysis estimates for
2000c)].
2.
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).
4Calculated as (Moolgavkar (2003) estimate - core analysis estimate)/(core analysis estimate).
5Because "respiratory illness" was much more broadly defined in both Zanobetti and Schwartz (2009) and Bell et al. (2008) than in Moolgavkar (2003), a comparison between the Moolgavkar (2003) estimates
and the corresponding core analysis estimates is not shown.
                                                                                  F-40
September 2009
                                                                                                                                                Draft - Do Not Quote or Cite

-------
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 s 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 Minimum 2
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
Non-Accidental Mortality Associated with Short-Term Exposure to PM 2.5 — 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
267
(-34 - 566)
292
(0 - 582)
188
(-94 - 468)
-75
(-362-211)
-45
(-320 - 228)
-279
(-575- 15)
Max. positive est. =
292
Min. positive est. =
188
Percent diff. =
55%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2.s — Impact of Changing the Type of 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-linear, GLM, 100 df
Non-Accidental Mortality Associated with Short-Term Exposure to 1
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-linear, GLM, 100 df
Oday
Oday
0 day
0 day
none
none
none
none
267
(-34 - 566)
198
(-169-562)
159
(-1 1 2 - 427)
149
(-212-506)
Max. positive est. =
267
Min. positive est. =
149
Percent diff. =
79%
'/M 2.5 - Impact of Changing the Type of Model, with a 1-Day Lai
1 day
1 day
1 day
1 day
none
none
none
none
292
(0 - 582)
272
(-84 - 625)
50
(-229 - 326)
-5
(-494 - 479)
Max. positive est. =
292
Min. positive est. =
50
Percent diff. =
484%
79
(-113-270)
238%
270%
138%
-195%
-157%
-453%
r
79
(-113-270)
238%
151%
101%
89%
r
79
(-113-270)
270%
244%
-37%
-106%
Non-Accidental Mortality Associated with Short-Term Exposure to PM2.s — Impact of 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-linear, GLM, 100 df
1 day
1 day
1 day
CO
CO
CO
-264
(-656- 124)
-164
(-524- 192)
-164
(-585 - 252)

79
(-113-270)
-434%
-308%
-308%
September 2009
                                                                        F-41
Draft - Do Not Quote or Cite

-------
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 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 Minimum 2
Cardiovascular Mortality Associated with Short- Term Exposure to PM 2.s - Impact of Changing the Type of Model, with a 0-Day Lag
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
Oday
none
none
none
166
(16-314)
163
(24-301)
163
(-3 - 327)
Max. positive est. =
166
Min. positive est. =
163
Percent diff. =
2%
Incidence
Estimate Using
Core Analysis
Model 3
Percent
Difference
(Compared to
Core Analysis
Model)4
1
-30
(-136-74)
110%
106%
106%
Cardiovascular Mortality Associated with Short- Term Exposure to PM 2.s - Impact of Changing the Type of Model, with a 1-Day Lag
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
Mortality, short-term cardiovascular
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
1 day
1 day
1 day
none
none
none
173
(26-318)
134
(-6 - 273)
116
(-54 - 284)
Max. positive est. =
173
Min. positive est. =
116
Percent diff. =
49%
-30
(-136-74)
119%
70%
47%
Cardiovascular Mortality Associated with Short-Term Exposure to PM2.s - 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), 100 df
log-linear, GLM, 100 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
1 day
1 day
Respiratory Mortality Associated with Short-Term Exposure to PM2.s - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
log-linear, GAM (stringent), 30 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
Oday
Oday
Oday
Respiratory Mortality Associated with Short-Term Exposure to PM2S - Im
Mortality, short-term respiratory
(COPD+)
Mortality, short-term respiratory
(COPD+)
Mortality, short-term 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
CO
CO
CO
CO
297
(126-466)
314
(112-512)
153
(-21 - 325)
153
(-59-361)
pact of Changing the Type of Model
none
none
none
-14
(-78 - 47)
-36
(-99 - 24)
-31
(-106-41)
pact of Changing the Type of Model
none
none
none
10
(-54-71)
22
(-41 - 82)
5
(-73 - 80)
Max. positive est. =
314
Min. positive est. =
153
Percent diff. =
105%
-30
(-136-74)
276%
297%
94%
94%
, with a 0-Day Lag

5
—
—
—
, with a 1-Day Lag
Max. positive est. =
22
Min. positive est. =
5
Percent diff. =
340%

—
—
—
September 2009
                                                                       F-42
Draft - Do Not Quote or Cite

-------
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 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 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 PM2.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
798
(459-1134)
587
(256-916)
637
(228-1043)
Max. positive est. =
798
Min. positive est. =
587
Percent diff. =
36%
265
(3 - 526)
910%
643%
706%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM2.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
703
(349-1053)
572
(235 - 906)
607
(195-1015)
Max. positive est. =
703
Min. positive est. =
572
Percent diff. =
23%
265
(3 - 526)
790%
624%
668%
Cardiovascular Hospital Admissions Associated with Short-Term Exposure to PM2.s - Impact of a Copollutant Model
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
HA, cardiovascular
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100 df
log-linear, GAM (stringent), 100 df
log-linear, GLM, 100df
Oday
Oday
1 day
1 day
CO
CO
CO
CO
198
(-225-618)
295
(-209 - 792)
122
(-331 -571)
137
(-383-651)
Max. positive est. =
295
Min. positive est. =
122
Percent diff. =
142%
265
(3 - 526)
151%
273%
54%
73%
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - 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
325
(134-515)
269
(101 -436)
291
(80 - 498)
Max. positive est. =
325
Min. positive est. =
269
Percent diff. =
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
233
(44-419)
147
(-22-314)
151
(-53 - 353)
Max. positive est. =
233
Min. positive est. =
147
Percent diff. =
59%

—
—
—
September 2009
                                                                       F-43
Draft - Do Not Quote or Cite

-------
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 Concentrations1
Health Effects
Model
Lag
Other
Pollutants
in Model
Incidence
Associated with
PM2.5 Above Policy
Relevant
Background
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - Impact of Changing the 7j
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
2 day
2 day
2 day
none
none
none
360
(161 -557)
223
(42 - 402)
201
(-23 - 423)
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
rpe of Model, with a 2-Day Lag
Max. positive est. =
360
Min. positive est. =
201
Percent diff. =
79%

—
—
—
Respiratory Hospital Admissions Associated with Short-Term Exposure to PM2.s - 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), 100 df
log-linear, GAM (stringent), 100 df
log-linear, GAM (stringent), 100 df
log-linear, GAM (stringent), 100 df
Oday
1 day
2 day
3 day
NO2
NO2
NO2
NO2
82
(-180-340)
-8
(-320 - 298)
69
(-202 - 335)
-216
(-476 - 40)
Max. positive est. =
82
Min. positive est. =
69
Percent diff. =
19%

—
—
—
—
1The current primary PM25 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 Moolgavkar (2000a, 2000b, and
2000c)].
2.
                                                                                                                                             . 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).
4Calculated as (Moolgavkar (2003) estimate - core analysis estimate)/(core analysis estimate).
5Because "respiratory illness" was much more broadly defined in both Zanobetti and Schwartz (2009) and Bell et al. (2008) than in Moolgavkar (2003), a comparison between the Moolgavkar (2003) estimates
and the corresponding core analysis estimates is not shown.
                                                                                  F-44
September 2009
Draft - Do Not Quote or Cite

-------
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
Pittsburgh, PA
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
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):
15/35 2
256
(104-406)
254
(103-403)
-1%
38
(-60-135)
39
(-60-136)
3%
141
(-25 - 305)
146
(-26-315)
4%
85
(-121 -289)
94
(-135-323)
11%
714
(419- 1007)
726
(426-1023)
2%
135
(40 - 230)
163
(48 - 277)
21%
252
(73 - 429)
266
(77 - 452)
6%
13/35
224
(91 - 355)
230
(93 - 365)
3%
33
(-51 -115)
33
(-51 -115)
0%
129
(-23 - 280)
143
(-25-310)
11%
85
(-121 -289)
92
(-132-315)
8%
696
(408-981)
695
(408 - 979)
0%
135
(40 - 230)
158
(46 - 268)
17%
219
(63 - 373)
228
(66 - 389)
4%
12/35
206
(84 - 327)
211
(86 - 335)
2%
30
(-46-105)
30
(-46- 105)
0%
119
(-21 - 257)
132
(-23 - 284)
11%
80
(-114-273)
85
(-121 -289)
6%
639
(375 - 902)
638
(375 - 900)
0%
135
(40 - 230)
145
(43 - 246)
7%
201
(58 - 342)
210
(61 - 357)
4%
13/30
219
(89 - 348)
217
(88 - 345)
-1%
33
(-51 -115)
33
(-51 -115)
0%
121
(-21 - 260)
125
(-22 - 270)
3%
72
(-103-247)
81
(-116-276)
13%
611
(358-861)
621
(364 - 876)
2%
116
(34-197)
140
(41 - 237)
21%
215
(62 - 367)
228
(66 - 389)
6%
12/25
182
(74 - 289)
181
(73 - 287)
-1%
27
(-42 - 96)
28
(-43 - 98)
4%
100
(-18-216)
103
(-18-224)
3%
60
(-86 - 205)
67
(-96 - 229)
12%
507
(297-716)
515
(302 - 727)
2%
96
(28-163)
116
(34-197)
21%
179
(51 - 305)
191
(55 - 325)
7%
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.
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-45
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2.5 Concentrations:
           Comparison of Proportional and Hybrid Rollback Methods
Risk Assessment
Location
Baltimore, MD
Birmingham, AL
Detroit, Ml
Los Angeles, CA
New York, NY
Pittsburgh, PA
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
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
234
(95 - 372)
233
(94 - 369)
0%
38
(-59- 132)
38
(-59-132)
0%
117
(-21 - 254)
121
(-21 - 262)
3%
78
(-112-267)
87
(-125-299)
12%
654
(384 - 922)
664
(390 - 936)
2%
122
(36 - 208)
147
(43 - 249)
20%
214
(62 - 364)
225
(65 - 384)
5%
13/35
205
(83 - 325)
210
(85 - 334)
2%
32
(-50- 113)
32
(-50-113)
0%
108
(-19-233)
119
(-21 - 258)
10%
78
(-112-267)
85
(-122-291)
9%
637
(374 - 898)
635
(373 - 896)
0%
122
(36 - 208)
141
(41 - 240)
16%
185
(53-316)
194
(56 - 330)
5%
12/35
188
(76 - 299)
193
(78 - 307)
3%
29
(-45- 103)
29
(-45-103)
0%
99
(-17-214)
109
(-19-237)
10%
74
(-106-252)
78
(-112-268)
5%
585
(343 - 826)
584
(343 - 824)
0%
122
(36 - 208)
130
(38-221)
7%
170
(49 - 290)
178
(51 - 303)
5%
13/30
201
(81 -318)
199
(81 -316)
-1%
32
(-50- 113)
32
(-50-113)
0%
100
(-18-217)
104
(-18-224)
4%
67
(-96 - 228)
74
(-107-255)
10%
559
(328 - 789)
568
(333-801)
2%
104
(31 - 177)
125
(37-213)
20%
183
(53-311)
194
(56 - 330)
6%
12/25
167
(68 - 265)
165
(67 - 263)
-1%
26
(-41 - 94)
27
(-42 - 96)
4%
83
(-15-180)
86
(-15-186)
4%
55
(-79-189)
62
(-88-211)
13%
464
(272 - 655)
471
(276 - 665)
2%
87
(25- 147)
104
(30-177)
20%
151
(44 - 258)
162
(47 - 276)
7%
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.
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 (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
                                                                                   F-46
September 2009
Draft - Do Not Quote or Cite

-------
Table F-36. 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 2007 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
Pittsburgh, PA
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
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):
15/352
225
(91 - 357)
223
(90 - 354)
-1%
37
(-58- 131)
37
(-58- 132)
0%
112
(-20 - 242)
116
(-20 - 250)
4%
79
(-113-270)
88
(-126-302)
11%
659
(387 - 930)
670
(393 - 944)
2%
123
(36 - 209)
147
(43 - 250)
20%
222
(64 - 378)
234
(68 - 399)
5%
13/35
197
(80-312)
202
(82 - 320)
3%
32
(-49-112)
32
(-49- 112)
0%
102
(-18-222)
113
(-20 - 245)
11%
79
(-113-270)
86
(-123-294)
9%
642
(377 - 906)
641
(376 - 904)
0%
123
(36 - 209)
142
(42 - 242)
15%
193
(56 - 328)
201
(58 - 343)
4%
12/35
181
(73 - 287)
185
(75 - 294)
2%
29
(-45-102)
29
(-45- 102)
0%
94
(-16-204)
104
(-18-225)
11%
74
(-107-255)
79
(-113-270)
7%
590
(346 - 833)
589
(346-831)
0%
123
(36 - 209)
131
(38 - 222)
7%
177
(51 -301)
185
(53-315)
5%
13/30
192
(78 - 306)
191
(77 - 303)
-1%
32
(-49 - 1 1 2)
32
(-49 - 1 1 2)
0%
95
(-17-207)
99
(-17-214)
4%
67
(-97-231)
75
(-108-258)
12%
564
(331 - 795)
573
(336 - 808)
2%
105
(31 - 179)
126
(37-214)
20%
190
(55 - 323)
201
(58 - 343)
6%
12/25
160
(65 - 254)
158
(64 - 252)
-1%
26
(-41 - 93)
27
(-42 - 95)
4%
79
(-14- 171)
82
(-14-1 77)
4%
56
(-80-191)
62
(-89-213)
11%
468
(274-661)
476
(279-671)
2%
87
(26- 148)
104
(31 -178)
20%
157
(45 - 268)
168
(48 - 287)
7%
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.
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 (mortality based on hybrid rollbacks - mortality based on proportional rollbacks)/(mortality based on proportional rollbacks).
                                                                                   F-47
September 2009
Draft - Do Not Quote or Cite

-------
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 PM2.5 Concentrations that Just Meet the Current
          Standards, Based on Adjusting 2005 PM2 s 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
1432
(911 -1948)
—
1330
(1083-1572)
—
1775
(1131 -2412)
24%
1634
(1334-1927)
23%
2934
(1876-3972)
105%
2621
(2156-3067)
97%
3268
(2091 -4418)
128%
2894
(2386 - 3380)
118%
3526
(2179-4841)
146%
2652
(1878-3376)
99%
4121
(2551 - 5644)
188%
3061
(2180-3877)
130%
13713
(8814-18111)
858%
8336
(6475 - 9736)
527%
14193
(9141 - 18709)
891%
8540
(6662 - 9937)
542%
Philadelphia, PA
547
(349 - 743)
—
347
(284 - 409)
—
4
—
—
—
732
(468-991)
34%
571
(472 - 665)
65%
—
—
—
—
1201
(745-1641)
120%
614
(439 - 774)
77%
—
—
—
—
3896
(2520-5116)
612%
1599
(1258-1846)
361%
—
—
—
—
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.
" 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 (3099 -1257)/1257 =
147%.
4 Philadelphia was not among the risk assessment urban areas for which hybrid rollbacks were calculated.
                                                                                  F-48
September 2009
Draft - Do Not Quote or Cite

-------
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
1221
(776-1662)
—
1140
(927-1349)
—
1540
(980 - 2094)
26%
1426
(1163-1685)
25%
2738
(1749-3709)
124%
2459
(2020 - 2882)
116%
3048
(1949-4125)
150%
2716
(2236-3177)
138%
3130
(1931 -4303)
156%
2374
(1675-3034)
108%
3726
(2304-5112)
205%
2792
(1981 -3549)
145%
13459
(8639-17800)
1002%
8241
(6382-9651)
623%
13941
(8966-18402)
1042%
8449
(6571 - 9857)
641%
Philadelphia, PA
536
(341 - 727)
—
340
(279-401)
—
4
—
—
—
923
(591 -1246)
72%
564
(466 - 657)
66%
—
—
—
—
1182
(733-1617)
121%
606
(433 - 764)
78%
—
—
—
—
3875
(2506 - 5089)
623%
1592
(1252-1839)
368%
—
—
—
—
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.
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 (3099 -1257)/1257 =
147%.
4 Philadelphia was not among the risk assessment urban areas for which hybrid rollbacks were calculated.
                                                                                  F-49
September 2009
Draft - Do Not Quote or Cite

-------
Table F-39. Multi-Factor Sensitivity Analysis: Impact of Using a Log-Linear vs. a Log-Log Model, Estimating Incidence Down to the Lowest
          Measured Level (LML) in the Study vs. PRB, and Using a Proportional vs. a Hybrid Rollback to Estimate the Incidence of All Cause
          and Ischemic Heart Disease Mortality Associated with Long-Term Exposure to PM2 s Concentrations that Just Meet the Current
          Standards, Based on Adjusting 2007 PM2.5 Concentrations1
Modeling Choices:
Fixed Effects (FE) Log-linear vs.
Random Effects (RE) log-log model
Down to LML (5.8 ug/m3) vs. PRB
Proportional vs. hybrid rollback

All Cause Mortality
Percent Difference: 3
Ischemic Heart Disease Mortality
Percent Difference:

All Cause Mortality
Percent Difference:
Ischemic Heart Disease Mortality
Percent Difference
Incidence of Mortality Associated with Long -Term Exposure to PM2.5 Concentrations Using:2
FE Log-Linear
LML
Proportional
FE Log-Linear
LML
Hybrid
FE Log-Linear
PRB
Proportional
FE Log-Linear
PRB
Hybrid
RE Log-Log
LML
Proportional
RE Log-Log
LML
Hybrid
RE Log -Log
PRB
Proportional
RE Log -Log
PRB
Hybrid
Los Angeles, CA
1257
(799-1711)
—
1173
(954-1388)
—
1582
(1007-1582)
26%
1464
(1193-1464)
25%
2783
(1778-3770)
121%
2498
(2053 - 2927)
113%
3099
(1982-4193)
147%
2759
(2272 - 3227)
135%
3206
(1979-4407)
155%
2429
(1715-3102)
107%
3807
(2354-5221)
203%
2849
(2022 - 3620)
143%
13590
(8725-17970)
981%
8313
(6440 - 9732)
609%
14076
(9054-18576)
1020%
8523
(6631 - 9939)
627%
Philadelphia, PA
530
(338-719)
—
337
(276 - 397)
—
4
—
:::
—
917
(588-1239)
73%
561
(464 - 654)
66%
:::
—
:::
—
1174
(728-1605)
122%
602
(430 - 759)
79%
:::
—
:::
—
3870
(2502 - 5083)
630%
1591
(1251 -1838)
372%
:::
—
:::
—
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.
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 (3099 -1257)/1257 =
147%.
4 Philadelphia was not among the risk assessment urban areas for which hybrid rollbacks were calculated.
                                                                                  F-50
September 2009
Draft - Do Not Quote or Cite

-------
  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 l'2
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)
...
38
(-60-135)
—
141
(-25 - 305)
—
85
(-121 -289)
—
714
(419-1007)
—
135
(40 - 230)
...
252
(73 - 429)
—
All Year
Hybrid
254
(103-403)
-1%
39
(-60-136)
3%
146
(-26-315)
4%
94
(-135-323)
11%
726
(426-1023)
2%
163
(48 - 277)
21%
266
(77 - 452)
6%
Sum of Four Seasons
Proportional
222
4
-13%
78
105%
153
9%
-28
-133%
717
0%
151
12%
243
-4%
Sum of Four Seasons
Hybrid
222
-13%
78
105%
159
13%
-32
-138%
728
2%
182
35%
256
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.

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 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)/225 = -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
September 2009
Draft - Do Not Quote or Cite

-------
  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 l'2
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
234
(95 - 372)
...
38
(-59-132)
—
117
(-21 - 254)
—
78
(-112-267)
—
654
(384 - 922)
—
122
(36 - 208)
...
214
(62 - 364)
—
All Year
Hybrid
233
(94 - 369)
0%
38
(-59-132)
0%
121
(-21 - 262)
3%
87
(-125-299)
12%
664
(390 - 936)
2%
147
(43 - 249)
20%
225
(65 - 384)
5%
Sum of Four Seasons
Proportional
202
4
-14%
81
113%
139
19%
-29
-137%
656
0%
137
12%
209
-2%
Sum of Four Seasons
Hybrid
201
-14%
80
111%
143
22%
-32
-141%
667
2%
164
34%
221
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.

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 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)/225 = -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
September 2009
Draft - Do Not Quote or Cite

-------
   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
             1,2
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
225
(91 - 357)
—
37
(-58-131)
—
112
(-20 - 242)
—
79
(-113-270)
...
659
(387 - 930)
...
123
(36 - 209)
...
222
(64 - 378)
—
All Year
Hybrid
223
(90 - 354)
-1%
37
(-58-132)
0%
116
(-20 - 250)
4%
88
(-126-302)
11%
670
(393 - 944)
2%
147
(43 - 250)
20%
234
(68 - 399)
5%
Sum of Four Seasons
Proportional
206
4
-8%
106
186%
148
32%
-11
-114%
797
21%
234
90%
244
10%
Sum of Four Seasons
Hybrid
192
-15%
81
119%
124
11%
-7
-109%
674
2%
162
32%
231
4%
 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.
 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)1225 = -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
September 2009
Draft - Do Not Quote or Cite

-------
     APPENDIX G: SUPPLEMEMNT TO THE NATIONAL-SCALE
  ASSESSMENT OF LONG-TERM  MORTALITY RELATED TO PM2 5
                          EXPOSURE
                              G-l
September 2009                                         Draft - Do Not Quote or Cite

-------
 1    Appendix G  National-Scale Assessment Of Long-Term Mortality Related To PM2.s
 2   	Exposure (additional technical detail regarding inputs used in the analysis)	
 3
 4
 5          This technical appendix includes additional details regarding the inputs to the national-
 6    scale current conditions health impact analysis. Below we present air quality modeling, exposure
 7    and risk information.
 8
 9          Air Quality Modeled Inputs
10          The Community Model for Air Quality (CMAQ) model was used to estimate annual
11    PM2.5 concentrations for the year 2005 for the continental US.  These data were then combined
12    with ambient monitored PM2.5 measurements to create "fused" spatial surfaces supplied to
13    BenMAP.
14          CMA Q Model Application and Evaluation
15          CMAQ is a non-proprietary computer model that simulates the formation and fate of
16    photochemical oxidants, including PM2.5 and ozone, for given input sets of meteorological
17    conditions and emissions. This analysis employed a version of CMAQ based on the latest
18    publicly released version (i.e. CMAQ version 4.72).
19          Model Domain and Grid Resolution
20          The CMAQ modeling analyses were performed for two domains covering the continental
21    United States, as shown in Figure 4-1.  These domains consist of a horizontal grid  of 36 km
22    covering the entire continental US and a finer-scale 12-km grid covering the Eastern U.S. The
23    model extends vertically from the surface to 100 millibars (approximately  15 km) using a sigma-
24    pressure coordinate system.  The 36-km grid was used to establish the incoming air quality
25    concentrations along the boundaries of the 12-km  grids.  Table G-l provides some basic
26    geographic information regarding the CMAQ domains. The 36-km and both 12-km CMAQ
27    modeling domains were modeled for the entire year of 2005.  All 365 model days were used in
28    the annual average levels of PM2.s.
29
30
31
32
33
            2CMAQ version 4.7 was released on December 1, 2008. It is available from the Community Modeling and
      Analysis System (CMAS) at: http://www.cmascenter.org.

                                                G-2
      September 2009                                                          Draft - Do Not Quote or Cite

-------
                  Table G-l. Geographic Information for Modeling Domains


Map
Projection
Grid
Resolution
Coordinate
Center
True
Latitudes
Dimensions
Vertical
extent
CMAQ Modeling Configuration
National Grid
Eastern U.S.
Grid
Fine
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
September 2009
                                         G-3
Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
           36km Domain
                                                  Specs:
                                                .  x,y: -1008000.-1620000
                                    12km Domain couow  279,240
Figure G-l.  Map of the CMAQ Modeling Domain (Note, the black outer box denotes the
          36-km national modeling domain; the red inner box is the 12-km Eastern U.S. fine
          grid).

       CMAQ 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)
       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
                                          G-4
September 2009
                                                                          Draft - Do Not Quote or Cite

-------
 1    SO2 and NOx emissions plus hourly heat input amounts. These three values are reported to the
 2    database by the largest electric generating facilities, usually based upon Continuous Emissions
 3    Monitors (CEMs). For all pollutants except the directly monitored SO2 and NOx, the ratio of the
 4    Acid Rain heat input for 2005 to the Acid Rain heat input for 2002 was used as the adjusting
 5    ratio to estimate the 2005 emissions.
 6          Other Stationary Sources (Point and Nonpoint)
 7          Emission estimates for other stationary sources including both point and nonpoint
 8    stationary sources were held constant at the level in Version 3 of the 2002 NEI. The only
 9    exception to this was that some information on plants that closed after 2002 was incorporated
10    into the emissions modeled. Emissions for plants that closed were set to zero.  U.S. EPA, 2008c
11    provides complete documentation on the development of the 2002 NEI.
12          Onroad and Nonroad Mobile Sources
13          Emission estimates for all pollutants were developed using EPA's National Mobile
14    Inventory Model (NMIM), which uses MOBILE6 to calculate onroad emission factors.  A full
15    VMT database at the county, roadway type, and vehicle type level of detail was developed from
16    Federal Highway Administration (FHWA) information. However, state and local agencies had
17    the opportunity to provide model inputs (vehicle populations, fuel characteristics, VMT, etc) for
18    2002 and 2005. If the state or local area submitted 2005 VMT estimates, these data were used.
19    However, if the state or local area only provided 2002 VMT estimates that were incorporated in
20    the 2002 NEI, the 2002 NEI VMT  data were grown to 2005 using growth factors developed from
21    the FHWA data, and these grown VMT data replaced the baseline FHWA-based VMT data.
22    Otherwise, the FHWA-based VMT data were  used.
23          Emission estimates for NONROAD model engines were developed using EPA's National
24    Mobile Inventory Model (NMIM), which incorporates NONROAD2005. Where states provided
25    alternate nonroad inputs, these data replaced EPA default inputs, as described above.  For more
26    information on how NMIM is run,  refer to the 2005 NEI documentation posted at
27    ftp://ftp.epa.gov/EmisInventory/2005_nei/mobile/2005_mobile_nei_version_2_report.pdf
28          Fires
29          Fires in the 2005 emissions inventory were modeled with the same methodology as used
30    for the 2002 NEI (U.S. EPA, 2008). However, as described in Raffuse et al., 2008, the wildland
31    fire emission inventories for 2005 were produced using the BlueSky framework for the
                                               G-5
      September 2009                                                          Draft - Do Not Quote or Cite

-------
 1    conterminous United States, which used the Satellite Mapping Automatic Reanalysis Tool for
 2    Fire Incident Reconciliation (SMARTFIRE) as the fire information source.  SMARTFIRE is an
 3    algorithm and database system designed to reconcile these disparate fire information sources to
 4    produce daily fire location and size information (Sullivan et al., 2008).
 5          Biogenic Emissions
 6          Biogenic emissions were computed for CMAQ based on 2005 meteorology data using the
 7    BEIS3.13 model (Schwede, et. al, 2005) from the Sparse Matrix Operator Kernel Emissions
 8    (SMOKE). The BEIS3.13 model creates gridded, hourly, model-species emissions from
 9    vegetation and soils. It estimates CO, VOC, and NOX emissions for the U.S., Mexico, and
10    Canada. The inputs to BEIS include:
11                 • temperature data at 10 meters which were obtained from the CMAQ
12          meteorological input files, and
13                 • land-use data from the Biogenic Emissions Landuse Database, version 3
14          (BELD3), which provides data on the 230 vegetation classes at 1 km resolution over most
15          of North America.
16          Meteorological Input Data:
17          The gridded meteorological input data for the entire year of 2005 were derived from
18    simulations of the Pennsylvania State University / National Center for Atmospheric Research
19    Mesoscale Model. This model, commonly referred to as MM5, is a limited-area, nonhydrostatic,
20    terrain-following system that solves for the full set of physical and thermodynamic equations
21    which govern atmospheric motions (Grell et al., 1994). Meteorological model input fields were
22    prepared separately for both of the domains shown in Figure G-l using MM5 version 3.7.4.  The
23    MM5 simulations were run on the same map projection as CMAQ.
24          Both meteorological model runs were configured similarly. The selections for key MM5
25    physics options are shown below:
26             •   Pleim-Xiu PEL and land surface schemes
27             •   Kain-Fritsh 2 cumulus parameterization
28             •   Reisner 2 mixed phase moisture scheme
29             •   RRTM longwave radiation scheme
30             •   Dudhia shortwave radiation scheme
31
                                               G-6
      September 2009                                                         Draft - Do Not Quote or Cite

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
       Three dimensional analysis nudging for temperature and moisture was applied above the
boundary layer only.  Analysis nudging for the wind field was applied above and below the
boundary layer. The 36 km domain nudging weighting factors were 3.0 x 104 for wind fields and
temperatures and 1.0 x 105 for moisture fields. The 12 km domain nudging weighting factors
were 1.0 x 104 for wind fields and temperatures and 1.0 x 105 for moisture fields.
       All model runs were conducted in 5.5 day segments with 12 hours of overlap for spin-up
purposes. Both domains contained 34 vertical layers with an approximately 38m deep surface
layer and a  100 millibar top. The MM5 and CMAQ vertical structures are shown in Table G-2
and do not vary by horizontal grid resolution.
        Table G-2. Vertical Layer Structure for MM5 and CMAQ (heights are layer top).
CMAQ
Layers
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
MM5
Layers
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
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
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
Approximate
Height (m)
0
38
77
115
154
232
310
389
469
550
631
712
794
961
1,130
1,303
1,478
1,657
1,930
2,212
2,600
3,108
3,644
4,212
4,816
5,461
Approximate
Pressure (mb)
1000
995
991
987
982
973
964
955
946
937
928
919
910
892
874
856
838
820
793
766
730
685
640
595
550
505
September 2009
                                          G-7
                                                                          Draft - Do Not Quote or Cite

-------
CMAQ
Layers

21
22
23
24
MM5
Layers
26
27
28
29
30
31
32
33
34
Sigma P
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Approximate
Height (m)
6,153
6,903
7,720
8,621
9,625
10,764
12,085
13,670
15,674
Approximate
Pressure (mb)
460
415
370
325
280
235
190
145
100
 2          The meteorological outputs from the MM5 sets were processed to create model-ready
 3    inputs for CMAQ using the Meteorology-Chemistry Interface Processor (MCIP), version 3.4, to
 4    derive the specific inputs to CMAQ.
 5          Before initiating the air quality simulations, it was important to identify the biases and
 6    errors associated with the meteorological modeling inputs. The 2005 MM5 model performance
 7    evaluations used an approach which included a combination of qualitative and quantitative
 8    analyses to assess the adequacy of the MM5 simulated fields.  The qualitative aspects involved
 9    comparisons of the model-estimated synoptic patterns against observed patterns from historical
10    weather chart archives. Additionally, the evaluations compared spatial patterns of monthly
11    average rainfall  and monthly maximum planetary boundary layer (PEL) heights.  Qualitatively,
12    the model fields closely matched the observed synoptic patterns, which is not unexpected given
13    the use of nudging. The operational evaluation included statistical comparisons of
14    model/observed pairs (e.g., mean normalized bias, mean normalized error, index of agreement,
15    root mean square errors, etc.) for multiple meteorological parameters, including temperature,
16    humidity, shortwave downward radiation, wind speed, and wind direction  (Baker and Dolwick,
17    2009a, Baker and Dolwick, 2009b). It was ultimately determined that the bias and error values
18    associated with the 2005  meteorological data were generally within the range of past
19    meteorological modeling results that have been used for air quality applications.
20          Initial and Boundary Conditions:
21          The lateral boundary and initial species concentrations are provided by a three-
22    dimensional global atmospheric chemistry model, the GEOS-CHEM model (Yantosca, 2004).
23    The global GEOS-CHEM model simulates atmospheric chemical and physical processes driven
24    by assimilated meteorological observations from the NASA's Goddard Earth Observing System
25    (GEOS).  This model was run for 2002 with a grid resolution of 2.0 degrees x 2.5  degrees
26    (latitude-longitude) and 24 vertical layers. The 2005 CMAQ 36km simulation used non-year
                                                G-8
      September 2009
Draft - Do Not Quote or Cite

-------
 1    specific GEOS-CHEM data, which was created by taking the median value for each month in
 2    each individual grid cell of the 2002 GEOS-CHEM data described above. The predictions were
 3    used to provide one-way dynamic boundary conditions and an initial concentration field for the
 4    CMAQ simulations.  More information is available about the GEOS-CHEM model and other
 5    applications using this tool at: http://www-as.harvard.edu/chemistry/trop/geos.
 6          CMA Q Model Performance Evaluation
 1          An operational model performance evaluation for PM2.5 and its related speciated
 8    components was conducted for 2005 using state/local monitoring sites data in order to estimate
 9    the ability of the CMAQ modeling system to replicate the concentrations for the 12-km Eastern
10    domain and 36-km domain in the west. The principal evaluation statistics used to evaluate
11    CMAQ performance included two bias metrics, normalized mean bias and fractional bias; and
12    two error metrics, normalized mean error and fractional error. For the 12-km Eastern domain,
13    performance evaluation statistics were computed for the entire domain as well as its subregions.
14    For the 36-km domain, evaluation focuses on the parts of the US not covered by the 12-km
15    Eastern domain by computing performance evaluation statistics for the states included in the
16    Western Regional Air Partnership (WRAP).
17          The PM2.s evaluation focuses on PM2 5 total mass and its components, including sulfate
18    (SO4), nitrate (NO3), total nitrate (TNO3 = NO3 + HNO3), ammonium (NH4), elemental carbon
19    (EC), and organic carbon (OC).  PM2.5 ambient measurements for 2005 were obtained from the
20    following networks for model evaluation: Speciation Trends Network (STN), Interagency
21    Monitoring of PROtected Visual Environments (IMPROVE), and Clean Air Status and Trends
22    Network (CASTNET).  For PM2.5 species that are measured by more than one network, we
23    calculated separate sets of statistics for each network. Table G-3 provides annual model
24    performance statistics for PM2 5  and its component species. Based on the bias and error values
25    associated with the 2005 CMAQ-modeled PM2.5 concentration data,  it was determined that the
26    annual average PM2 5 data were  generally within the range of past modeling results used for air
27    quality applications and are applicable to be used for this national-scale current conditions
28    analysis.
29
30            Table G-3.  CMAQ modeled performance evaluation statistics for PM2.5 for 2005.
31
CMAQ 2005 Annual
PM2.5 Total
Mass
STN
12-km BUS
Northeast
No. of
Obs.
11622
2795
NMB (%)
-2.2
4.2
NME (%)
39.1
41.3
FB (%)
-4.7
3.4
FE (%)
40.3
39.5
      September 2009
                                               G-9
Draft - Do Not Quote or Cite

-------
CMAQ 2005 Annual

Sulfate
Nitrate

IMPROVE
STN
IMPROVE
CASTNet
STN
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
No. of
Obs.
2318
2960
2523
3082
10534
2464
668
1963
2768
10,122
13317
3247
2495
3499
2944
3450
10164
2393
622
1990
2640
9693
3170
786
615
1099
300
1112
12186
3248
NMB (%)
4.3
-13.0
-2.2
-35.1
-9.4
5.3
-4.6
-20.8
-10.5
-21.0
-17.1
-13.7
-10.9
-19.2
-25.7
-21.9
-21.8
-14.6
-19.0
-25.2
-27.9
-5.2
-16.5
-11.7
-13.6
-18.4
-29.4
-12.6
20.1
28.7
NME (%)
35.2
37.5
43.1
50.7
44.3
48.6
38.2
42.8
42.8
56.0
34.0
32.4
33.9
32.8
38.7
46.4
36.4
35.5
34.5
35.9
38.0
45.2
22.9
20.5
21.4
22.9
32.5
34.5
67.8
70.2
FB (%)
5.0
-15.9
-8.4
-40.3
-13.8
2.3
-7.3
-25.9
-12.9
-24.4
-13.5
-9.4
-4.4
-16.8
-23.1
-15.0
-13.2
-6.6
-9.4
-22.3
-22.0
9.6
-15.6
-9.8
-11.2
-19.6
-30.3
-3.2
-10.1
-3.7
FE (%)
34.1
41.1
45.6
57.4
48.6
46.2
40.8
51.3
47.7
57.6
37.0
34.3
34.9
35.8
43.5
46.5
41.1
38.6
36.7
41.1
42.4
47.6
26.0
22.6
22.2
25.7
36.1
36.7
76.3
74.1
September 2009
                                                 G-10
Draft - Do Not Quote or Cite

-------
CMAQ 2005 Annual

Total Nitrate
(NO3+HNO3)
Ammonium
Elemental
Carbon

IMPROVE
CASTNet
STN
CASTNet
STN
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
No. of
Obs.
2495
3499
1812
15,533
10157
2388
622
1990
2640
17,452
3170
786
615
1099
300
4065
13317
3247
2495
3499
2944
16,680
3170
786
615
1099
300
4065
13460
3230
NMB (%)
20.2
23.5
8.1
15.2
30.1
67.0
14.0
37.4
17.3
33.1
24.6
36.5
23.3
23.6
10.6
37.7
1.8
7.1
7.1
-2.1
-7.6
8.1
2.2
9.2
10.9
-9.2
1.5
12.8
19.7
20.8
NME (%)
61.0
84.0
60.2
79.3
85.2
108.9
67.9
104.6
70.8
99.1
39.7
43.0
36.5
42.2
35.5
51.9
41.9
42.9
40.5
40.5
44.0
47.2
35.4
38.1
35.3
33.3
36.9
39.6
63.5
61.9
FB (%)
9.2
-25.0
-5.9
-15.6
-32.5
0.5
-24.1
-46.2
-19.3
-41.9
17.8
30.3
23.9
12.8
5.0
24.2
8.3
18.9
16.4
2.9
-4.0
12.8
3.1
13.3
14.8
-9.7
3.0
13.0
11.9
14.6
FE (%)
63.0
87.2
72.4
85.9
99.1
93.4
88.9
105.9
89.6
109.9
38.0
40.6
33.2
40.5
35.0
45.1
45.6
45.7
41.4
43.3
51.4
48.9
36.5
36.6
33.7
37.6
40.2
40.1
53.9
52.0
September 2009
                                                 G-ll
Draft - Do Not Quote or Cite

-------
CMAQ 2005 Annual

Organic Carbon

IMPROVE
STN
IMPROVE
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-km BUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
12-kmEUS
Northeast
Midwest
Southeast
Central
36-km
West
WRAP
No. of
Obs.
2502
3495
3107
16,700
10244
2341
696
1995
2626
17,289
12118
3083
2385
3442
2164
15,397
10210
2336
696
1993
2622
17,295
NMB (%)
7.3
10.2
47.6
2.6
-29.0
-17.8
-26.7
-45.6
-22.9
-16.6
-36.5
-29.1
-42.5
-42.6
-30.6
-41.2
-34.7
-21.0
-41.3
-40.4
-34.1
-22.5
NME (%)
46.1
60.2
88.2
56.7
49.7
49.2
41.9
53.3
49.2
53.4
53.6
53.1
52.6
53.5
57.7
56.1
53.7
52.2
47.6
53.7
52.8
57.5
FB (%)
10.8
3.0
23.0
2.6
-39.1
-25.6
-39.6
-58.5
-31.3
-23.4
-40.6
-27.6
-41.7
-55.6
-39.6
-45.7
-53.0
-29.2
-55.7
-64.0
-52.7
-40.8
FE (%)
44.9
50.6
64.9
55.0
61.3
57.7
55.7
69.8
56.8
60.2
66.5
64.2
65.3
70.2
66.5
69.2
70.0
58.4
63.6
74.2
68.1
67.6
1
2
3
4
5
6
7
       "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 1. 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
                                         G-12
September 2009
                                                                        Draft - Do Not Quote or Cite

-------
 1   monitor concentrations. The "fused" spatial fields are calculated by adjusting the interpolated
 2   ambient data (in each grid cell) up or down by a multiplicative factor calculated as the ratio of
 3   the modeled concentration at the grid cell divided by the modeled concentration at the nearest
 4   neighbor monitor locations (weighted by distance).
 5          To create the spatial surfaces for use in BenMAP, the 2005 CMAQ-modeled annual
 6   average PM2.5 concentrations were "fused" with 2005 total PM2.5 ambient monitoring data from
 7   STN, IMPROVE, and CASTNET sites. This was done for both the 36km national domain and
 8   the 12km eastern US domain.  The spatial surface of annual average PM2.5 air quality
 9   concentrations produced by this technique is shown in Figure G-2 for the continental U.S. Where
10   available, the 12km spatial  surface was used to supply BenMAP with annual  average PM2.5
11   concentrations. In the western part of the U.S., annual average PM2 5 concentrations were
12   supplied from the 36km domain.
13
14          Figure G-2: 2005 Predicted Annual Mean PM2.s Levels
15
16
17
                                                     2005 Fused Surface Baseline Concentrations (ug/m3)
                                                     ^B >-°3 to 4.2
                                                     ^H 4.3 to 6.5
                                                         6.6 to 9.34
                                                         9.35 to 1130
                                                     ^B l2-31 to 20.57
                                                     ^H 20.58 to 59.42
                                               G-13
     September 2009
Draft - Do Not Quote or Cite

-------
 1          Advantages and Limitations
 2          As compared to using monitored data alone, an advantage of using the CMAQ model
 3    output for comparing with health outcomes is that it has the potential to provide more complete
 4    spatial and temporal coverage. In addition, "fusing" the CMAQ data with ambient monitoring
 5    data allows for an improvement over non-fused fields (Timin et al., 2009). Doing so allows for a
 6    combination of the advantages of both sets of data: better spatial coverage and more accurate air
 7    quality estimates. Of course, the more accurate the model estimates of PM2.5, the better the
 8    performance of the "fused" spatial fields. Therefore, it is important to use model outputs that
 9    have adequate PM2.5 performance. As discussed above, we believe that the 2005 CMAQ-
10    modeled PM2.5 concentration data showed adequate model performance to be used for this
11    national-scale current conditions analysis.
12          As with any model estimate of air quality, there are limitations. For example, the
13    emissions and meteorological data used in CMAQ can each have large uncertainties, in particular
14    for unusual emission or meteorological events. There are  also uncertainties associated with the
15    chemical  transformation and fate process algorithms used  in air quality models.  For these
16    reasons, CMAQ predicts best on longer time scale bases (e.g., synoptic, monthly, and annual
17    scales). These limitations have led us to use modeled air quality estimates in this analysis that
18    are "fused" with measured ambient data and averaged over an annual scale.
19          Air Quality Estimates
20          Figures G-3 through G-6 below illustrate the spatial distribution of air quality impacts.
21    Figure 1 illustrates the modeled 2005 PM2.5 air quality levels across the U.S.  Figures 2 and 3
22    display the PM2.5 air quality levels after being adjusted so that the maximum level is no higher
23    than the LML reported in the Krewski et al. (2009) and Laden et al. (2006) studies. Figure G-4
24    displays the PRB by region of the county.
                                                G-14
      September 2009                                                          Draft - Do Not Quote or Cite

-------
        Figure G-3: 2005 Predicted Annual Mean PM2.s Levels
                                                     2005 Fused Surface Baseline Concentrations (ug/m3)
                                                     ^B I-°3 to 4.2
                                                     ^H 4.3 co 6.5
                                                          6.6 to 9.34
                                                          9.35 to 12.30
                                                     ^H 12.31 to 20.57
                                                        • 20.58 to 59.42
September 2009
                                               G-15
Draft - Do Not Quote or Cite

-------
       Figure G-4:   2005 Predicted Annual Mean PM2.s Levels Adjusted for LML of the
                      Krewski et al. (2009) study
                                                    2005 Adjusted Fused Surface Concentrations (ug/m3)
                                                    ^H 1.03 to 2.74
                                                    ^B 2.75 to 3.53
                                                        3.54 to 4.23
                                                        4.24 to 4.85
                                                    ^B 4-86 to 5-46
                                                      • 5.47 to 5.80
                                             G-16
September 2009
Draft - Do Not Quote or Cite

-------
       Figure G-5: 2005 Predicted Annual Mean PM2.s Levels Adjusted for LML of  the
                   Laden et al. (2006) study
                                                   2005 Adjusted Fused Surface Concentrations (ug/m3)
                                                   ^B I-03 to 3.08
                                                   ^H 3.09 to 4,28
                                                       4.29 to 5.58
                                                       S.59to 7.16
                                                   ^B 7.17 to 8.97
                                                   ^H 8.98 to 10.00
September 2009
                                             G-17
Draft - Do Not Quote or Cite

-------
       Figure G-6: PRB by Geographic Area in the U.S.
                                                            Policy Relevant Background Level (ug/m3)
                                                                0.74
                                                                0.84
                                                                0.86
                                                                1.01
                                                                1.72
       Figure G-7 displays the distribution of grid cells at different baseline PM2.5 air quality
levels. Figures G-8 through G-10 displays the distribution of grid cells according to the
incremental change in PM2.5 air quality for each of three scenarios: current conditions to 10
|ig/m3, current conditions to 5.8 |ig/m3 and current conditions to PRB.
                                            G-18
September 2009
Draft - Do Not Quote or Cite

-------
       Figure G-7: The Number of Grid Cells at Each Level of PMi.s Concentration in
                    2005 Current Conditions Air Quality Modeling Run








'
, 11.- i

yp


it

.i- jg.









1


<
; Jl |

|! .
I " '
! , "
($11 *
!
j -
i
If

f ^
ii i j

•
%
u
»ti-
tj

'!

1
K

' I " '
1
' F i?
. P ,,
I?
f"
' • ^
1 "
i. i' .. i
?

; "

^ 1,
! •<






f
f




J

I

1




i
(hv
1 ^
] U
J ». , t" ft a , tffi. ._..,.., r .
                     0123456
              Maximum value = 31.3 {.ig/
              Minimum value =1,5 ng/rn3
                                          iO 11 12 13 14 15 16 17 1G 19 20 21 22 23 24 25 26 27 28 29  30 31 32

                                               Basefine Ambient PM; 5 (ng/m3)
September 2009
                                              G-19
Draft - Do Not Quote or Cite

-------
       Figure G-8: The Number of CMAQ Grid Cells Experiencing an Incremental

                    Change in Annual Mean PM2.5 (jig/m3) (Current Conditions - 10

                    ug/m3)
              o  2 500  -»"'
              a
                             n
                    1  2  3  4 5 6  7  S  9 10 ii 12 13 14 lr J.i i~ 1  AJ -io U  22 23 24 25 26  27 28 25 30 31


                                               Change in Ambient PM? 5 ^|jg/m5)



              Maximum change =21,3 ^g/rn3

              Number of cells'.vith no change: 26..000
                                              G-20
September 2009                                                                Draft - Do Not Quote or Cite

-------
       Figure G-9: The Number of CMAQ Grid Cells Experiencing an Incremental
                    Change in Annual Mean PMi.s (jig/m3) (Current Conditions - 5.8
                    ug/m3)
                                     S  10 11 12 13 14 15  1,  1- 1  1

                                             Change in Ambient PM; 5 ^|ig
              Maximum change = 31.3 ftg/m3
              Number of cellswith no change: 10,000
                                            G-21
September 2009
Draft - Do Not Quote or Cite

-------
       Figure G-10: The Number of CMAQ Grid Cells Experiencing an Incremental
                    Change in Annual Mean PMi.s (ug/m3) (Current Conditions - Policy
                    Relevant Background)
                   12345
                                    y 10 11 12 13 14 15  1'3 17 13 19 20 21 22

                                            Change in Ambient PM; 5 (ng/m5)
             Maximum change = 31 ^g/m"^
             Number of cellsvvith tic change: 0
       Figure G-l 1 displays the cumulative distribution of grid cells at each baseline
concentration. Figures G-12 through G-14 display the cumulative distribution of grid cells
experiencing an incremental air quality change.
                                           G-22
September 2009
Draft - Do Not Quote or Cite

-------
        Figure G-ll: Cumulative Distribution of Baseline PM2.s Concentrations (ug/m3)
                   i  2  3 4  5  6  7  8  9  10 11  12 13 14  15 16 17 18 19 20 21 22 23 24  25 26 27 28 29 30  31 32 33



                                              Baseline Ambient PM2 5 Level (Mg/rc5)
                                               G-23
September 2009
Draft - Do Not Quote or Cite

-------
        Figure G-12: Cumulative Distribution of PM2.s (ug/m ) Changes (Baseline - 10
                       ug/m3)
                                          10 11 12 13  14 15 16 17  IS 19 20 21 22 23 24 25 26 27  28 29 30 31  32
                                                   Change m Ambient {|ig/msj
              "10 j.tgv'm5 represents the lowest measured level in the 6-dties cohort
                                                  G-24
September 2009
Draft - Do Not Quote or Cite

-------
        Figure G-13: Cumulative Distribution of PM2.s (ug/m ) Changes (Baseline - 5.8
                       ug/m3)
                   123456
                                        10 11 12 13 14 15 16 17 If  l->

                                                Change in Ambient PM^ ^ (
              '5.8 ^ig/m3 represents the lowest measured level in the ACS cohort
                                                G-25
September 2009
Draft - Do Not Quote or Cite

-------
       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-4
summarizes the population-weighted air quality change occurring among populations 30-99 (the
age range considered in the ACS cohort) for each scenario.
       Population-weighted air quality change is the average per-person change in PM2 5. It is
estimated by calculating the summation of the population in each grid cell multiplied against the
change in annual mean PM2.5 concentration in that grid cell and then dividing by the total
population.
                                          G-26
September 2009                                                           Draft - Do Not Quote or Cite

-------
       Table G-4.   Estimated Change in Annual Mean Population-Weighted PM2.s by
                    Model Scenario
           Model scenario
Population-weighted air quality change or
             baseline
Current conditions to 10 |ig/m
Current conditions to 5.8 |ig/m3
Current conditions to PRB
Current conditions
                2.6 ng/m
                6.3 |ig/m3
                11 |ig/m3
                12 |ig/m3
       Health Impact Estimates

       Figure G-15 through G-17 illustrate the distribution of total mortality attributable to
PM2.5 exposure for each of three scenarios: current conditions to 10 |ig/m3, 5.8 |ig/m3 and PRB.
                                         G-27
September 2009
                         Draft - Do Not Quote or Cite

-------
         Figure G-15: The Percentage of Total Mortality Attributable to PM2.5 Exposure:
                          Baseline -10  ug/m3
                I 3 000
                                                Percentage of Mortality Attributable to PM; ; Exposure

               Attributable mortality calculated usingKf e'vvski et al. (2009; risk estimate based on '99-00 follow-up period.
              Number of grid cells in which the percentage of attributable mortality is equal to 0: 23..000
                                                     G-28
September 2009
Draft - Do Not Quote or Cite

-------
         Figure G-16:  The Percentage of Total Mortality Attributable to PM2.5 Exposure:
                         Baseline - 5.8 ug/m3
                a
                I 4.000
                I  3 000
                                                                                            14°*   15%   16'
                                                 Percentage of Mortality Attributable to PM: ; Exposure


               Attributable mortality calculated using Krewski et al, (2009/ risk estimate based on ''99-'QQ follow-up period.
              Number of grid cells in which the percentage of attributable mortality is equal to 0: 11,000
                                                      G-29
September 2009
Draft - Do Not Quote or Cite

-------
       Figure G-17: The Percentage of Total Mortality Attributable to PM2.5 Exposure:
                    Baseline - Policy Relevant Background
             I 3 000
                                       Percentage of Mortality Attributable to PM: 5 Exposure
       Figures G-18 through G-20 illustrate the cumulative distribution of total mortality
attributable to PM2.5 exposure for each of three scenarios: current conditions to 10 |ig/m3, 5.8
|ig/m3andPRB.
                                           G-30
September 2009
Draft - Do Not Quote or Cite

-------
        Figure G-18: The Cumulative Distribution of the Percentage of Total Mortality
                       Attributable to PM2.s Exposure: Baseline - 10 ug/m3
                                            Percent of Total Mortality Attributable to PEVt; =. Exposure

            "Attributable mortality calculated using Krewski et al, (2009) risk estimate based on '99-'00 follow-up period.
                                                  G-31
September 2009
Draft - Do Not Quote or Cite

-------
        Figure G-19: The Cumulative Distribution of the Percentage of Total Mortality
                        Attributable to PM2.5 Exposure: Baseline - 5.8 ug/m3
                                            Percentage of Total Mortality Attributable to PM; --, Exposure


           'Attributable mortality calculated using Krewski et al. {2009} risk estimate based on '99-'00 follow-up period.
                                                  G-32
September 2009
Draft - Do Not Quote or Cite

-------
Figure G-20: The Cumulative Distribution of the Percentage of Total Mortality
                Attributable to PMi.s Exposure: Baseline - Policy Relevant Background
                                                     -«            i    li

                                              Percentage of Mortality Attributable to PM2 5
          "Attributable mortality calculated using Krevvski et aL (2009) risk estimate based on ''99-'GG follow-up period.
                                                G-33
September 2009
Draft - Do Not Quote or Cite

-------

-------

-------
United States                Office of Air Quality Planning and Standards          Publication No.
Environmental Protection      Health and Environmental Effects Division        EPA-452/P-09-006
Agency                            Research Triangle Park, NC                 September, 2009

-------